INTERPLAY BETWEEN GENE-EXPRESSION PROFILING AND ADJUVANT SYSTEMIC TREATMENT DECISION-MAKING IN EARLY STAGE BREAST CANCER PATIENTS.

Transcription

1 INTERPLAY BETWEEN GENE-EXPRESSION PROFILING AND ADJUVANT SYSTEMIC TREATMENT DECISION-MAKING IN EARLY STAGE BREAST CANCER PATIENTS by Anne Kuijer

2 Publication of this thesis was financially supported by / dit proefschrift werd mede mogelijk gemaakt met financiele steun van: The Dutch Cancer Society/Koningin Wilhelmina Fonds (KWF; grant number DU ), Jo Kolk Studiefonds, Michael van Vloten Fonds, Wetenschapsstichting Diakonessenhuis, Cancer Center University Medical Center Utrecht, van An kappers, IkbenFrits, ABN Amro, Chipsoft, Agendia, Pfizer, Integraal Kanker Centrum Nederland. Copyright A. Kuijer All rights reserved. No parts of this thesis may be reproduced, stored in a retrieval system of any nature or transmitted in any form or by any means, without prior written consent of the author. The copyright of the articles that have been published has been transferred to the respective journals. ISBN: Cover: Printed by: xxxx-xxxx-xxxx-xxxx Mileen van Ling ProefschriftMaken Proefschriftmaken.nl

3 Voor Astrid en Henk - en alle anderen die tegen kanker strijden of hebben gestreden -

4

5 Interplay Between Gene Expression Profiling and Adjuvant Systemic Therapy Decision-Making in Early Stage Breast Cancer Patients Gen Expressie Profielen en Adjuvante Systeem Therapie Besluitvorming in Vroeg Stadium Borstkanker Patiënten (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van rector magnificus, prof. dr. G.J. van der Zwaan, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op donderdag 26 oktober 2017 des middags te 2.30 uur door Anne Kuijer geboren op 14 augustus 1989 te Utrecht

6 Promotoren: Copromotoren: Prof. dr. I.H.M. Borel Rinkes Prof. dr. M.A.A.J. van den Bosch Dr. Th. Van Dalen Dr. S.G. Elias

7 TABLE OF CONTENTS Chapter 1: PART I: Chapter 2: Chapter 3: PART II: Chapter 4: Chapter 5: Chapter 6: Introduction and outline Adjuvant systemic treatment guidelines in the Netherlands Adjuvant systemic therapy in early breast cancer: impact of guideline changes and clinicopathological factors associated with nonadherence at a nation-wide level. A.M. Verschoor, A. Kuijer, J. Verloop, C.H. van Gils, G.S. Sonke, A. Jager, T. van Dalen, S.G. Elias. Breast Cancer Res Treat 2016;159: The influence of socio-economic status and ethnicity on adjuvant systemic treatment guideline adherence for early stage breast cancer in the Netherlands. A. Kuijer, J. Verloop, O. Visser, G. Sonke, A. Jager, C.H. van Gils, T. van Dalen, S.G. Elias. Ann Oncol 2017;28: Gene-expression profiling and adjuvant systemic treatment decisionmaking Factors associated with the use of gene-expression profiles in estrogen receptorpositive early-stage breast cancer patients: a nationwide study. A. Kuijer, K. Schreuder, S.G. Elias, C.H. Smorenburg, E.J.Th. Rutgers, S. Siesling, T. van Dalen. Public Health Genomics 2016;19: Using a gene-expression signature when controversy exists regarding the indication of adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. A. Kuijer, A.C.M. van Bommel, C.A. Drukker, M. van der Heiden-van der Loo, C.H. Smorenburg, P.J. Westenend, S.C. Linn, E.J.Th. Rutgers, S.G. Elias, T. van Dalen. Genet Med 2016:18; Changes over time in the impact of gene-expression profiles on the administration of adjuvant chemotherapy in estrogen receptor positive early stage breast cancer patients: a nationwide study. A. Kuijer, C.A. Drukker, S.G. Elias, C.H. Smorenburg, E.J.Th. Rutgers, S. Siesling, T. van Dalen. Int J Cancer 2016;139:

8 Chapter 7: Chapter 8: Chapter 9: PART III: Chapter 10: Chapter 11: Chapter 12: Chapter 13: Chapter 14: Impact of 70-gene signature use on adjuvant chemotherapy decisions in estrogen receptor positive early breast cancer patients: results of a prospective cohort study. A. Kuijer, M.E. Straver, B. den Dekker, A.C.M. van Bommel, S.G. Elias, C.H. Smorenburg, J. Wesseling, S.C. Linn, E.J.Th. Rutgers, S. Siesling, T. van Dalen. J Clin Oncol 2017;35: Impact of gene-expression profiling in patients with early breast cancer when applied outside the guideline directed indication area. K. Schreuder, A. Kuijer, E. J. Th. Rutgers, C.H. Smorenburg, T. van Dalen, S. Siesling. Eur J Cancer 2017;84: Characterization of multifocal breast cancer using the 70-gene signature in clinical low-risk patients enrolled in the EORTC 10041/BIG03-04 MINDACT trial. A. Kuijer, K.C. Aalders, M.E. Straver, L. Slaets, S. Litiere, G. Viale, L.J. van t Veer, A.M. Glas, M. Delorenzi, T. van Dalen, K. Tyfonidis, M.J. Piccart, F. Cardoso, E.J.Th. Rutgers, TRANSBIG Consortium and MINDACT Investigators. Eur J Cancer 2017;79: Molecular subtypes in breast cancer Comparison of molecular subtyping by conventional local pathology assessment or by microarray analysis using an 80-gene signature in estrogen receptor positive early stage breast cancer patients. J. van Steenhoven, A. Kuijer, M.E. Straver, S.G. Elias, C.H. Smorenburg, J. Wesseling, S.C. Linn, E.J.Th. Rutgers, S. Siesling, T. van Dalen. Concluding remarks and future perspectives Age, molecular subtypes and local therapy decision-making. A. Kuijer and T.A. King. Breast 2017 Breast 2017;34:S Summary in Dutch Appendices Review committee List of publications Acknowledgements About the author

9

10

11 CHAPTER 1 Introduction and outline of this thesis

12 Chapter 1 While surgery had been the exclusive treatment for primary breast cancer for nearly a century, the landmark trials of Fisher and Bonadonna in the mid 1970 s proved that a combination of chemotherapy regimens improved survival when used in addition to surgical resection of locoregional disease, especially in patients with more advanced breast cancer. 1 2 At the same time, the observation that young women generally suffered of more aggressive disease and that removing the ovaries or adrenal gland improved survival, supported the idea that there was an association between hormone responsiveness of breast tumors and outcome 3 which led to the development of receptor assays and endocrine therapy regimens in the beginning of the 1980 s. 4 Shortly after the introduction of endocrine therapy it was discovered that a mutated Her2 gene in breast cancer cells could stimulate excessive cell growth and that this was linked to a higher likelihood of metastasis or relapse. 5 This novel discovery paved the way for use of targeted therapies in breast cancer and approximately 15 years after discovery of this gene in 1984, the first therapeutic monoclonal antibody (an anti-her2 molecule) was developed and approved for clinical use. 6 The sprout of adjuvant systemic treatment use The first guideline advocating adjuvant systemic treatment for breast cancer was published in the United States in 1985: the National Institutes of Health Consensus Conference Statement. 7 The first national Dutch guideline did not follow until approximately 15 years later, in Initially, these guidelines reserved the use of adjuvant systemic treatment for breast cancer patients who were regarded to be at high risk of disease recurrence or distant metastasis such as young women, patients with extensive metastatic lymph-node involvement or large tumors. After release of the first Dutch guideline on breast cancer treatment, national guideline changes have rapidly followed one another. The guideline was adjusted in 2004, 2005, 2008 and 2012, echoing developments in adjuvant systemic treatment, as well as a perspiring expansion of the indication area for adjuvant systemic treatment. Based on unchanged clinicopathological factors categories of patients with an ever lower risk of developing metastases were considered candidates for adjuvant systemic therapy in subsequent guideline adjustments The increased use of adjuvant systemic treatment has contributed to improved outcome in breast cancer patients 12, but this increase comes at a the considerable price of a growing number of patients who are treated unnecessarily and suffer from side-effects. In more recent years focus has shifted towards optimal patient selection to determine in which patients the benefits of adjuvant systemic treatment, in terms of improved outcome, outweigh the negative effects. The last comprehensive national guideline (2012) reflects this paradigm shift by identifying categories of patients in whom controversy exists regarding the benefit of adjuvant systemic therapy. 12

13 Introduction and outline Novel roll of tumor biology in adjuvant systemic treatment decision-making The importance of tumor biology in relation to breast cancer outcome and patient selection for adjuvant systemic treatment benefit has been increasingly recognized over recent years. Perou et al. 13 identified molecular portraits of human breast tumors based on their gene-expression patterns. Four distinct intrinsic molecular subtypes were discriminated more than a decade ago: Luminal A, Luminal B, HER2-enriched and Basal-type tumors. Luminal type tumors were associated with more favorable outcome and endocrine therapy benefit, whereas Her2 and Basal-type tumors were associated with poorer prognosis and appeared more sensitive to chemotherapy. Distinction between these intrinsic molecular subtypes was incorporated in breast cancer guidelines in order to come to a more nuanced decision-making regarding adjuvant systemic treatment. 14 Chapter 1 Similarly, breast cancer gene-expression profiles have drawn attention as an adjunct or alternative to clinicopathological prognostic factors to predict outcome in patients with breast cancer. Concurrent with Perou s subclasification, several gene-expression profiles have been developed to guide adjuvant chemotherapy decision-making These gene-expression profiles assess the risk of recurrence or dissemination of disease based on the expression of, predominantly tumor proliferation related, genes. As such, they can be used to guide adjuvant chemotherapy decision-making. The 70-gene signature (MammaPrint) 15 is the most commonly used gene expression test in the Netherlands. It was developed using frozen tumor samples of 78, predominantly chemotherapy naïve, early stage breast cancer patients surgically treated between at the Netherlands Cancer Institute. Microarray technology was used to assess gene-expression and correlation coefficients with disease outcome were calculated. A risk set of 70 genes was identified by supervised classification and an algorithm providing an index score was developed. A cut-off value, corresponding with a sensitivity of 90% for detecting distant metastasis, was chosen to assign patients to the 70-gene signature low- or highrisk category. 15 The prognostic value of this gene-expression profile was retrospectively validated in lymph-node negative patients 18 19, lymph-node positive patients 20, postmenopausal patients 21 and patients with Her2-positive disease. 22 The feasibility of implementation of the 70-gene signature in the community-based setting was confirmed by the prospective observational RASTER study. 23 This study was conducted in the Netherlands between 2004 and 2006 and provided the first prospective evidence of the prognostic value of the 70-gene signature. 24 Following this national study, a large international randomized trial (EORTC 10041/BIG MINDACT) was designed to test the hypothesis whether this gene-expression profile would be superior to classic clinicopathological factors to select patients for adjuvant chemotherapy and between 2006 and patients were enrolled. 25 Based on the results of retrospective validation studies and this observational prospective 13

14 Chapter 1 study, use of a gene-expression profile was first suggested in the Dutch breast cancer treatment guideline of 2012 in patients with hormone-receptor positive breast cancer in whom controversy existed regarding chemotherapy benefit based on clinicopathological factors alone. 11 Aim and outline of this thesis In the context of developments in adjuvant systemic treatment regimens, expanding guideline indication area s and the novel role of tumor biology and gene-expression profiles in adjuvant systemic treatment decision-making, the aim of this thesis is to assess how these developments interacted and impacted on adjuvant systemic therapy decision-making and use in early stage breast cancer patients in the Netherlands. The first part of this thesis focuses on the use of adjuvant systemic therapy in Dutch breast cancer patients in relation to respective national treatment guidelines for adjuvant systemic therapy. In chapter 2 we investigate trends in adjuvant systemic treatment prescription in Dutch early stage breast cancer patients treated between 1990 and 2012 in relation to subsequent clinical guidelines. In chapter 3 we assess whether socioeconomic status and ethnicity affected adjuvant systemic therapy guideline adherence in recently treated Dutch breast cancer patients. In part two of this thesis we focus on the use of gene-expression profiles in Dutch clinical practice and assess the impact of these tests on adjuvant chemotherapy use and decision-making in Dutch early stage breast cancer patients. In chapter 4 we evaluate the uptake of gene-expression profiles in hormone receptor positive early stage breast cancer patients treated between 2011 and 2014 in the Netherlands and assess which patient-, tumor-, treatment- and hospital characteristics are associated with use of a gene-expression profile. In chapter 5, we assess what the impact of the 70-gene signature was on the proportion of patients who received adjuvant chemotherapy at a nation-wide level. In chapter 6 we evaluate to what extent this impact of gene-expression profile use on chemotherapy administration varies between two time periods in which two distinct clinical treatment guidelines were effective. In chapter 7 we present the results of a prospective observational multicenter study that assessed the influence of 70-gene signature on adjuvant chemotherapy decision-making in hormone receptor positive breast cancer patients. In chapter 8 we assess the influence of gene expression profile use on chemotherapy administration in patients for whom guidelines already clearly recommend to administer or withhold chemotherapy, and thus gene-expression profile use is not considered indicated. While current adjuvant systemic treatment guidelines for breast cancer disregard multifocal disease as part of the clinical decision-making process, we investigate the association between multifocal disease and the 70-gene 14

15 Introduction and outline signature result in chapter 9, evaluating whether multifocal breast cancer should be an argument to perform a gene-expression profile test in case of low clinical risk estimation. Chapter 1 In the last part of this thesis we compare the distinction between the four intrinsic tumor subtypes based on conventional pathology assessment and an 80-gene signature (chapter 10). 15

16 Chapter 1 REFERENCE LIST 1. Fisher B, Carbone P, Economou SG, et al. 1-Phenylalanine mustard (L-PAM) in the management of primary breast cancer. A report of early findings. N Engl J Med. 1975;292(3): doi: /nejm Bonadonna G, Brusamolino E, Valagussa P, et al. Combination chemotherapy as an adjuvant treatment in operable breast cancer. N Engl J Med. 1976;294(8): doi: /nejm Knight WA, Livingston RB, Gregory EJ, McGuire WL. Estrogen receptor as an independent prognostic factor for early recurrence in breast cancer. Cancer Res. 1977;37(12): Available at: Accessed April 19, Fisher B, Redmond C, Brown A, et al. Treatment of primary breast cancer with chemotherapy and tamoxifen. N Engl J Med. 1981;305(1):1 6. doi: / NEJM Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science. 1987;235(4785): Available at: pubmed/ Accessed April 19, Ménard S, Tagliabue E, Campiglio M, Pupa SM. Role of HER2 gene overexpression in breast carcinoma. J Cell Physiol. 2000;182(2): doi: /(sici) (200002)182:2<150::aid-jcp3>3.0.co;2-e. 7. Panel NI of HCD. National Institutes of Health Consensus Development Conference Statement: Adjuvant Chemotherapy for Breast Cancer. September 9-11, CA Cancer J Clin. 1986;36(1):42 7. Available at: Accessed April 19, Rutgers EJT, Nortier JWR, Tuut MK, et al. [Dutch Institute for Healthcare Improvement guideline, "Treatment of breast cancer"]. Ned Tijdschr Geneeskd. 2002;146(45): Available at: Accessed April 19, Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom.; Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05)

17 Introduction and outline 13. Perou CM, Sørlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature. 2000;406(6797): doi: / Goldhirsch A, Wood WC, Coates AS, Gelber RD, Thurlimann B, Senn H-J. Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol. 2011;22(8): doi: /annonc/mdr van t Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature. 2002;415(6871): doi: /415530a. 16. Paik S, Shak S, Tang G, et al. A multigene assay to predict recurrence of tamoxifentreated, node-negative breast cancer. N Engl J Med. 2004;351(27): doi: / NEJMoa Parker JS, Mullins M, Cheang MCU, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27(8): doi: / JCO van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: / NEJMoa Buyse M, Loi S, van t Veer L, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst. 2006;98(17): doi: /jnci/djj Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol Off J Eur Soc Med Oncol. 2010;21(4): doi: /annonc/mdp Wittner BS, Sgroi DC, Ryan PD, et al. Analysis of the MammaPrint breast cancer assay in a predominantly postmenopausal cohort. Clin Cancer Res. 2008;14(10): doi: / ccr Knauer M, Cardoso F, Wesseling J, et al. Identification of a low-risk subgroup of HER-2- positive breast cancer by the 70-gene prognosis signature. Br J Cancer. 2010;103(12): doi: /sj.bjc Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective communitybased feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Drukker CA, Bueno-de-Mesquita JM, Retèl VP, et al. A prospective evaluation of a breast cancer prognosis signature in the observational RASTER study. Int J cancer. 2013;133(4): doi: /ijc Cardoso F, Van t Veer L, Rutgers E, Loi S, Mook S, Piccart-Gebhart MJ. Clinical application of the 70-gene profile: the MINDACT trial. J Clin Oncol. 2008;26(5): doi: /jco Chapter 1 17

18

19 PART I Adjuvant systemic treatment guidelines in the Netherlands

20

21 CHAPTER 2 Adjuvant systemic therapy in early breast cancer: impact of guideline changes and clinicopathological factors associated with nonadherence at a nation-wide level. A.M. Verschoor, A. Kuijer, J. Verloop, C.H. van Gils, G.S. Sonke, A. Jager, T. van Dalen, S.G. Elias. Breast Cancer Research and Treatment 2016;159:

22 Chapter 2 ABSTRACT Purpose: Over recent years, adjuvant systemic treatment guidelines (AST) for early stage breast cancer have changed considerably. We aimed to assess the impact of these guideline changes on the administration of AST in early stage breast cancer patients and to what extent these guidelines are adhered to at a nation-wide level. Methods: We used Netherlands Cancer Registry data to describe trends in AST prescription, adherence to AST-guidelines, and to identify clinicopathological determinants of non-adherence. Results: Between , 231,648 Dutch patients were diagnosed with early breast cancer, of whom 124,472 received AST. Adjuvant endocrine treatment (ET) use increased from 23% of patients (1990) to 56% (2012), and chemotherapy from 11% to 44%. In , 8% of patients received ET and 3% received chemotherapy without guideline indication. Conversely, 10% and 29% of patients did not receive ET and chemotherapy respectively, despite a guideline indication. Unfavorable clinicopathological characteristics generally decreased the chance of undertreatment and increased the chance for overtreatment. Remarkable were the increased chance of ET undertreatment in younger women (RR <35 versus years 1.79; 95%CI ) and in women with HER2+ disease (RR 1.64; 95%CI ). Conclusions: Over the years, AST guidelines expanded resulting in much more Dutch early breast cancer patients receiving AST. In the majority of cases AST administration was guideline-concordant but the high frequency of chemotherapy undertreatment in some subgroups suggests limited AST-guideline support in these patients. 22

23 Factors associated with guideline nonadherence INTRODUCTION Breast cancer is the most common cancer type and the leading cause of cancer death among females worldwide. 1 Besides surgery, sometimes complemented with radiotherapy, treatment of early breast cancer may include adjuvant systemic therapy (AST). AST options traditionally encompass chemo- or endocrine therapy (CT or ET), to which HER2-targeted therapy was added in more recent years. Since the first National Institutes of Health (NIH) Consensus Conference in 1985, both national and international guidelines progressively expanded AST indications for early breast cancer patients towards patients with increasingly more favorable prognostic profiles; a trend that has continued into recent years. 2 5 Chapter 2 In the Netherlands, the first national AST guideline was published in 2002, and in line with the international trend, AST indications progressively expanded since then (Figure 1). 6 8 Following the most recent substantial Dutch guideline change in 2008, also patients aged 35 or older with 2-3 cm grade I or 1-2 cm grade II breast cancer became eligible for AST, and Dutch AST indications now encompass the vast majority of early breast cancer patients. At the same time, prospects for breast cancer patients improved substantially: 5-year breast cancer specific survival rates increased from 77% in 1990 to 87% in 2010 in the Netherlands. Increased AST prescription levels imposed by broadened indications have likely contributed to this survival improvement Nevertheless, expanding AST indications towards increasingly lower risk breast cancers inevitably increases the number of patients that are treated unnecessarily while still being exposed to the sometimes severe side effects. And as such, the cost-benefit of for instance adjuvant CT in ER+ early breast cancer patients is debated. 12 Furthermore, this trend of AST indication expansion as stated in clinical guidelines gives rise to the question to what extent clinicians and patients are willing to adhere to these guidelines. In the current study we therefore investigated trends in AST prescription in Dutch early breast cancer patients between at a nation-wide level, evaluated the uptake of the most recent substantial guideline change in daily practice and studied clinicopathological determinants of AST guideline non-adherence. PATIENTS AND METHODS Data collection From the Netherlands Cancer Registry (NCR) we obtained patient, tumor-, and treatment characteristics of all Dutch women surgically treated for primary, unilateral, invasive breast cancer in the period of 1990 through Data included age at diagnosis, tumor morphology, tumor grade, tumor size, hormone receptor (HR) status, HER2-status, TNM stage, axillary lymph-node involvement, multifocality, type of surgery, radiotherapy, and type of AST. We categorized tumor size in clinically relevant 23

24 Chapter 2 subgroups ( 1 cm, 1-2 cm, 2-3 cm and 3 cm) for which we used exact tumor size (registered since 2005), pathological T-stage or clinical T-stage in that order based on availability. The same approach was used to categorize regional lymph-node involvement (pn0 or pn1a). AST guidelines We distinguished four calendar periods of diagnosis ( , , & ) for which national AST guidelines differed. Currently, the 2012 guideline is effective in the Netherlands. Since it is too early to evaluate the incorporation of this guideline in daily practice and as guideline changes were minimal, in the current study the 2002, 2004 and 2008 guidelines were evaluated. Trends in AST administration To evaluate trends in AST administration over the last decade, we plotted percentages of patients receiving adjuvant ET, CT or no AST against year of diagnosis. To demonstrate the impact of expanding AST indications on the number of patients eligible for AST, we applied all guidelines to the same patients diagnosed between Uptake of AST guideline indications in daily practice We assessed the uptake of the 2008 guideline change by plotting the percentage of patients receiving AST between 2004 and 2012 in the group of patients newly eligible for AST after the 2008 guideline change (i.e. in whom AST was not indicated in 2004 but was indicated since 2008). This patient subgroup consists of lymph-node negative patients (pn0), >35 years of age (<70 years of age for CT) with a low-grade (BR I) tumor sized 2-3 cm or a tumor of moderate malignancy grade (BRII) sized 1-2 cm (and positive hormone-receptor status for ET; Suppl. Figure 1). Clinicopathological determinants of AST guideline non-adherence We assessed AST guideline non-adherence in patients diagnosed between and evaluated the relation between clinicopathological factors and non-adherence by univariable and multivariable Poisson regression analysis with robust standard error estimation (obtaining relative risks with 95% confidence intervals). Although guideline discordant (non-)treatment may be a justified clinical decision, for brevity we refer to AST guideline non-adherence as undertreatment or overtreatment. We assessed the relation between clinicopathological factors and over- respectively undertreatment both for CT and ET. We used the Nagelkerke s R 2 to evaluate how well the multivariable models explain the variation in over- and undertreatment. Furthermore, to show how well these models discriminate between patients who were and were not treated according the AST guideline, we used receiver operating characteristics (ROC) curves (with an area 24

25 Factors associated with guideline nonadherence under the curve (AUC) of 0.5 meaning no discrimination and of 1.0 meaning perfect discrimination). A two-sided p-value of <0.05 was considered statistically significant. All analyses were performed using R version (R Foundation for Statistical Computing, Vienna, Austria). RESULTS Breast cancer occurrence and AST administration over time Between 1990 and 2012, 231,648 early breast cancer patients were surgically treated in the Netherlands (sup. Table 1). The annual number of breast cancer patients almost doubled during this time, and simultaneously the percentage of patients treated with ET and/or CT increased: from 23% and 11% receiving ET or CT in 1990 to 56% and 44% in 2012 respectively (Figure 1). Over these years, AST indications have expanded substantially: approximately one third of contemporary breast cancer patients would have had an AST indication according to the 1990 guideline, which increases to approximately two thirds according to the 2008 guideline (Figure 2). Chapter 2 Uptake of the most recent substantial guideline change As a result of the guideline change in 2008, 15% of all patients diagnosed between became newly eligible for ET, which was actually prescribed to 85% of these patients in 2009 (Suppl. Figure 2A). Similarly, 12% of patients became newly eligible for CT, with 41% of them actually receiving a prescription in 2009 (Suppl. Figure 2B). A marked difference between age-groups in uptake of the new CT indication was observed: whereas 69% of newly eligible patients <50 years received CT one year following the guideline change, this was only 32% for patients aged years (Suppl. Figure 2B). Clinicopathological determinants of contemporary guideline-discordant treatment Overall, 10% and 29% of patients diagnosed between were undertreated for ET or CT respectively based on the 2008 guideline. ET and CT undertreatment was more frequent in patients whose clinicopathological profile was on the borderline of eligibility. ET undertreatment was particularly high in patients aged 35 years with a 2-3cm large grade I tumour, without axillary lymph node involvement (20%, sup. Table 2). In multivariable analyses, younger age, HER2+ disease, limited axillary lymph-node involvement, lower grade, smaller tumor size and concomitant administration of CT were independently associated with an increased risk for ET undertreatment (Table 1). A similar pattern was seen for CT undertreatment, with the exception for younger age and HER2+ disease which both decreased the risk of CT undertreatment (Table 2). These multivariable models explained 5% respectively 43% of the variation in ET and CT undertreatment with AUCs of 0.65 and

26 Chapter 2 Figure 1. Trends in the administration of AST in Dutch early stage breast cancer patients (stage I-IIIa) between Note. The percentage of patients receiving ET (A) or CT (B) per year in the Netherlands, overall and according to age. The three marked time-periods indicate distinct periods in which respectively the 2002, 2004, and 2008 national guidelines were effective. Figure 2. The proportion of contemporary ( ) early stage (stage I-IIIa) breast cancer patients eligible for AST according to historical (i.e. 1990, 1998, 2002, 2004) and contemporary (i.e. 2008) national guidelines. The grey shaded parts show the incremental proportion of patients eligible for AST according to each subsequent guideline, whereas the white part is the proportion of patients remaining ineligible for AST according to the 2008 guideline. Data are shown for endocrine therapy indication (A), chemotherapy indication (B), and endocrine and/or chemotherapy indication (C). 26

27 Factors associated with guideline nonadherence The grey shaded parts show the incremental proportion of patients eligible for AST according to each subsequent guideline, whereas the white part is the proportion of patients remaining ineligible for AST according to the 2008 guideline. Data are shown for endocrine therapy indication (A), chemotherapy indication (B), and endocrine and/or chemotherapy indication (C). Overtreatment was less frequent; 8% and 3% of all patients received ET or CT in the absence of a guideline-based indication. Again, guideline discordant treatment was most frequent for patients at the borderline of ineligibility (with young age and grade III particularly related to overtreatment; sup. Table 4-5). In multivariable analyses, younger age, increasing grade, increasing tumour size (1-2 vs. 1) and HER2+ disease independently increased the risk for both overtreatment with ET or CT (Table 3 and 4). Together, the evaluated clinicopathological variables explained 20% respectively 45% of the variation in ET and CT overtreatment with AUCs of 0.76 and 0.94). Chapter 2 DISCUSSION In line with international trends, Dutch AST indications as stated in clinical guidelines for breast cancer treatment have gradually expanded between resulting in approximately two-thirds of patients currently eligible for AST compared to onethird if the 1990 guideline would still be used nowadays. Using nation-wide data, we observed a rapid uptake of the most recent substantial ET and CT guideline change in clinical practice. Nevertheless, a substantial proportion of patients diagnosed between did not receive AST despite a guideline-based indication and especially CT undertreatment was frequent (up to 66% in subgroups at borderline of CTeligibility). Administration of AST in absence of a guideline-based indication is rare. As expected, unfavorable clinicopathological characteristics generally decreased the risk of undertreatment but were associated with a higher risk for overtreatment. We observed that clinicopathological factors were strongly related with AST guideline non-adherence, especially for CT overtreatment (45% explained variation; AUC 0.94), and undertreatment (43% explained variation; AUC 0.86). ET guideline nonadherence was less frequent as compared to CT guideline non-adherence and the same clinicopathological factors were less able to explain ET guideline non-adherence, suggesting that additional factors, such as patient and physician preference, play a more substantial role. An emerging clinicopathological factor particularly influencing CT decisions in ER+/HER2- breast cancer 13 we did not take into account is geneexpression profiling. 27

28 Chapter 2 Table 1. Adjuvant ET undertreatment in relation to clinicopathological factors in stage I-IIIa Dutch breast cancer patients diagnosed between with an ET indication. Univariable Multivariable Total N Undertreated % RR [95% CI] RR [95% CI] P-value Age (yrs) < [ ] 0.94 [ ] 0.98 [ ] [ ] < [ ] < [ ] Reference Reference Positive lymph nodes 1.61 [ ] < [ ] <0.001 Negative Reference Reference > Grade 0.72 [ ] 0.54 [ ] 0.57 [ ] < [ ] <0.001 < [ ] <0.001 < [ ] 0.02 I Reference Reference II III Size (cm) 0.75 [ ] 0.70 [ ] < [ ] <0.001 < [ ] < Reference Reference [ ] 0.58 [ ] < [ ] <0.001 < [ ] <

29 Factors associated with guideline nonadherence [ ] < [ ] <0.001 Chemotherapy No Reference Reference Chapter 2 Yes HER [ ] < [ ] <0.001 Negative Reference Reference Positive [ ] < [ ] <0.001 Note. Patients with T4 tumors, multifocal disease, receiving neo-adjuvant therapy, or having incomplete data for the considered clinicopathological factors were excluded, leaving patients for analysis, all with an indication for ET according to the 2008 national guideline. In the multivariable model we corrected mutually for all clinic-pathological factors. In all patients <70 years of age adjuvant CT was also recommended as the indication areas for ET and CT coincide in the 2008 guideline for hormone receptor positive patients. 29

30 Chapter 2 Table 2. Adjuvant CT undertreatment in relation to clinicopathological factors in stage I-IIIa Dutch breast cancer patients diagnosed between with a CT indication. Total N Undertreated Univariable Multivariable % RR [95% CI] P-value RR [95% CI] P-value Age (yrs) < [ ] < [ ] < [ ] < [ ] < [ ] < [ ] < Reference Reference Positive lymph nodes Negative Reference Reference [ ] < [ ] < [ ] < [ ] <0.001 > [ ] < [ ] <0.001 Grade I Reference Reference II [ ] < [ ] <0.001 III [ ] < [ ] <0.001 Size (cm) Reference Reference [ ] [ ] < [ ] < [ ] < [ ] < [ ] <0.001 Hormone receptor status and endocrine therapy HR Reference Reference HR+ / ET-* HR+ / ET+* [ ] < [ ] < [ ] < [ ] <

31 Factors associated with guideline nonadherence HER2 Negative Reference Reference Positive [ ] < [ ] <0.001 Chapter 2 Patients with T4 tumors, multifocal disease, neo-adjuvant therapy, or incomplete data were excluded, leaving patients for analysis, all with an CT indication according to the 2008 national guideline. In the multivariable model we mutually corrected for all clinicopathological factors. *Since ET and CT indications coincide in the 2008 guideline all patients with HR+/ET+ disease in this model received ET in accordance with the guideline and patients with HR+/ET- were undertreated for ET. Table 3. Adjuvant ET overtreatment in relation to clinicopathological factors in stage I-IIIa Dutch HR+ breast cancer patients diagnosed between without an ET indication. Total N Overtreated Univariable Multivariable % RR [95% CI] P-value RR [95% CI] P-value Age (yrs) < [ ] < [ ] < [ ] < [ ] < [ ] [ ] Reference Reference [ ] [ ] 0.42 Grade I Reference Reference II [ ] < [ ] <0.001 III [ ] < [ ] <0.001 Size (cm) Reference Reference [ ] [ ] <0.001 Chemotherapy 31

32 Chapter 2 No Reference Reference Yes HER [ ] < [ ] <0.001 Negative Reference Reference Positive [ ] < [ ] <0.001 Patients with T4 tumors, multifocal disease, neo-adjuvant therapy, or having incomplete data were excluded, leaving 8794 patients for analysis, all without an indication for ET according to the 2008 Dutch national guideline but with HR+ disease. In the multivariable model we mutually corrected for all clinicopathological factors. All patients without an indication for ET, also had no indication for CT. Therefore, all patients included in this analyses who received CT were concomitantly over-treated with CT. Table 4. Adjuvant CT overtreatment in relation to clinicopathological factors in stage I-IIIa Dutch breast cancer patients <70 years diagnosed between without a CT indication. Univariable Multivariable Total N Overtreated % RR [95% CI] P-value RR [95% CI] P-value Age (yrs) < [ ] < [ ] < [ ] < [ ] < [ ] < [ ] < Reference Reference Grade I Reference Reference II [ ] < [ ] 0.02 III [ ] < [ ] <0.001 Size (cm) Reference Reference 32

33 Factors associated with guideline nonadherence [ ] < [ ] 0.08 Hormone receptor status and endocrine therapy HR Reference Reference HR+ / ET- HR+ / ET+ HER [ ] < [ ] < [ ] [ ] <0.001 Chapter 2 Negative Reference Reference Positive [ ] < [ ] <0.001 Patients with T4 tumors, multifocal disease, neo-adjuvant therapy, or having incomplete data were excluded, leaving 7370 patients for analysis, all without an indication for CT according to the 2008 Dutch national guideline and < 70 years at diagnosis. In the multivariable model there was mutually corrected for all clinicopathological factors. Although clinical use of such tests was very limited before , between 2007 and August 2011 several Dutch hospitals participated in the MINDACT trial in which patients with a discordant clinical and genomic prognostic profile according to the70- gene signature were randomized to receive or withhold adjuvant CT. 15 Participation in MINDACT may therefore have contributed to the high frequency of CT undertreatment in patients at borderline CT-eligibility. However, reanalysis of the data while excluding hospitals that participated in MINDACT yielded similar results (with up to 60% CT undertreatment). Our observation that a poor clinicopathological profile is associated with a decreased risk of undertreatment (and vice versa) is not surprising. A remarkable exception, however, is the observed higher risk of both ET over- as well as undertreatment in younger women, indicating an altogether more patient-tailored approach to the ET guidelines in younger patients. In the current study 12% of all women <35 years of age did not receive ET despite a guideline-based indication. Fertility concerns may play a role in the decision to withhold ET in younger women. Initiatives on reducing the negative effects of ET on fertility are currently ongoing (e.g. POSITIVE trial). Our findings further endorse the importance of future research in order to achieve better adherence to ET in this highrisk population. 33

34 Chapter 2 Literature on AST prescription trends in early breast cancer patients in recent years is scarce. AST trends on a population-based scale have been reported for the Netherlands ( ) 16, the South of the Netherlands ( ), South East England ( ) 17, the US ( and ) 2,18 and Sweden ( ) 19. All of these studies report the increased use of AST in breast cancer patients over time as a result of the expansion of the indication area in clinical guidelines with some international differences. 20 Actual guideline-adherence was not addressed by any of these authors, although some stress a need for a more effective practical implementation of AST guidelines In summary, between indications for AST have expanded substantially resulting in a drastic increase in AST administration in the Netherlands. Guidelinediscordant overtreatment with AST was rare in recent years, but the high frequency of chemotherapy undertreatment in some patient subgroups with borderline AST eligibility suggests limited AST-guideline support in these patients. 34

35 Factors associated with guideline nonadherence Suppl. Figure 1. Overview of the 2002, 2004 and 2008 national guideline recommendation for adjuvant systemic therapy (CT and/or ET) in patients without metastatic lymphnode involvement (N0). Size (cm) Grade I Grade II Grade III Year Age (y) > > > * * * * * * * * * * * 70 Guideline-based indication for ET and CT Guideline-based indication for ET only No indication for AST 2008 < * In patients years of age CT was indicated in case of HR- disease, in patients with HR+ disease administration of CT could be considered. In the 2008 guideline, the lower age limit was adjusted from 35 to < 35 years. Note. In patients with metastatic lymph-node involvement ( N1a), in principle both CT and ET were indicated in all time periods with the following exceptions: ET was only indicated in patients with HR+ disease; during all periods patients 70 years of age were not eligible for CT and in patients years of age with HR+ disease CT was not strictly indicated but could be considered according to the guideline of 2002 and Chapter 2 35

36 Chapter 2 Supplementary Table 1. Patients- and tumor characteristics of all Dutch early stage breast cancer patients (TNM stage I-IIIc) surgically treated between , overall and stratified according to AST guideline period. Regional guidelines National guidelines Total n (%) n (%) n (%) n (%) n (%) (46) (9) (19) (26) Patient characteristics Age (median, minmax) 59 (17-102) 58 (19-99) 58 (20-101) 59 (18-98) 59 (17-102) 35 year 3,460 (3) 602 (3) 1,177 (3) 1,546 (3) 6,785 (3) year 24,454 (23) 4,761 (23) 10,169 (23) 12,736 (21) 52,120 (22) year 49,986 (47) 10,379 (49) 22,062 (50) 32,112 (54) 114,539 (49) 70 year 28,796 (27) 5,259 (25) 10,522 (24) 13,627 (23) 58,204 (25) Tumor characteristics Morphology Ductal 73,009 (68) 15,447 (74) 32,437 (74) 45,991 (77) 166,884 (72) Lobular 11,832 (11) 2,263 (11) 4,827 (11) 6,434 (11) 25,356 (11) Mixed 4,598 (4) 869 (4) 1,676 (4) 1,886 (3) 9,029 (4) Other 17,257 (16) 2,422 (11) 4,990 (11) 5,710 (10) 30,379 (14) Multifocality No 16,380 (86) 10,426 (87) 33,177 (84) 48,950 (83) 108,933 (84) 36

37 Factors associated with guideline nonadherence Yes 2,599 (14) 1,499 (13) 6,418 (16) 10,148 (17) 20,664 (16) Missing 87,717 9,076 4, ,051 Pathological tumor size (median, minmax)* N/A N/A 17 (0-250) 16 (0-210) 16 (0-250) T1 57,165 (54) 12,127 (58) 26,309 (60) 37,749 (63) 133,350 (58) Chapter 2 T2 39,577 (38) 7,576 (36) 15,203 (35) 19,238 (32) 81,594 (36) T3 3,985 (4) 729 (3) 1,594 (4) 2,135 (4) 8,443 (4) T4 4,536 (4) 473 (2) 646 (1) 703 (1) 6,358 (3) Missing 1, ,903 Invasive tumor grade Grade I 8,487 (14) 3,666 (20) 9,000 (22) 13,239 (24) 34,392 (20) Grade II 24,334 (41) 8,200 (45) 18,229 (45) 24,536 (45) 75,299 (44) Grade III 26,617 (45) 6,297 (35) 13,398 (33) 16,302 (30) 62,614 (36) Missing 47,258 2,838 3,303 5,944 59,343 Pathological axillary status (pn)* N1a 41,140 (39) 8,571 (41) 17,097 (39) 21,525 (36) 88,333 (38) N0 64,358 (61) 12,299 (59) 26,735 (61) 38,431 (64) 141,823 (62) Missing 1, ,492 Hormone receptor status Negative 577 (21) 2,138 (19) 7,349 (18) 9,772 (17) 19,836 (17) Positive 2,155 (79) 9,183 (81) 33,676 (82) 49,350 (83) 94,364 (83) 37

38 Chapter 2 Missing 103,964 9,680 2, ,448 HER2 status Negative N/A 1,820 (76) 25,857 (85) 49,792 (86) 77,469 (85) Positive N/A 568 (24) 4,483 (15) 8,173 (14) 13,224 (15) Missing 106,696 18,613 13,590 2, ,955 Treatment characteristics Axillary surgery No axillary surgery 17,601 (16) 8,912 (42) 22,525 (51) 38,149 (64) 87,187 (38) Axillary lymph node dissection Surgery 89,095 (84) 12,089 (58) 21,405 (49) 21,872 (36) 144,461 (62) Ablative 56,366 (53) 10,443 (50) 20,108 (46) 26,314 (44) Breast-conserving 50,330 (47) 10,558 (50) 23,822 (54) 33,707 (56) 113,231 (49) 118,417 (51) Neo-adjuvant therapy Hormonal therapy 590 (1) 170 (1) 429 (1) 1,030 (2) 2,219 (1) Chemotherapy 1,127 (1) 620 (3) 1,953 (4) 4,837 (8) 8,537 (4) * Pathological classification according to the TNM staging system and, if missing, the clinical estimates. HR status was based on IHC and was considered positive when ER and/or PR were positive. Her2 is based on IHC and/or FISH, and was considered positive when the IHC result was positive (3+) or, in case of a dubious IHC result (2+) a positive FISH result. Abbreviations: IHC: immunohistochemistry FISH: fluorescence in situ hybridization, HR: Hormone receptor, ER: Estrogen receptor, PR: Progesterone receptor, Her2: human epidermal growth factor receptor 2, N/A: Not applicable. 38

39 Factors associated with guideline nonadherence Suppl. Figure 2. Uptake of the expanded AST indication by the 2008 Dutch national guideline. Chapter 2 (A) Proportion of patients with a new indication for ET according to the 2008 national guideline receiving ET before and after this guideline became effective (i.e. HR+, age >35, N0, Grade I and 2-3 cm, or Grade 2 and 1-2 cm; affected ~1300 patients per year between ) (B) Proportion of patients with a new indication for CT according to the 2008 national guideline receiving CT before and after this guideline became effective (i.e. age 35-70, N0, Grade I and 2-3 cm, or Grade 2 and 1-2 cm; affected ~1000 patients per year between ). 39

40 Chapter 2 Supplementary Table 2. ET undertreatment in patients with a clinicopathological profile on the borderline of eligibility for ET according to the national guideline of Patient categories pn* Grade Size Age (y) HR Total (n) Undertreated (%) Overall Total (10) N (12) N (8) Elderly patients (> 70 years of age) N (18) N (9) Patients with a borderline indication N0 I 1-2 < (17) N0 II 1 < (29) N0 III 1 < (20) N0 I (20) N0 II (13) N0 III (10) Patients with borderline ER over-expression ER 10-20% (19) ER>20% (9) * Pathological classification according to the TNM staging system and, if missing, the clinical estimates. Percentages ER expression registered since

41 Factors associated with guideline nonadherence Supplementary Table 3. CT undertreatment in patients with a clinicopathological profile on the borderline of eligibility for CT according to the national guideline of Patient categories pn* Grade Size (cm) Age (y) Total (n) Undertreated (%) Overall Total (28) N (38) N (17) Chapter 2 Patients with a borderline indication N0 I 1-2 < (25) N0 II 1 < (29) N0 III 1 < (14) N0 I (66) N0 II (61) N0 III (23) * Pathological classification according to the TNM staging system and, if missing, the clinical estimates. 41

42 Chapter 2 Supplementary Table 4. ET overtreatment in patients with a clinicopathological profile on the borderline of eligibility for ET according to the national guideline of Patient categories pn Grade Size (cm) Age HR Total (n) Overtreated (%) Overall Total (8) Patients on the borderline of no indication N0 I 1 < (38) N0 I (8) N0 II (10) N0 III (31) Patients with extreme over expression of ER ER10-80% (11) ER 80% (8) Percentages of ER expression registered since * Pathological classification according to the TNM staging system and, if missing, the clinical estimates. 42

43 Factors associated with guideline nonadherence Supplementary Table 5. CT overtreatment in patients with a clinicopathological profile on the borderline of eligibility for CT according to the national guideline of Patient categories Grade Size (cm) Age (y) Total (n) Overtreated (%) Chapter 2 Overall Total (3) Patients on the borderline of no indication N0 I 1 < (22) N0 I (2) N0 II (3) N0 III (16) Elderly patients(>70 year of age) N (2) N (7) * Pathological classification according to the TNM staging system and, if missing, the clinical estimates. 43

44 Chapter 2 REFERENCE LIST 1. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, CA Cancer J Clin. 2015;65(2): doi: /caac Mariotto A, Feuer EJ, Harlan LC, Wun L-M, Johnson KA, Abrams J. Trends in use of adjuvant multi-agent chemotherapy and tamoxifen for breast cancer in the United States: J Natl Cancer Inst. 2002;94(21): Available at: nlm.nih.gov/pubmed/ Accessed April 19, Panel NI of HCD. National Institutes of Health Consensus Development Conference Statement: Adjuvant Chemotherapy for Breast Cancer. September 9-11, CA Cancer J Clin. 1986;36(1):42 7. Available at: Accessed April 19, National Institutes of Health Consensus Development Panel. National Institutes of Health Consensus Development Conference statement: adjuvant therapy for breast cancer, November 1-3, J Natl Cancer Inst Monogr. 2001;(30):5 15. Available at: Accessed April 19, National Institutes of Health Consensus Development Panel. NIH consensus conference. Treatment of early-stage breast cancer. JAMA. 1991;265(3): Available at: Accessed April 19, Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom.; Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom de Gelder R, Heijnsdijk EAM, Fracheboud J, Draisma G, de Koning HJ. The effects of population-based mammography screening starting between age 40 and 50 in the presence of adjuvant systemic therapy. Int J Cancer. 2015;137(1): doi: / ijc Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05) Berry DA, Cronin KA, Plevritis SK, et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N Engl J Med. 2005;353(17): doi: / NEJMoa Goldhirsch A, Ingle JN, Gelber RD, et al. Thresholds for therapies: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol. 2009;20(8): doi: /annonc/mdp

45 Factors associated with guideline nonadherence 13. Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective communitybased feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Kuijer A, Drukker CA, Elias SG, et al. Changes over time in the impact of gene-expression profiles on the administration of adjuvant chemotherapy in estrogen receptor positive early stage breast cancer patients: A nationwide study. Int J Cancer. 2016;139(4): doi: /ijc Cardoso F, Van t Veer L, Rutgers E, Loi S, Mook S, Piccart-Gebhart MJ. Clinical application of the 70-gene profile: the MINDACT trial. J Clin Oncol. 2008;26(5): doi: /jco Vervoort MM, Draisma G, Fracheboud J, van de Poll-Franse L V, de Koning HJ. Trends in the usage of adjuvant systemic therapy for breast cancer in the Netherlands and its effect on mortality. Br J Cancer. 2004;91(2): doi: /sj.bjc Tataru D, Robinson D, Møller H, Davies E. Trends in the treatment of breast cancer in Southeast England following the introduction of national guidelines. J Public Health (Oxf). 2006;28(3): doi: /pubmed/fdl Harlan LC, Clegg LX, Abrams J, Stevens JL, Ballard-Barbash R. Community-Based Use of Chemotherapy and Hormonal Therapy for Early-Stage Breast Cancer: J Clin Oncol. 2006;24(6): doi: /jco Kemetli L, Rutqvist LE, Jonsson H, Nyström L, Lenner P, Törnberg S. Temporal trends in the use of adjuvant systemic therapy in breast cancer: a population based study in Sweden Acta Oncol. 2009;48(1): doi: / Kelly E, Lu CY, Albertini S, Vitry A. Longitudinal trends in utilization of endocrine therapies for breast cancer: an international comparison. J Clin Pharm Ther. 2015;40(1): doi: /jcpt Thuerlimann B, Koeberle D, Senn H-J. Guidelines for the adjuvant treatment of postmenopausal women with endocrine-responsive breast cancer: Past, present and future recommendations. Eur J Cancer. 2007;43(1): doi: /j.ejca Chapter 2 45

46

47 CHAPTER 3 The influence of socio-economic status and ethnicity on adjuvant systemic treatment guideline adherence for early stage breast cancer in the Netherlands. A. Kuijer, J. Verloop, O. Visser, G. Sonke, A. Jager, C.H. van Gils, T. van Dalen, S.G. Elias. Annals of Oncology 2017 Apr 29 Annals of Oncology 2017;28:

48 Chapter 3 ABSTRACT Purpose: We aimed to assess whether socio-economic status (SES) and ethnicity affect adjuvant systemic therapy (AST) guideline adherence in early breast cancer patients in a health care setting with assumed equal access to care. Methods: Data from all female patients surgically treated for primary unifocal early breast cancer between January 2005 and December 2014 were retrieved from the Netherlands Cancer Registry. We assessed the association between SES, ethnicity and non-adherence to adjuvant chemotherapy (CT) or endocrine therapy (ET) guideline indications with Poisson regression models, adjusting for clinicopathological variables. Results: 104,201 patients were included in the current analysis. Of patients without an indication, 4% and 13% received adjuvant CT or ET ( overtreatment ), whereas 39% and 14% of patients with an indication did not receive CT or ET ( undertreatment ). Medium and low SES patients were 1.01 (95%CI ) and 1.01 (95%CI ) more likely to be undertreated and 0.85 (95%CI ) and 0.67 (95%CI ) times more likely to be overtreated with CT compared to high SES patients (resulting in an overall relative risk of CT use of 0.94 (95%CI ) and 0.85 (95%CI ), respectively). No association between SES and ET guideline adherence or ethnicity and CT/ET guideline adherence was observed. Conclusion: in the Netherlands, minimal SES disparities in CT guideline adherence were observed: low SES patients are less likely be overtreated and marginally more likely to be undertreated with CT resulting in an overall decreased risk of receiving CT. No ethnical disparities in AST guideline adherence were observed. 48

49 The infuence of socio-economic status and ethnicity on guideline adherence INTRODUCTION Adjuvant systemic treatment (AST) is an important pillar in the management of early stage breast cancer. Since 1990 several (inter)national guidelines recommend AST in patients at higher risk of distant recurrence in order to improve clinical outcome. 1,2 Appropriate use of AST decreases breast cancer mortality 3, and patients with a poor clinicopathological profile should in essence receive AST. On the other hand, in patients with a good prognostic profile AST benefits do not always outweigh its negative effects, such as associated (long-term) side effects. Therefore, both under- and overtreatment with AST should ideally be avoided to optimize breast cancer treatment while maintaining optimal quality of life. Chapter 3 Several studies from the U.S. report that actual AST prescription varies by insurance status 4 7, income 6,8, socio-economic status (SES) 9,10 and ethnicity. 8,11 13 These observed disparities could be explained by educated and/or more affluent women being more likely to seek medical care proactively whereas less affluent patients may not communicate as well with physicians and are less informed about treatment options Furthermore, minority and low SES patients may receive care from different providers due to financial inequalities, geography, or insurance status, and between-provider variation in quality of care may cause socio-economic or racial disparities in the availability of appropriate systemic therapies. 10,17 In the Netherlands, healthcare insurance is mandatory and AST is fully reimbursed by all health insurers. Hence, in contrast to other healthcare systems, AST is available for all Dutch breast cancer patients irrespective of their background. The aim of the current population-based study is to assess whether socio-economic and ethnical disparities in AST guideline adherence in early stage breast cancer are also present in the Dutch equal access healthcare system. While financial incentives in the Netherlands are less likely to underlie AST decision-making, other SES or ethnicity related factors such as views and beliefs might still be of influence. This study can thus provide important clues for the relative importance of financial and non-financial SES/ethnicity associated determinants of AST adherence and could thereby aid in reducing breast cancer treatment disparities and increasing appropriate AST use globally. PATIENTS AND METHODS Data collection From the Netherlands Cancer Registry (NCR) we obtained patient-, tumor-, treatment and hospital-characteristics of all women surgically treated for primary, unilateral, invasive breast cancer in the period of 2005 through

50 Chapter 3 We used a SES indicator, which ranks neighborhoods defined by postal code based on the aggregated value of houses and household income (average of 17 households/code), which was categorized in low (1 th -3 th decile), medium (4 th -7 th decile) and high (8 th -10 th decile). 18 Country of birth is registered by the NCR and corrected or supplemented with information from municipal population registries. For patients born <1970 without known country of birth ( 25% patients) a name-based approach was used to assess country of birth. Non-Dutch names are specific for 90-95% of immigrants from Turkey and Morocco (the largest immigrant countries) and for 50-75% of immigrants from Surinam and the Netherlands Antilles (third and fourth immigrant countries). This approach was not used for patients born >1969 because of the large number of second generation immigrants following immigration in the 1960s. Patients were categorized into Native Dutch, Non-Western Immigrant or Western Immigrant according to the classification of Statistics Netherlands. 19 AST guideline recommendation CT/ET indication was based on Dutch AST guidelines at time of treatment (three guidelines were active during our study-period; , and ). We assessed guideline non-adherence as follows: CT/ET administration without such guideline indication (CT or ET overtreatment ) or refrain CT/ET despite a guideline indication to administer CT/ET (CT or ET undertreatment ). Risk of ET under- or overtreatment was evaluated in hormone-receptor positive patients only. Patients with multifocal disease were excluded from the analysis, since guidelines are unclear about AST recommendation in such patients. Patients treated with (neo) adjuvant therapy were excluded from the primary analysis, since AST eligibility is based on clinical assessment of the tumor, but were included in sensitivity analyses. Risk of guideline discordant treatment To obtain appropriate estimates of the relative risk of CT/ET under- or overtreatment in relation to SES/ethnicity, we used Poisson regression models instead of logistic regression (the latter estimates odds ratios which are more extreme than relative risks if the outcome of interest is common). We first assessed the unadjusted association between SES/ethnicity and guideline discordant CT/ET (Model 1). Subsequently, we adjusted for year of diagnosis, age, morphology, grade, tumor size, axillary lymphnode involvement, ER-, PR- and Her2-status (Model 2). Hereafter we also adjusted for hospital-characteristics, and included a random intercept per hospital resulting in a mixed-effect model that takes patient clustering at a hospital-level into account (Model 3). Lastly, we corrected for all clinicopathological and hospital-characteristics and mutually adjusted for SES and ethnicity (Model 4). Models were conducted in patients 50

51 The infuence of socio-economic status and ethnicity on guideline adherence with and without an indication for CT/ET to assess the risk for undertreatment and overtreatment. Model 4 was also used to assess the association between SES/ethnicity and CT/ET, regardless of guideline recommendation. All variables were included as categorical variables, except for age which was treated as a continuous variable after assessment of the assumed log-linear relation by model transformations and comparing Akaike s information criteria. Furthermore, all models were well calibrated. To evaluate a possible interaction between SES and ethnicity, an interaction term was added to Model 4 and a likelihood-ratio test was performed to assess statistical significance. The same approach was used to assess a possible interaction between SES and year of birth, year of diagnosis or age. Robust standard errors and appropriate 95%CI s were calculated with the Sandwich method. Chapter 3 A two-sided p-value of <0.05 was considered statistically significant. Analyses were performed using R version For the mixed-effect Poisson regression analysis we used STATA version 14.1 supplemented with GLLAMM for obtaining robust 95%CI s using the Sandwich estimator of the covariance matrix. 20 RESULTS Between 2005 through 2014, early breast cancer patients were surgically treated in the Netherlands (Sup. Figure 1). A total of (24%) patients were excluded from the analysis (due to multifocality and/or neo-adjuvant therapy). Baseline differences by SES and ethnicity High SES patients were younger at time of diagnosis (59 vs. 64 years) and more often of Dutch origin (95% vs. 89%) as compared to low SES patients (Table 1). We observed no notable differences in tumor characteristics. Native Dutch patients were older, of higher SES and had tumors with more favorable characteristics (ER+/Her2-, low malignancy grade without axillary lymph-node involvement) as compared to immigrant patients (Table 2). Guideline discordant treatment Overtreatment was rare, 1780 (4%) and 3958 (13%) patients received CT/ET without such recommendation. Undertreatment was more frequent: (38%) and 6740 (12%) patients did not receive CT/ET despite a guideline recommendation (Table 3/4, respectively). Five percent (n=565) of high SES patients and 3% (n=362) of low SES patients received CT without a recommendation. Medium and low SES patients had a 0.85 (95%CI ) and 0.67 (95%CI ) higher risk of CT overtreatment compared 51

52 Chapter 3 to high SES patients, respectively. When analyzing the relation between SES and CT overtreatment per SES deciles, this revealed a similar trend towards a decreasing risk for CT overtreatment with decreasing SES (Sup. Figure 2). Table 1. Patient-, tumor-, hospital- and treatment characteristics of all Dutch female patients surgically treated for primary breast cancer stage I-IIIc between (n = ) by socio-economic status (SES). Patient characteristics Low SES Medium SES High SES n=31134 (26%) n=49345(41%) n=38428 (32%) Year of diagnosis (28.5) (27.0) (26.9) (40.4) (40.4) (40.0) (31.2) (32.6) (33.0) Age in years, mean (SD) 63.8 (14.7) 59.8 (13.1) 59.4 (12.2) Ethnicity Native Dutch (88.5) (93.6) (95.0) Non-Western Immigrant 2623 (8.6) 2101 (4.3) 1106 (2.9) Western Immigrant 896 (2.9) 1020 (2.1) 776 (2.0) Missing Hospital characteristics Type of hospital District (48.8) (51.2) (49.4) Top clinical (45.2) (42.9) (43.7) Academic 1624 (5.9) 2722 (5.9) 2512 (6.9) Missing Tumor characteristics Morphology Ductal (74.5) (74.9) (74.8) Lobular 3373 (10.8) 5595 (11.3) 4506 (11.7) Ductoobular 842 (2.7) 1634 (3.3) 1344 (3.5) Adenocarcinoma 1037 (3.3) 1227 (2.5) 856 (2.2) Other types 2699 (8.7) 3927 (8.0) 2987 (7.8) 52

53 The infuence of socio-economic status and ethnicity on guideline adherence Invasive tumor grade Grade (23.7) (24.5) 8538 (25.2) Grade (46.0) (45.6) (45.9) Grade (30.3) (30.0) 9797 (28.9) Missing Tumor size in cm, mean (SD) 1.9 (1.4) 1.8 (1.4) 1.8 (1.4) Missing Chapter 3 Hormone receptor status ER and PR negative 4854 (15.9) 7778 (16.0) 5883 (15.6) ER or PR positive (84.1) (84.0) (84.4) Missing HER2-status Negative (86.6) (86.2) (86.6) Positive 3803 (13.4) 6402 (13.8) 4840 (13.4) Missing Axillary lymph-node involvement N (63.5) (64.6) (64.6) N (26.3) (25.7) 9496 (26.1) N (6.5) 2909 (6.3) 2160 (5.9) N (3.8) 1609 (3.5) 1219 (3.4) Missing Treatment characteristics Type of Surgery Breast-conserving (51.8) (58.5) (60.9) Ablative (39.6) (37.3) (35.6) Type of surgery unknown 2691 (8.6) 2095 (4.2) 1336 (3.5) Axillary Surgery No axillary surgery 5386 (17.3) 5868 (11.9) 4610 (12.0) ALND 8363 (26.9) (26.0) 9607 (25.0) SNP (55.8) (62.1) (63.0) Gene-expression profile use* No GEP use (95.4) (93.8) (92.2) GEP use 580 (4.6) 1304 (6.2) 1275 (7.8) 53

54 Chapter 3 Missing Adjuvant chemotherapy (34.0) (42.1) (42.9) Adjuvant endocrine therapy (55.5) (53.6) (53.5) Neo-adjuvant chemotherapy 2091 (6.7) 3736 (7.6) 3151 (8.2) Neo-adjuvant endocrine therapy 583 (1.9) 717 (1.5) 574 (1.5) * The use of gene-expression profiles (GEP) was only registered since 2012, patients treated before 2012 were coded as missing. Abbreviations: SD = standard deviation, SES = socio-economic status, ER= estrogen receptor, PR = progesterone receptor, ALND = axillary lymph-node dissection, SNP = sentinel node procedure, GEP = gene-expression profile 54

55 The infuence of socio-economic status and ethnicity on guideline adherence Low or medium SES patients were marginally more likely to be undertreated with CT as compared to high SES patients (RR %CI & RR %CI , respectively). CT administration regardless of indication was 0.85 (95%CI ) times more likely in low SES patients as compared to high SES patients (Sup. Table 2). No association between SES and ET guideline discordant treatment was observed (Table 4 and Sup. Table 2). We observed no association between ethnicity and the risk of guideline discordant treatment. Immigrant patients were as likely as Native Dutch patients to be overtreated with CT in the fully adjusted models (RR 0.91; 95%CI , and RR 0.75; 95%CI for Non-Western and Western Immigrants). The same holds for the risk of CT undertreatment (RR 1.00; 95%CI , and RR %CI for Non-Western and Western Immigrants). We observed no association between ethnicity and ET guideline discordant treatment or overall CT/ET administration (Table 4 and Sup. Table 2). Chapter 3 We observed no differences in the association between SES and CT/ET over- or undertreatment by ethnicity (p value for interaction: CT overtreatment p 0.211, CT undertreatment p 0.976, ET overtreatment p and ET undertreatment p 0.067). In addition, there was no statistically significant interaction between SES and year of birth, year of diagnosis or age and the risk of CT overtreatment. Sensitivity analysis Similar percentages of patients were treated with neo-adjuvant CT across SES categories. A higher incidence of neo-adjuvant CT was observed in Non-Western Immigrants (11.9%) as compared to Native Dutch or Western Immigrant Patients (7.1% and 8.6%, respectively). Sensitivity analysis in which both adjuvant and neo-adjuvant treated patients were included yielded similar results for the association between SES and risk of CT under- or overtreatment (Sup. Table 1). In addition, no association between ethnicity and guideline discordant CT administration observed (data not shown). 55

56 Chapter 3 Table 2. Patient-, tumor-, hospital- and treatment characteristics of all Dutch female patients surgically treated for primary breast cancer stage I-IIIc between (n = ) by ethnicity. Native Dutch Non-Western Immigrant Western Immigrant n= (94%) n=6569 (5%) n=2022(2%) Patient characteristics Year of diagnosis (27.7) 1745 (26.1) 803 (25.8) (40.0) 2773 (41.5) 1302 (41.8) (32.3) 2161 (32.4) 1013 (32.5) Age, mean (SD) 61.7 (13.8) 56.6 (13.6) 59.5 (14.4) Socio-economic status Low SES (24.8) 2591 (45.1) 653 (36.9) Medium SES (41.9) 2068 (36.0) 669 (37.8) High SES (33.2) 1083 (18.9) 448 (25.3) Missing Hospital characteristics Type of hospital District (50.2) 2687 (44.7) 1356 (49.0) Top clinical (43.7) 2818 (46.9) 1189 (42.9) Academic 7050 (6.2) 506 (8.4) 224 (8.1) Missing Tumor characteristics Morphology Ductal (74.2) 5200 (77.9) 2375 (76.2) Lobular (11.6) 503 (7.5) 315 (10.1) Ductal and Lobular 4022 (3.2) 174 (2.6) 93 (3.0) Adenocarcinoma 3544 (2.8) 212 (3.2) 96 (3.1) Other types (8.1) 590 (8.8) 239 (7.7) Invasive tumor grade Grade (24.7) 1125 (20.0) 640 (24.3) Grade (46.0) 2366 (42.1) 1202 (45.7) Grade (29.3) 2131 (37.9) 787 (29.9) Missing Tumor size in cm (SD) 1.8 (1.4) 1.8 (1.5) 1.8 (1.4) Missing

57 The infuence of socio-economic status and ethnicity on guideline adherence Hormone receptor status ER and PR negative (15.5) 1351 (20.7) 502 (16.5) ER or PR positive (84.5) 5174 (79.3) 2541 (83.5) Missing HER2-status Negative (86.8) 5119 (82.3) 2466 (85.9) Positive (13.2) 1102 (17.7) 406 (14.1) Missing Chapter 3 Axillary lymph-node involvement N (64.5) 3714 (60.1) 1768 (62.9) N (25.8) 1725 (27.9) 737 (26.2) N (6.1) 483 (7.8) 198 (7.0) N (3.5) 258 (4.2) 106 (3.8) Missing Treatment characteristics Type of Surgery Breast-conserving (55.9) 3934 (58.9) 1719 (55.1) Ablative (37.5) 2374 (35.5) 1146 (36.8) Type of surgery unknown 8282 (6.6) 371 (5.6) 253 (8.1) Axillary Surgery No axillary surgery (14.6) 1100 (16.5) 577 (18.5) ALND (25.4) 2050 (30.7) 836 (26.8) SNP (60.0) 3529 (52.8) 1705 (54.7) Gene-expression profile use* No GEP use (93.6) 2734 (95.6) 1228 (94.4) GEP use 3332 (6.4) 125 (4.4) 73 (5.6) Missing Adjuvant chemotherapy (38.4) 3424 (51.3) 1274 (40.9) Adjuvant endocrine therapy (54.6) 3566 (53.4) 1704 (54.7) Neo-adjuvant chemotherapy 8911 (7.1) 794 (11.9) 267 (8.6) Neo-adjuvant endocrine therapy 2008 (1.6) 106 (1.6) 53 (1.7) * The use of gene-expression profiles was only registered since 2012, patients treated before 2012 were coded as missing. Abbreviations: SD = standard deviation, SES = socio-economic status, ER= estrogen receptor, PR = progesterone receptor, ALND = axillary lymph-node dissection, SNP = sentinel node procedure, GEP = gene-expression profile 57

58 Chapter 3 Table 3. Relative risk (RR) for adjuvant chemotherapy (CT) over- or under treatment in Dutch surgically treated patients with stadium I-IIIc unifocal breast cancer who did not receive neo-adjuvant systemic treatment. Chemotherapy overtreatment Number of patients without an indication for adjuvant CT* Number of patients overtreated (%) Unadjusted RR (95% CI) p-value Model 2 RR (95%CI) Model 3 RR (95%CI) SES High SES (5%) 1 [ref] 1 [ref] 1 [ref] 1 [ref] Medium SES (4%) 0.86 [ ] [ ] 0.61 [ ] < [ ] [ ] < [ ] Model 4 RR (95%CI) [ ] < [ ] Low SES (3%) Ethnicity Native Dutch (4%) 1 [ref] 1 [ref] 1 [ref] 1 [ref] Non-Western Immigrant (4%) 1.06 [ ] Western Immigrant (4%) 0.95 [ ] Chemotherapy undertreatment [ ] 0.82 [ ] [ ] 0.77 [ ] [ ] [ ] p-value <0.001 <

59 The infuence of socio-economic status and ethnicity on guideline adherence SES Number of patients with an indication for adjuvant CT* Number of patients undertreated (%) High SES (39%) 1 [ref] 1 [ref] 1 [ref] 1 [ref] Medium SES (37%) 0.99 [ ] < [ ] 0.99 [ ] [ ] [ ] < [ ] [ ] < [ ] Low SES (38%) Ethnicity Native Dutch (39%) 1 [ref] 1 [ref] 1 [ref] 1 [ref] <0.001 <0.001 Non-Western Immigrant (32%) 0.95 [ ] Western Immigrant (40%) 1.01 [ ] [ ] 1.02 [ ] [ ] 1.02 [ ] [ ] [ ] < Model 1: Unadjusted Poisson regression analysis for SES or ethnicity; Model 2: Model 1 + adjusting for the following clinicopathological factors: incidence year, age, morphology, invasive tumor grade, tumor size in categories (<=1cm/1-2cm/>=3cm), ER, PR, Her2 and axillary lymph-node involvement; Model 3: Mixed effect Poisson regression analysis including all variables of Model 2, a random intercept per hospital and correcting for type of hospital (district/teaching/ academic/topclinical) and region of treatment (12 specific regions in the Netherlands); Model 4: Fully adjusted mixed effect Poisson regression analysis including all clinicopathological and hospital factors, a random intercept per hospital and mutually adjusting for SES and ethnicity. Chapter 3 59

60 Chapter 3 Table 4. Relative risk (RR) for adjuvant endocrine therapy (ET) over- or under treatment in hormone-receptor positive surgically treated Dutch early stage breast cancer patients with stadium I-IIIc unifocal disease who did not receive neo-adjuvant systemic treatment.model 1: Unadjusted Endocrine therapy overtreatment Number of patients without an indication for adjuvant ET* Number of patients overtreated (%) Unadjusted RR (95% CI) p-value Model 2 RR (95%CI) p-value Model 3 RR (95%CI) p-value Model 4 RR (95%CI) p-value SES High SES (11%) 1 (ref) 1 (ref) 1 [ref] 1 [ref] Medium SES (11%) Low SES (15%) Ethnicity 1,04 [ ] [ ] 1.40 [ ] < [ ] [ ] [ ] [ ] [ ] Native Dutch (13%) 1 (ref) 1 (ref) 1 [ref] 1 [ref] Non- Western Immigrant (13%) 0.95 [ ] Western Immigrant (15%) 1.14 [ ] [ ] [ ] [ ] 0.94 [ ] [ ] [ ] ,223 0,357 60

61 The infuence of socio-economic status and ethnicity on guideline adherence Endocrine therapy undertreatment Number of patients with an indication for adjuvant ET* Number of patients undertreated (%) SES High SES (12%) 1 (ref) 1 (ref) 1 [ref] 1 [ref] Medium SES (12%) Low SES (13%) Ethnicity 1.00 [ ] [ ] 1.01 [ ] [ ] [ ] [ ] [ ] [ ] Native Dutch (13%) 1 (ref) 1 (ref) 1 [ref] 1 [ref] Non- Western Immigrant (13%) 1.00 [ ] Western Immigrant (13%) 1.01 [ ] [ ] [ ] [ ] 1.01 [ ] [ ] [ ] ,389 0,625 Poisson regression analysis including all clinicopathological and hospital factors, a random intercept per hospital and mutually adjusting for SES and ethnicity. * ET indication based on applicable treatment guideline at time of treatment. Patients with an indication for ET: hormone receptor positive patients with lymphnode positive disease or lymph-node negative disease and unfavourable clinicopathological features. Chapter 3 61

62 Chapter 3 DISCUSSION In this nation-wide study, we observed minimal socio-economic variations in adjuvant CT use in Dutch early stage breast cancer patients. While low SES was associated with a relative 15% lower risk of receiving CT, the observed differences merely resulted from the higher chance of high SES patients to receive CT without such indication combined with a marginally higher risk for CT undertreatment in low SES patients. No association between SES and ET guideline adherence was observed and we did not observe any ethnical disparities. Although the observed disparities in guideline adherence between SES categories on a relative risk scale are small in absolute terms, AST undertreatment is associated with mortality and AST overtreatment leads to unnecessary side-effects therefore both should be avoided to optimize curation chances while maintaining optimal quality of life. Several studies, predominantly conducted in the US, report a lower frequency of CT in low SES patients 10,21, patients with Medicaid insurance 4,6,7 or patients with lower income levels. 8 In a large study, using data of the National Program of Cancer Registries, 34% and 1% of early stage breast cancer patients were respectively under- or overtreated with adjuvant CT and patients with Medicaid insurance were less likely (OR %CI ) to receive guideline concordant CT care compared to privately insured patients. 11 Others did not observe a significant association between SES and CT. 12,22 We did not observe any socio-economic disparities in ET guideline adherence which is in line with most studies addressing this association. 11,13,21 24 Wu et al. did observe a borderline significant increased risk for ET guideline discordance in patients living in high-poverty areas but found no association between insurance type or education level. 11 Others report a lower risk of ET in women without insurance 8 or with Medicaid insurance as compared to women who were privately insured. 12 The present study is conducted in an equal access healthcare-system mandating insurance for all with full reimbursement of AST. Therefore, it is unlikely that the observed SES associated disparities CT administration are attributable to financial incentives. Our results might suggest that high SES patients are more inclined to undergo all therapeutic options, even if there is no certain benefit and these could be potentially harmful, to overcome disease. This reasoning is supported by a recent population based study from the Netherlands in which high SES breast cancer patients were more likely to undergo axillary dissection. 25 In other forms of cancer this tendency to more aggressive treatment in high SES patients is also observed, such as a higher likelihood of receiving surgery for pancreatic or esophageal cancer in Dutch high SES patients. 26,27 In other countries this trend is also observed, for example illustrated by a higher incidence of lymphnode dissection in high SES colon cancer patients from Louisiana. 28 Together with the fact that high SES patients play a more proactive role in decision-making 14, this trend 62

63 The infuence of socio-economic status and ethnicity on guideline adherence might explain the observed higher incidence of CT overtreatment. In the US financial disparities could also underlie these findings, whereas in the Dutch system this is less likely. In this line of reasoning, physicians should be more responsive towards this tendency to opt for more aggressive treatment in high SES patients. Another possible explanation could be a difference in physician s contributions to CT decision-making between SES categories. However, as our study design precludes detailed analysis on this observation, this remains speculation. Chapter 3 Our results indicate that ethnical disparities in AST guideline adherence can be excluded on an aggregated level. Some authors report higher incidence of AST administration in black 12 or Hispanic 12,24 patients whereas others observe a lower risk of AST administration in minority patient groups. Our findings are in line with two large recent studies who also did not observe ethnical disparities in AST administration. 8,11 Based on our study-size and narrow 95%CI s found for the effect estimates, we can conclude that even the presence of very small ethnical disparities in AST guideline non-adherence can be excluded. Our population-based approach allows a comprehensive overview of socio-economic and ethnic variation in AST guideline adherence. The association between SES or ethnicity and guideline discordant AST at a nation-wide level in a healthcare system that seeks to provide equal access to care has not been assessed before. Nevertheless, use of this comprehensive real-life data reflecting routine clinical care comes with the cost of limited availability of detailed information per patient. First, reasons to deviate from guidelines and information on co-morbidities were unknown. It is important to note that deviation from AST guideline recommendations does not equal poor medical care. In addition, we had to rely on postal code information for determination of SES. Although this method has been used before 23 and has shown to accurately reflect SES of individual patients 29, this is prone to misclassification and could have diluted the observed association. Regarding ethnicity, for 25% of patients country of birth was estimated based on a name-based approach. Although in this method is very specific for the largest immigrant countries a certain error measure could be included which could also have led to a diluted observed association. Then again, sensitivity analysis, in which we only included patients for whom country of birth was actually registered, also revealed no association between ethnicity and CT or ET under- or overtreatment. In this study, conducted in an equal access care system, SES was not a limiting factor in obtaining appropriate care, which suggests that without financial disparities in health care reimbursement, better equality in the administration of AST is achieved. This finding is of great importance given that breast cancer is the most common cancer type and leading cause of cancer death among females worldwide. 63

64 Chapter 3 In conclusion, although there is a relation between guideline discordant CT administration and SES in early stage breast cancer patients treated in an equal access health care setting, the observed associations were rather small. Low SES patients are only slightly more likely to be undertreated and less likely to be overtreated. We observed no ethnical disparities in AST guideline adherence in the Netherlands. Sup. Figure 1. Flowchart of the analysed study-population. Dutch female patients surgically treated for primary invasive breast cancer (stage I-IIIc) between n = Excluded from the analysis: - Patients with multifocal disease or information on focality unknown (n = ) - Patients with multifocal disease who received neo-adjuvant CT or ET (n = 3524) - Patients with unifocal disease who received neo-adjuvant CT or ET (n = 8253) Dutch female patients surgically treated for primary invasive unifocal breast cancer (stage I-IIIC) between who did not receive neo-adjuvant treatment n =

65 The infuence of socio-economic status and ethnicity on guideline adherence Sup. Figure 2. Relative Risk (RR) for adjuvant chemotherapy (CT) overtreatment in surgically treated Dutch early stage breast cancer patients (stadium I-IIIc) with unifocal disease without an indication for adjuvant CT according to socio-economic status in deciles according to the fully adjusted mixed-effect Poisson regression analysis (Model 4). Chapter 3 * Represents a p-value of <0.05. The highest (10 th ) SES decile is the reference category. 65

66 Chapter 3 Sup. Table 1. Relative risk (RR) for chemotherapy (CT) (i.e. both neo-adjuvant as adjuvant) over- or under treatment in Dutch surgically treated breast cancer patients with stadium I-IIIc unifocal disease including patients who received neo-adjuvant systemic treatment. Unadjusted RR (95% CI) p-value Model 4 RR (95%CI) p-value Chemotherapy overtreatment SES Number of patients without an indication for adjuvant CT* Number of patients overtreated (%) High SES (6%) 1 [ref] 1 [ref] Medium SES (5%) 0.84 [ ] < [ ] Low SES (4%) 0.64 [ ] < [ ] <0.001 Ethnicity Native Dutch (5%) 1 [ref] 1 [ref] Non-Western Immigrant (7%) 1.40 [ ] < [ ] Western Immigrant (5%) 0.94 [ ] [ ] Chemotherapy undertreatment Number of patients with an indication for adjuvant CT* Number of patients undertreated (%) 66

67 The infuence of socio-economic status and ethnicity on guideline adherence SES High SES (35%) 1 [ref] 1 [ref] Medium SES (34%) 0.99 [ ] < [ ] Low SES (34%) 0.99 [ ] [ ] <0.001 Ethnicity Native Dutch (35%) 1 [ref] 1 [ref] Non-Western Immigrant (28%) 0.95 [ ] < [ ] Western Immigrant (36%) 1.00 [ ] [ ] Model 4: Fully adjusted mixed effect Poisson regression analysis correcting for year of diagnosis, age, region of treatment, type of hospital, tumor grade, morphology, tumor size, ER/PR and Her2 status and axillary lymph-node involvement including a random intercept per hospital and mutually adjusting for SES and ethnicity. * CT (both adjuvant and neo-adjuvant) indication based on applicable treatment guideline at time of treatment. Patients with an indication for CT: patients under the age of 70 with lymph-node positive disease or lymph-node negative disease and unfavourable clinicopathological features. Chapter 3 67

68 Chapter 3 Sup. Table 2. Relative risk (RR) for chemotherapy or endocrine therapy initiation (i.e. regardless of guideline recommendation) in patients with stadium I-IIIc unifocal disease surgically treated between in the Netherlands. Chemotherapy initiation Number of patients Patients received chemotherapy Unadjusted RR (95% CI) p value Model 4 RR (95%CI) SES High SES (36%) 1 [ref] 1 [ref] Medium SES (36%) 0.99 [ ] [ ] <0.001 Low SES (28%) 0.79 [ ] < [ ] <0.001 Ethnicity Native Dutch (32%) 1 [ref] 1 [ref] Non-Western Immigrant (44%) 1.35 [ ] < [ ] 0,567 Western Immigrant (34%) 1.07 [ ] 0, [ ] 0,714 Endocrine therapy initiation* Number of patients Patients received endocrine therapy SES High SES (41%) 1 [ref] 1 [ref] Medium SES (40%) 1.02 [ ]] [ ] 0,880 Low SES (38%) 1.07 [ ]] < [ ] 0,145 Ethnicity Native Dutch (61%) 1 [ref] 1 [ref] Non-Western Immigrant (63%) 1.03 [ ] 0, [ ] 0,260 Western Immigrant (61%) 1.01 [ ] 0, [ ] 0,300 p value Model 4: Fully adjusted mixed effect Poisson regression analysis correcting for year of diagnosis, age, region of treatment, type of hospital, tumor grade, morphology, tumor size, ER/PR and Her2 status and axillary lymph-node involvement including a random intercept per hospital and mutually adjusting for SES and ethnicity. * Only includes hormone-receptor positive patients. 68

69 The infuence of socio-economic status and ethnicity on guideline adherence REFERENCE LIST 1. National Institutes of Health Consensus Development Panel. NIH consensus conference. Treatment of early-stage breast cancer. JAMA. 1991;265(3): Available at: Accessed April 19, Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom.; Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05) Warren JL, Butler EN, Stevens J, et al. Receipt of chemotherapy among medicare patients with cancer by type of supplemental insurance. J Clin Oncol. 2015;33(4): doi: /jco Lipscomb J, Gillespie TW, Goodman M, et al. Black-white differences in receipt and completion of adjuvant chemotherapy among breast cancer patients in a rural region of the US. Breast Cancer Res Treat. 2012;133(1): doi: /s Griggs JJ, Hawley ST, Graff JJ, et al. Factors associated with receipt of breast cancer adjuvant chemotherapy in a diverse population-based sample. J Clin Oncol. 2012;30(25): doi: /jco Simon MS, Lamerato L, Krajenta R, et al. Racial Differences in the Use of Adjuvant Chemotherapy for Breast Cancer in a Large Urban Integrated Health System. Int J Breast Cancer. 2012;2012:1 8. doi: /2012/ Freedman RA, Virgo KS, He Y, et al. The association of race/ethnicity, insurance status, and socioeconomic factors with breast cancer care. Cancer. 2011;117(1): doi: /cncr Yen TWF, Czypinski LK, Sparapani RA, et al. Socioeconomic factors associated with adjuvant hormone therapy use in older breast cancer survivors. Cancer. 2011;117(2): doi: /cncr Popescu I, Schrag D, Ang A, Wong M. Racial/Ethnic and Socioeconomic Differences in Colorectal and Breast Cancer Treatment Quality: The Role of Physician-level Variations in Care. Med Care. 2016;54(8): doi: /mlr Wu X-C, Lund MJ, Kimmick GG, et al. Influence of race, insurance, socioeconomic status, and hospital type on receipt of guideline-concordant adjuvant systemic therapy for locoregional breast cancers. J Clin Oncol. 2012;30(2): doi: / JCO Hassett MJ, Schymura MJ, Chen K, Boscoe FP, Gesten FC, Schrag D. Variation in breast cancer care quality in New York and California based on race/ethnicity and Medicaid enrollment. Cancer. 2016;122(3): doi: /cncr Chapter 3 69

70 Chapter Bickell NA, Wang JJ, Oluwole S, et al. Missed opportunities: racial disparities in adjuvant breast cancer treatment. J Clin Oncol. 2006;24(9): doi: / JCO Degner LF, Kristjanson LJ, Bowman D, et al. Information needs and decisional preferences in women with breast cancer. JAMA. 1997;277(18): Available at: ncbi.nlm.nih.gov/pubmed/ Accessed April 21, Peele PB, Siminoff LA, Xu Y, Ravdin PM. Decreased use of adjuvant breast cancer therapy in a randomized controlled trial of a decision aid with individualized risk information. Med Decis Making. 2005;25(3): doi: / x Ashing-Giwa KT, Padilla G, Tejero J, et al. Understanding the breast cancer experience of women: A qualitative study of African American, Asian American, Latina and Caucasian cancer survivors. Psychooncology. 2004;13(6): doi: /pon Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary Care Physicians Who Treat Blacks and Whites. N Engl J Med. 2004;351(6): doi: / NEJMsa C VDKI. Sociaal-economische status indicator op postcode niveau. Maand Stat van Bevolk. 2002;50: Standaard definitie allochtonen. Cent Bur voor Stat. 2010;10: Rabe-Hesketh S. Maximum likelihood estimation of limited and descrete dependent variable models with random effects. J Econom. 2005;128: Guy GP, Lipscomb J, Gillespie TW, Goodman M, Richardson LC, Ward KC. Variations in Guideline-Concordant Breast Cancer Adjuvant Therapy in Rural Georgia. Health Serv Res. 2015;50(4): doi: / Banerjee M, George J, Yee C, Hryniuk W, Schwartz K. Disentangling the effects of race on breast cancer treatment. Cancer. 2007;110(10): doi: /cncr Aarts MJ, Voogd AC, Duijm LEM, Coebergh JWW, Louwman WJ. Socioeconomic inequalities in attending the mass screening for breast cancer in the south of the Netherlands--associations with stage at diagnosis and survival. Breast Cancer Res Treat. 2011;128(2): doi: /s z. 24. Livaudais JC, Hershman DL, Habel L, et al. Racial/ethnic differences in initiation of adjuvant hormonal therapy among women with hormone receptor-positive breast cancer. Breast Cancer Res Treat. 2012;131(2): doi: /s Aarts MJ, Hamelinck VC, Bastiaannet E, et al. Small but significant socioeconomic inequalities in axillary staging and treatment of breast cancer in the Netherlands. Br J Cancer. 2012;107(1):12 7. doi: /bjc Bakens MJAM, Lemmens VEPP, de Hingh IHJT. Socio-economic status influences the likelihood of undergoing surgical treatment for pancreatic cancer in the Netherlands. HPB doi: /j.hpb

71 The infuence of socio-economic status and ethnicity on guideline adherence 27. van Vliet EPM, Eijkemans MJC, Steyerberg EW, et al. The role of socio-economic status in the decision making on diagnosis and treatment of oesophageal cancer in The Netherlands. Br J Cancer. 2006;95(9): doi: /sj.bjc Hsieh M-C, Velasco C, Wu X-C, Pareti LA, Andrews PA, Chen VW. Influence of socioeconomic status and hospital type on disparities of lymph node evaluation in colon cancer patients. Cancer. 2012;118(6): doi: /cncr Bos V, Kunst A, Mackenbach J. De omvang van sociaal-economische verschillen gemeten op buurtniveau: vergelijkingen met schattingen op basis van informatie op individueel niveau. Sociaal-economische gezondheidsverschillen van verklaren naar verkleinen. 2001:8 20. Chapter 3 71

72

73 PART II Gene-expression profiling and adjuvant systemic treatment decision-making

74

75 CHAPTER 4 Factors associated with the use of gene expression profiles in estrogen receptor positive early-stage breast cancer patients: a nationwide study. A. Kuijer, K. Schreuder, S.G. Elias, C.H. Smorenburg, E.J.Th. Rutgers, S. Siesling, T. van Dalen. Public Health Genomics 2016;19:

76 Chapter 4 ABSTRACT: Background: Breast cancer guidelines suggest the use of gene-expression profiles (GEPs) in estrogen-receptor positive (ER+) breast cancer patients in whom controversy exists regarding adjuvant chemotherapy benefit based on traditional prognostic factors alone. We evaluated current use of GEPs in these patients in the Netherlands. Patients and methods: Primary breast cancer patients treated between eligible for GEP use according to Dutch guideline were identified in the Netherlands Cancer Registry: ER+ patients <70 years with grade 1 > 2cm; or grade cm tumors without overt lymph node metastases (pn0-nmi). Mixed-effect logistic regression analysis was performed to associate characteristics of patients, tumors and hospitals with GEP use. Results: GEPs were increasingly deployed: 12% of eligible patients received a GEP in 2011 vs. 46% in Lobular vs. ductal morphology (OR %CI ), pn1mi status (vs. pn0 OR %CI ), tumor size (>3 cm vs. >2 cm OR %CI ) were inversely associated with GEP use. High socioeconomic status (OR %CI ) and younger age (OR 0.96/year increasing age 95%CI ) were positively associated with GEP use. GEP use per hospital did vary, but no predefined institutional factors remained independently associated with GEP use. Conclusion: GEP use increased over time and was influenced by patient- and tumor associated factors, as well as by socioeconomic status. 76

77 Factors associated with the use of gene expression profiles INTRODUCTION Breast cancer prognosis has improved over the last two decades for an important part due to the administration of adjuvant systemic therapy (AST). 1 The indication area for AST has gradually expanded, now recommending adjuvant chemotherapy (CT) and/ or endocrine therapy in the majority of all breast cancer patients. Then again, there is a growing awareness that this is associated with a considerable risk of CT overtreatment. Especially in estrogen-receptor positive (ER+) early-stage breast cancer patients controversy exists regarding the benefit of adjuvant CT. Several gene-expression profiles (GEPs) were developed and validated to improve prognosis prediction and hereby reduce CT overtreatment in these patients. 2 5 Currently, both national 6 and international 7,8 treatment guidelines suggest the use of a GEP as an adjunct to clinicopathological factors to guide decisions on adjuvant CT for ER+ early-stage breast cancer. Chapter 4 The Dutch national breast cancer guideline (2012) suggests the use of a validated GEP in ER+ breast cancer patients in whom controversy exists regarding the benefit of administering adjuvant CT based on traditional prognostic factors alone. 6 In 2011 the 70-GS became available in Dutch clinical practice followed by the 21-recurrence score which became available in the Netherlands in GEPs are available for every Dutch ER+ breast cancer patient since health insurance is mandatory and practically all health insurance companies fully reimburse GEPs. We have recently reported that using GEPs leads to a decrease in CT administration in Dutch ER+ early-stage breast cancer patients. In this study we noticed that only a modest proportion of Dutch breast cancer patients for whom GEP is considered worthwhile actually received a GEP. 9 Therefore, the aim of the present study is to gain insight in factors associated with the use of a GEP in daily practice in patients with ER+ early breast cancer in the Netherlands. PATIENTS AND METHODS Study-population The Dutch national guideline (2012) suggests the use of a GEP in ER+ breast cancer patients in whom controversy exists regarding the benefit of adjuvant CT based on traditional prognostic factors alone. 6 According to the directives in the Dutch guidelines this category consists of patients year of age with either grade 1 ER+ invasive breast cancer > 2 cm or grade 2 ER+ invasive breast cancer 1-2 cm with no or limited axillary lymph-node involvement (pn0 or pn1mi). In the absence of GEP use national guidelines would advocate the administration of CT for this category of patients. Patients within this guideline-directed indicated area for GEP use who were surgically treated between January 1, 2011, and December 31, 2014, were identified in the Netherlands 77

78 Chapter 4 Cancer Registry (NCR). Patients with a prior malignancy and those receiving neoadjuvant systemic treatment were excluded. Data collection and variable categorization The NCR prospectively registers demographic and clinicopathological information of all cancer patients treated in the Netherlands since Demographic information included age and postal code. The postal code at the time of diagnosis was used to determine socio-economic status (SES). This SES-indicator uses individual fiscal data based on a combination of the mean value of the home and mean house-hold income and is provided at an aggregated level for each postal code (covering an average of 17 households). Postal codes were categorized to one of three predefined socio-economic status categories: low (first to third decile), medium (fourth to seventh decile) and high (eight to tenth decile). Patients living in a care-providing institution were categorized into a fourth category that was not included in the present analysis. Furthermore, common clinicopathological variables on all Dutch cancer patients are prospectively collected by the NCR. The NCR started registering use of a GEP since 2011 of both the 70-gene signature (70-GS) and the 21-recurrence score (21-RS). Information obtained on hospital characteristics consisted of type of hospital, hospital localization and volume of delivered breast cancer care. Institutional patient volume was categorized based on the annual number of patients treated for primary breast cancer (<100, , > 200 breast cancer patients per year), hospital localization as a location in the North, Middle or South of the Netherlands). There are currently 26 top clinical hospitals in the Netherlands. These hospitals, without an academic affiliation, focus on improvement of quality of care, education and medical research and are inspected every 5-year on strict quality criteria to obtain or preserve this quality mark. Hospital type was categorized as teaching hospital for surgical and/or internal medicine residents (yes/no) and as district hospitals, university hospitals or top clinical hospitals (not affiliated with a medical university). Statistical analysis The distributions of patient-, tumor- and hospital characteristics were compared between patients who did or did not receive a GEP with a Chi-square test for categorical variables and a Wilcoxon rank sum test for the non-normal distributed continuous variables age, volume of breast cancer care and tumor size. To assess the increase in GEP use over time, percentages of patients within the guideline directed-indicated area for GEP use actually receiving a GEP, categorized according to SES, was plotted against calendar year. A mixed-effect logistic regression analysis was performed to assess the association between patient-/tumor- and hospital characteristics and GEP use, taking into account patient clustering within hospitals. For this purpose, we included a random intercept per 78

79 Factors associated with the use of gene expression profiles hospital (thus taking baseline differences in GEP use among hospitals into account). We adjusted for age and tumor size (continuously) and tumor morphology, invasive tumor grade, PR receptor status, axillary status, incidence year, socio-economic status, volume of breast cancer care in treating hospital and type of hospital and region (categorically). A p-value < 0.05 was considered to be statistically significant. All analyses were performed by using STATA, version 12.0, and in R version using the lme4 package for the mixed-effect model. RESULTS Based on the guideline advocated use of GEPs 5110 patients were eligible for GEP use during our study period and 1360 of them (27%) received a GEP. In most patients (1321) the 70-gene signature was used whereas 39 patients (3%) received the 21-recurrence score. Over time, GEPs were increasingly used: in 2011, 12% of all eligible patients received a GEP, compared to 13%, 33% and 46% in 2012, 2013 and 2014, respectively. Of patients who received a GEP 66% were assigned to the low-risk and 29% to the high-risk category by the 70-GS. In 5% of patients who received the 70-GS no risk profile was recorded. The 21-recurrence score assigned 67% of patients to the low-, 15% to the intermediate and 18% of patients to the high-risk category. The test result was adhered to (i.e. no administration of CT in case of a low-risk profile and administration of CT in case of a high-risk profile) in 89% of all patients. Chapter 4 Factors associated with GEP use Characteristics of patients, tumor and hospital according to the use of GEP are depicted in Table 1. Tumor characteristics significantly associated in univariable analysis with the use of a GEP were invasive ductal carcinoma as opposed to lobular carcinoma, the absence of axillary micro-metastases (pn0), small tumor size and intermediate malignancy grade (as opposed to low malignancy grade). Patients who received a GEP were on average younger and of higher socio-economic status compared to patients who did not receive a GEP. In addition, GEP testing was more frequent in patients treated in district hospitals, in hospitals with a higher volume of breast cancer care and in hospitals situated in the Northern part of the Netherlands (Table 1). 79

80 Chapter 4 Table 1. Patient-, tumor- and hospital characteristics of patients within the guideline-directed indicated area for gene-expression profile (GEP) use (ER+/Her2- disease without axillary lymph-node involvement and grade I tumours > 2 cm or grade II tumours 1-2 cm). Tumor characteristics No GEP GEP (n = 3750) (n = 1360) p-value n (%) n (%) Morpholgy Ductal 2801 (75%) 1133 (83%) Lobular 679 (18%) 162 (12%) Mixed 143 (4%) 42 (3%) Other 127 (3%) 23 (2%) < Tumorsize in mm (mean)* 23,8 16,2 0,002 Invasive tumor grade Grade I 462 (12%) 105 (8%) Grade II 3288 (88%) 1255 (92%) < Progesterone receptor status Negative 3173 (85%) 1151 (85%) Positive 565 (15%) 208 (15%) 0,299 pn status pn (88%) 1248 (92%) pn1mi 438 (12%) 112 (8%) < Patient characteristics Age in years (mean)* 57,4 55,6 < < 35 5 (0%) 1 (0%) (24%) 400 (29%) (76%) 959 (71%) 0,001 Socioeconomic status Low 1055 (28%) 328 (24%) Medium 1561 (42%) 522 (38%) High 1134 (30%) 510 (38%) <

81 Factors associated with the use of gene expression profiles Hospital characteristics Volume of breast cancer care per year (mean)* 138,7 144,8 < < 100 patients 1010 (27%) 310 (23%) patients 2132 (57%) 776 (57%) > 200 patients 608 (16%) 274 (20%) < Type of hospital District 1065 (28%) 424 (31%) Top clinical 2424 (65%) 872 (64%) University 261 (7%) 64 (5%) 0,005 Teaching hospital No 2457 (66%) 875 (64%) Yes 1293 (34%) 485 (36%) 0,433 Region North 1375 (37%) 577 (43%) Middle 1489 (40%) 491 (36%) South 883 (23%) 292 (21%) 0,002 * Wilcoxon rank sum test, other values represent chi-square values. Chapter 4 81

82 Chapter 4 We plotted the proportion of patients receiving a GEP for the three different SES categories over time to assess whether there was a relation between time since reimbursement of GEPs and the association between SES and GEP use (Figure 1). Patients of lower SES had similar tumor- and patient characteristics as patients of medium or high SES. However, patients of high SES were slightly more often treated in hospitals with a large volume of breast cancer care, top clinical hospitals and hospitals situated in the Northern part of the Netherlands (Suppl. Table 1). Before reimbursement (2011), there was no significant difference in GEP use between the SES categories (11%; 13% vs. 12%, of patients received a GEP in the low-, medium and high SES category in 2011, p 0.589). After reimbursement GEPs were more frequently deployed in patients of high SES compared to low- or medium SES (Figure 1). Figure 1. Gene-expression profile (GEP) use in Dutch breast cancer patients within the guideline-directed indication area for GEP use over time, by socio-economic status (SES) categories. 82

83 Factors associated with the use of gene expression profiles Mixed-effect logistic regression analysis Mixed-effect logistic regression analysis, including a random intercept per hospital to correct for patient clustering at a hospital level (thus taking baseline difference in GEP use between individual hospitals into account), demonstrated an independently decreased probability of GEP use for patients with invasive lobular carcinomas (vs. ductal OR 0,58, 95%CI ), larger tumor size (>3cm vs. <2cm OR %CI ) and presence of axillary lymph node micro-metastasis (vs. pn0 OR %CI ). Patient characteristics independently associated with GEP use were younger age (OR 0.96/year increase in age 95%CI ), high SES (vs. low SES OR %CI ) and diagnosis in a more recent year (2014 vs OR %CI ). In the mixed-effect analysis none of the aforementioned institutional characteristics remained independently associated with GEP use (Table 2). An interaction term for SES at a patient level and percentage of patients of high SES per hospital was added to the model to assess whether there was a difference in the association between SES and GEP use in hospitals with a high SES patient population but this interaction term was not statistically significant. Chapter 4 DISCUSSION In this population based study in a country where GEPs are available for every ER+ breast cancer patient, we observed an increase in GEP use over time with considerable variation in GEP use in eligible patients. In % of eligible patients according to the Dutch breast cancer treatment guideline received a GEP. As expected, tumor factors pertaining to an intermediate clinical risk profile were independently associated with GEP use. Surprisingly, we observed a lower probability of GEP deployment in patients of low SES. Furthermore, the proportion of eligible patients receiving a GEP differed between individual hospitals. However, this inter-hospital variation could not be explained by hospital size, - type, - region or presence of an educational program. 83

84 Chapter 4 Table 2. Patient-, tumor and hospital characteristics associated with gene-expression profile (GEP) use in patients within the guideline-directed indication area for GEP in a multilevel logistic regression model including a random intercept per hospital. Odds Ratio 95% CI p-value Tumor characteristics Morpholgy Ductal 1 (ref) Lobular 0, < Mixed 0, Other 0, Tumor size < 2 cm 1 (ref) 2 3 cm 0, > 3 cm 0, Invasive tumor grade Grade I 1 (ref) Grade II 1, Progesterone receptor status Negative 1 (ref) Positive 0, pn status pn0 1 (ref) pn1mi 0, < Patient characteristics Age in years 0, < Incidence year (ref) , , < , < Socioeconomic status Low 1 (ref) Medium 1, High 1,

85 Factors associated with the use of gene expression profiles Hospital characteristics Volume of breast cancer care per year < 100 patients 1 (ref) patients 0, > 200 patients 0, Type of hospital District 1 (ref) Top clinical 0, University 0, Teaching hospital No 1 (ref) Yes 0, Region North 1 (ref) Middle 0, South 0, Chapter 4 85

86 Chapter 4 In accordance with previous reports we observed a higher probability of GEP testing in patients with an intermediate clinical risk profile: smaller tumors of low or intermediate grade without axillary lymph-node involvement This finding is in itself not surprising since in these patients most controversy exists regarding CT benefit. GEP use in this category can lead to the decision to omit CT while national guidelines would otherwise advocate administration of CT. We observed lower percentages of patients with axillary micro-metastasis receiving a GEP, reflecting a reluctant attitude of clinicians to consider patients with axillary micro-metastasis as clinical intermediate risk. Furthermore, patients with an invasive ductal carcinoma were more likely to receive GEP testing compared to patients with tumors of lobular pathology which coheres to the knowledge that GEP validation studies did not include analysis of histologic subtypes and controversy exists regarding the deployment of GEPs in tumors other than of ductal morphology. Considerable inter-hospital variation in GEP use was observed. In univariable analysis, GEPs were more frequently deployed in regional hospitals with a large patient volume situated in the Northern part of the Netherlands. However, this association between institutional factors and GEP use did not remain significant after correction for baseline difference among hospitals in GEP use in a mixed effect logistic regression model, thus taking patient clustering at a hospital level into account. This finding indicates that the chance of receiving a GEP depends on the hospital in which a patient is diagnosed but this difference is not attributable to the type of hospital, the volume of breast cancer care per hospital or the region. Until date, there are limited reports on inter-hospital variation regarding deployment of a GEP. Enewold et al. reported no association between hospital ownership, presence of a residency program or hospital bed size and 21-RS use in the United states. 13 Based on the present data we conclude that baseline attitudes of hospitals towards GEPs vary but this attitude is not associated with hospital type, size, region or presence of an educational program. Some patient factors were independently associated with the receipt of a GEP in the current study. We observed a higher incidence of GEP use in younger patients. Several reports endorse this finding 13,14,16 whereas others report an increased use of GEPs in patients between years of age. 11,15 Since the added value of CT is inversely related with age, it is possible that the observed preferential GEP use in younger women is explained by the fact that GEPs are mainly used to seek reassurance in withholding CT in these women in whom guidelines advocate administration of CT. 17 In a previous study, conducted by this research group, we observed a higher baseline propensity to administer CT in younger women eligible for GEP use compared to patients of older age. In the latter study GEP use was independently associated with a decreased probability of receiving CT in younger patients whereas in older patients a reverse relationship was 86

87 Factors associated with the use of gene expression profiles observed. 9 The conceivably more aggressive attitude in younger women might explain the increased use of GEPs in order to come to a substantiated decision to omit CT in younger women. Noteworthy, Dutch patients with lower SES were less likely to receive GEP testing in the current study compared to patients of high SES. Previous studies, mainly conducted in a US health care setting, report contradictory results on the association between race, median income or educational status and GEP use. Some observed disparities 13,15,18 whereas others found no differences in GEP uptake. 11,12 In a US health-care setting disparities in GEP uptake between different socio-economic classes may well be explained by financial inequalities. In a Dutch health care setting these financial motives cannot explain difference in GEP uptake as GEPs are fully reimbursed for every breast cancer patient. The fact that only a limited proportion of patients received a GEP illustrates that GEP use within the guideline-directed indicated area is not yet self-evident and one can hypothesize that GEP use is driven by patient request to some extent. DeFrank et al. 12 report higher incidence of GEP use in patients who played an active role in their treatment decision-making style; these patients commonly are younger patients and of a higher educational level. 19 The retrospective observational design of the present study precludes firm conclusions. Chapter 4 The population-based character of the current study makes this work unique and enables us to give an overview of GEP use in the Dutch health care setting. Both 21-RS use and 70-GS use were incorporated into the current study in contrast to other reports on GEP uptake. Data on income or education on an individual level was not available and we therefore used mean value of the home and household income on an aggregated level as a proxy for SES. Although this method is adapted by others and has shown to give a fair estimation of SES on an individual level 20, care must be taken when interpreting the association between SES and GEP use. Furthermore, the current study design precludes detailed analysis on possible explanations for the observed disparities in GEP use uptake between different SES categories and therefore our findings merit further study. In conclusion, substantial variation was observed in the deployment of GEPs in breast cancer patients eligible for GEP use in the Dutch health-care setting. In 2014 nearly half of all patients for whom GEPs are considered worthwhile received a GEP. Tumor characteristics pertaining to an intermediate clinical risk-profile were associated with the use of GEP. Older patients and patients of low SES were less likely to receive GEP testing, the latter coming as a surprise. As GEP use within this guideline-directed indicated area for GEP use has been shown to decrease the proportion of patients receiving CT and hence prevents overtreatment and two-thirds of the tests come out as low risk, efforts should be made to diminish the disparities in GEP use. 87

88 Chapter 4 Acknowledgements This work was supported by the Dutch Cancer Society (KWF). 88

89 Factors associated with the use of gene expression profiles Sup. Table 1. Patient-, tumor and treatment characteristics of patients within the guideline-directed indicated area for gene-expression profile use stratified by socioeconomic status (SES). Tumor characteristics Low SES Medium SES High SES (n = 1383) (n = 2083) (n = 1644) p-value n (%) n (%) n (%) Morpholgy Ductal 1086 (79%) 1603 (80%) 1245 (76%) Lobular 208 (15%) 338 (16%) 295 (18%) Mixed 41 (3%) 83 (4%) 61 (4%) Other 48 (4%) 59 (3%) 43 (3%) Tumorsize in mm (mean)* Chapter 4 Invasive tumor grade Grade I 175 (13%) 216 (10%) 176 (11%) Grade II 1208 (87%) 1867 (90%) 1468 (89%) PR-status Negative 208 (15%) 309 (15%) 256 (16%) Positive 1174 (85%) 1771 (85%) 1379 (84%) Missing pn status pn (90%) 1865 (90%) 1450 (88%) pn1mi 138 (10%) 218 (10%) 194 (12%) GEP use No 1055 (76%) 1561 (75%) 1134 (69%) Yes 328 (24%) 522 (25%) 510 (31%) <0.001 Patient characteristics Age in years (mean)* < 35 2 (0%) 2 (0%) 2 (0%) (24%) 517 (25%) 452 (28%) (76%) 1564 (75%) 1190 (72%) Incidence year (24%) 497 (23%) 372 (23%) (23%) 492 (24%) 388 (24%) (24%) 509 (24%) 388 (24%) (22%) 464 (22%) 403 (25%) Hospital characteristics Volume of breast cancer care p/y (mean)* <

90 Chapter 4 <100 patients 402 (29%) 517 (25%) 401 (24%) patients 789 (57%) 1183 (57%) 936 (57%) >200 patients 192 (14%) 383 (18%) 307 (19%) < Type of hospital District 475 (34%) 647 (31%) 367 (22%) Top clinical 801 (58%) 1336 (64%) 1159 (71%) University 107 (8%) 100 (5%) 118 (7%) < Teaching hospital No 505 (37%) 725 (35%) 548 (33%) Yes 878 (63%) 1358 (65%) 1096 (67%) Region North 630 (46%) 720 (35%) 602 (37%) Middle 424 (31%) 763 (37%) 793 (48%) South 328 (24%) 599 (29%) 248 (15%) <

91 Factors associated with the use of gene expression profiles REFERENCE LIST 1. Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05) van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: / NEJMoa Paik S, Shak S, Tang G, et al. A multigene assay to predict recurrence of tamoxifentreated, node-negative breast cancer. N Engl J Med. 2004;351(27): doi: / NEJMoa Parker JS, Mullins M, Cheang MCU, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27(8): doi: / JCO Filipits M, Rudas M, Jakesz R, et al. A New Molecular Predictor of Distant Recurrence in ER-Positive, HER2-Negative Breast Cancer Adds Independent Information to Conventional Clinical Risk Factors. Clin Cancer Res. 2011;17(18): doi: / ccr Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Senkus E, Kyriakides S, Penault-Llorca F, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2013;24(suppl 6):vi7 vi23. doi: /annonc/mdt Goldhirsch A, Wood WC, Coates AS, et al. Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol Off J Eur Soc Med Oncol. 2011;22(8): doi: /annonc/mdr Kuijer A, van Bommel ACM, Drukker CA, et al. Using a gene expression signature when controversy exists regarding the indication for adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. Genet Med. 2016;18(7): doi: /gim Siesling S, Louwman WJ, Kwast A, et al. Uses of cancer registries for public health and clinical research in Europe: Results of the European Network of Cancer Registries survey among 161 population-based cancer registries during Eur J Cancer. 2015;51(9): doi: /j.ejca Roberts MC, Weinberger M, Dusetzina SB, et al. Racial Variation in the Uptake of Onco type DX Testing for Early-Stage Breast Cancer. J Clin Oncol. 2016;34(2): doi: /jco Chapter 4 91

92 Chapter DeFrank JT, Salz T, Reeder-Hayes K, Brewer NT. Who gets genomic testing for breast cancer recurrence risk? Public Health Genomics. 2013;16(5): doi: / Enewold L, Geiger AM, Zujewski J, Harlan LC. Oncotype Dx assay and breast cancer in the United States: usage and concordance with chemotherapy. Breast Cancer Res Treat. 2015;151(1): doi: /s Zhu X, Dent S, Paquet L, Zhang T, Graham N, Song X. Factors influencing Oncotype DX use in the management of early breast cancer: a single centre experience. Eur J Cancer. 2014;50(15): doi: /j.ejca Hassett MJ, Silver SM, Hughes ME, et al. Adoption of gene expression profile testing and association with use of chemotherapy among women with breast cancer. J Clin Oncol. 2012;30(18): doi: /jco Haas JS, Liang S-Y, Hassett MJ, Shiboski S, Elkin EB, Phillips KA. Gene expression profile testing for breast cancer and the use of chemotherapy, serious adverse effects, and costs of care. Breast Cancer Res Treat. 2011;130(2): doi: /s Carlson JJ, Roth JA. The impact of the Oncotype Dx breast cancer assay in clinical practice: a systematic review and meta-analysis. Breast Cancer Res Treat. 2013;141(1): doi: /s z. 18. Lund MJ, Mosunjac M, Davis KM, et al. 21-Gene recurrence scores: racial differences in testing, scores, treatment, and outcome. Cancer. 2012;118(3): doi: / cncr Frosch DL, Kaplan RM. Shared decision making in clinical medicine: past research and future directions. Am J Prev Med. 1999;17(4): Available at: nlm.nih.gov/pubmed/ Accessed April 22, Bos V, Kunst A, Mackenbach J. De omvang van sociaal-economische verschillen gemeten op buurtniveau: vergelijkingen met schattingen op basis van informatie op individueel niveau. Sociaal-economische gezondheidsverschillen van verklaren naar verkleinen. 2001:

93 Factors associated with the use of gene expression profiles Chapter 4 93

94

95 CHAPTER 5 Using a gene-expression signature when controversy exists regarding the indication of adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. A. Kuijer, A.C.M. van Bommel, C.A. Drukker, M. van der Heiden-van der Loo, C.H. Smorenburg, P.J. Westenend, S.C. Linn, E.J.Th. Rutgers, S.G. Elias, T. van Dalen. Genetics in Medicine 2016:18;720-6.

96 Chapter 5 ABSTRACT Purpose: The Dutch national guideline advises use of gene-expression signatures, such as the 70-gene signature (70-GS), in case of ambivalence regarding the benefit of adjuvant chemotherapy (CT). In this nationwide study, the impact of 70-GS use on the administration of CT in early breast cancer patients with a dubious indication for CT is assessed. Methods: Patients within a national guideline directed indication area for 70-GS use who were surgically treated between November 2011 and April 2013 were selected from the Netherlands Cancer Registry database. The effect of 70-GS use on the administration of CT was evaluated in guideline- and age delineated subgroups addressing potential effect of bias by linear mixed-effect modeling and instrumental variable analyses. Results: 2043 patients within the indicated area for 70-GS use were included of which 298 received a 70-GS. 45% of patients in whom no 70-GS was used received CT versus 35% of patients who did receive the 70-GS. 70-GS use was associated with a 9.5% decrease in CT administration (95%CI: to -3.3%) in linear mixed-effect model analyses and instrumental variable analyses showed similar results (-9.9%, 95%CI to -0.4). Conclusions: In patients in whom the Dutch national guidelines suggest the use of a gene-expression profile 70-GS use is associated with a 10% decrease in the administration of adjuvant CT. 96

97 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy INTRODUCTION The mortality rates in breast cancer patients have decreased dramatically in the past two decades mainly due to the administration and improvement of adjuvant systemic therapy. 1,2 Clinical decision-making regarding administration of adjuvant systemic therapy in breast cancer patients is guided by conventional prognostic factors such as age, tumor size, tumor grade, status of axillary lymph nodes and hormone receptor status. Clinicopathological risk indexes, such as Adjuvant!Online (AOL) and the Nottingham Prognostic Index (NPI), use these factors to estimate the risk of recurrence and the benefit of adjuvant systemic treatment. 3,4 Nevertheless, breast cancer tumors with comparable clinicopathological characteristics may have considerable different outcomes, reflecting the heterogeneity of the disease. 5 Several gene expression profiles were developed to predict the risk of dissemination in breast cancer patients, and these gene expression profiles have drawn attention as an accurate alternative or adjunct for predicting outcome in individual breast cancer patients. The European Society of Medical Oncology (ESMO) suggests the use of gene expression profiles to gain additional prognostic and/or predictive information to complement pathology assessment particularly in patients with estrogen-receptor (ER) positive (+) breast cancer. 6 The current Dutch national (NABON) guideline for breast cancer (2012) recommends the use of a validated gene expression profile in patients with an invasive ductal carcinoma with ER positive disease and a questionable indication for adjuvant chemotherapy (CT) based on conventional prognostic factors. 7 Chapter 5 To date, research of gene expression profiles mainly focuses on the value of the test for the individual patient. The 70-gene signature (70-GS; MammaPrint ) is the most commonly used gene expression profile in the Netherlands. 8 The aim of the present study is to evaluate the impact of the 70-GS on the administration of adjuvant CT at a nation-wide level in a subgroup of patients in whom the additional value of CT is debated and national guidelines suggest the use of a gene expression profile. 97

98 Chapter 5 PATIENTS AND METHODS Data on patient, tumor- and treatment characteristics were obtained from the Netherlands Cancer Registry (NCR). The NCR is a nationwide database, managed by the Netherlands Comprehensive Cancer Organization (IKNL), which prospectively registers clinicopathological and treatment characteristics of all cancer patients treated in the Netherlands since The NCR started registering the use of gene expression profiles in Between February 2007 July 2011 the 70-GS became available in the Netherlands and was offered to patients enrolled in the Microarray in Node-Negative Disease May Avoid Chemotherapy (MINDACT) trial. 9 In this trial patients were randomized between administration of adjuvant CT based on the gene signature or conventional prognostic clinicopathological factors. Since accrual of the MINDACT trial ended, the 70-GS is increasingly used as an adjunct to conventional clinicopathological factors for clinical decision-making. It is the most widely used gene-expression profile in the Netherlands. Although OncotypeDX recently has become commercially available in the Netherlands, this gene expression profile was not frequently used during the study period, and its use was therefore not taken into account. Study population Female patients with primary breast cancer, over 17 years of age, surgically treated between November 2011 and October 2013, with no prior history of malignancy, neoadjuvant treatment or distant metastasis upon diagnosis were identified in the NCR. According to the Dutch national guidelines, adjuvant CT should be administered to all lymph node positive patients ( N1a) and to patients without lymph node involvement but with unfavorable clinicopathological tumor features (grade III tumors >1cm, any tumor > 2 cm or, HER2 + tumors) as well as to patients <35 years. The current Dutch guideline advices against the administration of CT in patients who do not fulfill the aforementioned criteria, in patients >70 years and in patients with grade I tumors <2cm. According to the Dutch guidelines validated gene expression profiles may be used in individual cases with a hormone receptor sensitive invasive ductal carcinoma, if there is doubt about the indication for adjuvant chemotherapy on the basis of traditional prognostic factors. 7 In line with this guideline we identified three groups of patients, all < 70 years with ER + and HER2 invasive ductal carcinoma, where controversy exists regarding the administration of adjuvant CT: N0, grade I, >2 cm (group A); N0, grade II, >1 cm (group B) and N1mi, grade I/II (group C). 98

99 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy Statistical analysis Frequencies of baseline characteristics were compared between patients who received a 70-GS versus patients who did not receive the test within the indicated area for 70- GS use, i.e. in whom controversy existed regarding the benefit of adjuvant CT, using a chi square test for differences in categorical data. For normally distributed continuous variables (age and size) means were calculated and a t-test was performed. Concordance of the test result with the administration of CT in the overall study population and the aforementioned subgroups was assessed by dividing the sum of patients with a low risk test result in whom adjuvant CT was omitted and patients with a high risk test result who received adjuvant CT by all patients with a known test result. For the whole group and the three categories of patients the proportions receiving CT were assessed in relation to 70-GS use and adherence to the low- or high-risk test result was evaluated. To provide the most valid results using observational data taking patient clustering in hospitals into account and to address potential confounding by indication, linear mixed-effect regression models and instrumental variable analyses were used. Chapter 5 Linear mixed-effects regression analysis The association between 70-GS use and the administration of adjuvant CT was assessed in a linear mixed model adjusting for possible observed confounders, taking into account patient clustering within hospitals. For this, we included both a random intercept per hospital (thus taking baseline differences in CT use among hospitals into account), and a random 70-GS slope per hospital (thus taking potential differences in het effect of the 70-GS on CT administration between hospitals into account). We adjusted for age and tumor size (continuously), and for grade, axillary status and incidence year (categorically). Linearity of the relation between the continuous variables and CT use was inspected using a LOWESS smoother, and was concluded to be linear for both age and size (the latter after truncation of 0.5% of the data above 5 cm). As the dependent variable (CT) was coded as 0 (no) and 1 (yes), the results of the linear mixed-effects analyses are on the risk difference scale, i.e. showing absolute differences in proportion adjuvant CT use. The proportions of absolute differences were multiplied by a hundred in order to present the results of these linear mixed-effects models as percentages in absolute risk differences. Besides investigating the association between 70-GS use on the administration of adjuvant CT in the overall study population, we assessed the effect of 70-GS within the aforementioned subgroups of patients (A,B and C) and age categories (< 50 years, years and years). To test for differences in the association between 70-GS and CT use between these subgroups, we calculated P-values for interaction by including interaction terms to the models. 99

100 Chapter 5 Instrumental variable (IV) analyses In an attempt to optimally control confounding by indication and further assess the validity of the linear mixed-effect models, we performed instrumental variable (IV) analyses (two-stage least square [2SLS] using ordinary linear regression). Confounding by indication is a well-known phenomenon in studies with observational data, which may only be partly resolved by multivariable regression analyses, such as mentioned above, since these methods cannot adjust for unmeasured confounders. An IV may serve as a substitute for randomization (pseudo-randomization) in non-randomized studies under the assumption that the IV is: (1) strongly associated with the exposure (in our case 70-GS use), (2) unrelated to confounders and (3) has no direct association with outcome (the administration of adjuvant CT). 10 We considered percentage of 70-GS use within the indicated area per hospital per year (IV 1), 70-GS use in the previous patient within the indicated area treated in the same hospital (IV 2) and a combination of both in the first stage of 2SLS analyses as IV s. All are measures for hospital 70-GS preference. First, the association of these IV s with 70-GS use, the administration of adjuvant CT or possible confounders was assessed by univariable logistic regression analysis. We considered using incidence year as an IV, however incidence year was strongly related with CT use, also in patients treated in hospitals were the 70GS was never used in our study period, and therefore failed the third assumption. We present the results of the IV analyses with and without adjustment for the same potential confounders as in the abovementioned linear mixed-effects models (these were then included in both stages of the 2SLS approach). The results of the 2SLS analysis can be interpreted as the absolute change in percent CT use due to 70GS use. All P-values and 95% confidence intervals (CI s) for the linear mixed-effects and the 2SLS analyses were based on 5000-fold bootstrap resampling. A p-value < 0.05 was considered statistically significant. All analyses were performed in R (version for mac OS), using the lme4 package (version 1.1-7) for the linear mixed-effects models. RESULTS Between November 2011 and October 2013, 2043 women with primary breast cancer, without prior malignancy or having received neo-adjuvant treatment, within the indicated area for 70-GS use were surgically treated in the Netherlands. Of these patients, 298 (15%) actually received a 70-GS. Patients who received the 70-GS were younger, had more limited axillary involvement and suffered from smaller tumors compared to patients within the indicated area who did not receive the 70-GS (table 1). Furthermore, an increase in 70-GS use over time was observed. 100

101 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy Table 1. Patient and tumor characteristics according to 70-gene signature use in 2043 patients within the indicated area for 70-GS use (all < 70 years of age with ER+/HER2-, invasive ductal carcinoma). 70-GS used n 70-GS not used = 298 n n = 1745 (%) n (%) p-value Patient characteristics Age at diagnosis in years, mean (SD)* 55.6 (8.3) 56.2 (8.8) 0.24 < (26%) 415 (24%) (74%) 1330 (76%) 0.41 Incidence year (6%) 227 (13%) (63%) 1252 (72%) (31%) 266 (15%) < Tumor characteristics Chapter 5 Pathological axillary status (pn) a pn0 (i-/i+) 238 (80%) 1346 (77%) pn1mi 60 (20%) 399 (23%) 0.33 Pathological tumor size (cm) Mean (SD)* 1.7 (0.6) 1.8 (0.8) (77%) 1214 (70%) > 2 68 (23%) 531 (30%) < 0.01 Invasive tumor grade Grade I 41 (14%) 253 (14%) Grade II 257 (86%) 1492 (86%) 0.80 * T-test; other data represent chi-square test values a pn0(i-/i+): no axillary lymph node involvement or isolated tumor cells; pn1mi; micro-metastases. Adherence to test result The majority of patients was assigned to a low-risk test result (64%) and high adherence rates to the 70-GS test result in the administration of adjuvant CT were observed in the overall study-population (86%). In the predefined guideline delineated subgroups A, B and C, the majority of the patients were assigned to the low-risk group by the 70-GS 101

102 Chapter 5 (83%, 63% and 63%, respectively), and the administration of CT was in line with the 70-GS result in 89%, 86% and 83% of the patients in subgroup A, B and C respectively (table 2). The 70-GS did not affect the administration of adjuvant endocrine therapy within these subgroups. Table 2. Administration of adjuvant CT and ET in patients within the indicated area for 70-GS use who received the 70-gene signature, i.e. in whom controversy exists regarding the administration of adjuvant CT based on conventional prognostic factors. N CT (%) ET (%) All patients 70-gene signature % 90% (Group A, B and C combined) Low risk % 88% High risk 95 80% 91% Risk unknown 34 24% 97% Adherence to test result 86% Group A 70-gene signature 19 21% 100% (N0, grade I, > 2 cm) Low risk 15 7% 100% High risk 3 67% 100% Risk unknown 1 100% 100% Adherence to test result 89% Group B 70-gene signature % 89% (N0, grade II, >1 cm) Low risk 120 8% 88% High risk 72 76% 88% Risk unknown 28 36% 96% Adherence to test result 86% Group C 70-gene signature 59 39% 88% (pn1mi, grade I/II) Low risk 34 18% 82% High risk 20 85% 100% Risk unknown 5 0% 80% Adherence to test result 83% CT = adjuvant chemotherapy; ET = adjuvant endocrine therapy. Adherence to risk profile = proportion of patients with a low risk test result and omission of CT or a high risk test result and administration of CT, of all patients with known 70GS test results. 102

103 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy Table 3. The effect of 70-GS use on the administration of adjuvant CT as observed with linear mixed-effect modeling, in the overall study population (n = 2043), guideline and age delineated subgroups of patients within the indicated area for 70-GS use, with and without adjustment for potential confounders. Unadjusted Fully adjusted Absolute % change in CT (95% CI) p-value Absolute % change in CT (95% CI) p-value Overall (-16.9 to -4.1) (-15.7 to -3.3) Guideline delineated subgroups Subgroup A (N0, grade I, > 2 cm) (-48.4 to - 7.0) (-46.5 to -1.6) Subgroup B (N0, grade II, > 1 cm) -7.8 (-15.2 to -0.4) (-13.0 to 1.3) 0.11 Chapter 5 Subgroup C (N1mi, grade I/II) (-26.2 to 0.2) (-28. to -4.0) Age categories (years) < (-47.0 to -22.9) < (-38.3 to -15.6) < (-23.4 to -3.4) (-18.7 to 0.3) (-6.1 to 13.6) (-5.9 to 12.9) 0.48 P-values for interaction between 70-GS use and subgroups; subgroup 70-GS*group B p 0.122; 70-GS*group C p P-values for interaction between 70-GS use and age categories; 70-GS*50-69 years of age p 0.017; 70-GS*60-69 years of age p <0.001.* Age (per year), incidence year (2011, 2012, or 2013), size (per mm), axillary involvement (N0i-/i+ vs. N1mi) and invasive tumor grade (grade I vs. II) were included in the fully adjusted model as covariables. 103

104 Chapter 5 Subgroup analyses revealed a significant reduction in the administration of adjuvant CT after 70-GS use in subgroup A and C (-24.0% 95% CI to -1.6, p and -16.0% 95%CI to -4.0, p 0.009, respectively) but no significant reduction in subgroup B (-5.8% 95%CI-13.0 to 1.3, p 0.11). Although the association between 70-GS use and the administration of adjuvant CT is lower in subgroup B compared to subgroup A, this difference in association was not significant (p for interaction 70GS*group B: 0.122), see figure 1. The strongest relation between 70-GS use and the administration of adjuvant CT was seen in younger patients. In patients < 50 years of age a 26.9% reduction in the administration of CT was observed in patients who received the 70-GS (95% CI to -15.6, p < 0.001). In patients between years of age, 70-GS use led to a 9.2% reduction in the administration of adjuvant CT however this was not significant (95%CI to 0.3, p 0.058, table 3). In the older age category a reverse relation is observed, 70-GS use resulted in a 3.5% (95%CI -5.9 to 12.9, p 0.48) increase in the administration of adjuvant CT (p for interaction 70GS* year of age: <0.001) (see supplementary figure 1). Instrumental variable analysis To assess the validity of the two IV s as an instrument in this study we confirmed that both percentage of 70-GS use within the indicated area per hospital per year (IV1) as use of the 70-GS in the previous patient within the indicated area (IV2) had a strong positive association with 70-GS use. The 70-GS was used in 5.6% of the patients within the indicated area in hospitals who less frequently used the 70-GS (0-25% 70-GS use within the indicated area per year), which was 47.9% for patients treated in hospitals who used the 70-GS in >25% of the patients within the indicated area (OR 14). If the 70-GS was used in the previous patient within the indicated area, 33.6% of the subsequently treated patients also received a 70-GS compared to 12.2% if the prior patient did not receive a 70-GS (OR 18). 104

105 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy Figure 1. Absolute risk of receiving adjuvant CT according to the (fully adjusted) linear mixed model in guideline delineated subgroups of patients within the indicated area for 70-GS use. All patients < 70 years of age with ER+/Her2- invasive ductal carcinoma (subgroup A: N0, BRI, > 2 cm; subgroup B: N0, BRII, > 1 cm; subgroup C: N1mi, BRI/II). Chapter 5 CT = chemotherapy; 70 GS = 70-gene signature. * Calculated using the fully adjusted linear mixed model; age, tumor size, invasive tumor grade, axillary involvement and incidence year were included as covariables in the model. Projected values are the percent chemotherapy use for those without and with the 70-GS as derived from the linear mixed models regression equations filled-in at the mean of all other covariables for the entire (sub-)group (i.e. the projected difference between 70-GS use and non-use is adjusted for confounding). ** P-values for interaction between 70-GS use and the subgroups. Combining IV1 and IV2 in the first stage of the two-stage least square (2SLS) regression method showed the strongest association with 70-GS use (5.2% vs. 45.3%, OR 22, comparing the best scoring 25% of patients for this IV versus the rest). Both IV1 and IV2 were generally less associated with potential confounders (age at diagnosis and tumor characteristics) than actual 70GS use. Incidence year was however not equally distributed between levels of the three IVs (supplementary table 2). The results of the IV analyses are shown in table 4, with and without adjustment for potential confounders. The analyses of both IV1 and IV2 revealed a significant reduction in the administration of adjuvant CT in patients who received a 70-GS (-27.6%, 95%CI: to -0.6%, p and -10.2% 95%CI to -0.8, p 0.033, for IV 1 105

106 Chapter 5 and 2 respectively). The results of the instrumental variable analysis with the combined IV s were similar to the results of the adjusted linear mixed-effects regression model (-9.9% 95%CI to -0.4, p 0.040). Table 4. The effect of 70-GS use on the administration of adjuvant CT, in 2043 patients within the indicated area for 70-GS use, as derived from different instrumental variable (IV) analyses (two-stage least square regression analysis). 70-GS use in the previous patient within the indicated area(iv 1), percentage of 70-GS use within the indicated area per hospital per year (IV 2) and a combination of both were used as IV s. Unadjusted Absolute % change in CT (95% CI) p-value Fully adjusted Absolute % change in CT (95% CI) p-value IV (-75.0 to ) (-54.5 to -0.6) IV (-31.2 to -10.0) < (-19.6 to -0.8) IV 1 and IV 2 combined (-29.7 to -8.3) < (-19.3 to -0.4) CT = adjuvant chemotherapy; 95%CI = 95% Confidence Interval; 70-GS = 70-gene signature. *Age (per year), incidence year (2011, 2012, or 2013), size (per mm), axillary involvement (N0i-/i+ vs. N1mi) and invasive tumor grade (grade I vs. II) were included in the fully adjusted model as covariables. DISCUSSION In this nation-wide study patients in whom the 70-GS was used received 10% less adjuvant CT compared to patients who did not receive the 70-GS in a cohort of Dutch breast cancer patients in whom controversy exists regarding the benefit of CT based on clinicopathological characteristics. Furthermore, in this selection of patients compliance with the test result was high. We observed a significant and clinically relevant decrease in the administration of adjuvant CT after use of the 70-gene signature in patients with an uncertain indication for adjuvant CT based on conventional prognostic factors after correction of all measured confounders. Until date, there is only limited evidence concerning the impact of the 70-gene signature on the administration of CT. Various studies have reported the impact of gene expression profiles within cohorts of patients that received the test. 11,12 A theoretical change in adjuvant systemic treatment decisions was reported in approximately 30% of patients after use of the 70-gene signature, resulting in a 106

107 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy more tailored adjuvant systemic treatment plan for the individual patient. 13 Current evidence is based on a few impact studies with relatively small numbers of patients, predominantly describing a hypothesized decrease in the administration of ACT after 70-GS use in early breast cancer patients. 12,13 Similar results were reported regarding the impact of the 21-gene Recurrence Score (OncotypeDx) on the administration of adjuvant CT In addition, high compliance rates to the test result in patients with early stage ER+, node negative breast cancer have been reported. 17 Yet, to our knowledge there are no nation-wide studies assessing the impact on the administration of adjuvant CT for either OncotypeDx or the 70-gene signature. Recently, Cusumano et al. conducted a European inter-institutional impact study of the 70-gene signature (n= 194). Cases were presented to clinician panels in four different countries asking for an adjuvant treatment advice, first without and then with knowledge of the 70-gene test result. In the subset of ER + and HER2Neu patients (n = 100), a 21% (73% vs. 52%) expected absolute decrease in the administration of adjuvant CT was reported when the 70-gene signature result was taken into account by a Dutch multidisciplinary team. 18 Our results are in line with this report, although we see a smaller decrease in the administration of adjuvant CT when the 70-gene signature was used. This difference may be explained by differences in the design of the present study in which patients were categorized by a combination of hormone status, HER2- status, tumor size, differentiation grade and lymph node involvement. In addition, the smaller proportion of patients receiving adjuvant CT in the control group in our study (44% vs 73%) illustrates the different national attitudes towards the administration of adjuvant systemic in the absence of a gene signature test. In fact, the relative reduction of administered adjuvant CT was comparable between our population-based and Cusumano s questionnaire study. Chapter 5 Approximately 45% of patients in whom controversy exists regarding the administration of adjuvant CT received adjuvant CT without use of the 70-gene signature. This reflects the current controversy regarding the administration of adjuvant CT in these patients. High compliance rates to the 70-gene signature test result were seen demonstrating the propensity to adhere to a reproducible advice in this group of patients. The 70- gene signature assigned the majority of these patients to the low risk category (83% in group A, 63% in group B and 63% in group C). In particular in group B, consisting of patients with grade II breast cancer, the proportion of low risk test results was higher than reported in previous studies assigning approximately 50% to the high-risk category and 50% to the low-risk category. 19 Especially in subgroup A (N0, BRI, >2cm) and C (N1mi, BRI/II) 70-GS use was associated with less administered CT, whereas in subgroup B (N0,BRII,>1cm) only a non-significant trend was observed. Nevertheless, the p-values for interaction between the subgroups indicate that the effect of 70-GS 107

108 Chapter 5 use on the administration of CT does not significantly differs between the various subgroups. Presumably, also considering the prevalence of genomic high-risk patients, the current sample size of subgroup B is to small to reach significance. This is the first nation-wide study to report the independent association between the use of the 70-gene signature and the administration of adjuvant CT in patients in whom controversy exists according to the national guidelines, i.e. patients with ER-positive, Her2-negative invasive ductal tumors of low or intermediate malignancy grade in the absence of overt lymph node metastases. This was supported by linear mixed-effect modelling which has as main advantage that no independence is assumed amongst observations, allowing correlated observations within a unit or cluster. Since hospital- or even clinicians preference plays an important role in 70-GS use and the administration of adjuvant CT, patient clustering might have influenced the results. Therefore, this statistical approach is ideally suited for assessment of the association between 70-GS use and the administration of adjuvant CT and makes the observed association more valid. In this study, other factors also influenced the decision to administer CT in the subset of patients where CT was considered controversial. In particular the effect of age was remarkable. The high proportion of young women who received adjuvant CT in the controversy group reflects an understandably more aggressive attitude in younger women. Nonetheless, while the overwhelming majority of younger women received adjuvant CT when a 70-GS was not used (83%, supplementary figure 2), 53% of these young women received adjuvant CT when the 70-GS was used, resulting in the largest relative and absolute reduction of administered CT. In contrast, in the older age category (60-69 years of age) 70-GS use was associated with an increased administration of adjuvant CT (supplementary figure 2). Although this effect in the older age category is not statistically significant in the linear mixed-effect analyses, the p-value for interaction (p <0.001) implies that the effect of 70-GS use on the administration of CT in older patients is different from those in patients < 50 years of age. These findings might indicate that the physician s intention for 70-GS use is associated with age but the retrospective observational design of the current study precludes detailed analyses of these findings. Confounding by indication is a well-known phenomenon in studies using observational data. In an attempt to control this form of bias we performed additional IV analyses. 10 An IV is an external factor which influences outcome (in our case the administration of CT) exclusively through its effect on exposure (70-GS use), unrelated to potential confounders. We decided to use facility-prescribing patterns as IVs, an approach adopted previously by others. 20 In our case a high-quality IV should be strongly related to 70-GS use, which was particular the case when combining both IVs. Furthermore, a robust IV should not be associated with other confounders of the administration of CT. Since confounders can be measured or unmeasured, it is impossible to assess the direct relation 108

109 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy between an IV and all possible confounders. Therefore, most authors using IVs report on the relation between IVs and measured confounders (for example patient characteristics) under the assumption that if the IV is not related to a measured confounder it likely is also not related to unmeasured confounders. In our case, all IVs (IV1: 70-GS use in the previous patient within the indicated area and IV 2: proportion of 70-GS use per hospital within the indicated area) and a combination of both, resulted in a better balance for measured confounders compared to actual 70-GS use (supplementary table 2) but a disbalance remained for the variable incidence year. Therefore, we also adjusted the IV effects for the measured potential confounders, and the combined IV analysis resulted in a strikingly similar absolute risk reduction for the administration of CT after 70-GS as observed in the standard linear mixed-effect regression analyses. The prognostic value of the 70-GS has been validated in multiple retrospective and one prospective patient series. 8,13,21 24 Evidence supporting the predictive value of the 70-GS on chemotherapy benefit is limited and therefore it s clinical utility mainly lies in it s prognostic capacity Currently the MINDACT trial is being conducted, of which the first results are expected by the end of this calendar year, which further assesses the predictive capacity of the 70-GS in a randomized prospective setting. 9 In the present study, i.e. in patients who were treated within the indicated area for GEPs characterized as having a fairly good prognosis, it was the prognostic value of the 70-GS that was used to discern a group of patients with such good outcome that no substantial benefit was to be expected of adjuvant CT. The observed effect in the present study is not attributable to a hitherto unproven predictive value of the 70-GS. In conclusion, In Dutch early stage breast cancer patients with a dubious indication for adjuvant CT based on clinicopathological factors 70-GS use is associated with a 10% decrease in the administration of adjuvant CT. Chapter 5 ACKNOWLEDGEMENTS This work was supported by the Dutch Cancer Society (KWF). 109

110 Chapter 5 Supplementary Table 1. Unadjusted and fully adjusted results of the linear mixed-effect modelling using random intercept per hospital and a random 70-GS slope per hospital. Unadjusted Fully adjusted Absolute % change in CT (95% CI) p-value Absolute % change in CT (95% CI) p-value 70-gene signature use 70-GS not used reference reference 70-GS used (-16.9 to -4.1) (-15.7 to -3.3) Patientcharacteristics Age per year -2.6 (-2.8 to -2.4) < (-2.7 to -2.3) < Incidence year 2011 reference reference (-12.6 to 0.4) (-9.2 to 1.7) (-22.6 to -7.0) < (-16.3 to -3.2) Tumorcharacteristics Size per mm 1.9 (1.6 to 2.2) < (1.7 to 2.2) < Pathological axillary status (pn) pn0(i-/i+) reference reference pn1mi 10.2 (5.1 to 15.3) < (11.3 to 20.4) < Invasive tumor grade Grade I reference reference Grade II 12.6 (6.9 to 18.3) < (20.1 to 30.7) < CT = adjuvant chemotherapy; 95%CI = 95% Confidence Interval; 70-GS = 70-gene signature 110

111 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy Supplementary Table 2. Odds ratio s for a positive instrumental variable (or actual 70-gene signature use) per observed confounder. Patient characteristics Actual 70-GS use IV1 Previous patient received 70GS IV2 Percent 70GS use per hospital per year Yes No Yes No >25% 0-25% top 25% (n = 298) (n = 1745) (n = 277) (n = 1662) (n = 432) (n = 1611) (n = 483) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) IV1 and 2 combined by linear regression bottom 75% (n = 1454) Age (years) < 50 reference reference reference reference ( ) 0.99 ( ) 1.00 ( ) 0.97 ( ) Incidence year < 2013 reference reference reference reference ( ) 2.27 ( ) 2.46 ( ) 2.89 ( ) Tumor characteristics Axillary status (pn) pn0(i-/i+) reference reference reference reference pn1mi 0.85 ( ) 0.96 ( ) 1.03 ( ) 1.06 ( ) Size (mm) 2 cm reference reference reference reference > 2 cm 0.68 ( ) 0.91 ( ) 0.91 ( ) 0.83 ( ) Invasive tumor grade Grade I reference reference reference reference Grade II 1.06 ( ) 0.87 ( ) 1.26 ( ) 1.13 ( ) IV = instrumental variable, 70-GS = 70-gene signature, OR = Odds Ratio, 95%CI = 95% Confidence Interval Chapter 5 111

112 Chapter 5 Supplementary Figure 1. Absolute risk of receiving adjuvant CT according to the (fully adjusted) linear mixed model in age delineated subgroups of patients within the indicated area for 70-GS use. All patients < 70 years of age with ER+/Her2- invasive ductal carcinoma. CT = chemotherapy; 70 GS = 70-gene signature * Calculated using the fully adjusted linear mixed model; age, tumor size, invasive tumor grade, axillary involvement and incidence year were included as covariables in the model. Projected values are the percent chemotherapy use for those without and with the 70-GS as derived from the linear mixed models regression equations filled-in at the mean of all other covariables for the entire (sub-)group (i.e. the projected difference between 70-GS use and non-use is adjusted for confounding). ** P-values for interaction between 70-GS use and the subgroups. 112

113 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy REFERENCE LIST 1. Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05) Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Early Breast Cancer Trialists Collaborative Group. Lancet (London, England). 1992;339(8784):1 15. Available at: Accessed April 23, Ravdin PM, Siminoff LA, Davis GJ, et al. Computer program to assist in making decisions about adjuvant therapy for women with early breast cancer. J Clin Oncol. 2001;19(4): doi: /jco Galea MH, Blamey RW, Elston CE, Ellis IO. The Nottingham Prognostic Index in primary breast cancer. Breast Cancer Res Treat. 1992;22(3): Available at: Accessed April 23, Mook S, Van t Veer LJ, Rutgers EJT, Piccart-Gebhart MJ, Cardoso F. Individualization of therapy using Mammaprint: from development to the MINDACT Trial. Cancer Genomics Proteomics. 4(3): Available at: Accessed April 19, Senkus E, Kyriakides S, Penault-Llorca F, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2013;24(suppl 6):vi7 vi23. doi: /annonc/mdt Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: /nej- Moa Rutgers E, Piccart-Gebhart MJ, Bogaerts J, et al. The EORTC 10041/BIG MIN- DACT trial is feasible: Results of the pilot phase. Eur J Cancer. 2011;47(18): doi: /j.ejca Greenland S. An introduction to instrumental variables for epidemiologists. Int J Epidemiol. 2000;29(4): Available at: Accessed April 23, Torrisi R, Garcia-Etienne CA, Losurdo A, et al. Potential impact of the 70-gene signature in the choice of adjuvant systemic treatment for ER positive, HER2 negative tumors: a single institution experience. Breast. 2013;22(4): doi: /j.breast Chapter 5 113

114 Chapter Exner R, Bago-Horvath Z, Bartsch R, et al. The multigene signature MammaPrint impacts on multidisciplinary team decisions in ER+, HER2- early breast cancer. Br J Cancer. 2014;111(5): doi: /bjc Drukker CA, Bueno-de-Mesquita JM, Retèl VP, et al. A prospective evaluation of a breast cancer prognosis signature in the observational RASTER study. Int J cancer. 2013;133(4): doi: /ijc Albanell J, González A, Ruiz-Borrego M, et al. Prospective transgeicam study of the impact of the 21-gene Recurrence Score assay and traditional clinicopathological factors on adjuvant clinical decision making in women with estrogen receptor-positive (ER+) node-negative breast cancer. Ann Oncol Off J Eur Soc Med Oncol. 2012;23(3): doi: /annonc/mdr Ademuyiwa FO, Miller A, O Connor T, et al. The effects of oncotype DX recurrence scores on chemotherapy utilization in a multi-institutional breast cancer cohort. Breast Cancer Res Treat. 2011;126(3): doi: /s Partin JF, Mamounas EP. Impact of the 21-gene recurrence score assay compared with standard clinicopathologic guidelines in adjuvant therapy selection for node-negative, estrogen receptor-positive breast cancer. Ann Surg Oncol. 2011;18(12): doi: /s z. 17. McVeigh TP, Hughes LM, Miller N, et al. The impact of Oncotype DX testing on breast cancer management and chemotherapy prescribing patterns in a tertiary referral centre. Eur J Cancer. 2014;50(16): doi: /j.ejca Cusumano PG, Generali D, Ciruelos E, et al. European inter-institutional impact study of MammaPrint. Breast. 2014;23(4): doi: /j.breast Knauer M, Mook S, Rutgers EJT, et al. The predictive value of the 70-gene signature for adjuvant chemotherapy in early breast cancer. Breast Cancer Res Treat. 2010;120(3): doi: /s Chen Y, Briesacher BA. Use of instrumental variable in prescription drug research with observational data: a systematic review. J Clin Epidemiol. 2011;64(6): doi: /j.jclinepi Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective community-based feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Buyse M, Loi S, van t Veer L, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst. 2006;98(17): doi: /jnci/djj Wittner BS, Sgroi DC, Ryan PD, et al. Analysis of the MammaPrint breast cancer assay in a predominantly postmenopausal cohort. Clin Cancer Res. 2008;14(10): doi: / ccr

115 Using a gene-expression signature reduces the proporttion of patients receiving chemotherapy 24. Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol Off J Eur Soc Med Oncol. 2010;21(4): doi: /annonc/mdp Straver ME, Glas AM, Hannemann J, et al. The 70-gene signature as a response predictor for neoadjuvant chemotherapy in breast cancer. Breast Cancer Res Treat. 2010;119(3): doi: /s Esserman LJ, Berry DA, Cheang MCU, et al. Chemotherapy response and recurrence-free survival in neoadjuvant breast cancer depends on biomarker profiles: results from the I-SPY 1 TRIAL (CALGB /150012; ACRIN 6657). Breast Cancer Res Treat. 2012;132(3): doi: /s Whitworth P, Stork-Sloots L, de Snoo FA, et al. Chemosensitivity Predicted by Blue- Print 80-Gene Functional Subtype and MammaPrint in the Prospective Neoadjuvant Breast Registry Symphony Trial (NBRST). Ann Surg Oncol. 2014;21(10): doi: /s y. Chapter 5 115

116

117 CHAPTER 6 Changes over time in the impact of gene-expression profiles on the administration of adjuvant chemotherapy in estrogen receptor positive early stage breast cancer patients: a nationwide study. A. Kuijer, C.A. Drukker, S.G. Elias, C.H. Smorenburg, E.J.Th. Rutgers, S. Siesling, T. van Dalen. International Journal of Cancer 2016;139:

118 Chapter 6 ABSTRACT Ten years ago gene-expression profiles were introduced to aid adjuvant chemotherapy decision-making in breast cancer. Since then subsequent national guidelines gradually expanded the indication area for adjuvant chemotherapy. In this nation-wide study the evolution of the proportion of patients with estrogen-receptor positive (ER+) tumors receiving adjuvant chemotherapy in relation to gene-expression profile use in patient groups that became newly eligible for chemotherapy according to national guideline changes over time is assessed. Data on all surgically treated early breast cancer patients diagnosed between and were obtained from the Netherlands Cancer Registry. ER+/Her2- patients with tumor-characteristics making them eligible for geneexpression testing in both cohorts and a discordant chemotherapy recommendation over time (2004 guideline not recommending and 2012 guideline recommending chemotherapy) were identified. We identified 3864 patients eligible for gene-expression profile use during both periods. Gene-expression profiles were deployed in 5% and 35% of the patients in the respective periods. In both periods the majority of patients was assigned to a low genomic risk-profile (67% and 69% respectively) and high adherence rates to the test result were observed (86% and 91% respectively). Without deploying a gene-expression profile 8% and 52% (p <0.001) of the respective cohorts received chemotherapy while 21% and 28% of these patients received chemotherapy when a gene-expression profile was used (p 0.177). In conclusion, in ER+/Her2- early stage breast cancer patients gene-expression profile use was associated with a consistent proportion of patients receiving chemotherapy despite an adjusted guideline-based recommendation to administer chemotherapy. 118

119 Change over time in the impact of gene-expression profiles on chemotherapy administration INTRODUCTION Worldwide, the administration of adjuvant systemic therapy is associated with improved breast cancer survival. 1,2 While systemic treatment was previously reserved for relatively young patients or those with metastatic lymph-node involvement, indication areas have broadened in the last decade. Guidelines, such as the Dutch NABON guideline, now recommend adjuvant systemic therapy in the majority of ER+ lymph-node negative patients 3 5, albeit that considerable controversy exists regarding the benefit of adjuvant chemotherapy in selections of patients. 2,6,7 Concomitantly, gene-expression profiles were developed for better outcome prediction in early stage breast cancer patients, such as the 21-gene-recurrence score (Oncotype DX, Genomic Health Inc. Redwood city, CA) and the 70-gene signature (MammaPrint TM, Agendia Inc., Amsterdam). The 70-gene signature (70-GS) 8 first became eligible for patients enrolled in the RASTER study 9 in the Netherlands in 2004, and its prognostic value has been validated in several retrospective and a single prospective patient-series The 21-gene recurrence score (21-GS) is clinically validated and assesses the risk of distant recurrence based on the expression of 21 genes. 7 Since 2013 the 21-RS is also used in Dutch daily practice. Chapter 6 Current international guidelines advise the use of gene-expression profiles to predict outcome in early stage ER+ breast cancer patients. 3,14,15 The Dutch guideline advocates the use of a gene-expression profile in ER+/Her2- patients in whom benefit of adjuvant chemotherapy is questionable based on clinic-pathological factors alone. This target population for gene-expression profile use coincides with patient categories that have become eligible for chemotherapy during the last decade as guidelines broadened the indication area, i.e. patients with ER+ breast cancers and no or limited metastatic lymph-node involvement. The aim of the present study is to assess the time-dependent changes in the proportion of ER+/Her2- patients receiving adjuvant chemotherapy in relation to gene-expression profile use in patient groups that became newly eligible for chemotherapy according to national guideline changes over time. PATIENTS AND METHODS To evaluate the effect of clinical guideline changes in relation to gene-expression profile use on the administration of chemotherapy, females surgically treated for primary, unilateral breast cancer between (cohort I) and between (cohort II) were identified in the Netherlands Cancer Registry (NCR). The NCR contains prospectively registered data on patient-, tumour- and treatment characteristics 119

120 Chapter 6 of all cancer patients treated in the Netherlands. 16 Patients with a prior history of malignancy, multifocal disease, distant metastasis or neo-adjuvant systemic treatment (1849 and 4472 patients in the two cohorts, respectively) were excluded. The effect of guideline changes on the administration of chemotherapy was addressed in the selection of ER+/Her2- patients eligible for gene-expression profile use in both Figure 1. Flowchart of the various subgroups of patients analysed. The administration of adjuvant systemic treatment was assessed in the three subgroups categorized by the indication for adjuvant chemotherapy (CT) according to the 2004 and 2012 clinical guideline recommendation. *The impact of gene-expression profiles on adjuvant CT administration was assessed within these subgroups of ER+/Her2- patients with a discordant guideline recommendation between the two periods (N0 or N1mi and grade I disease >2cm or grade II disease 1-2 cm) who were elegible for receiving a GEP during both periods (<61 years of age). periods. For that purpose, patients were categorized into three groups: guidelines advocating chemotherapy in both cohorts, guidelines not recommending chemotherapy in both cohorts and patients eligible for gene-expression profile use during both periods (ER+/Her2- patients < 61 years of age in which guidelines not recommended chemotherapy in cohort I and recommend chemotherapy in cohort II, figure 1). Within the subgroup of ER+/Her2- patients considered eligible to receive chemotherapy based on guideline adjustment the effect of gene-expression profile use on the administration of chemotherapy and endocrine therapy was assessed in the two cohorts. In both cohorts gene-expression profiles were part of the decision-making regarding the eventual administration of chemotherapy. Since gene-expression profiles were only deployed in patients enrolled in the RASTER study in the first time period, a 61 years of age upper limit was chosen for both cohorts concordant with the inclusion criteria of this study. 120

121 Change over time in the impact of gene-expression profiles on chemotherapy administration Guidelines for adjuvant systemic therapy in the Netherlands in the two time periods In cohort I, the national guideline of 2004 was operative at time of treatment. 17 All patients with N1a disease 60 years were eligible for receiving adjuvant systemic treatment (chemotherapy and endocrine therapy in case of hormone receptor (HR) positive disease). Additional adjuvant systemic therapy recommendations for patients without or micro-metastatic lymph node involvement (N0 or N1mi) are listed in table 1. For patients in cohort II the guideline of 2012 was applicable at time of treatment. 3 The indicated area to administer chemotherapy was expanded in patients without overt lymph node involvement (N0/N1mi). The 2012 guideline recommendation for chemotherapy completely coincides with the endocrine therapy recommendations in patients with HR+ disease. The 70-GS was offered to patients enrolled in the prospective microarray-prognosticsin-breast-cancer (RASTER) study between 2004 and In this observational feasibility study 18 women < 61 years of age with primary breast cancer (ct1-4n0m0) were eligible to receive 70-GS testing and the medical oncologist was free to follow or neglect the genomic result when deciding on the administration of chemotherapy. Dutch health insurance companies reimburse gene-expression profiles since In addition, in 2013 the 21-RS became available in the Netherlands. Between February 2007 and July 2011 the 70-GS was offered to patients enrolled in the Microarray in Node-Negative Disease May Avoid Chemotherapy (MINDACT) trial, where adherence to the test result was dictated by randomization. Therefore this period was not evaluated. Chapter 6 121

122 Chapter 6 Table 1. Adjuvant systemic treatment recommendation in lymph-node negative patients (pn0) or patients with limited metastatic lymph-node involvement (pn1mi) with unfavourable tumor characteristics according to the Dutch clinical guideline of 2004 and Guideline of 2004 Guideline of 2012 Adjuvant systemic treatment indication <60 years of age (N0 or N1mi) 1 : <70 years of age (N0 or N1mi) 2 : Tumor size > 3 cm, any grade Tumor size 2 cm, any grade 3 Tumor size 2 cm, grade II Tumor size > 1 cm,, grade II 3 Tumor size >1 cm, grade III Patients 35 years are always eligible for AST, except for patients with grade I tumours 1 cm 1 In patients years of age with HRdisease adjuvant CT could be considered. Patients 60 were eligible for ET. Patients <35 years are always eligible for AST, except for patients with grade I tumours 1 cm 2 CT could also be considered in patients with N1mi disease and favorable tumor characteristics. Patients 70 were eligible for ET. 3 In case of Her2+ disease and a tumor 0.5 cm CT can be considered regardless of other tumor features Note. Adjuvant endocrine therapy was merely indicated in case of hormone receptor positive disease in both periods. HR = hormone receptor; CT= chemotherapy; ET= endocrine therapy; AST = adjuvant systemic treatment. 122

123 Change over time in the impact of gene-expression profiles on chemotherapy administration Statistical analyses Baseline characteristics for the selection of patients with a new indication for chemotherapy due to clinical guideline changes (i.e. eligible for gene-expression profile use during both periods) were compared using a Chi-square test for categorical variables and a student s t-test for normally distributed continuous data. The administration of chemotherapy was evaluated in relation to gene-expression profile use by a Chi-square test. Furthermore, adherence to the gene-expression profile result was calculated for both cohorts by the sum of patients who did receive chemotherapy in accordance with a highrisk profile and patients in whom chemotherapy was omitted concordant with a low-risk test result subsequently divided by the total number of patients who with a known 70- GS result. For the 21-RS only high- or low- risk results were taken into account, since the intermediate risk score is non-directive in whether or not chemotherapy should be administered. All analyses were performed by using STATA, version RESULTS Baseline characteristics A total of female patients surgically treated for primary, unilateral breast cancer between or were identified in the NCR. In 41% (n = 23305) of patients chemotherapy was recommended in both periods, in 43% (n = 24583) clinical guidelines in both periods advised to omit adjuvant chemotherapy and in 13% (n = 7209) of patients clinicopathological characteristics were reason to advise against the administration of chemotherapy in cohort I and to advocate the administration of chemotherapy in cohort II (patients years of age with no or limited lymph-node involvement and grade I tumours of 2-3 cm or grade II tumours 1-2 cm, figure 1). The remaining patients could not be categorized due to missing values. Chapter 6 Of patients who were considered candidates to receive chemotherapy in the second time period due to guideline adjustment, 3864 patients (1912 in cohort I and 1952 in cohort II) were eligible for receiving a gene-expression profile during both periods (ER+, Her2-, < 61 years of age). Patient-, tumor- and treatment characteristics of these patients are listed in table 2. Age at diagnosis was slightly higher and lobular carcinomas were more frequent in the second time period. Gene-expression profiles were used in 103 patients in the first cohort and in 680 in the second. Sixteen institutions deployed gene-expression profiles in the first period (17%). In the second period 79 institutions (86%) deployed at least one gene-expression of which 56 institutions (61% of all Dutch hospitals) frequently used gene-expression profiles ( 10 gene-expression profiles during ). Patients who received a gene-expression profile in cohort I were younger compared to patients who received a gene-expression profile in cohort II, but no significant differences in tumor-characteristics were observed (sup. Table 1). 123

124 Chapter 6 Administered systemic treatment in relation to the respective guidelines in the two time periods In patients newly eligible for chemotherapy based on guideline adjustments an increase in the administration of chemotherapy was seen: 9% in cohort I and 40% in cohort II received chemotherapy (p < 0.001, Figure 2). Concomitantly, 22% of the patients with an adjusted indication for endocrine therapy received endocrine therapy in cohort I and 86% in cohort II (p < 0.001, sup. Figure 1). An increase in the administration Table 2. Patient-, tumour- and treatment characteristics of ER+/Her2- breast cancer patients with changed indication for adjuvant chemotherapy between 2004 and 2012 eligible for receiving a gene-expression profile during both time periods. Cohort I n= 1912 n (%) Cohort II n = 1952 n(%) P-value Patientcharacteristics Age at diagnosis in years, mean (SD)* 50.6 (0.14) 51.6 (0.13) < (42%) 652 (33%) (58%) 1298 (67%) <0.001 Tumorcharacteristics Histology Ductal carcinoma 1546 (81%) 1529 (78%) Lobular carcinoma 223 (12%) 314 (16%) Mixed 104 (5%) 62 (3%) Other 39 (2%) 47 (2%) <0.001 Axillary status (pn) a pn (82%) 1565 (80%) pn1mi 347 (18%) 386 (20%) Tumor size (pt) b pt (88%) 1749 (90%) pt2 220 (12%) 186 (10%)

125 Change over time in the impact of gene-expression profiles on chemotherapy administration Invasive tumor grade Grade I 374 (20%) 340 (17%) Grade II 1538 (80%) 1612 (83%) Progesterone receptor status Negative 251 (14%) 231 (12%) Positive 1571 (86%) 1719 (88%) Her2 status Negative 1227 (64%) 1946 (100%) Dubious 63 (3%) 2 (0%) <0.001 Unknown 622 (33%) 4 (0%) Chapter 6 * Students t-test; other data represent Chi-square values. a Pathological lymph node involvement (pn) according to the TNM staging system b Pathological tumour size (pt) according to the TNM staging system. of chemotherapy was also seen in patients who had an indication for chemotherapy in both periods: 58% of patients in cohort I and 86% in cohort II received chemotherapy (p < 0.001), while the proportion of patients receiving endocrine therapy remained stable (84% and 86% in the respective cohorts, p 0.001). Gene-expression profile use and the administration of chemotherapy and endocrine therapy In patients eligible for gene-expression profile use in both cohorts a significant increase in the administration of chemotherapy over time was seen (9% in cohort I vs 40% in cohort II; p <0.001), but no significant difference in chemotherapy administration was observed in patients in whom a gene-expression profile was applied (21% vs 28% received chemotherapy in the respective cohorts, p 0.191, figure 3). While the use of a gene-expression profile was associated with a higher proportion of patients receiving endocrine therapy in cohort I (41% vs. 21% of patients in whom no gene-expression profile was used; p <0.001), the vast majority of patients received endocrine therapy irrespective of the use of a gene-expression profile in cohort II (85 vs. 83% respectively; p 0.662, suppl. Figure 2). 125

126 Chapter 6 Figure 2. The administration of adjuvant chemotherapy (CT) in the two cohorts for patients categorized according to national guideline recommendations of 2004 and Figure 3. Administration of adjuvant chemotherapy (CT) stratified for both cohorts in relation to geneexpression profile use (GEP) in estrogen-receptor positive patients with a new indication for the latter according to the guideline of 2012 (i.e. according to the 2004 guideline adjuvant CT was not indicated) and T1-2cN0M0 disease. 126

127 Change over time in the impact of gene-expression profiles on chemotherapy administration Adherence to the test result The majority of patients in whom a gene-expression profile was applied were assigned to the low-risk category in both cohorts (67% of patients in cohort I and 71% in cohort II were assigned to a low-risk profile and 55% of patients in cohort II were low-risk according to the 21-RS). The administration of chemotherapy was in line with the 70-GS result in 85% and 91% of the patients treated in cohort I and II, respectively. A similar adherence rate of 89% was seen in patients who received the 21-RS in cohort II (table 3). Adherence to the 70-GS was higher in patients assigned to the low-risk category compared to patients assigned to the high-risk category (withholding chemotherapy in 99% vs. 94% in case of a genomic lowrisk profile and chemotherapy administration in 62% vs. 83% in case of a genomic high-risk profile, for the respective periods). In the latter period there was no significant association between gene-expression profile use or the test result and the administration of endocrine therapy. Only patients assigned to the low-risk category in cohort I received significantly less endocrine therapy as compared to high-risk patients (suppl. Table 2). Table gene signature (70-GS) and 21-recurrence score (21-RS) test results and the administration of adjuvant chemotherapy stratified for the two time periods. Cohort I: Cohort II: n CT (%) n CT (%) All* % % Low risk % % High risk % % Chapter 6 Unknown % Adherence to test result 85.60% 90.80% 21-recurrence score All* n.a % Low risk 6 0% Intermediate 2 0% High risk 3 67% Adherence to test result 88.80% Adherence to test result = proportion of patients with a low risk result and omission of chemotherapy or a high risk result and administration of chemotherapy. * Patients with a new indication for adjuvant chemotherapy according to the national guideline of 2012 and ER+/Her2-disease, <61 years of age. 127

128 Chapter 6 DISCUSSION In the present nation-wide study into the use of a gene-expression profile in two different time periods in ER+/Her2- patients with clinicopathological characteristics that were reason to advice against chemotherapy in 2004 and administer chemotherapy in 2012, using a gene-expression profile was associated with high adherence to the test result in both periods and a consistent proportion of patients receiving chemotherapy over time. Indication areas for the administration of chemotherapy, as stated in national and international clinical treatment guidelines, have increasingly broadened over the past decade, in the present study resulting in 13% more patients eligible for chemotherapy according to the most recent guideline. In this group of patients, we observed (an expected) significant increase in the administration of adjuvant chemotherapy. In 2002, the first Dutch national guideline for adjuvant systemic treatment was introduced into clinical practice advocating adjuvant chemotherapy merely in early breast cancer patients with metastatic involved regional lymph nodes. Two years earlier the National Institutes of Health Consensus Development Conference of 2000 had already recommended adjuvant systemic treatment in the majority of early stage breast cancer patients regardless of nodal, menopausal, or HR status. 19 In the US introduction of this consensus guideline increased the administration of adjuvant systemic treatment in all early-stage, node-negative, breast cancer patients < 70 years of age, from approximately 70% in 1995 to 80% in The increased administration of chemotherapy in patients with an adjusted indication over time is in itself not remarkable, but the absence of this trend in patients in whom a gene-expression profile was applied is notable. In the latter group, no significant difference in chemotherapy administration was observed over time. In fact gene-expression profile use was associated with a higher probability to administer chemotherapy in cohort I, and a decreased administration in cohort II. Adherence to the test result, in terms of administration or omission of chemotherapy in line with the test result, was high in both periods. Reports on the time-dependent impact of gene-expression profiles are scarce. McVeigh et al. 21 conducted a single-center cohort study in patients (N0/ER+) diagnosed between , in which they compared chemotherapy use in subsequent cohorts stratified for 21-RS availability and reported a stepwise reduction in the administration of chemotherapy since the introduction of the 21-RS. Our observations also make clear that guideline adjustment with broadening indications for adjuvant chemotherapy is not automatically followed. The mere 52% of patients newly considered eligible to receive chemotherapy did receive chemotherapy without the use of a gene-expression profile in the period , and as such reflects 128

129 Change over time in the impact of gene-expression profiles on chemotherapy administration considerable reluctance of Dutch clinicians to administrate chemotherapy in these patients. Clinicians may feel endorsed by the fact that the majority of patients in whom a gene-expression profile was applied were assigned to a genomic low-risk profile. Together with the observed higher adherence to a low-risk result than to a high-risk result our results suggest that Dutch clinicians tend to use gene-expression profiles in this group of patients for a substantiated decision to omit chemotherapy. In contrast, it is notable that there is a much higher compliance to the renewed endocrine therapy guideline: 80% of newly eligible patients for endocrine therapy actually received the latter in period II. Several authors observed higher adherence to guideline recommendations regarding endocrine therapy than chemotherapy in early stage breast cancer in patients with intermediate grade tumors sized 1-3 cm which resembles our studypopulation. Explanations for this discrepancy between chemotherapy and endocrine therapy in the adherence to similar guideline recommendations are absent Further research focussing on guideline adherence is needed to identify underlying reasons for guideline discordant administration of adjuvant systemic treatment. In the current study, use of gene-expression profiles did affect the proportion of patients receiving endocrine therapy in cohort I but not in cohort II. Literature underscoring that the 70- GS is not validated to aid endocrine therapy decision-making 11 apparently has had its effect, which is also acknowledged in the most recent systemic treatment guidelines. 3,14 Chapter 6 The results of the present study illustrate the effects of guideline changes and geneexpression profile use on the administration of chemotherapy. Previous studies addressed the impact of gene-expression profiles on chemotherapy decisions, but these were mainly smaller studies with a retrospective design evaluating the 21-RS, and reported contradictory results. Some report an increase in chemotherapy administration 18 while others described a significant decrease in the administration of chemotherapy in ER+ early stage breast cancer patients after gene-expression profile use These studies reported changes in chemotherapy decisions in 19% to 44% of the cases when a gene-expression profile had been used. The present study illustrates that much of the observed varying impact of gene-expression profiles is the mere result of the different chemotherapy recommendations as stated in time-dependent clinical guidelines. In contrast to clinical guidelines gene-expression profiles provide a reproducible prognostic measure and appear to be associated with a consistent use of adjuvant chemotherapy. A strength of this study is the nation-wide population-based character. However, this observational design also brings the potential risk of bias, especially confounding by indication. Therefore, only associations between gene-expression profile use and chemotherapy administration can be described and no assumptions on a causal relationship are made in the current paper. In addition, in the first period gene- 129

130 Chapter 6 expression profiles were only used as part of the RASTER study, hence limiting our analysis assessing the impact of gene-expression profiles on chemotherapy decisions for both time periods to patients who met the inclusion criteria of this study (< 61 years of age). A substantial number of patients with a dubious indication for chemotherapy aged between years of age (n = 2659) could not be included in the analyses for this reason. Furthermore, participation in the RASTER study in the first period could have led to bias by the propensity to use the test and adhere to the test result as result of trial participation. Then again, in current practice the use of gene-expression profiles is merely suggested (and not dictated) by the national guidelines and used by a selection of institutions and doctors that are likewise driven by a certain attitude towards deploying gene-expression profiling. The latter is underscored by the fact that only one third of patients eligible for gene-expression profile use in the second period actually received a gene-expression profile and that 61% of the hospitals frequently deployed gene-expression profiles. The observed similar high adherence rates in both periods reflect a comparable attitude towards the interpretation of the result of a 70-GS or 21-RS in both periods. In conclusion, in Dutch ER+ early stage breast cancer patients newly eligible for chemotherapy due to clinical guideline changes gene-expression profile use is associated with high adherence rates to the gene-expression profile result and more consistent administration of CT over time. ACKNOWLEDGEMENTS This work was supported by the Dutch Cancer Society (KWF). 130

131 Change over time in the impact of gene-expression profiles on chemotherapy administration Supplementary Table 1. Patient-, tumour- and treatment characteristics of the 1132 ER+/Her2- patients who were eligible for receiving a GEP during both periods, who received a gene-expression profile. Cohort I (n= 107) n (%) Cohort II (n = 1025) n(%) P-value Patientcharacteristics Age at diagnosis in years, mean (SD)* 48.2 (0.54) 55.8 (0.24) <0.001 < 35 0 (0%) 1 (0%) (56%) 228 (22%) (44%) 806 (78%) <0.001 >70 na na Tumorcharacteristics Histology Ductal carcinoma 81 (76%) 869 (85%) Lobular carcinoma 18 (17%) 118 (11%) Mixed 5 (5%) 27 (3%) Other 3 (3%) 14 (2%) Chapter 6 Axillary status (pn) pn0 92 (86%) 891 (87%) pn1mi 15 (14%) 137 (13%) Tumor size (pt) pt1 99 (93%) 944 (93%) pt2 8 (7%) 67 (7%) Invasive tumor grade Grade I 18 (17%) 119 (12%) Grade II 89 (83%) 909 (88%) Progesterone receptor status Negative 14 (13%) 159 (15%) Positive 93 (87%) 868 (85%) Her2 status Negative 97 (92%) 995 (97%) Positive 7 (7%) 28 (3%) Dubious 0 (0%) 1 (0%) Unknown 1 (1%) 2 (0%) * Students t-test; other data represent chi-square values. a Pathological lymph node involvement (pn) according to the TNM staging system. b Pathological tumour size (pt) according to the TNM staging system 131

132 Chapter 6 Supplementary Table gene signature (70-GS) and 21-recurrence score (21-RS) test results and the administration of adjuvant endocrine therapy stratified for the two time periods. Cohort I: Cohort II: n ET (%) n ET (%) All* % % Low risk % % High risk % % Unknown % 21-recurrence score All* n.a % Low risk 6 100% Intermediate 2 50% High risk 3 33% * Patients with a new indication for adjuvant systemic therapy according to the national guideline of 2012 and ER+/Her2-disease, <61 years of age. 132

133 Change over time in the impact of gene-expression profiles on chemotherapy administration Supplementary Figure 2. Administration of endocrine therapy (ET) stratified for both cohorts in relation to gene-expression profile (GEP) use in patients with a new indication for the latter according to the guideline of 2012 and T1-2cN0M0 disease, < 61 years of age. Chapter 6 133

134 Chapter 6 REFERENCE LIST 1. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Early Breast Cancer Trialists Collaborative Group. Lancet (London, England). 1992;339(8784):1 15. Available at: Accessed April 23, Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet (London, England). 2005;365(9472): doi: /s (05) Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Glick JH, Gelber RD, Goldhirsch A, Senn HJ. Meeting highlights: adjuvant therapy for primary breast cancer. J Natl Cancer Inst. 1992;84(19): Available at: Accessed May 19, Goldhirsch A, Glick JH, Gelber RD, Coates AS, Senn H-J. Meeting Highlights: International Consensus Panel on the Treatment of Primary Breast Cancer. J Clin Oncol. 2001;19(18): doi: /jco Mook S, Van t Veer LJ, Rutgers EJT, Piccart-Gebhart MJ, Cardoso F. Individualization of therapy using Mammaprint: from development to the MINDACT Trial. Cancer Genomics Proteomics. 4(3): Available at: Accessed April 19, Paik S, Shak S, Tang G, et al. A multigene assay to predict recurrence of tamoxifentreated, node-negative breast cancer. N Engl J Med. 2004;351(27): doi: / NEJMoa van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: / NEJMoa Drukker CA, Bueno-de-Mesquita JM, Retèl VP, et al. A prospective evaluation of a breast cancer prognosis signature in the observational RASTER study. Int J cancer. 2013;133(4): doi: /ijc Drukker CA, van Tinteren H, Schmidt MK, et al. Long-term impact of the 70-gene signature on breast cancer outcome. Breast Cancer Res Treat. 2014;143(3): doi: /s Buyse M, Loi S, van t Veer L, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst. 2006;98(17): doi: /jnci/djj

135 Change over time in the impact of gene-expression profiles on chemotherapy administration 12. Bueno-de-Mesquita JM, Linn SC, Keijzer R, et al. Validation of 70-gene prognosis signature in node-negative breast cancer. Breast Cancer Res Treat. 2009;117(3): doi: /s Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol Off J Eur Soc Med Oncol. 2010;21(4): doi: /annonc/mdp Goldhirsch A, Wood WC, Coates AS, Gelber RD, Thurlimann B, Senn H-J. Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol. 2011;22(8): doi: /annonc/mdr Senkus E, Kyriakides S, Penault-Llorca F, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2013;24(suppl 6):vi7 vi23. doi: /annonc/mdt Siesling S, Louwman WJ, Kwast A, et al. Uses of cancer registries for public health and clinical research in Europe: Results of the European Network of Cancer Registries survey among 161 population-based cancer registries during Eur J Cancer. 2015;51(9): doi: /j.ejca Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom.; Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective communitybased feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Eifel P, Axelson JA, Costa J, et al. National Institutes of Health Consensus Development Conference Statement: adjuvant therapy for breast cancer, November 1-3, J Natl Cancer Inst. 2001;93(13): Available at: pubmed/ Accessed May 19, Harlan LC, Clegg LX, Abrams J, Stevens JL, Ballard-Barbash R. Community-Based Use of Chemotherapy and Hormonal Therapy for Early-Stage Breast Cancer: J Clin Oncol. 2006;24(6): doi: /jco McVeigh TP, Hughes LM, Miller N, et al. The impact of Oncotype DX testing on breast cancer management and chemotherapy prescribing patterns in a tertiary referral centre. Eur J Cancer. 2014;50(16): doi: /j.ejca Wu X-C, Lund MJ, Kimmick GG, et al. Influence of race, insurance, socioeconomic status, and hospital type on receipt of guideline-concordant adjuvant systemic therapy for locoregional breast cancers. J Clin Oncol. 2012;30(2): doi: / JCO van de Water W, Bastiaannet E, Dekkers OM, et al. Adherence to treatment guidelines and survival in patients with early-stage breast cancer by age at diagnosis. Br J Surg. 2012;99(6): doi: /bjs Chapter 6 135

136 Chapter Kimmick GG, Camacho F, Mackley HB, et al. Individual, Area, and Provider Characteristics Associated With Care Received for Stages I to III Breast Cancer in a Multistate Region of Appalachia. J Oncol Pract. 2015;11(1):e9 e18. doi: / JOP Kuijer A, van Bommel ACM, Drukker CA, et al. Using a gene expression signature when controversy exists regarding the indication for adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. Genet Med. 2016;18(7): doi: /gim Ademuyiwa FO, Miller A, O Connor T, et al. The effects of oncotype DX recurrence scores on chemotherapy utilization in a multi-institutional breast cancer cohort. Breast Cancer Res Treat. 2011;126(3): doi: /s Asad J, Jacobson AF, Estabrook A, et al. Does oncotype DX recurrence score affect the management of patients with early-stage breast cancer? Am J Surg. 2008;196(4): doi: /j.amjsurg Klang SH, Hammerman A, Liebermann N, Efrat N, Doberne J, Hornberger J. Economic implications of 21-gene breast cancer risk assay from the perspective of an Israeli-managed health-care organization. Value Health. 2010;13(4): doi: / j x. 29. Hornberger J, Chien R, Krebs K, Hochheiser L. US Insurance Program s Experience With a Multigene Assay for Early-Stage Breast Cancer. J Oncol Pract. 2011;7(3S):e38s e45s. doi: /jop

137 Change over time in the impact of gene-expression profiles on chemotherapy administration Chapter 6 137

138

139 CHAPTER 7 Impact of 70-gene signature use on adjuvant chemotherapy decisions in estrogen receptor positive early breast cancer patients: results of a prospective cohort study. A. Kuijer, M.E. Straver, B. den Dekker, A.C.M. van Bommel, S.G. Elias, C.H. Smorenburg, J. Wesseling, S.C. Linn, E.J.Th. Rutgers, S. Siesling, T. van Dalen. Journal of Clinical Oncology 2017;35:

140 Chapter 7 ABSTRACT Purpose: Gene-expression profiles are increasingly used in addition to conventional prognostic factors to guide adjuvant chemotherapy (CT) decisions. The Dutch guideline suggests use of validated gene-expression profiles in estrogen-receptor positive (ER+) early stage breast cancer patients without overt lymph node metastases. We aimed to assess the impact of 70-gene signature (70-GS) on CT decisions in ER+ early stage breast cancer patients. Patients and methods: In a prospective observational multicenter study in patients < 70 years who had undergone surgery for ER+ early breast cancer physicians were asked whether they intended to administer adjuvant CT before deployment of the 70-GS and after test result became available. Results: Between October And December patients, treated in 33 hospitals, were enrolled. Fifty-one percent of patients had pt1cn0, BRII, Her2- breast cancer. Based on conventional clinicopathological characteristics physicians would recommend CT in 270/660 (41%) of the patients and withhold CT in 107/660 (16%). In the remaining 43% of patients the physicians felt unsure to give an advice before 70-GS testing. In patients in whom CT was initially recommended or not recommended 56% and 59% were assigned to a low-risk profile by the 70-GS respectively (ĸ =0.02, 95%CI to 0.11). After disclosure of the 70- GS test result the preliminary advice was changed in 51% of patients in whom a recommendation was given before testing; the definitive CT recommendation of the physician was in line with the 70-GS result in 96% of patients. Conclusions: In this prospective multicenter study in a selection of patients with ER+ early breast cancer 70-GS use changed the physicians intended recommendation to administer CT in half of the patients. 140

141 Impact of the 70-gene signature on chemotherapy decision-making INTRODUCTION During the last two decades treatment guidelines for adjuvant systemic therapy in breast cancer have changed considerably, today advising chemotherapy (CT) in the majority of patients. The recommendation to administer adjuvant CT is based on clinicopathological prognostic factors. There is growing awareness that conventional factors do not accurately estimate prognosis and benefit of CT in estrogen-receptor (ER) positive (+) breast cancer patients which is reflected by a more reticent attitude towards CT administration in these patients in more recent treatment guidelines. 1 Several gene-expression profiles have been developed to aid adjuvant CT decisionmaking in ER+ early stage breast cancer. 2,3 It was validated in several retrospective and a community based feasibility study. 10,11 Very recent, the results of the MINDACT trial provided the first level 1A evidence that omission of CT in patients assigned to the 70-GS low-risk category is safe. 12 Awaiting the implementation of the results of randomized clinical trials comparing the value of gene-expression profiles to clinicopathological factors when adjuvant CT is considered, national 13 and international 1,14 guidelines advise the use of a gene-expression profile in a selection of early stage ER+ patients. The Dutch guideline recommends use of a validated gene-expression profile since 2012 in patients with ER+ invasive ductal carcinoma in whom doubt exists regarding CT benefit based on clinicopathological factors. 13 This guideline considers CT beneficial in patients with an expected 10-year breast cancer specific mortality of at least 15% as these patients would have an absolute overall survival gain of 4-5% of adjuvant chemotherapy administration. In common practice, gene-expression profiles in the Netherlands are until now mainly used in patients with ER+/HER2- patients without overt lymph node metastases (pt1c-2n0-1mi). 15 In the Netherlands, the 70-gene signature (70-GS) is mostly used and accountable for 97% of all deployed gene-expression tests. 16 Chapter 7 The aim of the present prospective observational multicenter study is to assess the impact of the 70-GS on individual physician s decisions when there was doubt about the benefit of adjuvant CT in surgically treated ER+ early stage breast cancer patients. PATIENTS AND METHODS Study design We conducted an observational prospective multicenter study to assess impact of 70- GS use on adjuvant CT decision-making. Patients in whom 70-GS deployment was considered as part of routine clinical practice were eligible for participation. The Dutch national guideline suggests deployment of a validated gene-expression profile in patients 141

142 Chapter 7 with ER+ ductal carcinomas in whom doubt exists regarding CT benefit. This guideline also states that in patients with an expected 10-year overall survival of 85% CT is indicated. It was expected that gene-expression profiles would merely be deployed in patients with ER+/Her2-, low- or intermediate grade tumors staged as T1c-2N0-1mi. 15 Patients treated between January and December in 31 participating hospitals were enrolled following informed consent before deployment of 70-GS. Exclusion criteria were a history of malignancy, distant metastasis and neo-adjuvant systemic treatment. The study was approved by the Medical Ethics Committee of the University Medical Center Utrecht (12-450) and by institutional review boards of participating centers. The study protocol was registered in the clinicaltrial.gov database (NCT ). Eligible patients were identified during postoperative multi-disciplinary teammeetings where a preliminary CT recommendation was formulated based on the clinicopathological results (Figure 1). The treating physician completed the first clinical report form (CRF) in which information on clinicopathological characteristics and preliminary CT recommendation was registered: to administer CT, withhold adjuvant CT or to give no advice being unsure. Estimated 10-year overall survival without CT based on the UK-PREDICT tool 17 was calculated for every patient to objectify the prognostic perspective of the preliminary CT recommendation. In comparison to Adjuvant online, PREDICT incorporates Her2-status in its model and tumor size is entered more detailed too. 18 After completion of CRF 1 the tumor sample was sent for 70-GS analysis and the result was disclosed to the oncologist within ten working days. The post-test CT recommendation and actually administered CT were recorded in CRF2. Endpoints The primary endpoint was the percentage of patients in whom 70-GS use led to an altered oncologists adjuvant CT recommendation. Secondary endpoints were the relation between the physician preliminary recommendation and the proportion of patients actually receiving CT. Both outcomes were also calculated basing the recommendation of adjuvant CT on the estimated 10-year overall survival of 85% as a surrogate cut-off value, since the national guideline considers an estimated 10-year survival of <85% as a reason to recommend CT. 142

143 Impact of the 70-gene signature on chemotherapy decision-making Figure 1. Flowchart of study-inclusion between October 2013 and January Female patient with breast cancer seen at out-patient clinic Surgery Tissue sent to pathologists for classic histopathological analysis Multi-disciplinary team meeting Preliminary CT recommendation physician Inclusion criteria: - ER+, invasive ductal carcinoma Exclusion criteria: - Previous malignancies or metastatic disease Patient seen at out-patient clinic Informed consent obtained CRF1 Chapter 7 Post-test CT recommendation physician Actual administered CT CRF2 Tissue sent off for 70-GS test Statistical analysis Baseline characteristics of all patients were summarized in a baseline table. The difference in estimated median 10-year overall survival and CT benefit in relation to the recommendation before deployment of the test and in relation to the test result was compared by Kruskall-Wallis test or Mann-Whitney U test, respectively. Percentage of CT recommendation change was assessed with McNemar test. Agreement between the preliminary CT recommendation and the 70-GS result was assessed by Cohens Kappa statistic which can be interpreted as follows: values 0 as indicating no agreement, as none to slight, as fair, as moderate, as substantial, and as almost perfect agreement. All analyses were performed in R version

144 Chapter 7 RESULTS A total of 698 ER+ early stage breast cancer patients were enrolled and the mean number of patients per participating hospital was 21 (range 2 60). Thirty-eight patients were excluded from the study: in five patients no written informed consent was received and in 33 patients data on either preliminary or post-test CT recommendation was not complete at the closing date of the study. A total of 660 patients with a median age of 57 years were evaluable. Postmenopausal women with unifocal Her2-negative low grade tumors > 2 cm or intermediate grade tumors 1-2 cm without axillary lymph-node involvement (pn0 or pn1mi) prevailed (n=405; 67%). Based on clinicopathological characteristics median estimated overall 10-year survival according to PREDICT was 86.1% (40.2%-95.3%, IQR 7.4%) and expected CT benefit 1,6% (0.4%-8.5%, IQR 1.0%). Before deployment of the 70-GS, oncologists would recommend CT in 41% and advise against in 16% of patients. In the remaining 43% of patients, no preliminary CT recommendation was made as physicians preferred to await the 70-GS result (Figure 2). Patients in whom CT was recommended were younger, had larger tumors of higher grade compared to patients in whom physicians advised against adjuvant CT or to patients in whom oncologists gave no advice (Table 1). There was no significant difference in median 10-year overall survival according to PREDICT among the three pre-test CT recommendation groups (p 0.231). The 70 GS assigned 41% of all patients to the high-risk category. In patients in whom CT was initially advised 56% was assigned to the low-risk category by the 70-GS compared to 59% of the patients in whom CT was not recommended. The preliminary advice was in line with 70 GS risk-category in 48% of patients. There was none to slight agreement between this preliminary CT advice and the 70-GS result (ĸ %CI to 0.11, p 0.342, Figure 2). Subgroup analysis in patients with tumors >2 cm (n=146) or patients with axillary lymph-node involvement ( pn1mi, n = 103) yielded similar proportions of patients assigned to the risk categories with a discordance between the preliminary CT advice and the 70-GS result in 56% and 62% of patients (ĸ %CI to 0.17 and ĸ %CI to

145 Impact of the 70-gene signature on chemotherapy decision-making Figure 2. Concordance between pre-test CT recommendation of the oncologist and the 70-gene signature (GS) test result. Chapter 7 Note. Agreement between pre-test oncologist CT recommendation and the 70-GS test result: Cohen s Kappa 0.02 (95%CI , p ). * Median estimated 10-year benefit of adjuvant CT based on the UK-PREDICT tool per pre-test CT recommendation category. The oncologist adhered to the 70-GS result in 96% of patients: in eighteen 70-GS lowrisk patients CT was recommended while in nine 70-GS high-risk patients oncologists advised against CT. Actually administered CT was the same as oncologists post-test recommendation in 94% of patients. Thirty patients in whom CT was recommended, 29 of them with a high-risk 70-GS result, did not receive CT eventually. Conversely, eight patients received CT despite the oncologist advice to withhold CT of whom seven had a low-risk 70-GS result. Eventual administered CT was in line with the 70-GS result in 91% of patients. In patients in whom the clinician formulated a clear recommendation prior to the test (n=377), this recommendation was changed after 70-GS use in 51% of patients (95%CI 46% - 56%, p < 0,001; Table 2). Actually administered CT deviated from the preliminary CT recommendation in 52% of the patients. 145

146 Chapter 7 Table 1. Baseline characteristics of all patients included in the study (n = 660), all patients had ER+ disease. Patient characteristics All patients n=660 Pre-test chemotherapy advice Advice CT Advice no CT No advice given n = 270 n = 107 n = 283 Age in years (mean, SD) 57.0 (8.1) 56.2 (7.9) 58.5 (8.0) 57.2 (8.3) Menopausal status Pre- or peri- 218 (34%) 102 (39%) 28 (27%) 88 (32%) Post 428 (66%) 162 (61%) 76 (73%) 190 (68%) Missing Tumor characteristics Unifocal disease 596 (90%) 250 (93%) 97 (91%) 249 (88%) Invasive tumor grade Grade I 96 (15%) 29 (11%) 25 (23%) 42 (15%) Grade II 481 (73%) 195 (72%) 77 (72%) 209 (74%) Grade III 82 (12%) 46 (17%) 5 (5%) 31 (11%) Tumor size in mm (mean, SD) a 16.4 (7.0) 16.9 (5.7) 16.1 (12.0) 16.2 (5.3) pt1 534 (81%) 212 (79%) 96 (90%) 226 (80%) pt2 125 (19%) 58 (22%) 10 (9%) 57 (20%) pt3 1 (0%) - 1 (1%) - PR positive disease 571 (87%) 231 (86%) 90 (84%) 250 (88%) Her2 negative disease 638 (97%) 259 (96%) 105 (98%) 274 (97%) Axillary lymph-node involvement b pn0 552 (84%) 224 (83%) 95 (89%) 233 (82%) pn1mi 65 (10%) 27 (10%) 5 (5%) 33 (12%) pn1a 35 (5%) 17 (6%) 6 (6%) 12 (4%) >pn1a 3 (1%) (1%) Nx 5 (1%) 2 (1%) 1 (1%) 2 (1%) Treatment characteristics Surgery Breast conserving surgery 532 (81%) 211 (78%) 91 (85%) 230 (81%) Mastectomy 128 (19%) 59 (22%) 16 (15%) 53 (19%) 146

147 Impact of the 70-gene signature on chemotherapy decision-making Axillary surgery SNP 621 (94%) 258 (96%) 96 (90%) 267 (94%) ALND 6 (2%) 1 (0%) 4 (4%) 1 (0%) SNP + ALND 12 (1%) 3 (1%) 1 (1%) 8 (3%) No axillary surgery 21 (3%) 8 (3%) 6 (6%) 7 (3%) Estimated 10-year survival (median, IQR) 86.1 (7.4) 86.1% (7.6) 87.1% (6.6) 85.8% (7.9) a pathological tumour size (pt) according to the TNM-staging system. b pathological lymph-node involvement according to the TNM staging system. PR = progesterone receptor; SNP = sentinel node procedure; ALND = axillary lymph-node dissection. Using the 10-year estimated survival of 85% as cut-off value to administer CT or not, CT would have been considered beneficial in 289 (44%) patients. In 37% of patients (n = 139) with estimated 10-years survival > 85% the 70-GS indicated a high-risk test result and in 46% of patients (n = 132) with an estimated 10-year survival <85% the 70- GS indicated a high-risk test. Again, no or only slight agreement between the 10-year survival categories and the 70-GS result was observed (ĸ 0.08, 95%CI ). A recommendation based on the estimated 10-year survival did also not correspond to the 70-GS test result in almost half of the patients (45%; 95%CI: 41-49%). The physician adhered to the 70-GS test result in 96% (n = 357) and 96% (n = 276) of patients with an estimated 10-year survival of >85% or <85%, respectively. Chapter 7 147

148 Chapter 7 Table 2. CT recommendation before vs. after obtaining the 70-GS test result and the actual administration of CT. Post-test recommendation Adherence to test-result* Actually administered CT Adherence to testresult* Preliminary recommendation no. patients No CT CT % No CT CT % No CT (65%) 38 (35%) 94% 156 CT 270 (58%) 114 (42%) 97% 173 Unsure 283 (61%) 110 (39%) 95% 73 (68%) 162 (60%) 185 (65%) 34 (32%) 91% 108 (40%) 90% 98 (35%) 91% *Percentage of patients in whom the post-test recommendation/actual administered CT was in line with the 70-GS test result (i.e. no CT in case of a low-risk profile and CT in case of a high-risk profile. Note. There was in change in CT recommendation in 51% (95%CI 46%-56%, p < 0.001) of patients with a clear pre-test CT recommendation (i.e. CT or no CT, n = 377). Actual administered CT differed from the preliminary CT recommendation in 52% (95%CI , p <0.001) of patients with a clear pre-test CT recommendation. DISCUSSION In this prospective multicenter study we demonstrate that 70-GS use has substantial impact on CT decision-making in ER+/Her2- early stage breast cancer patients: 70-GS use changed oncologists premeditated advice regarding adjuvant CT treatment in half of patients. In addition, physicians felt unsure to give an advice without the 70-GS in a substantial proportion of patients. In the current study, 70-GS use led to an altered CT recommendation in the 51% of patients and CT was administered in contradiction with the preliminary recommendation in a similar proportion of patients. Adherence of physicians and patients to the 70-GS result was high, which is in line with recent studiesthat show that a gene-expression profile result has the highest attribute to patients preferences in CT decision-making. 19 Several, mainly single institution, prospective studies report a change in treatment recommendation in 16%-38% of patients predominantly changing from CT to no CT Similar results regarding the impact of OncotypeDx or Prosigna use on CT decision-making have been reported. In a recent meta-analysis of Carlson and Roth a mean pooled CT recommendation change of 38% (ranging from 23% - 43% in eight prospective studies) was reported after OncotypeDx use in ER+/Her2- early stage 148

149 Impact of the 70-gene signature on chemotherapy decision-making breast cancer patients. 24 Two prospective multicenter studies assessing the influence of Prosigna use on CT decision-making report a discordanc between clinical and genomic risk assessment in approximately half of ER+/Her2- early stage breast cancer patients, leading to an altered CT advice in 18-20% of patients. 25,26 In the present study, the oncologist could refrain from giving an advice before deployment of the 70-GS which might have led to overestimation of the proportion of patients in whom the treatment recommendation was changed. Then again, our results are in line with a study of Levine et al, who performed a large equivalent population-based study in ER+/Her2- early stage breast cancer patients which also included the option to refrain from pre-test CT recommendation, and reported 50% treatment change in patients after OncotypeDx use. 27 In addition, an estimated 10-year overall survival of 85%, retrospectively estimated by PREDICT, was applied to the study group as a cut-off value to administer adjuvant CT or not. Still, even when using this as a cut-off value, in almost half of patients the 70-GS result would differ from this resulting surrogate recommendation. The oncologist recommended administration of adjuvant CT in 41% of all patients prior to 70-GS testing. In a nationwide study conducted by this research group using observational Netherlands Cancer Registry data, 45% of all (pt1-2n0-i) ER+/Her2- Dutch breast cancer patients in whom no 70-GS was deployed, received adjuvant CT. 15 The latter observation illustrates the current controversy regarding CT benefit in these patients. Chapter 7 This is the largest prospective study evaluating the impact of 70-GS use on CT decisionmaking. The design of the current study mimics routine clinic practice. Nevertheless, the pre-test CT recommendation by the oncologists remains an artificial statement which is made with the prospect of obtaining the 70-GS result. In addition, due to the observational design of the study it is possible that oncologists used the 70-GS only in a selection of eligible patients which might have overestimated adherence rates or percentages of treatment change. Despite these limitations the substantial impact of 70- GS use on CT decision-making in ER+/Her2- patients is clearly demonstrated. The impact of the 70-GS on CT decision-making was assessed in ER+/Her2- patients in whom controversy regarding CT benefit led to deployment of the 70-GS. It is important to note that in the current study the prognostic value of the 70-GS was used to guide CT decisions. Future studies should be conducted to provide outcome data that support the altered decisions. For that purpose, we intend to merge the data of the present study with the Netherlands Cancer Registry database. In the meantime, more robust results 149

150 Chapter 7 of the recently presented EORTC 10041/BIG MINDACT trial indicate that 46% of breast cancer patients who are identified as high-risk for recurrence according to clinicopathological factors are assigned to the 70-GS low-risk category and these patients are unlikely to derive any significant benefit from CT. 12 Expectedly, the indication area for 70-GS use as stated in the present clinical guidelines will further extend to the clinical high-risk patient categories. The present study demonstrates that the 70-GS is supportive for CT-decision making in assumed high-risk categories such as patients with larger tumors or lymph node metastases. At the same time, in patients identified as low-risk for recurrence, according to clinicopathological factors, assigned to the 70-GS high-risk category no significant difference in the 5-year distant metastasis free survival was found between patients who received adjuvant CT or not. As there is currently no evidence of clinical utility of the 70-GS in clinical low-risk patients this may probably lead to less deployment of the 70-GS in these patients. In the present study in breast cancer patients with ER+/Her2- low- or intermediate grade tumors staged as T1c-2N0-1mi, 70-GS use changed the physicians intended recommendation to administer CT in half of the patients. The weak correlation between the pre-test CT recommendation and the 70-GS result implies that oncologists encounter difficulties in the recommendation to administer CT or not without deployment of a gene-expression profile in a patient group, mainly consisting of patients with ER+/ Her2- intermediate grade tumors staged as T1c-2 N0-1mi. While previous studies have demonstrated existing controversy regarding the use of CT in this group of patients, the present study demonstrates the inability of physicians to make an accurate CT recommendation based on conventional prognostic factors alone. ACKNOWLEDGEMENTS The authors thank all patients for participation in this study and Marianne Deelen for her logistic support in performing this study. The authors thank all principal investigators of the participating hospitals for their collaboration: dr. A. Imholz (Deventer Ziekenhuis), dr. A. Honkoop (Isala ziekenhuis, Zwolle), dr. A. Timmer-Bonte (Alexander Monro, Bilthoven), dr. P. Nieboer (Wilhelmina ziekenhuis, Assen), dr. S. Hovenga (Ziekenhuis Nij Smellighe, Dronten), dr. J. Hunting (Antoniusziekenhuis, Nieuwegein), dr. T. Smilde (Jeroen Bosch ziekenhuis, Den Bosch), dr. E. Vriens (Ter Gooi ziekenhuis, Hilversum), dr. H. Zuetenhorst (St. Fransiscus Gasthuis, Rotterdam), dr. A. van der Velden (Martini ziekenhuis, Groningen), dr. B. de Valk (Spaarne ziekenhuis, Hoofddorp), dr. B. Spaansen (Gemini ziekenhuis, Den Helder), dr. Q. van Rossum (Vlietland ziekenhuis, Sliedrecht), dr. M.W.A. van Tilburg (Sint Jansdal, Harderwijk), dr. A. van der Pas (Lange Land ziekenhuis, Zoetermeer), dr. 150

151 Impact of the 70-gene signature on chemotherapy decision-making A. Haringhuizen (Ziekenhuis Gelderse Vallei, Ede), dr. W. Lastdrager (Gelre ziekenhuis, Apeldoorn), dr. C. Blanken (Rijnstate ziekenhuis, Arnhem), dr. H. Rijna (Kennemer Gasthuis, Haarlem), dr. R. van Doorn (Zuwe Hofpoort ziekenhuis, Woerden), dr. J. de Boer (Tjongerschans, Heereveen), dr. S. Vrijaldenhoven (Medisch Centrum Alkmaar), dr. J. Bollen (Medisch Centrum Zuiderzee, Lelystad), dr. L. de Widt (Waterland ziekenhuis, Purmerend), dr. M. de Roos (Ziekenhuis Rivierenland, Tiel), dr. G. Tetteroo (IJsselland ziekenhuis, Capelle aan de Ijssel), dr. C. van Riel (Antoniusziekenhuis, Sneek), dr. S. Muller (Zaans Medisch Centrum), dr. S. Dohmen (Boven IJ ziekenhuis, Amsterdam), dr. J. de Waard (West Fries Gasthuis, Hoorn), dr. M. Jagers op Akkerhuis (Ropke Zweers, Hardenberg), dr. J. Ketel (Gelre ziekenhuis, Zutphen) and dr. Meerum Terwogt (Onze Lieve Vrouwe Gasthuis, Amsterdam). Chapter 7 151

152 Chapter 7 REFERENCE LIST 1. Goldhirsch A, Wood WC, Coates AS, et al. Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol Off J Eur Soc Med Oncol. 2011;22(8): doi: /annonc/mdr van t Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature. 2002;415(6871): doi: /415530a. 3. van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: /nej- Moa Buyse M, Loi S, van t Veer L, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst. 2006;98(17): doi: /jnci/djj Bueno-de-Mesquita JM, Linn SC, Keijzer R, et al. Validation of 70-gene prognosis signature in node-negative breast cancer. Breast Cancer Res Treat. 2009;117(3): doi: /s Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol Off J Eur Soc Med Oncol. 2010;21(4): doi: /annonc/mdp Knauer M, Mook S, Rutgers EJT, et al. The predictive value of the 70-gene signature for adjuvant chemotherapy in early breast cancer. Breast Cancer Res Treat. 2010;120(3): doi: /s Mook S, Knauer M, Bueno-de-Mesquita JM, et al. Metastatic Potential of T1 Breast Cancer can be Predicted by the 70-gene MammaPrint Signature. Ann Surg Oncol. 2010;17(5): doi: /s x. 9. Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol. 2010;21(4): doi: /annonc/mdp Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective community-based feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Drukker CA, Bueno-de-Mesquita JM, Retèl VP, et al. A prospective evaluation of a breast cancer prognosis signature in the observational RASTER study. Int J cancer. 2013;133(4): doi: /ijc Cardoso F, van t Veer LJ, Bogaerts J, et al. 70-Gene Signature as an Aid to Treatment Decisions in Early-Stage Breast Cancer. N Engl J Med. 2016;375(8): doi: / NEJMoa

153 Impact of the 70-gene signature on chemotherapy decision-making 13. Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom Senkus E, Kyriakides S, Penault-Llorca F, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2013;24(suppl 6):vi7 vi23. doi: /annonc/mdt Kuijer A, van Bommel ACM, Drukker CA, et al. Using a gene expression signature when controversy exists regarding the indication for adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. Genet Med. 2016;18(7): doi: /gim Kuijer A, Drukker CA, Elias SG, et al. Changes over time in the impact of gene-expression profiles on the administration of adjuvant chemotherapy in estrogen receptor positive early stage breast cancer patients: A nationwide study. Int J Cancer. 2016;139(4): doi: /ijc Wishart GC, Bajdik CD, Dicks E, et al. PREDICT Plus: development and validation of a prognostic model for early breast cancer that includes HER2. Br J Cancer. 2012;107(5): doi: /bjc Drukker CA, Nijenhuis M V, Bueno-de-Mesquita JM, et al. Optimized outcome prediction in breast cancer by combining the 70-gene signature with clinical risk prediction algorithms. Breast Cancer Res Treat. 2014;145(3): doi: /s Marshall DA, Deal K, Bombard Y, Leighl N, MacDonald K V, Trudeau M. How do women trade-off benefits and risks in chemotherapy treatment decisions based on gene expression profiling for early-stage breast cancer? A discrete choice experiment. BMJ Open. 2016;6(6):e doi: /bmjopen Exner R, Bago-Horvath Z, Bartsch R, et al. The multigene signature MammaPrint impacts on multidisciplinary team decisions in ER+, HER2- early breast cancer. Br J Cancer. 2014;111(5): doi: /bjc Cusumano PG, Generali D, Ciruelos E, et al. European inter-institutional impact study of MammaPrint. Breast. 2014;23(4): doi: /j.breast Torrisi R, Garcia-Etienne CA, Losurdo A, et al. Potential impact of the 70-gene signature in the choice of adjuvant systemic treatment for ER positive, HER2 negative tumors: a single institution experience. Breast. 2013;22(4): doi: /j.breast DeFrank JT, Salz T, Reeder-Hayes K, Brewer NT. Who gets genomic testing for breast cancer recurrence risk? Public Health Genomics. 2013;16(5): doi: / Carlson JJ, Roth JA. The impact of the Oncotype Dx breast cancer assay in clinical practice: a systematic review and meta-analysis. Breast Cancer Res Treat. 2013;141(1): doi: /s z. Chapter 7 153

154 Chapter Martín M, González-Rivera M, Morales S, et al. Prospective study of the impact of the Prosigna assay on adjuvant clinical decision-making in unselected patients with estrogen receptor positive, human epidermal growth factor receptor negative, node negative early-stage breast cancer. Curr Med Res Opin. 2015;31(6): doi: / Wuerstlein R, Sotlar K, Gluz O, et al. The West German Study Group Breast Cancer Intrinsic Subtype study: a prospective multicenter decision impact study utilizing the Prosigna assay for adjuvant treatment decision-making in estrogen-receptor-positive, HER2-negative early-stage breast cancer. Curr Med Res Opin. 2016;32(7): doi: / Levine MN, Julian JA, Bedard PL, et al. Prospective Evaluation of the 21-Gene Recurrence Score Assay for Breast Cancer Decision-Making in Ontario. J Clin Oncol. 2016;34(10): doi: /jco

155 Impact of the 70-gene signature on chemotherapy decision-making Chapter 7 155

156

157 CHAPTER 8 Impact of gene-expression profiling in patients with early stage breast cancer when applied outside the guideline directed indication area. K. Schreuder, A. Kuijer, E. J. Th. Rutgers, C.H. Smorenburg, T. van Dalen, S. Siesling. European Journal of Cancer 2017;84:

158 Chapter 8 ABSTRACT Purpose: implications of gene expression profiling (GEP) in breast cancer patients for whom guidelines already clearly recommend to administer or withhold adjuvant chemotherapy (CT) are assessed based on a clinical high or low risk. Material and Methods: clinical low- and high risk patients, according to current guidelines, diagnosed between were selected from the Netherlands Cancer Registry (NCR). Influence of GEP use and GEP test result on CT administration was assessed with logistic regression. Results: overall, 26,425 patients were identified; 4.8% of patients with clinical low-risk (444/ 9,354), 7.5% of the patients with a clinical high risk cancer (1,281/ 17,071) received a GEP. GEP use was associated with a significantly increased odds of CT administration in clinical low-risk patients (OR= %CI: ). In clinical high-risk patients GEP was associated with a decreased frequency of CT administration (OR=0.55, 95%CI: ). Adherence to the GEP result was higher in clinical high-risk patients with a discordant GEP result as compared to clinical low-risk patients with a discordant GEP result: 71.7% vs. 52.2%, respectively. Conclusion: GEPs are frequently used outside the indicated area and significantly influenced the administration of adjuvant CT, although adherence to the test-result was limited. 158

159 Impact of gene-expression profiling when applied outside the guideline directed indication area INTRODUCTION The use of adjuvant systemic therapy has considerably improved the prognosis of patients with breast cancer over the last two decades. 1 However, there is also a growing awareness that this broad application of adjuvant chemotherapy (CT) increases the risk of overtreatment as the threshold to use CT is difficult to determine. 2 Different biologic and clinical clues suggest that not all patients derive substantial benefit from CT. 3 Especially in estrogen-receptor (ER) positive (+) early-stage breast cancer patients doubt exists regarding the benefit of adjuvant CT. Because of negative side effects of systemic therapies, effective use is important. 4 Gene expression profiles (GEPs) were developed a decade ago to enable a prediction of prognosis in addition to the prognostic information of conventional clinicopathological factors. Although the predictive value of GEPs in terms of a quantified benefit of administering CT is still disputed, national and international treatment guidelines currently suggest the use of a GEP complementary to clinicopathological factors in ER+ early stage breast cancer patients. 3,5-9 The Dutch guideline (2012) suggests the use of a validated GEP in early breast cancer patients, in whom benefit of CT is controversial based on traditional prognostic factors alone. 3,9 In a previous study it was demonstrated that this category, in which GEP use is highest, consists of patients with estrogen receptor (ER) positive (+)/HER2-Neu negative (-) disease without overt lymph-node metastasis (pt1c-2n0-1mi). 10 Since all insurance companies fully reimburse GEP use in the Netherlands, and healthcare insurance is mandatory, GEPs are available for every Dutch breast cancer patient. Within the guideline directed indicated area an increase in GEP use over recent years and high adherence rates to the GEP test-result were observed. 11 An unexpected observation in a previous population-based study was the frequent use of GEPs outside the guideline-intended indicated area, i.e. in patients in whom clinical guidelines state a clear recommendation to administer or withhold CT based on clinicopathological factors alone. 12 GEP use in this patient group raises the question whether the GEP test results influenced CT administration in these patients. Chapter 8 The aim of the current study is to evaluate the clinical implications (CT administration) of GEP use (MammaPrint 70-gene signature) and GEP test results when used outside the guideline intended GEP indication area. In this group, clinical risk estimation and the GEP test-result was compared and adherence rates to the test-result were determined in case of discordance between the clinical and genomic risk assessment. 159

160 Chapter 8 MATERIAL AND METHODS Data source Data was derived from the Netherlands Cancer Registry (NCR) database. Since 1989, the NCR registers data on patient-, tumour-, diagnostic- and treatment characteristics of all Dutch cancer patients, obtained by data-managers directly from patient records. All surgically treated female patients diagnosed with primary non-metastatic invasive breast cancer between January 1st 2011 and December 31st 2014 were identified. Study population Patients with a prior history of malignancy or initially treated with CT or endocrine therapy prior to surgical treatment were excluded from the analysis. Patients >70 years of age were excluded since guidelines are inconclusive about the benefit of adjuvant CT advice in these patients. For the present study, patients were excluded for whom the current guideline advises to use a GEP as an adjunct to clinicopathological factors to guide adjuvant CT decision-making, i.e. patients with ER positive /HER2-Neu negative (-) disease without overt lymph-node metastasis (pt1c-2n0-1mi). The 70- GS is accountable for 97% of all deployed GEPs in the Netherlands, and we therefore decided to focus on the MammaPrint 70-gene signature only. Patients for whom the current Dutch treatment guidelines states a clear advice to administer or withhold CT, so without an indication to perform a GEP, were included in the study. This includes patients 70 years of age, regarded as clinical low- risk, for which adjuvant CT is not recommended or high-risk based with recommendation to administer CT according to the Dutch breast cancer treatment guideline (Supplementary Table 1). 13 Statistical Analyses Clinical low- and high-risk group were identified and further classified into different subcategories according to the Dutch guidelines based on grade, tumor size and lymphnode involvement. For both the clinical low- and high-risk group patient and tumor characteristics as well as hospital type (district, teaching and university) were compared between patients who did and did not received GEP testing by chi-square tests and an independent t-test for the normally distributed continuous variables age and size. Proportions of patients receiving a GEP in relation to the frequencies of the listed low- and high risk categories are summarized and listed with the respective GEP test results and proportions of patients receiving adjuvant CT. Implications of GEP use, in terms of discordance between the clinical and genomic risk estimate and adherence to the test-result reflected in adjuvant CT administration were evaluated in both the clinical low- and high-risk patients and 160

161 Impact of gene-expression profiling when applied outside the guideline directed indication area the various subcategories. Subsequently, logistic regression analysis was performed to assess if GEP use was independently associated with the administration of adjuvant CT in clinical low- or high-risk patients after correction for confounders. The same approach was used to assess whether the GEP test result was independently associated with CT administration in clinical low- or high-risk patients who received GEP testing. Results are presented as Odds Ratio s (OR) and 95% confidence intervals (95% CI). A p-value of <0.05 was considered to be statistically significant. All statistical analyses were performed in STATA (version , Texas). RESULTS Study population A total of 26,425 patients were identified in the NCR database: 35.4% of these patients were considered as clinical low-risk and 64.6% as clinical high-risk according to the guideline (figure 1). Overall, 3.9% patients in the clinical low-risk group received CT and 79.7% of clinical high-risk patients. A total of 1,725 GEPs (6.5%) were deployed in the study-population: in 4.8% (n=444) of the patients in the clinical low-risk group and in 7.5% (n=1,281) of patients in the clinical high-risk group received a GEP. Overall, 68.5% of patients with a discordant clinical and genomic risk estimation were treated in line with the GEP test result. GEP use in clinical low-risk patients GEPs assigned 20.3% of the clinical low-risk patients to a high genomic risk category. Clinical low-risk patients who received a GEP were younger, more often had ER+/ HER2- tumors of limited size without axillary lymph-node involvement as compared to patients who did not receive a GEP. Furthermore, GEPs were more often deployed in patients treated in teaching hospitals (Table 1). GEP use was highest (32.2%) in the clinical low-risk patients <35 years of age with HER2-negative, grade 1 tumours 1cm without axillary lymph-node involvement (group 1, Supplementary Table 2). Chapter 8 161

162 Chapter 8 Figure 1. Flowchart describing discordance between the clinical and genomic risk estimation and adherence to the genomic test-result reflected in adjuvant CT administration. 162

163 Impact of gene-expression profiling when applied outside the guideline directed indication area Overall, 12.6% of clinical low-risk patients in whom a GEP was deployed received CT compared to 3.4% who did not receive GEP testing (p<0.05) (Table 1). GEP use was independently associated with an increased risk of receiving CT in clinical low-risk patients on multivariate logistic regression analysis (OR=2.12, 95%CI: , data not shown). The presence of axillary micro-metastases was the only clinicopathological factor that remained independently associated with CT administration in clinical lowrisk patients who received GEP testing (pnmi vs. pn0, OR= %CI: , Table 2). In the subset of clinical low risk patients with discordance between clinical and genomic risk assessment (n=90; i.e. the GEP assigned patients to the high-risk category) CT was administered in 52.2% of patients (Figure 1). GEP use in clinical high-risk patients The GEP assigned 449 patients to a low genomic risk category (35%). Clinical highrisk patients who received a GEP were slightly older than clinical high-risk patients who did not and more often had ER+/Her2- tumors <3 cm without axillary node involvement (Table 1). Overall, 80.8% of clinical high-risk patients in whom no GEP was used received CT compared to 65.3% of patients in whom a GEP was deployed (p<0.001, Table 1 ). GEP use in clinical high-risk patients remained independently associated with a decreased risk of CT administration in multivariate logistic regression analysis (OR= %CI: , Data not shown). Chapter 8 163

164 Chapter 8 Table 1. Patient and tumour characteristics as well as hospital type were compared between patients who did and did not received GEP testing for both the clinical low- and high-risk group. Clinical low risk (n=9,354) Clinical high risk (n=17,071) No GEP received (n=8,910) GEP received (n=444) p-value* No GEP received (n=15,790) GEP received (n=1,281) p-value* Age in years (mean, SD) < % 5 1.1% % % , % % 5, % % , % % <0,05 9, % % <0,05 Tumour size in mm (mean, SD) <10 6, % % % % , % % 6, % % % 0 0.0% 5, % % > % 0 0.0% 2, % % Unknown 0 0.0% 0 0.0% <0, % % <0,05 Estrogen receptor ER % % 3, % % ER+ 8, % % 11, % 1, % Unknown % 1 0.2% <0, % 1 0.1% <0,05 Progesterone receptor PR- 1, % % 5, % % PR+ 6, % % 9, % % Unknown % 1 0.2% <0, % 2 0.2% <0,05 164

165 Impact of gene-expression profiling when applied outside the guideline directed indication area Her2-Neu status Her2-8, % % 11, % 1, % Her % 6 1.4% 3, % % Unknown % 4 0.9% <0, % % <0,05 Axillary lymph node status Negative 8, % % 7, % % N1Mi** % % 1, % % N % 0 0.0% 5, % % N % 0 0.0% 1, % % N % 0 0.0% % % Unknown % 3 0.7% <0, % % <0,05 Grade 1 5, % % 1, % % 2 2, % % 6, % % % % 7, % % Unknown % 0 0.0% <0, % 9 0.7% 0.05 Multifocality No 7, % % 12, % 1, % Yes % % 2, % % Unknown % 0 0.0% % 1 0.1% <0,05 Hospital of surgery District 2, % % 5, % % Teaching 5, % % 9, % % University % % <0,05 1, % % <0,05 Chemo/Targeted therapy No 8, % % 3, % % Yes % % <0,05 12, % % <0,05 * Chi-square test Chapter 8 165

166 Chapter 8 Table 2. Association between GEP test result and the administration of adjuvant CT in clinical low- risk patients. No CT (n=307) CT (n=52) Univariate Multivariate n % n % OR 95% CI OR 95% CI GEP result Low Risk % 5 9.6% ref ref High Risk % % * Age in years < % 2 3.8% ref ref % % % % Tumour size in mm < % % ref ref % % Estrogene receptor ER % % ref ref ER % % Unknown 0 0.0% 0 0.0% Progesterone receptor PR % % ref ref PR % % Unknown 0 0.0% 0 0.0% Her2 Neu Her % % ref ref Her % 1 1.9%

167 Impact of gene-expression profiling when applied outside the guideline directed indication area Unknown 3 1.0% 0 0.0% omitted Node state Negative % % ref ref N1Mi % % * Unknown 2 0.7% 0 0.0% omitted Grade % % ref ref % % % % Unknown 0 0.0% 0 0.0% omitted Multifocality No % % ref ref Yes % % Unknown 0 0.0% 0 0.0% Hospital of surgery District % % ref ref Teaching % % University % 4 7.7% * Significant OR Chapter 8 167

168 Chapter 8 Table 3. Association between GEP result and the administration of adjuvant CT in clinical high-risk patients. No CT (n=401) CT (n=650) Univariate Multivariate n % n % OR 95% CI OR 95% CI GEP result Low Risk % % ,04-0,08 0,05* 0,03-0,07 High Risk % % ref ref Age in years < % % ref ref % % ,19-1, ,07-1, % % ,09-0,78 0,12* 0,03-0,54 Tumour size in mm < % % ref ref % % ,26-4,47 2,82* 1,11-7, % % ,64-2,31 3,10* 1,17-8,25 > % % ,29-5,86 6,84* 2,21-21,23 Unknown 4 1.0% 5 0.8% ,32-5, ,73-37,74 Estrogen receptor ER % % ref ref ER % % ,09-0, ,25-1,16 Unknown 0 0.0% 1 0.2% omitted Progesterone receptor PR % % ref ref PR % % ,23-0, ,48-1,29 Unknown 0 0.0% 2 0.3% omitted Her2-Neu status Her % % ref ref Her % % ,78-4,25 3,30* 1,68-6,52 168

169 Impact of gene-expression profiling when applied outside the guideline directed indication area Unknown 8 2.0% 2 0.3% ,04-0, ,02-5,11 Axillary lymph-node status Negative % % ref ref N1Mi % % ,48-1, ,95-3,60 N % % ,88-1,65 7,48* 4,27-13,13 N % % omitted omitted N % % ,03-61,72 28,01* 3,15-249,41 Unknown 3 0.7% 7 1.1% ,40-6, ,23-5,95 Grade % % ref ref % % ,87-2,10 2,19* 1,17-4, % % ,66-6,59 3,93* 1,94-7,96 Unknown 1 0.2% 7 1.1% ,09-77, ,24-36,72 Multifocality No % % ref ref Yes % % ,85-1,76 Unknown 0 0.0% 1 0.2% omitted Hospital of surgery District % % ref ref Teaching % % ,99-1,71 University % % ,35-1,33 * Significant OR Chapter 8 169

170 Chapter 8 In clinical high-risk patients who received a GEP, a low-risk GEP result was strongly associated with a decreased risk of CT administration (OR=0.05, 95%CI: ). In 71.7% (n = 322) of these discordant patients the administration of adjuvant CT was in line with the low-risk GEP test-result (i.e. no CT was administered, Figure 1). Young age, larger tumor size, higher grade, Her2+ disease and (micro-)metastatic lymph-node involvement remained independently associated with an increased risk of CT administration in these patients (Table 3). DISCUSSION While the Dutch guideline suggests the use of a validated GEP in ER+ early breast cancer patients, in whom benefit of CT is controversial based on traditional prognostic factors alone, 3,9 in the present population based study GEPs were used in 4.7% and 7.5% of patients who were considered as clinical low-risk and high-risk respectively. In these groups a discordance between the clinical and genomic risk-estimation was observed in 20.3% and 35.1% respectively. GEP use significantly influenced CT administration in these patients. To our knowledge this is the first report on the clinical impact of GEP use in patients in whom a GEP should be superfluous as the recommendation to administer or withhold CT is clear based on clinicopathological factors. The observed frequency of 6.5% in the present group is remarkable and compares to a 15% deployment of GEPs in the category of patients in whom GEPs were advocated. 10 The relatively high incidence of GEP use in the present study and the apparent impact of GEP use on CT administration in these patients, suggests limited support among clinicians and patients for the current clinical guideline recommendations. The mere frequency of unintended GEP use underscores that clinicians need reproducible and objective measures for the decision to administer CT. In both clinical low- as high-risk patients GEP use was more frequent in patients of younger age with ER+/Her2- intermediate grade tumors of limited size, indicating controversy regarding CT administration especially in these subgroups of patients. Patients with micro-metastatic axillary lymph-node involvement were more likely to receive a GEP in the clinical low-risk group while GEPs were deployed mere frequently deployed in node-negative patients in the clinical high-risk group. When a GEP was deployed we observed an overall discordance between clinical and genomic risk estimation in 31,3% of patients assigned to the clinical low- or highrisk category. One out of three clinical high-risk patients were assigned to the lowrisk category by GEPs which led to omission of CT, despite a guideline indication to administer CT, in approximately 72% of these patients. The results of the MINDACT trial support the omission of adjuvant CT in stage I-III early stage clinical high-risk 170

171 Impact of gene-expression profiling when applied outside the guideline directed indication area breast cancer patients with up to three axillary lymph-node metastasis when the GEP categorizes these patients as having a low genomic risk.. 14 On the other hand, in the MINDACT trial clinical utility of 70-gene signature use was not demonstrated for clinical low-risk patients as clinical low-risk patients assigned to the genomic high-risk profile who did not receive CT had similar 5-year disease free survival rates as patients who did receive CT. Therefore, the indication area for GEP use as stated in current clinical practice guidelines will probably be further broadened to clinical high-risk patients in coming years while its use will be discommended in clinical low-risk patients. Overall, 68.5% of patients with a discordant clinical and genomic risk estimation were treated in line with the GEP test result (52,2% in low en 71,7% in high). This is substantially lower as compared to patients within the guideline intended area for GEP use, in whom adherence rates to the GEP result of up to 89% have been reported. 10 This observation is on the one hand not surprising since the level of evidence for GEP use in clinical low- or high risk patients was modest during our study-period. On the other hand it remains strange that the test was deployed for some reason and subsequently not adhered to in 47,8% of patients with a low and 28.3% of patients with a high risk test result. This may be explained by deployment of a GEP on a patients request. On the other hand physicians may seek more support for the recommendation or avoidance of CT instead of being in true doubt when deploying a GEP in the guideline intended indication area. The results of the MINDACT trial will probably strengthen the motivation for GEP use in clinical high-risk patients, and may lead to a higher adherence to the low-risk GEP result. The observed higher adherence to the GEP result in clinical high risk patients assigned to the low-risk GEP category (71.7%) in comparison to clinical low risk patients assigned to the high GEP category (52.2%) is in line with previous studies which also report on GEPs being mainly used for a substantiated decision to withhold CT. The population-based character of the present study makes our work unique and enables us to provide a nation-wide overview of GEP use (MammaPrint 70-gene signature). Implications of GEP use in ER+/Her2- early stage breast cancer patients in whom controversy exists regarding CT benefit based on traditional prognostic factors alone are increasingly studied. Reports on implications of GEP testing at a nation-wide level or in patients outside this guideline intended indication area are scarce. The strength of the population based design is the weakness of the study as well. Although we assessed the association between GEP use and CT administration in multivariable logistic regression analysis correcting for all known clinicopathological characteristics, confounding by indication cannot be ruled out completely. Chapter 8 171

172 Chapter 8 CONCLUSION GEPs are relatively quite frequently used to aid adjuvant CT decision-making in patients with a clear clinical guideline recommendation to administer or withhold CT in the Netherlands. Although adherence to the test result is limited in the categories of patients who are considered as having a low- or high clinical risk of developing metastases, GEP use significantly influenced CT decision-making in these patients illustrating the clinicians need for reproducible and objective measures for the decision to administer CT. Supl. Table 1. Patients outside the guideline intended GEP indicated area: distinction between clinical low- or high-risk early stage breast cancer patients based on established prognostic factors according to the Dutch breast cancer treatment guideline. This guideline recommends to withhold chemotherapy in patients considered as clinical low-risk and administer chemotherapy in clinical high-risk patients. Clinical low-risk: Clinical high-risk*: No axillary lymph-node involvement (pn0 or pn1mi), > 35 years of age, Her2- disease and: Tumour size < 1 cm (Group 2) o ER+ All patients with metastatic axillary lymph-node involvement ( pn1a) (Group I) All patients <35 years of age with tumours > 1cm o ERo Grade III o Grade I/II Grade I and tumour size 1-2 cm (Group 3) All patients with Her2+ disease (except for tumours cm) Or: Patients without metastatic axillary lymph-node involvement (pn0 or pn1mi) and: <35 years of age, grade I, 0-1 cm (Group I) Tumour size > 2 cm Her2+, tumour size cm (Group 4) Grade II or III tumours > 1 cm *Note: adjuvant chemotherapy is only recommended in clinical high-risk patients <70 years of age 172

173 Impact of gene-expression profiling when applied outside the guideline directed indication area Supl. Table 2. Proportions of patients receiving a GEP in relation to the frequencies of the listed low- and high risk categories are summarized and listed with the respective GEP test results and proportions of patients receiving adjuvant CT. Clinical Low risk (n=9,354) Total Gep test GEP result Adjuvant CT administration n n % Low High Risk Unknown Patients with no GEP GEP Low Risk Risk GEP high risk Group 1 n % n % n % n % n % n % p1n0, MS or 1M, age <35, % % % 0 0.0% 1 9.1% 1 25% % grade 1, size <11 mm and her2 negative disease Group 2 p1n0, MS or 1M, age >34, % % % % % 1 8% % grade 1, 2 or 3, size <11 mm and her2 negative disease Group 3 p1n0, MS or 1M, age >35, grade 1, size <21 mm and her2 negative disease Group 4 p1n0, MS or 1M, age >35, grade 1 or 2, size <11 mm and her2 negative disease Group 5 p1n0, MS or 1M, >34, size <6 mm and her2 disease Group 6 p1n0, MS or 1M, age <35, grade 1, size <11 mm % % % % % 2 1% % % % % % % 1 1% % % % % 3 3.0% % 0 0% % % 0 0.0% % % % Chapter 8 173

174 Chapter 8 Clinical High risk (n=17,071) Total Gep test GEP result Adjuvant CT administration n n % Low High Risk Unknown Patients with no GEP GEP Low Risk Risk GEP high risk Group 1 n % n % n % n % n % n % All patients with metastatic % % % % % % % axillary lymph-node involvement ( pn1a) Group 2 p1n0, MS or 1M, grade I or % % % % % % % II, size > 10 mm Group3 p1n0, MS or 1M, age >35, grade 1, size > 20mm Group 4 p1n0, MS or 1M, size > 0.5 mm and her2 positive disease Group 5 p1n0, MS or 1M, < 35 years % % % % % 0 0.0% % % % % % % % % % % % % % % % 174

175 Impact of gene-expression profiling when applied outside the guideline directed indication area RERERENCE LIST 1. Early Breast Cancer Trialists Collaborative, G., Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet, (9472): p Goldhirsch, A., et al., Thresholds for therapies: highlights of the St Gallen International Expert Consensus on the primary therapy of early breast cancer Ann Oncol, (8): p Paik, S., et al., Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol, (23): p Beisecker, A., et al., Side effects of adjuvant chemotherapy: perceptions of node-negative breast cancer patients. Psychooncology, (2): p Senkus, E., et al., Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol, Suppl 6: p. vi Goldhirsch, A., et al., Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol, (8): p van de Vijver, M.J., et al., A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med, (25): p Filipits, M., et al., A new molecular predictor of distant recurrence in ER-positive, HER2-negative breast cancer adds independent information to conventional clinical risk factors. Clin Cancer Res, (18): p Kwaliteitsinstituut voor de gezondheidszorg CBO VvIK, in Risicoprofilering. Richtlijn mammacarcinoom p Kuijer, A., et al., Using a gene expression signature when controversy exists regarding the indication for adjuvant systemic treatment reduces the proportion of patients receiving adjuvant chemotherapy: a nationwide study. Genet Med, (7): p Kuijer, A., et al., Changes over time in the impact of gene-expression profiles on the administration of adjuvant chemotherapy in estrogen receptor positive early stage breast cancer patients: A nationwide study. Int J Cancer, (4): p Kuijer, A., et al., Factors Associated with the Use of Gene Expression Profiles in Estrogen Receptor-Positive Early-Stage Breast Cancer Patients: A Nationwide Study. Public Health Genomics, (5): p Landelijke Richtijn Oncoline. 2012; 2.0:[Available from: php?pagina=/richtlijn/item/pagina.php&id=34737&richtlijn_id= Cardoso, F., et al., 70-Gene Signature as an Aid to Treatment Decisions in Early-Stage Breast Cancer. New England Journal of Medicine, (8): p Chapter 8 175

176

177 CHAPTER 9 Characterization of multifocal breast cancer using the 70-gene signature in clinical low-risk patients enrolled in the EORTC 10041/BIG MINDACT trial. *A. Kuijer, *K.C. Aalders, M.E. Straver, L. Slaets, S. Litiere, G. Viale, L.J. van t Veer, A.M. Glas, M. Delorenzi, T. van Dalen, K. Tryfonidis, M.J. Piccart, F. Cardoso, E.J. Rutgers and on behalf of the TRANSBIG Consortium & the MINDACT Investigators. * Co-authorship, both authors contributed equal to this manuscript. European Journal of Cancer 2017;79:

178 Chapter 9 ABSTRACT Background: In multifocal (MF) breast cancer, guidelines recommend basing adjuvant systemic treatment decisions on characteristics of the largest lesion, disregarding multifocality as an independent prognosticator. We assessed the association between MF disease and both the 70-gene signature (70-GS), and distant metastasis-free survival (DMFS) in clinical low-risk breast cancer patients enrolled in the EORTC 10041/BIG MINDACT trial. Patients and methods: The analysed population consisted of enrolled patients in the MINDACT trial with clinical low-risk disease, defined by a modified Adjuvant!Online cut-off for the 10-year risk of recurrent disease or death. Eligibility criteria of MINDACT dictate that patients with MF disease could be included if the different lesions had similar pathological characteristics. The presence of MF disease was deducted from the CRF-question for sum of diameter for all invasive tumor foci. Clinicopathological characteristics and gene expression of patients with unifocal and MF (largest lesion) disease were compared. Subsequently, the association between MF disease and the 70-GS was evaluated as well as the association between multifocality and 5-year DMFS. Results: The study included 3,090 clinical low-risk patients with unifocal and 238 patients with MF disease. Apart from a higher prevalence of lobular tumours (21.8% vs. 10.8%, by local pathology), we did not observe differences in baseline characteristics between MF and unifocal tumours. Patients with MF tumours were more likely to be at high genomic risk as compared to patients with unifocal tumours (22.7% vs. 17.3%, OR 1.45, 95% CI , p=0.038). We did not find a significant association between tumour focality and DMFS (97.1% for unifocal vs. 96.9% for multifocal, HR=1.55, 95% CI , P=0.172), nor a signal for a potential interaction between the prognostic effect of the 70-GS and focality of the tumour regarding DMFS. Conclusion: In the group of clinical low-risk MINDACT patients, MF tumours were more likely to have a high-risk 70-GS profile compared to unifocal tumours. We did not observe a significant interaction between multifocality and the 70-GS with respect to survival without distant metastasis in these patients. 178

179 Characterization of multifocal breast cancer using the 70-gene signature INTRODUCTION Multifocal breast cancer, generally defined as the presence of multiple invasive tumour foci in the same quadrant of the breast, has a wide-ranged incidence varying from 6%- 77%, depending on the definition and method of diagnosis 1 3. Multifocal disease is more often seen in lobular carcinomas and has been associated with an increased incidence of lymph-node involvement, poor differentiation grade, HER2 positivity and lymphovascular invasion as compared to unifocal tumours 2,4 8. Due to improvements in diagnostic imaging and increased use of MRI, multifocal disease is diagnosed more often 2,9. Nevertheless, the prognostic relevance of multifocality remains largely unclear 1. Current guidelines recommend basing adjuvant systemic treatment (AST) decisions on pathological characteristics of the largest lesion, thus assuming that outcome in multifocal disease depends entirely on the prognostic features of this lesion and the extent of lymph-node involvement This approach might result in omission of AST in patients who are regarded as low-risk for disease recurrence based on clinicopathological assessment of their largest lesion. Furthermore, as multifocality has been suggested to be a sign of high tumour burden, which in turn has been associated with a greater tendency to metastasize, disregarding multifocality as an independent prognosticator may result in under treatment 4,13. Over the last few years, several gene expression profiles have been developed to better predict clinical outcome compared to standard assessment based on clinicopathological characteristics 14. The prospective MINDACT study showed that the 70-gene signature/ MammaPrint (70-GS) could accurately differentiate between patients with a low and high risk of distant metastases and death at 5 years, thereby providing valuable information for determining the potential benefit of adjuvant chemotherapy 15. Chapter 9 The aim of the current study was to assess whether multifocal disease is associated with an increased rate of having a high genomic risk as assessed by the 70-GS in clinical lowrisk patients enrolled in the EORTC 10041/BIG MINDACT trial. In addition, we evaluated the association between tumour focality, the 70-GS result and distant metastasis-free survival (DMFS) to determine whether multifocality in clinical low-risk patients would be an argument for performing the 70-GS to better select patients for systemic treatment. METHODS Study design and eligible patients The EORTC 10041/BIG MINDACT 15 trial (NCT ) enrolled women aged diagnosed with histologically proven unilateral primary early-stage (ct1-2 or operable T3) breast cancer with 0-3 positive lymph nodes, that had their risk 179

180 Chapter 9 of distant disease recurrence assessed by both the 70-GS (Genomic) and a modified version of Adjuvant!Online (Clinical) 16. Patients with C-High/G-High risk assessment received adjuvant chemotherapy (CT), while those with a C-Low/G-low risk profile did not. Patients with discordant results for the two risk assessments were randomized to follow either the genomic or clinical risk for the decision regarding chemotherapy administration. For this study, only patients with clinical low-risk disease and a known focality status were included. Clinical low-risk, as per the modified Adjuvant!Online, was defined as a 10-year breast cancer survival probability of >88% for ER+ disease without systemic therapy, and >92% for ER- breast cancer accounting for an average 4% absolute benefit of adjuvant endocrine treatment in ER+ disease 15. Patients with multifocal disease were eligible to be included in the MINDACT trial if the different tumour foci were of similar histopathology (histological subtype, grade, ER, PgR and HER2 status). The genomic risk assessment had to be performed on the largest lesion. To select our population, the presence of multifocal disease was deduced from the mandatory baseline form CRF-question for sum of diameter for all invasive tumour foci, a question that only needed to be answered in the presence of multifocal disease (Appendix A). In the case of multifocal breast cancer, clinicopathologic and genomic characteristics of the largest lesion were considered for analysis. Whenever the clinical or genomic risk changed after enrolment, e.g. due to incorrect reporting of LN status or logistical errors, we used the corrected risk status 15. Objectives and end-points The primary objective of this substudy was to evaluate the association between multifocality and genomic risk result (70-gene signature) in clinical low-risk patients. Secondary objectives included 1) assessment of the association between routine clinicopathological characteristics (including age, stage, grade, ER, PgR, HER2, histology) and focality of the tumour in clinical low risk patients, 2) evaluating the association between multifocal disease and 5-year DMFS, within the group of clinical low risk patients, and 3) to study a potential interaction between multifocality and 70-gene risk in relation to outcome (DMFS) in clinical low risk patients. DMFS was defined as the time until first distant metastatic recurrence or death from any cause, whichever occurred first. Patients without a DMFS event at cut-off date were censored at the date of last disease assessment. Statistical analysis We hypothesized that in clinical low risk patients with multifocal disease the percentage 180

181 Characterization of multifocal breast cancer using the 70-gene signature of patients at high genomic risk according to the 70-GS would be higher than in patients with unifocal breast cancer; 15% vs. 7% 17. This would correspond to a relative risk increase of 114% and an absolute risk increase of 8%. With 3088 clinical low risk unifocal tumours and 238 clinical low risk multifocal tumours in the MINDACT population, there would be a 97% power to detect the hypothesized association at a significance level (alpha) of 5%. The association between multifocality and genomic risk was evaluated using Fisher s exact test. This association was further examined using multivariate logistic regression adjusting for age, pathological tumour and nodal status, grade, hormone receptor (ER and PgR) and HER2 status, and histology as per local assessment. The distribution of baseline patient and tumour characteristics were compared by tumour focality and presented in percentages. Subsequently, the association between multifocality and DMFS was evaluated using multivariate Cox regression adjusting for the abovementioned clinicopathological factors as well as adjuvant chemotherapy and endocrine treatment. Patients with missing information for (one of) the considered variables were excluded from this analysis (n=82). Following the primary results of MINDACT 15, the 70-GS was not included as a factor in the main model but a sensitivity analysis was conducted which adjusted for the prognostic effect of the 70-gene risk (as a potential confounder). Finally, Cox regression analyses were performed to assess a potential interaction between multifocality and 70-GS result in relation to outcome (DMFS). All analyses were performed using SAS software, version 9.4 (SAS Institute). A significance level of 5% was considered for all analyses. RESULTS Patient population Out of the 6,693 patients enrolled in MINDACT, a total of 3,328 patients had a clinical low risk and known focality status and were included in this study. We excluded 9 patients with missing information on the focality of the tumour. Of the included patients, 238 (7%) were registered as having multifocal and 3,090 (93%) unifocal breast cancer. Baseline clinicopathological characteristics of patients are presented in Table 1. Apart from a higher incidence of lobular tumours (21.8% for multifocal vs. 10.8% for unifocal tumours), we did not observe any differences in baseline characteristics between multifocal and unifocal breast tumours. The vast majority of patients (94%) had node-negative (micrometastases of 0.2-2mm were considered as pn+ and isolated tumour cells as pn0) and hormone receptor positive disease (98% ER+), while 5% were HER2+. Patients with multifocal disease were less often treated with breast-conserving surgery (51% vs. 91%). Chemotherapy was administered in 10% of patients with unifocal and 15% of patients with multifocal tumours. Chapter 9 181

182 Chapter 9 Primary endpoint: Multifocality and 70-gene signature There was a significant association between the 70-GS result and focality of the tumour in the group of clinical low-risk patients (P=0.043, Table 2). In patients with unifocal disease 17.3% were assigned to the genomic high-risk profile, while this was 22.7% in patients with multifocal tumours. This corresponds to an absolute increase by 5.4% and a relative increase of 31%, which is smaller than hypothesized. In multivariable regression analysis, adjusting for age, pathological tumour and nodal status, grade, ER status, PgR status, HER-2 status and histology, multifocality remained significantly associated with a high-risk 70-GS profile (OR 1.45, 95% CI , P=0.038). Table 1. Baseline patient and tumour (by local assessment of largest lesion) characteristics of included patients according to focality of the tumour. Unifocal (N=3,090) Multifocal (N=238) Total (N=3,328) N (%) N (%) N (%) Age (median with range) 56 (26-71) 53 (26-56 (26-71) 71) Age (categories) < (0.3) 1 (0.4) 11 (0.3) 30-< (4.0) 15 (6.3) 140 (4.2) 40-< (24.2) 73 (30.7) 820 (24.6) 50-<60 1,081 (35.0) 83 (34.9) 1,164 (35.0) 60 and more 1,127 (36.5) 66 (27.7) 1,193 (35.8) Pathological tumour size 1 cm 788 (25.5) 64 (26.9) 852 (25.6) 1 cm - 2 cm 2,179 (70.5) 164 (68.9) 2,343 (70.4) 2 cm - 3 cm 123 (4.0) 10 (4.2) 133 (4.0) Lymph node status Node negative 2,916 (94.4) 223 (93.7) 3,139 (94.3) 1 positive LN 133 (4.3) 7 (2.9) 140 (4.2) 2 positive LN 24 (0.8) 5 (2.1) 29 (0.9) 3 positive LN 17 (0.6) 3 (1.3) 20 (0.6) Tumour grade Well differentiated 1,250 (40.5) 81 (34.0) 1,331 (40.0) 182

183 Characterization of multifocal breast cancer using the 70-gene signature Unifocal (N=3,090) Multifocal (N=238) Total (N=3,328) N (%) N (%) N (%) Moderately differentiated 1,717 (55.6) 148 (62.2) 1,865 (56.0) Poorly differentiated or undifferentiated 111 (3.6) 8 (3.4) 119 (3.6) Undefined 12 (0.4) 1 (0.4) 13 (0.4) ER status Negative 58 (1.9) 7 (2.9) 65 (2.0) Positive 3,032 (98.1) 231 (97.1) 3,263 (98.0) PgR status Negative 417 (13.5) 30 (12.6) 447 (13.4) Positive 2,647 (85.7) 208 (87.4) 2,855 (85.8) Missing 26 (0.8) 0 (0.0) 26 (0.8) HER2 status Negative 2,935 (95.0) 224 (94.1) 3,159 (94.9) Positive 148 (4.8) 13 (5.5) 161 (4.8) Missing 7 (0.2) 1 (0.4) 10 (0.2) Tumour histology Ductal 2,546 (82.4) 165 (69.3) 2,711 (81.5) Lobular 333 (10.8) 52 (21.8) 385 (11.6) Mixed 93 (3.0) 15 (6.3) 108 (3.2) Other 115 (3.7) 6 (2.5) 121 (3.6) Missing 3 (0.1) 0 (0.0) 3 (0.1) Local treatment BCS alone 16 (0.5) 2 (0.8) 18 (0.5) BCS + radiotherapy 2,754 (89.1) 119 (50.0) 2,873 (86.3) BCS radiotherapy unknown 34 (1.1) 1 (0.4) 35 (1.1) Mastectomy 255 (8.3) 84 (35.3) 339 (10.2) Mastectomy + radiotherapy 23 (0.7) 29 (12.2) 52 (1.6) Mastectomy radiotherapy unknown 8 (0.3) 3 (1.3) 11 (0.3) Chemotherapy Yes 317 (10.3) 35 (14.7) 352 (10.6) No 2,767 (89.5) 202 (84.9) 2,969 (89.2) Missing 6 (0.2) 1 (0.4) 7 (0.2) Endocrine treatment Chapter 9 183

184 Chapter 9 Unifocal (N=3,090) Multifocal (N=238) Total (N=3,328) N (%) N (%) N (%) Yes 2,461 (79.6) 191 (80.3) 2,652 (79.7) No 573 (18.5) 43 (18.1) 616 (18.5) Missing 56 (1.8) 4 (1.7) 60 (1.8) Trastuzumab treatment Yes 67 (2.2) 6 (2.5) 62 (1.9) No 2,967 (96.0) 226 (95.0) 3,193 (95.9) Missing 56 (1.8) 6 (2.5) 62 (1.9) Percentages may not add up to 100% due to rounding Table 2. Association between focality and genomic risk as assessed by the 70-GS Unifocal (N=3090) Multifocal (N=238) Total (N=3328) N (%) N (%) N (%) p-value* Genomic risk (corrected) Low risk 2,554 (82.7) 184 (77.3) 2,738 (82.3) High risk 536 (17.3) 54 (22.7) 590 (17.7) *Fisher exact test for association 184

185 Characterization of multifocal breast cancer using the 70-gene signature Secondary endpoints: Outcome The 5-year DMFS rate was 97.1% (95% CI ) for patients with unifocal and 96.9% (95% CI ) for patients with multifocal tumours (Figure 1). We did not find a significant association between tumour focality and DMFS (HR=1.55, 95% CI , P=0.172). A similar result was found in the sensitivity analysis adjusting for the result of the 70-GS (HR 1.49, 95% CI , P=0.217). Additionally, the Cox regression model did not demonstrate a signal for a potential interaction between the prognostic effect of the 70-GS and tumour focality regarding DMFS (P=0.411; Figure 2). We did observe a prognostic effect of the 70-GS on DMFS in both unifocal (HR= 1.85, 95% CI ) and multifocal tumours (HR 3.14, 95% CI ), although with 11 events in 238 patients the latter trend was not powered to be conclusive (Figure 2). Figure 1. Distant metastasis-free survival according to focality of the tumour Chapter 9 185

186 Chapter 9 DISCUSSION In this study, multifocal disease was independently associated with a high genomic risk according to the 70-GS in clinical low-risk patients, albeit smaller than hypothesized. This could have been a reflection of possible limitations in the current staging strategy for multifocal breast cancer, as current guidelines do not take into account the higher tumour burden that is generally attributed to multifocal tumours 4,11 This study did not demonstrate a significant interaction between the prognostic effect of the 70-GS and multifocality with respect to patients outcome (DMFS). These results appear to be in accordance with the primary results of the MINDACT trial, indicating limited value of performing the 70-GS in clinical low-risk patients 15. It should however be noted that the interaction analysis of the present study was not adequately powered to answer this question. The prognostic value of the 70-GS signature in multifocal breast cancer will need to be confirmed with more follow-up data. Figure 2. Forest plot exploring the prognostic effect of the 70-gene signature on distant metastasis-free survival by focality of the tumour 186

187 Characterization of multifocal breast cancer using the 70-gene signature To our knowledge, this is the first study evaluating the association between tumour focality and gene expression in a large population of early-stage breast cancer patients. We evaluated theclinical low-risk patients in order to determine whether performing the 70-GS in these patients would improve risk assessment and therefore could have clinical implications. In clinical high-risk patients the decision to give AST would already have been made based on the clinicopathological assessment of largest lesion so the presence of multifocality will not impact the AST decision. Heterogeneity, not only inter- but also intra-tumoral, is one of the hallmarks of cancer, which complicates AST decisions 18. It is generally believed that multifocal disease arises from one type of cancer cell resulting in different lesions with largely identical phenotypes, though evolution can occur during proliferation 3,19,20. Previous research has shown multiple lesions in multifocal disease are largely concordant with respect to hormone receptor status 19,21, suggesting that consideration of only one lesion to determine hormonal treatment would be safe. However, there were differences demonstrated between different foci regarding histologic tumour type, differentiation grade, HER2 status and p53 expression[24, 25] while these characteristics are important when deciding on adjuvant chemotherapy. This heterogeneity in the case of multifocal disease could indicate different foci may display different genomic risks, raising the question whether multiple lesions should be assessed. We observed no differences in clinicopathological characteristics between patients with multifocal and unifocal disease apart from the expected higher incidence of lobular tumours in the multifocal group. This is likely the consequence of only selecting the clinical low-risk patients, as patients with multifocal disease and poor clinical prognostic features, occurring more frequently in case of multifocality, would have been classified as clinical high-risk. Chapter 9 There was a significant association between the 70-GS result and focality of the tumour in the group of clinical low-risk patients. Overall, the outcome of this clinical low-risk population was excellent (5-year DMFS 97%) as was the case for the whole MINDACT population 15 making it difficult to identify clinical and statistically significant differences between groups. In this study, multifocality was not an independent prognosticator for DMFS in this clinical low-risk population. This is in accordance with previous reports, although studies on the association between multifocality and outcome have contradictory results 1,3, Weissenbacher et al. 13 performed a matched-pair analysis comparing patients with unifocal (n=288) and multifocal/multicentric (n=288) disease, demonstrating a significant increase in the occurrence of distant metastasis in the latter group (21.2% vs. 12.5%, P=.004). Neri et al. 28 confirmed these results in 131 patients, 187

188 Chapter 9 also reporting that administration of adjuvant anthracyclines appeared to reduce this difference. In the largest series to date, including 1,554 patients with multifocal breast cancer, multifocality was associated with a decreased breast cancer specific survival yet without an influence on overall survival 29. The incidence of multifocal disease in this study was 7%, which is relatively low when compared other reports 1 3. The rate was the same in the overall MINDACT population with 497 out of 6,693 patients reported as multifocal (n=28 with unknown focality status). The relatively low rate of multifocality in this study is likely the result of the eligibility criteria of the MINDACT trial. The MINDACT protocol dictated only patients with multifocal disease whose separate foci displayed similar histolopathology could be included, meaning the subgroup of patients with more heterogeneity was missing. This could have led to an underestimation of the reported association between multifocal disease and DMFS ( confounding by indication ). Furthermore, in MINDACT, over 80% of patients underwent breast-conserving surgery which could have resulted in a lower detection rate of multifocality. Finally, we have no information about the use of MRI in the diagnostic process which might have impacted the detection rate. A limitation of this study is the deduction of the presence of multifocal disease from the CRF-question for sum of diameter for all invasive tumour foci (Appendix A). Multifocal breast cancer is generally defined as the presence of more than one invasive tumour lesion in one quadrant of the breast. However, for this study we did not possess detailed information on location of the various lesions. This means both multifocal and multicentric (multiple lesions in different quadrants of the breast) disease could be included in the multifocal population evaluated in this study. Then again, this distinction is increasingly considered arbitrary as is reflected by the current TNM staging manual establishing the term multiple cancers 11. Furthermore, we only had access to the clinicopathologic characteristics and 70-GS result of the largest lesion in the case of multifocal disease. In summary, while multifocality proved to be associated with an increased incidence of a high genomic risk as per the 70-GS, this study did not demonstrate a significant interaction between multifocality and the 70-GS with respect to survival without distant metastasis in patients regarded as clinical low-risk. ACKNOWLEDGEMENTS & FUNDING This trial has received grants from the European Commission Framework Programme VI (FP6-LSHC-CT ), the Breast Cancer Research Foundation, Novartis, F. Hoffman La Roche, Sanofi-Aventis, the National Cancer Institute (NCI), the EBCC- 188

189 Characterization of multifocal breast cancer using the 70-gene signature Breast Cancer Working Group (BCWG grant for the MINDACT biobank), the Jacqueline Seroussi Memorial Foundation (2006 JSMF award), Prix Mois du Cancer du Sein (2004 award), Susan G. Komen for the Cure (SG ), Fondation Belge Contre le Cancer (SCIE ), Dutch Cancer Society (KWF), Association Le Cancer du Sein, Parlons-en!, Deutsche Krebshilfe, the Grant Simpson Trust and Cancer Research UK. This trial was also supported by the EORTC Cancer Research Fund. Whole genome analysis was provided in kind by Agendia. We are grateful to all women participating in this study; all the investigators, surgeons, pathologists, and research nurses; the National Coordinating Centers/BIG Groups (BOOG, GOIRC, NCRI-BCG, SOLTI, UNICANCER, WSG); World Courier; the pharma companies Novartis, Roche, and Sanofi- Aventis; and Agendia. DISCLOSURES Ownership: LV is a founder of Agendia and has stock ownership. Corporate-sponsored Research: KA, LS and KT are employees of the EORTC which receives funding for the study. Other Substantive Relationships: AG is an employee of Agendia; GV has received consultancy fees/honorarium from Agendia; MP has received consultancy fees from Sanofi-Aventis, Novartis and Roche, honorarium from Roche and research grant from Roche and Novartis; FC has received consultancy fees/honorarium from Novartis, Sanofi-Aventis and Roche. Chapter 9 189

190 Chapter 9 REFERENCE LIST 1. Bendifallah S, Werkoff G, Borie-Moutafoff C, et al. Multiple synchronous (multifocal and multicentric) breast cancer: clinical implications. Surg Oncol. 2010;19(4):e115-e123. doi: /j.suronc Lynch SP, Lei X, Chavez-MacGregor M, et al. Multifocality and multicentricity in breast cancer and survival outcomes. Ann Oncol. 2012;23(12): doi: / annonc/mds Salgado R, Aftimos P, Sotiriou C, Desmedt C. Evolving paradigms in multifocal breast cancer. Semin Cancer Biol. 2015;31: doi: /j.semcancer Coombs NJ, Boyages J. Multifocal and multicentric breast cancer: Does each focus matter? J Clin Oncol. 2005;23(30): doi: /jco Lynch SP, Lei X, Meric-Bernstam F, et al. Breast cancer multifocality and multicentricity and locoregional recurrence. Oncologist. 2013;18(11): Donker M, Straver ME, van Tienhoven G, et al. Comparison of the sentinel node procedure between patients with multifocal and unifocal breast cancer in the EORTC AMAROS Trial: identification rate and nodal outcome. Eur J Cancer. 2013;49(9): doi: /j.ejca Mann RM, Hoogeveen YL, Blickman JG, Boetes C. MRI compared to conventional diagnostic work-up in the detection and evaluation of invasive lobular carcinoma of the breast: a review of existing literature. Breast Cancer Res Treat. 2007;107(1):1-14. doi: /s Li CI, Anderson BO, Daling JR, Moe RE. Trends in Incidence Rates of Invasive Lobular and Ductal Breast Carcinoma. 2003;289(11): Houssami N, Ciatto S, Macaskill P, et al. Accuracy and surgical impact of magnetic resonance imaging in breast cancer staging: Systematic review and meta-analysis in detection of multifocal and multicentric cancer. J Clin Oncol. 2008;26(19): doi: /jco Senkus E, Kyriakides S, Ohno S, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26(Supplement 5):v8-v30. doi: /annonc/mdv Edge S, Byrd D, Compton C, et al. AJCC Cancer Staging Manual (7th Ed). New York: Springer; Lester S, Bose S, Chen Y, et al. Protocol for the Examination of Specimens from Patients with Invasive Carcinoma of the Breast.; UCMCon/Contribution Folders/WebContent/pdf/cp-breast-invasive-16protocol pdf. 190

191 Characterization of multifocal breast cancer using the 70-gene signature 13. Weissenbacher TM, Zschage M, Janni W, et al. Multicentric and multifocal versus unifocal breast cancer: Is the tumor-node-metastasis classification justified? Breast Cancer Res Treat. 2010;122(1): doi: /s Sotiriou C, Pusztai L. Gene-expression signatures in breast cancer. N Engl J Med. 2009;360(8): doi: /nejmra Cardoso F, van t Veer LJ, Bogaerts J, et al. 70-Gene Signature as an Aid to Treatment Decisions in Early-Stage Breast Cancer. N Engl J Med. 2016;375(8): doi: / NEJMoa Rutgers E, Piccart-Gebhart MJ, Bogaerts J, et al. The EORTC 10041/BIG MINDACT trial is feasible: results of the pilot phase. Eur J Cancer. 2011;47(18): doi: /j.ejca Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective communitybased feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Zardavas D, Irrthum A, Swanton C, Piccart M. Clinical management of breast cancer heterogeneity. Nat Rev Clin Oncol. 2015;12(7):1-14. doi: /nrclinonc Choi Y, Kim EJ, Seol H, et al. The hormone receptor, human epidermal growth factor receptor 2, and molecular subtype status of individual tumor foci in multifocal/ multicentric invasive ductal carcinoma of breast. Hum Pathol. 2012;43(1): doi: 20. East EG, Pang JC, Kidwell KM, Jorns JM. Utility of Estrogen Receptor, Progesterone Receptor, and HER-2/ neu Analysis of Multiple Foci in Multifocal Ipsilateral Invasive Breast Carcinoma. Am J Clin Pathol. 2015;144(6): doi: / AJCPFWXP54OLILMU. 21. Garimella V, Long ED, O Kane SL, Drew PJ, Cawkwell L. Oestrogen and progesterone receptor status of individual foci in multifocal invasive ductal breast cancer. Acta Oncol. 2007;46(2): doi: / Wiechmann L, Sampson M, Stempel M, et al. Presenting features of breast cancer differ by molecular subtype. Ann Surg Oncol. 2009;16(10): doi: /s Buggi F, Folli S, Curcio a, et al. Multicentric/multifocal breast cancer with a single histotype: is the biological characterization of all individual foci justified? Ann Oncol. 2012;23(8): doi: /annonc/mdr Joergensen LE, Gunnarsdottir KA, Lanng C, Moeller S, Rasmussen BB. Multifocality as a prognostic factor in breast cancer patients registered in Danish Breast Cancer Cooperative Group (DBCG) Breast. 2008;17(6): doi: /j. breast Chapter 9 191

192 Chapter Pedersen L, Gunnarsdottir KA, Rasmussen BB, Moeller S, Lanng C. The prognostic influence of multifocality in breast cancer patients. Breast. 2004;13(3): doi: /j.breast Ustaalioglu BO, Bilici A, Kefeli U, et al. The importance of multifocal/multicentric tumor on the disease-free survival of breast cancer patients: single center experience. Am J Clin Oncol. 2012;35(6): doi: /coc.0b013e31822d9cd Chung AP, Huynh K, Kidner T, Mirzadehgan P, Sim M-S, Giuliano AE. Comparison of outcomes of breast conserving therapy in multifocal and unifocal invasive breast cancer. J Am Coll Surg. 2012;215(1): doi: /j.jamcollsurg Neri A, Marrelli D, Megha T, et al. Clinical significance of multifocal and multicentric breast cancers and choice of surgical treatment: a retrospective study on a series of 1158 cases. BMC Surg. 2015;15(1):1. doi: / Yerushalmi R, Kennecke H, Woods R, Olivotto IA, Speers C, Gelmon KA. Does multicentric/multifocal breast cancer differ from unifocal breast cancer? An analysis of survival and contralateral breast cancer incidence. Breast Cancer Res Treat. 2009;117(2): doi: /s

193 Characterization of multifocal breast cancer using the 70-gene signature Chapter 9 193

194

195 PART III Molecular subtypes in breast cancer

196

197 CHAPTER 10 Comparison of molecular subtyping by conventional local pathology assessment or by microarray analysis using an 80- gene signature in estrogen receptor positive early stage breast cancer patients. J. van Steenhoven, A. Kuijer, M.E. Straver, S.G. Elias, C.H. Smorenburg, J. Wesseling, S.C. Linn, E.J.Th. Rutgers, S. Siesling, T. van Dalen.

198 Chapter 10 ABSTRACT Background: gene expression profiling has become an accepted method to determine risk of developing metastases, but can also be used to determine intrinsic molecular breast cancer subtypes. The aim of the present study was to compare the classification of molecular tumor subtypes based on conventional pathology assessment to a classification based on an 80 gene-signature (80-GS). Patients and methods: in a prospective multicenter study into the effect of 70 genesignature (70-GS) use on chemotherapy decision making in clinical intermediate risk ER+ breast cancers, surrogate molecular subtypes (Luminal/HER2/Basal) were determined for all patients (n=595) by local pathology assessment using immunohistochemistry (IHC) and fluor in situ hybridization (FISH). For a subset of patients (n=48) Ki67 assessment was available and used to further stratify Luminal-type tumors into Luminal A and B. A 80-GS was performed to categorize cancers as luminal, HER2 or basal type and the 70-GS was used to distinguish between Luminal A and B type tumors. Concordance between molecular subtypes classified by IHC/FISH and the 80-GS was assessed. Results: between January and December , 595 patients, treated in 23 hospitals, were enrolled. Using local pathology as surrogate molecular subtyping; 98% of patients (n=578) were regarded as Luminal-type and 2% (n=12) as HER2- type. Comparison of surrogate molecular subtyping versus an 80-GS resulted in an overall concordance of 97%. In the subset of patients (n=48) in whom Ki67 was assessed 25% (n=12) and 75% (n=36) of patients were classified as Luminal A and Luminal B, respectively. Within this subgroup, 25% of patients (n=3) regarded as Luminal A by local pathology were reclassified as Luminal B by the 80-GS. Fiftyeight percent (n=21) of patients regarded as Luminal B by local pathology were reclassified as Luminal A by gene-expression subtyping (Kappa 0.11 [95%CI ]). Conclusion: in this group of predominantly ER+/HER2- Dutch breast cancer patients we observed high concordance between intrinsic molecular subtyping using conventional pathology assessment or gene-expression profiling. In the selection of patients in whom Ki67 expression was assessed, the sub stratification of Luminaltype tumors into A and B revealed high discordance between local pathology and microarray analysis. 198

199 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients INTRODUCTION Using hierarchical clustering, molecular portraits of human breast tumors have been identified according to their gene expression patterns and 4 distinct intrinsic molecular subtypes are discriminated: Luminal A (hormone receptor positive, low proliferation grade), Luminal B (hormone receptor positive, high proliferation grade), HER2 (HER2 receptor enriched) and Basal-type (hormone- and HER2 receptor negative) tumors. 1 While HER2- and Basal-like tumors are associated with relatively poor prognosis and require specific systemic treatment, Luminal type tumors benefit from endocrine therapy and have a more favorable outcome. 2-4 In patients with Luminal A tumors the value of chemotherapy over endocrine treatment is questionable. In routine practice, hormone receptors and HER2-status determined by immunohistochemistry (IHC) and fluorescent in situ hybridization (FISH) are used to determine surrogate intrinsic cancer subtypes. Further stratification of Luminal-type tumors into Luminal A and B is often based on the expression of the proliferation marker Ki67, 5 although there is concern regarding the accuracy of this method. 6,7 The variation in analytical assessment limits the role of Ki67 in routine clinical decisions making, since the cutoff values to designate Ki67 as high or low differ widely and the Ki67 level also varies within the tumor Gene-expression profiling has become an accepted means to stratify ER+ breast cancers into groups with a low or high risk of developing distant metastases Gene-expression testing can be used to determine breast cancer molecular subtypes too. Based on the expression of 80 genes a stratification of breast cancer tumors into three molecular subtypes (Luminal/HER2/Basal) can be made 14 and by adding the prognostic risk profile of the 70-GS a distinction between ER-receptor positive tumors into Luminal A or B cancers can be made. 2 In a prospective observational multicenter study in patients with intermediate risk ER+ breast cancers local pathology assessment was available as well as the gene-expression read-outs. The aim of the present study was to compare the classification of molecular tumor subtypes based on conventional pathology assessment to the 80-GS, Secondly, in a selection of patients in whom Ki67 levels were assessed, we compared Luminal A-type and Luminal B-type tumors using local pathology assessment versus the combined use of the 70-GS and the 80-GS. Chapter 10 PATIENT AND METHODS Patients As part of a prospective observational multicenter study regarding the influence of the 70-GS on adjuvant chemotherapy decisions in patients with clinical intermediate 199

200 Chapter 10 risk ER+ breast cancer data of conventional local pathology and microarray analysis data were obtained between January and December Within the trial (NCT ), patients suffering from ER+ invasive ductal breast cancer with an uncertain benefit of adjuvant chemotherapy based on traditional prognostic factors were eligible for inclusion. Twenty-three out of thirty-three participating hospitals offered patients the opportunity for their tumor samples to be additionally evaluated by an 80- GS. As a result 595 patients were included in the present study. Routine pathology comprised the local determination of hormone receptor status (ER and PR) and HER2-receptor status. ER and PR status were determined by immunohistochemistry (IHC) in all institutions and were considered positive if nuclear staining was present in 10% of breast cancer cells. HER2-expression was scored by IHC according to international guideline recommendation 15 : 0 if no staining was observed or membrane staining was incomplete and faint/barely perceptible and within less than 10% of the tumor cells, 1+ if staining was incomplete and faint/barely perceptible, but within more than 10% of the tumor cells, 2+ if more than 10% of the tumor cells displayed circumferential staining of moderate intensity or complete and circumferential staining within less than 10%, and 3+ for strong complete membrane staining within 30% of the tumor cells. A tumor was considered HER2-negative when a score of 0 or 1+ was found and positive when a score of 3+ was observed. Tumors with 2+ HER2- expression were additionally evaluated by FISH. The cut-off for HER2 low level and high level amplification was defined as >6 and >10 copies of the HER2 gene or clusters respectively. Ki67 was part of routine pathology assessment in a limited number of participating hospitals. When Ki67 had been determined, a Ki67 cutoff value of 14% was used for the designation of Ki67 into high or low, based on the 2011 recommendations of the St. Gallen s expert consensus panel. 16 Assessment of molecular subtypes Based on conventional pathology examinations, surrogate molecular tumor subtypes were determined as follows: Luminal-type (ER+/PR+, HER2-), HER2-type (ER&PR +/-, HER2+) and Basal-type (ER-, PR-, HER2-). Using Ki-67 assessment a further sub stratification of Luminal-type tumors into Luminal A-type (ER+/PR+, HER2-, Ki-67 low) and Luminal B-type (ER+/PR+, HER2-, Ki-67 high) could be performed. Molecular subtyping based on microarray analysis consisted of combining an 80-GS with the 70-GS. DNA was co-hybridized with a standard reference to the customdesigned diagnostic chip, each containing oligonucleotide probes for the profiles in triplicate or more. This 80-GS enabled stratification of tumors into Luminal, HER2, and Basal-type tumors, whereas the 70-GS further categorized the Luminal-type into 200

201 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients Luminal A (70-GS Low Risk) and Luminal B (70-GS High Risk). In the present study, concordance between the molecular subtypes based on conventional pathology and gene-expression profiling was evaluated for all patients. In addition, in the subset of patients for whom a Ki-67 level assessment was available, concordance between Luminal A-type and Luminal B-type tumors based on local pathology and microarray analysis was explored. Statistical Analysis Comparison of molecular and surrogate clinical subtyping was done with a two by three contingency table, including overall concordance. In addition, comparison of the sub stratification of Luminal-type tumors into A and B determined by local pathology assessment or gene-expression profiling was done with the Kappa statistic to evaluate the agreement between these two classifications on a nominal scale and accompanying 95% confidence intervals (CI) were calculated. 17,18 RESULTS Patients Classification of molecular subtypes by gene-expression profiling was performed in 590 patients, treated in 23 hospitals. According to local pathology assessment, all patients had ER+ tumors, 87% PR+ and 2% had HER2+ tumors. The majority of the patients had intermediate grade tumours (74%), with no or limited axillary lymph-node involvement (90% pn1mi, Table 1). Chapter

202 Chapter 10 Table 1 Baseline characteristics determined by local pathology assessment All patients (n = 595) Age in years, median (range) 58 (35-80) pt T1 480 T2 114 T3 1 pn N0 (i+) 499 N1mi 54 N1 37 Nx 5 Tumor grade Grade I 86 Grade II 438 Grade III 70 Unknown 1 ER status Positive 595 PR status Positive 518 Negative 76 Unknown 1 HER2-status Positive 12 Negative 576 Unknown 7 Ki67 <14% 12 >14% 36 Not assessed 547 ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth receptor 2, Ki-67 proliferation marker, N0 / N0 (i+)no axillary lymph node involvement / isolated tumor cells, Nmi micrometastasis, N1a-c metastasis in movable ipsilateral Level I, II axillary lymph node(s), Nx axillary lymph node status not assessed. 202

203 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients Molecular subtyping by local pathology and gene-expression profiling Using local pathology as a surrogate for molecular subtyping 98% of patients (n=578) were regarded as Luminal-type and 2% (n=12) of patients as HER2-type. The 80-GS classified 98% (n=583) of all enrolled (n=595) patients as having Luminal-type, 1% (n=7) as HER2-type and 1% (n=5) as Basal-like tumors. There was 98% concordance for the Luminal-type tumors (Table 2). Of the patients classified as Luminal-type by the 80-GS, 60% (n=347) were assigned to the low-risk (Luminal A) and 40% (n=236) to the high-risk (Luminal B) category by the 70-GS. Of the twelve patients that were classified as HER2-type by IHC/FISH, nine patients (75%) were reclassified as Luminal-type by the 80-GS, resulting in a concordance of 25% within the HER2 group. Table 2 Molecular subtyping with an 80-GS and IHC/FISH (n=590) Molecular subtypes Clinical subtypes Luminal(%) HER2(%) Basal(%) Total ER+/PR+, HER2-569(98) 4(1) 5(1) 578 ER&PR+/-, HER2+ 9(75) 3(25) 0 12 Total IHC immunohistochemistry, FISH fluorescence in situ hybridization, 80-GS 80-gene signature, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2. The overall concordance between IHC/FISH and GEP was 97%. In 48 patients treated in 4 hospitals Ki67 assessment was done and a distinction could be made between Luminal A and B. Twenty-five percent (n=12) of these patients were classified as Luminal A and 75% (n=36) of patients as Luminal B. Of the patients that were classified as Luminal A by local pathology three (25%) were reclassified as Luminal B by the 80-GS. Twenty-one (58%) of the 36 patients considered as having a Luminal B tumour by local pathology were reclassified as Luminal A by gene-expression subtyping (Kappa 0.11 [95%CI ] Table 3) resulting in a concordance of 50% for further distinguishing between Luminal A and B. Chapter

204 Chapter 10 DISCUSSION In this prospective multicenter study we observed high concordance between intrinsic molecular subtyping by conventional pathology assessment and microarray analysis using an 80-GS in predominantly ER+/HER2- Dutch breast cancer patients. Although the number of patients who were considered to have HER2-type tumors was limited most of them were reclassified as luminal type tumors by gene-expression profiling. When Ki67 expression assessment was used in addition to routine pathology to further distinguish between Luminal A and B type tumors and compared to gene-expression profiling, concordance was low. Table 3 Molecular subtyping of Luminal type tumors with GEP (80-GS and 70-GS) and IHC/FISH (n=48) IHC immunohistochemistry, FISH fluorescence in situ hybridization, 70-GS 70-gene signature, 80-GS 80-gene signature. ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, Ki-67 proliferation marker protein. The overall concordance between IHC/FISH and GEP was 50%. Kappa %CI Molecular subtypes Clinical subtypes Luminal A (%) Luminal B (%) Total (70-GS low risk) (70-GS high risk) ER+/PR+, HER2-9(75) 3(25) 12 Ki-67 <14% ER+/PR+, HER2-21(58) 15(42) 36 Ki-67 >14% Total The current study shows that molecular subtyping using IHC/FISH or an 80-GS results in classification of similar proportions Luminal-type tumours. The observed concordance of 98% is in line with a previous study conducted by Nguyen et al. (n=135) in which determination of Luminal-type tumours by IHC/FISH and an 80-GS showed high concordance (94%). 22 This implies that there is little or no additional value for routine use of gene-expression profiling in patients that are classified as ER+ and HER2- by routine pathology. However, in the small subset of patients categorized as having a HER2-type tumour according to local pathology assessment, gene-expression profiling did reveal a high proportion of tumors that were reclassified as Luminal-type breast cancers. Our results are in line with a previous study conducted by Viale et al. in which a relatively large proportion (42%) of clinical HER2+ patients were considered as having Luminal-type 204

205 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients (38%) or Basal-type (4%) by the 80-GS. 23 In contrast to our results, Nguyen et al. showed a concordance of 94% between IHC/FISH and 80-GS molecular subtyping for the HER2 subtype. 22 In the current study only a small proportion of patients suffered of HER2+ disease, which might explain the differences in results. In addition, sub stratification of Luminal-type tumors into A and B by local pathology based on the expression of proliferation marker Ki67 compared to the use of geneexpression profiling was associated with a high discordance rate. The results are similar to the results reported in previous studies, where molecular subtyping based on the expression of Ki67 versus gen-expression profiling resulted in a discordance rate of 33% 22 and 31%. 24 In the study of Nguyen et al. a distinction of Luminal-type tumours into A and B based on tumour grade was also associated with a discordance rate of 48%. 22 The pathological distinction between ER+ tumours into low risk (Luminal A-type) and high risk (Luminal B-type) of developing metastasis is mostly based on the assessment of proliferative activity. Two means to assess proliferative activity are mitotic count and the assessment of the growth fraction of a given cell population, known as the Ki67-labelling index. While mitotic count is a reliable diagnostic test and the clinical utility is implemented in a standardized protocol regarding the grading system for breast cancer differentiation, 24 a standardized methodology for Ki67 assessment is lacking and revisions within the St. Gallen expert panel recommendations for the most appropriate cut-off for high proliferative tumours are still pending. 19,25 Our results support previous studies in which intratumoral heterogeneity of Ki67 expression levels, interlaboratory and interobserver variability of Ki67 staining and differences in Ki67-Labelling Index values have been observed and cast doubt on the utility of this biomarker as a parameter of proliferative activity Until date, this is the largest prospective multicenter study in which molecular subtyping using local pathology assessment compared to gene-expression profiling is evaluated. The absence of ER- patients within our study population, precludes comparison of local pathology to gene expression profiling regarding the determination of basal like tumours. Similarly, the small number of HER2+ tumours within our study population, limits the evidence regarding the accuracy of pathology based HER2 receptor status assessment. The composition of our study population with mainly ER+/HER2- patients is relevant for comparison of these two alternative ways of molecular subtyping, since this is the subset of patients in which gene-expression profiling (70-GS) is commonly deployed. Chapter

206 Chapter 10 Lastly, in the present study in patients who were candidates for the 70-GS use, a 80-GS proved to be of little value when tumours were characterized as ER+ and HER2- by routine pathology assessment. However, gene profiling did reclassify more than half of tumours that were originally classified as HER2+ by local pathology and further distinction between luminal A and B also revealed little concordance with routine pathology enhanced by Ki-67 assessment. Since the therapeutic implications in the latter groups are large, gene-expression profiling is to be considered of additional value. 206

207 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients REFERENCE LIST 1. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000;406: Yao K, Goldschmidt R, Turk M et al. Molecular subtyping improves diagnostic stratification of patients with primary breast cancer into prognostically defined risk groups. Breast Cancer Res Treat. 2015, 154 (1): Glϋck S, de Snoo F, Peeters J, Stork-Sloots L, Somlo G. Molecular subtyping of earlystage breast cancer identifies a group of patients who do not benefit from neo-adjuvant chemotherapy. Breast Cancer Res Treat. 2013; 129: Whitworth P, Stork-Slooks L, de Snoo F et al. Chemosensitivity Predicted by BluePrint 80-Gene Functional Subtype and MammaPrint in the Prospective Neoadjuvant Breast Registry Symphony Trial (NBRST) Ann Surg Oncol. 2014; 21(10): MC Cheang, SK Chia, Voduc D et al. Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer J Natl Cancer Inst May 20; 101(10): Bueno-de-Mesquita JM, Nuyten DS,Wesseling J, et al. (2010) The impact of interobserver variation in pathological assessment of node-negative breast cancer on clinical risk assessment and patient selection for adjuvant systemic treatment. Ann Oncol 21: Oyama T, Ishikawa Y, Hayashi M, Arihiro K, Horiguchi J. The effects of fixation, processing and evaluation criteria on immunohistochemical detection of hormone receptors in breast cancer. Breast Cancer. 2007;14: Tang P, Gary M et al. Immunohistochemical Surrogates for Molecular Classification of Breast Carcinoma; Arch Pathol Lab Med-Vol 140 August Dowsett M, Nielsen T, A Hern R, et al. Assessment of Ki67 in Breast Cancer: Recommendations from the International Ki67 in Breast Cancer Working Group; JNCI J Natl Cancer Inst (2011) 10. Varga, Z., et al, Standardization for Ki67 assessment in moderately differentiated breast cancer. A retrospective analysis of the SAKK 18/12 study. PLoS. One, (4): p. e van t Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 415:530-6, van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. The New England journal of medicine 347: , Cardoso F, van t Veer LJ, Bogaerts J, et al. 70-gene Signature as an Aid to Treatment Decisions in Early-Stage Breast Cancer. N Engl J Med 375;717-29, Krijgsman O, Roepman P, Zwart W et al. A diagnostic gene profile for molecular subtyping of breast cancer associated with treatment response; Breast Cancer Res Treat (2012) 133:37 47 Chapter

208 Chapter Wolff A, Hammond ME, Schwartz JN, et al. American Society of Clinical Oncology/ College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. J Clin Oncol. 2007;25: Goldhirsch, a., et al., Strategies for subtypes-dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2011 Ann. Oncol, (8): p, IBM SPSS Statistics V23.0 for Windows 18. R studio version for Windows 19. Coates, A.S., et al., -Tailoring therapies-improving the management of early breast cancer: St. Gallen international Expert consensus on the Primary Therapy or Early Breast Cancer Ann Oncol, (8): p, Goldhirsch A, Ingle JN, Gelber RD, Coates AS, Thürlimann B, Senn HJ. Thresholds for therapies: highlights of the St. Gallen International Expert Consensus on the primary therapy of early breast cancer, Ann Oncol. 2009;20: Pinto AC, Ades F, de Azambuja E, Trastuzumab for patients with HER2 positive breast cancer: delivery, duration and combination therapies /j.breast Nguyen B, Cusomano PG, Deck k. Comparison of Molecular subtyping with BluePrint, MammaPrint, and TargetPrint to local clinical subtyping in breast cancer patients. Ann. Surg oncol 2012 Oct. 19(10): Viale G, Slaets L, de Snoo F et al. Comparison of molecular (BluePrint+MammaPrint) and pathological subtypes for breast cancer among the first 800 patients from the EORTC 10041/BIG 3-04 (MINDACT) trial. Journal of Clinical Oncology, Volume:30, Issue:27_suppl 24. Van Diest, P.J., et al., Reproducibility of mitosis counting in 2,469 breast cancer specimens: results from the Multicenter Morphometric Mammary Carcinoma Project. Hum. Pathol, (6): p Scholzen, T. and J. Gerdes, The Ki-67 protein: from the known and the unknown. J. Cell Physiol, (3): p Dowsett, M., et al., Assessment of Ki67 in breast cancer: recommendations from the International Ki67 in Breast cancer working group. J. Natl. Cancer Inst, (22): p Urriticoechea A., I.E Smith, and M. Dowsett, Proliferation marker Ki-67 in early stage breast cancer J. Clin. Oncol, (28): p Rakha EA, Ellis IO. An overview of assessment of prognostic and predictive factors in breast cancer needle core biopsy specimens. J. Clin. Pathol Dec; 60 (12): Goldhisch, A., et al., Personalizing the treatment of woman with early breast cancer: highlights of the St. Gallen International Expert consensus on the Primary Therapy of Early Breast Cancer 2013 Ann. Oncol, (9): p

209 Comparison of molecular subtyping by local pathology or an 80-gene signature in estrogen receptor positive patients Chapter

210

211 CHAPTER 11 Concluding remarks and future perspectives

212 Chapter 11 CONCLUDING REMARKS AND FUTURE PERSPECTIVES Trends in adjuvant chemotherapy in relation to national guideline adjustments and gene expression profiling Part I of this thesis illustrates the increased use of adjuvant systemic treatment in Dutch early stage breast cancer patients over the last twenty years in relation to the gradual expansion of the indication area for adjuvant systemic treatment as reflected by respective national treatment guidelines. The increased proportion of patients who actually received adjuvant systemic treatment well preceded the proportions that one would expect based on the respective national guideline adjustments. International guidelines for breast cancer treatment already expanded the indication area for adjuvant systemic therapy years before. The age-standardized mortality rate for breast cancer in the Netherlands was above the European average in the early 2000 s, but concurrent to the increased use of adjuvant systemic therapy, European and Dutch age-standardized mortality curves are converging. Since 2010 a downward trend is observed as less patients have received adjuvant chemotherapy (chapter 2). Again, this trend in clinical practice was observed prior to the guideline adjustment of The results of this chapter illustrate that, as opposed to what is expected of guidelines, they basically lag behind on movements in adjuvant systemic treatment use in clinical practice. This part of this thesis also illustrates that there is considerable variation in adjuvant systemic treatment guideline adherence in the Netherlands which cannot be explained by patient factors such as ethnicity or socio-economic status (chapter 3). In contrast with studies conducted in the United States, under treatment was not associated with socio-economic status. Part II of this thesis studies the use and impact of gene-expression profiling and illustrates that gene-expression profiles have affected the proportion of patients receiving adjuvant chemotherapy in early stage breast cancer in recent years (chapter 5, 7 and 8). This was not merely the effect of deployment of gene-expression profiles in individual patients since gene-expression profiles were used in a limited proportion of patients. Gene-expression profiles have also played an important role during a period of time when the acknowledgement of tumour biology on breast cancer outcome contributed to a tendency of physicians to become more reluctant to administer chemotherapy in subgroups of patients with luminal type breast cancers, i.e. candidates for geneexpression profile use. The impact of considerable doubt regarding the benefit of adjuvant chemotherapy in patients with estrogen receptor (ER) positive/her2 negative disease is illustrated in chapter 5: in the absence of gene-expression profile deployment half of all Dutch ER+/Her2- early stage breast cancer patients received adjuvant chemotherapy whereas the other half did not receive chemotherapy despite similar clinicopathological characteristics. In a prospective study, when clinicians were asked to 212

213 Concluding remarks and future perspectives state a chemotherapy recommendation in patients with comparable clinicopathological characteristics prior to knowledge about a gene-expression profile result, the chance of receiving chemotherapy was approximately 50% (chapter 7). Chapter 7 also illustrates that deployment of a gene-expression profile alters the adjuvant chemotherapy recommendation in half of patients which resulted in a different selection of patients who eventually received chemotherapy compared to when the clinician did not involve a gene-expression profile in the chemotherapy decision-making process. In chapter 6 we demonstrate that over the last decade, considerable variation existed in adjuvant chemotherapy administration in certain prespecified patient groups. In these groups, use of a gene-expression profile was associated with a more consistent chemotherapy administration policy over time. Chapter 4 and 8 illustrate that considerable variation also existed regarding the use of gene-expression profiles in the Netherlands. Geneexpression profiles were less frequently deployed in patients of low socio-economic status suggesting that patient preferences and awareness affect the deployment of the test by physicians. Lastly, we observed that unintended use of gene-expression profiles was quite common. The absolute number of patients for whom the test was deployed in the presence of a clear guideline recommendation to administer or withhold chemotherapy based on a high or low clinical risk and hence outside the indication area for geneexpression profiling, was larger than the number of patients who received the test in line with the suggested indication area (chapter 8). Current place of gene-expression profiles in early stage breast cancer patients Gene-expression profiles have played a role in the recognition of intrinsic tumour biology as the most important factor influencing breast cancer outcome. Combining the assessment of ER, progesterone receptor and Her2-status to define surrogate molecular subtypes has contributed more importantly to this acknowledgement of tumour biology. The use of a gene-expression profile is not necessary in every breast cancer patient, in a substantial proportion of patients an adequate chemotherapy recommendation can be made based on less expensive assessment of clinicopathological characteristics alone. The Dutch guideline of 2012 was the first to suggest the use of a validated gene-expression profile in patients with estrogen receptor positive disease in whom, based on conventional clinicopathological factors alone, controversy existed regarding adjuvant chemotherapy benefit. 1 This guideline recommendation was based on several retrospective 2 7, one prospective study 8 9, and awaiting the results of an international, multicentre, randomized controlled trial (Microarray in Node-negative and 1-3 lymph-node positive Disease may Avoid Chemotherapy EORTC 10041/BIG 3-04 MINDACT). 10 In the latter trial a total of 6693 early stage breast cancer patients (T1-2, N0-1) were enrolled between January 2007 and December A clinical risk estimation was made based on a risk prediction normogram and simultaneously Chapter

214 Chapter 11 a genomic risk estimation was made based on the 70-gene signature. Patients with a discordant clinical and genomic risk estimation were randomized between adjuvant chemotherapy administration based on the clinical risk or genomic risk estimation. The primary goal of this trial was to assess whether the lower boundary of the 95% CI 5-year distant metastasis free survival would be 92% among patients with a high clinical risk profile but low genomic risk profile who were treated according to the genomic risk strategy (i.e. did not receive chemotherapy). For clinical high-risk patients assigned to the low genomic risk category the 5-year distant metastasis free survival was 95.9% vs. 94.4% for patients who did and did not received chemotherapy, respectively (HR %CI ). Furthermore, use of the 70-gene signature to guide adjuvant chemotherapy decision-making in clinical high-risk patients would lead to a reduction in the use of chemotherapy in 46.2% of patients. Since the primary endpoint of the trial was met, the authors conclude that in clinical high-risk patients the 70-gene signature can differentiate patients who may avoid chemotherapy from those for whom chemotherapy is likely to provide a survival benefit. Furthermore, results of the MINDACT trial indicate that use of a gene-expression profile is not necessary in clinical low risk patients and chapter 9 of this thesis demonstrates that this also holds true in case of multifocal disease. 11 Since the publication of the MINDACT trial the discussion regarding the indication area for gene-expression profiles is far from being closed. Some suggest that the 70- gene signature should only be used in a selection of clinical high risk patients given that there were major clinicopathological differences between the characteristics of the patients at high clinical risk but low genomic risk and those at high risk in both categories, indicating that clinicopathological factors can be used to predict the 70-gene signature result and hereby determine in which patients use of the 70-gene signature is truly necessary. 12 The downside of such an approach is the high level of variability in assessment of some of these tumor characteristics, which is also illustrated by the results described in the last chapter of this thesis (chapter 10). A similar plea for a selective use of the 70-gene signature was made by others who underscore the fact that in some clinical high risk categories, such as patients with triple negative or hormone receptor negative/her2-positive disease, >90% of patients was assigned to the genomic high risk category. Their advice is to restrict 70-gene signature use to patients with hormone receptor positive/her2-negative breast cancer in whom survival benefit of chemotherapy on the 10-year survival rate is 2-4% (as calculated with New Adjuvant Online), assuming that patients consider an absolute survival benefit of 3% or more due to chemotherapy worthwhile. 13 On the other hand, critiques were raised about the statement in the MINDACT-publication that a low genomic risk score in clinical high risk patients would justify omission of chemotherapy given the fact that breast cancer 214

215 Concluding remarks and future perspectives patients are willing to accept adjuvant chemotherapy treatment for very small survival gains (<1%) A salient detail is that especially in the Netherlands, where the 70- gene signature was developed, considerable controversy and resistance exists regarding its deployment. All in all, the current advice dating from 2012 to deploy gene-expression profiles as an adjunct to clinicopathological factors in estrogen-receptor positive early breast cancer patients still makes sense. Then again, it is important to better discriminate clinicopathological characteristics of patient groups in whom doubt about additional benefit of chemotherapy warrants gene-expression profile use. As stated above, deployment of a gene-expression profile in patients with hormone receptor negative, Her2+ or triple negative cancers does not seem suitable. While, in patients with Luminal type tumours, gene-expression profiles do serve as a consistent and adequate tool to distinguish between luminal A and B, and to guide adjuvant chemotherapy decisionmaking. As such it is superior to grade or Ki67 testing. In addition, following delineation of the indication area for gene-expression profiles their use should be advocated more strongly. Future perspectives Adjuvant systemic treatment guidelines ideally serve adequate and uniform administration of adjuvant systemic treatment in early stage breast cancer patients. In addition, guideline recommendations regarding gene-expression profile use intend to promote adequate deployment of gene-expression profiles in patients in whom such test can be of additional benefit, annihilating disparities in gene-expression profile use and avoiding unnecessary use of gene-expression profiles. A new edition of the national breast cancer treatment guideline, including a revised section on the role of geneexpression profiles in early stage breast cancer patients, is expected in Against the background of a better defined indication area for adjuvant chemotherapy and geneexpression profile use, institutional variation or physicians believes should not be a factor to explain variation in adjuvant chemotherapy administration or gene-expression profile use,. In contrast, patient preferences might play a role in the existing variation. We recently started enrolment of patients in a prospective observational multicentre study (Triple A patients perspective study) in which we aim to gain insight in patients perspectives regarding gene-expression profile testing (70-gene signature), including items on patients understanding of the 70-gene signature and the impact of 70-gene signature use on the shared decision-making process regarding adjuvant chemotherapy. Chapter 11 Gene-expression profiles serve as means to get a better grip on tumour biology and the impact of biology on breast cancer treatment or outcome. The rapid uptake and 215

216 Chapter 11 fast increase of the use of gene-expression profiles over the last decade illustrates the need for more adequate predictors than routine clinicopathological factors to guide adjuvant systemic treatment decision-making in early stage breast cancer patients. Geneexpression profiles have improved patient selection for adjuvant systemic treatment but the quest for alternative adequate predictors of breast cancer outcome and treatment response continues. As the prognostic information of gene-expression profiles seems to lie in obtained information on proliferation related genes 16,17, we are currently evaluating the concordance between phosphohistone H3 (PPH3), an easy assessible and less expansive proliferation marker, and the risk of distant metastasis as assessed by the 70- gene signature. Nevertheless, we are convinced that genomic information, not merely obtained by gene-expression profiling, but perhaps whole genome sequencing, will become pivotal in determining prognosis and optimal treatment selection, both local and systemic, in early stage breast cancer patients. From this perspective, we reviewed the literature regarding the interplay between age, molecular subtypes and determining optimal local therapy strategies in early stage breast cancer patients (chapter 12). The OPTIMA Prelim Trial, in which agreement among five genomic assays was evaluated, showed significant disagreement between assays in identification of patients at genomic low-risk (only 39% of the 313 enrolled patients were classified uniformly by different gene-expression tests). 18 This poor concordance between assays indicates that a well validated assay cannot reliably be replaced with others with less validation and suggests genomic testing in itself should be optimized. Developments in incorporation of more individualized genomic information in breast cancer treatment selection follow each other in rapid succession. Gene-expression profiles provide an opportunity for accurate assessment of tumour biology than traditional clinicopathologic classifiers mainly in estrogen-receptor positive/her2 negative early stage breast cancer patients. Recently, in-depth genomic characterization of Her2 positive breast tumours revealed that, despite being clinically defined as Her2 positive based on a single gene amplification as assessed by pathology, Her2 positive tumours seem to melt into the whole luminalbasal breast cancer spectrum. 19 In addition, results of a recent study, in which whole genome sequencing was performed on pre- and post-treatment biopsies of Her2 positive breast tumours, suggest that it may be possible to predict, at time of diagnosis, whether Her2 positive breast cancer patients will respond to treatment with chemotherapy and trastuzumab. 20 These developments strengthen our believe that, in the near future, gene-expression profiles will probably be replaced more individualized genomic testing strategies. Changing clinical practice Although not the primary aim of the individual studies in this thesis a number of these studies illustrates that clinical guidelines for the treatment of early stage breast cancer 216

217 Concluding remarks and future perspectives more often follow then precede practice change and that the adherence to chemotherapy guidelines as well as gene-expression profile use varies considerably despite the presence of a national guideline (chapter 2, 4, 5 and 7). Similar variations are observed regarding other aspects of breast cancer treatment such as the implementation of neo-adjuvant treatment strategies in clinical practice 21 on the axillary management in early stage breast cancer. Recent landmark studies regarding axillary management have led to considerable guideline changes regarding management of the axilla, recommending less aggressive strategies in a substantial proportion of patients. However, large variation in the uptake of these guideline recommendations were observed between institutions in one country and between countries. 22,23 A correlation between the use of gene-expression profiles and more restrictive treatment of the axilla was observed as a secondary finding in chapter 4 of this thesis. In Dutch hospitals who did not deploy gene-expression profiles more breast cancer patients treated between received an axillary lymph-node dissection compared to patients treated in hospitals that frequently deployed a gene-expression profile. This suggests an institutional tendency to treat conservatively or liberally. Taking a broader view to health care, studies conducted in the United States and the Netherlands suggest that at least 30-40% of patients do not receive care according to current scientific evidence, while 20% of the care provided is unnecessary or can be harmful to patients. 24,25 The overarching question is how can we achieve faster implementation of novel health-care innovations in routine clinical practice in order to achieve optimal care for every patient and diminish variation. In this era in which health care innovations follow each other in a rapid succession, future research should also focus more on strategies for improving implementation of health care innovation into clinical practice. Chapter

218 Chapter 11 REFERENCE LIST 1. Kwaliteitsinstituut voor de Gezondheidszorg CbO VvIK. Landelijke richtlijn mammacarcinoom van t Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature. 2002;415(6871): doi: /415530a. 3. van de Vijver MJ, He YD, van t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25): doi: /nej- Moa Buyse M, Loi S, van t Veer L, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst. 2006;98(17): doi: /jnci/djj Wittner BS, Sgroi DC, Ryan PD, et al. Analysis of the MammaPrint breast cancer assay in a predominantly postmenopausal cohort. Clin Cancer Res. 2008;14(10): doi: / ccr Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature predicts early metastasis in breast cancer patients between 55 and 70 years of age. Ann Oncol Off J Eur Soc Med Oncol. 2010;21(4): doi: /annonc/mdp Knauer M, Cardoso F, Wesseling J, et al. Identification of a low-risk subgroup of HER-2-positive breast cancer by the 70-gene prognosis signature. Br J Cancer. 2010;103(12): doi: /sj.bjc Bueno-de-Mesquita JM, van Harten WH, Retel VP, et al. Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective community-based feasibility study (RASTER). Lancet Oncol. 2007;8(12): doi: / S (07) Drukker CA, Bueno-de-Mesquita JM, Retèl VP, et al. A prospective evaluation of a breast cancer prognosis signature in the observational RASTER study. Int J cancer. 2013;133(4): doi: /ijc Cardoso F, Van t Veer L, Rutgers E, Loi S, Mook S, Piccart-Gebhart MJ. Clinical application of the 70-gene profile: the MINDACT trial. J Clin Oncol. 2008;26(5): doi: /jco Cardoso F, van t Veer LJ, Bogaerts J, et al. 70-Gene Signature as an Aid to Treatment Decisions in Early-Stage Breast Cancer. N Engl J Med. 2016;375(8): doi: / NEJMoa Blok EJ, van de Velde CJ, Smit VT. 70-Gene Signature in Early-Stage Breast Cancer. N Engl J Med. 2016;375(22): doi: /nejmc de Boer M, Voogd AC, Tjan-Heijnen VCG. [A plea for selective use of the MammaPrint test]. Ned Tijdschr Geneeskd. 2017;161(0):D1160. Available at: nih.gov/pubmed/ Accessed May 12,

219 Concluding remarks and future perspectives 14. Thewes B, Prins J, Friedlander M. 70-Gene Signature in Early-Stage Breast Cancer. N Engl J Med. 2016;375(22): doi: /nejmc Hudis CA, Dickler M. Increasing Precision in Adjuvant Therapy for Breast Cancer. N Engl J Med. 2016;375(8): doi: /nejme Wirapati P, Sotiriou C, Kunkel S, et al. Meta-analysis of gene expression profiles in breast cancer: toward a unified understanding of breast cancer subtyping and prognosis signatures. Breast Cancer Res. 2008;10(4):R65. doi: /bcr Sotiriou C, Wirapati P, Loi S, et al. Gene Expression Profiling in Breast Cancer: Understanding the Molecular Basis of Histologic Grade To Improve Prognosis. JNCI J Natl Cancer Inst. 2006;98(4): doi: /jnci/djj Bartlett JMS, Bayani J, Marshall A, et al. Comparing Breast Cancer Multiparameter Tests in the OPTIMA Prelim Trial: No Test Is More Equal Than the Others. J Natl Cancer Inst. 2016;108(9):djw050. doi: /jnci/djw Ferrari A, Vincent-Salomon A, Pivot X, et al. A whole-genome sequence and transcriptome perspective on HER2-positive breast cancers. Nat Commun. 2016;7: doi: /ncomms Lesurf R, Griffith OL, Griffith M, et al. Genomic characterization of HER2-positive breast cancer and response to neoadjuvant trastuzumab and chemotherapy results from the ACOSOG Z1041 (Alliance) trial. Ann Oncol. 2017;28(5): doi: / annonc/mdx Vugts G, Maaskant-Braat AJG, Nieuwenhuijzen GAP, Roumen RMH, Luiten EJT, Voogd AC. Patterns of Care in the Administration of Neo-adjuvant Chemotherapy for Breast Cancer. A Population-Based Study. Breast J. 2016;22(3): doi: / tbj Maaskant-Braat AJ, Voogd AC, van de Poll-Franse L V., Coebergh JWW, Nieuwenhuijzen GA. Axillary and systemic treatment of patients with breast cancer and micrometastatic disease or isolated tumor cells in the sentinel lymph node. The Breast. 2012;21(4): doi: /j.breast Gondos A, Jansen L, Heil J, et al. Time trends in axilla management among early breast cancer patients: Persisting major variation in clinical practice across European centers. Acta Oncol (Madr). 2016;55(6): doi: / x Grol R. Successes and failures in the implementation of evidence-based guidelines for clinical practice. Med Care. 2001;39(8 Suppl 2):II Available at: nlm.nih.gov/pubmed/ Accessed May 19, Schuster MA, McGlynn EA, Brook RH. How good is the quality of health care in the United States? Milbank Q. 1998;76(4):517 63, 509. Available at: nih.gov/pubmed/ Accessed May 19, Chapter

220

221 CHAPTER 12 Age, molecular subtypes and local therapy decision-making A. Kuijer & T.A. King The Breast 2017 The Breast 2017;34:S70-77.

222 Chapter 12 ABSTRACT The relationship between age and breast cancer subtype is complex: both impact risk of locoregional recurrence (LRR) and survival. Young patients frequently present with aggressive tumors but the increased risk imparted by young age appears to differ among breast cancer subtypes. Dramatic improvements in local control among young women with breast cancer of all tumor subtypes have been observed, likely attributable to improved local therapy strategies, improvements in adjuvant therapies and implementation of subtype-specific targeted therapies. In the light of these improvements in local control, accumulating evidence demonstrates that there is no difference in LRR or survival between breast conserving therapy (BCT) and mastectomy in young patients. An increased risk of LRR in triple-negative cancers is apparent; yet this increased risk of LRR is present following surgical treatment with both BCT and mastectomy and does not significantly differ by age. Also, contralateral breast cancer rates remain low for all age groups and, although the use of contralateral prophylactic mastectomy (CPM) has increased, there is no evidence that CPM improves survival. At the other end of the age spectrum, there is a growing body of evidence demonstrating a favorable interaction between older age and molecular subtype such that many older women with estrogen receptor positive breast cancer may be spared axillary staging and/or radiation therapy without a detrimental impact on survival. Thus for both age and subtype, it appears that the intrinsic biology is the strongest predictor of outcome. Tumor biology, and not age, should be the driver in local therapy decision-making. 222

223 Age, molecular subtypes and local therapy decision-making INTRODUCTION The relationship between age and breast cancer subtype is complex. Both impact risk of locoregional recurrence (LRR) and survival; however, the increased risk imparted by young age appears to differ among breast cancer subtypes, and as more is learned about the heterogeneity of breast cancer, the absolute risk from young age alone will likely continue to lessen. At the other end of the age spectrum, there is a growing body of evidence demonstrating a favorable interaction between older age and molecular subtype such that many older women with estrogen receptor (ER) positive (+) breast cancer may be spared axillary staging and/or adjuvant radiation therapy (RT) without a detrimental impact on survival. BREAST CANCER IN YOUNG WOMEN It is well recognized that younger breast cancer patients present with tumors of more aggressive biology, illustrated by more aggressive clinicopathological features (i.e. ER negative (-) or HER2+ tumors of higher grade 1,2 or assignment to the high-risk group by molecular sub typing 3,4, which is accompanied by a higher risk of LRR and distant recurrence as compared to their older counterparts. Over recent years, it has become evident that tumor biology, indicated by molecular subtypes, largely overshadows the influence of young age with regard to breast cancer outcome. The distribution of the four distinct molecular subtypes (Luminal A, Luminal B, HER2 and basal-like) differs between younger (<40 years) and older patients ( 40 years of age) with a decreasing incidence of more aggressive molecular subtypes with increasing age. 5,6 Younger patients are more likely to be diagnosed with HER2-enriched and basal-like tumors and less likely to be diagnosed with Luminal tumors as compared to older patients. In addition, among patients with hormone receptor (HR) positive (+)+/Her2- disease by immunohistochemistry (IHC) there are proportionally less true Luminal cancers in young women. 3 In parallel with advances in systemic treatment, LRR and breast cancer mortality rates in young breast cancer patients continue to decline 7 and this decline is seen among all tumor subtypes. 8,9 In addition, there is a growing body of evidence that young age is no longer associated with inferior outcomes in HER2+ disease 10,11 or triple-negative breast cancer (TNBC) 12, yet remains an important prognostic factor in HR+disease. 13,14 Young age, molecular subtype and locoregional recurrence The dramatic improvement in local and regional control among young women with breast cancer is illustrated by two recent population-based studies from the Netherlands. Van Laar et al. 7 reported an overall 5-year local recurrence (LR) rate of 7.5% among 1143 women aged 40 years with early stage breast cancer (pt1-2/ct1-2,n0-2,m0) Chapter

224 Chapter 12 treated with breast conserving therapy (BCT) between ; however when evaluated over time, 5-year LR risk decreased dramatically from 9.8% for patients treated between to 3.3% for patients treated in later years ( ). Aalders et al. 8 confirmed this trend in decreasing risk of LRR in a cohort of 1000 Dutch women <35 years of age surgically treated for primary invasive breast cancer between The overall 5-year rate of LR and regional recurrence (RR) were 3.5% and 3.7% respectively, yet these rates varied significantly by tumor subtype in patients treated prior to the introduction of trastuzumab. When analyzed for the period after the introduction of trastuzumab these observed differences were no longer significant: <2% of patients with HR+ disease, regardless of HER2-status, 5.6% of patients with HR-/ HER2+ disease and 4.5% of patients with TNBC suffered a LR at 5-years of follow-up (p 0.24). Table 1 summarizes the studies evaluating the association between LR risk, age and tumor subtype. 6,8,9,15,16 Although most studies demonstrate that, younger age remains independently associated with increased LR risk even after correction for tumor biology, LR risk in young patients treated with either mastectomy or BCT appear to be acceptably low. Arvold et al. 9 assessed the association between age, tumor biology and LR risk in 1434 consecutive patients treated with BCT at two centers between On multivariate analyses, including adjustment for tumor subtype, increasing age remained independently associated with decreased risk of LR (HR 0.97/year increase 95%CI , p 0.01). However, at a median follow-up of 85 months the overall cumulative incidence of LR was 2.1% and was acceptably low in all age quartiles: 5.0% for patients years; 2.2% years; 0.9% years and 0.6% years. Luminal type tumors were associated with the lowest risk of LR and HER2 or TNBC with a higher crude LR risk in all age quartiles. In the youngest patients a LR risk of 4.7%, 8.1%, 3.0%, 13.3% and 10.2% was observed for Luminal A, Luminal B, Luminal-Her2, HER2 and TNBC, respectively. In an updated and expanded analysis of this dataset, now including 2233 consecutive patients who underwent BCT between , Braunstein et al. 15 reported a crude overall LR rate of 3.1% at a median follow-up of 106 months and the 8-year risk of LR was 1.8%, 5.5%, 2.2%, 11.7% and 9.8% for patients with Luminal A, Luminal B, Luminal-Her2, Her2 and TNBC, respectively. In the youngest patients (23-46 years of age, n = 504) 10-year LR Kaplan Meier estimates were 6.3%, 9.3%, 3.5%, 29.3% and 9.3% for Luminal A, Luminal B, Luminal-Her2, Her2 and TNBC patients, respectively. Luminal B subtype (vs. Luminal A; HR 2.64, p = 0.001), HER2 subtype (HR 5.42, p <0.01) and TNBC (HR 4.33, p <0.01) were independently associated with increased LR risk on multivariable analysis. Younger age 50 years (HR 0.56 for patients >50 year, p 0.01) and increasing number of positive nodes (HR 1.06 per involved node, p = 0.004) were also independently 224

225 Age, molecular subtypes and local therapy decision-making associated with increased LR risk. 15 It is important to note that none of the patients included in these studies received anti-her2 therapy. In the more recent populationbased study by Aalders et al., where all HER2+ patients treated after 2005 received trastuzumab, 5-year LR risk was <6% among all tumor subtypes and type of surgery was not associated with risk of LR. 8 Similarly low LR rates among all tumor subtypes were observed among 1994 patients 50 years of age treated with mastectomy without post-operative RT in a multiinstitutional cohort study by Truong et al. 16 After a median follow-up of 4.3 years the crude LR risk was 1.3%, 2.3%, 4.6%, 5.6% and 2.2% for Luminal A, Luminal B, Luminal-Her2, HER2 or Basal-type, respectively. Approximately half of the HER2+ patients included in this study received anti-her2 therapy. On multivariable analysis age was not independently associated with LRR risk, factors that conferred a higher LRR risk were tumor >2 cm (HR %CI ), lobular histology (HR %CI ) and close/positive surgical margins (HR %CI ). Chapter

226 Chapter 12 Table 1. Studies evaluating tumor biology and risk of local recurrence in young patients with breast cancer. Study No. Pts Study Period Type of Surgery Follow- Up Anti- Her2 Therapy Classification of Tumor Subtype Age Local Recurrence Overall LR (%) Luminal A type Luminal B type Luminal- HER2 type HER2-type Basal-type (TNBC) Braunstein et al. (2017) BCT 106 m No IHC receptor status + grade Overall (5y/10y) n.a./4.1% 1.0%/2.5% 3.0%/6% n.a./2.5% 10.1%/12% 6.1%/11% y n=504) 2.1%/6.3% 4.3%/9.3% n.a./3.5% 7.1%*/29.3% 7.2%/9.3% y (n=557) 0.6%/0.6% 2.1%/3.2% n.a./0% 4.2%*/17.3% 5.9%/10.8% y (n=603) 0.8%/1.7% 1.3%/3.0% n.a./2.2% 3.1%*/5.3% 5.1%/7.0% y (n=569) 0.8%/1.8% 4.3%/6.2% n.a./3.3% 2.5%*/0% 5.6%/20.5% Aalders et al. (2016) ± % BCT / 55% MT 5 y Yes IHC receptor status Overall (<35 y) 3.5% 1.0% 1.2% 6.7% 4.1% Treated > % 1.0% 5.6% 4.0% Truong et al. (2013) MT (no PMRT) 4.3 y Yes IHC receptor status + grade Overall 1.8% 3.1% 1.7% 1.9% 1.9% 50 y (n=585) 1.3% 2.3% 4.6% 5.6% 2.2% >50 y (n=1409) 2.0% 3.5% 0.0% 0.0% 1.7% 226

227 Age, molecular subtypes and local therapy decision-making Arvold et al. (2011) BCT 85 m No IHC receptor status + grade Overall 3.1% (FU=85m) 1.5% 4.0% 1.0% 10.9% 8.8% <46 y (n=290) 6.5% (FU=87m) 4.7% 8.1% 3.0% 13.3% 10.2% y (n=470) 2.6% (FU=85m) 0.5% 5.5% 0.0% 18.8% 8.9% y (n=376) 2.3% (FU=83m) 1.6% 0.0% 0.0% 6.7% 8.3% y (n=298) 0.9% (FU=85m) 0.4% 0.0% 0.0% 0.0% 6.5% Cancello et al. (2010) % BCT / 26% MT 5 y No IHC receptor status + ki67 < 35 y (n=315) 10.0% 1.0% 10.0% n.a. 18.0% 26.0% y (n=2655) 5.0% 2.0% 5.0% n.a. 10.0% 18.0% Abbreviations: LR = local recurrence; BCT = breast conserving therapy; MT = mastectomy; IHC = immunohistochemistry; TNBC = triple negative breast cancer; PMRT = post-mastectomy radiation therapy; n.a. = not assessed; y = years; FU = follow-up; m = months. Values for local recurrence risk represent 5-year (Braunstein et al., Aalders et al., Truong et al. & Cancello et al.) or 10-year (Braunstein et al.) LR risk Kaplan Meier estimates or crude LR risk after median follow-up (Arvold et al.). * For the 5-year KM estimate of LR no distinction was made between Luminal-Her2 and Her2-type tumors due to limited numbers of events. The 5-year/10-year results of the Braunstein study stratified for subtype and age are not published and were obtained through personal communication. ± The authors included 1000 patients but only 589 patients could be stratified for tumor subtype; in the remaining 418 patients ER/PR or HER2 status was missing. No further subdivision in Luminal A or B type was made. In 2005 anti-her2 directed therapy (trastuzumab) was introduced in Dutch clinical practice and all HER2+ patients treated after 2005 who were included in the analysis received anti-her2 therapy. Chapter

228 Chapter 12 Surgical decision making in younger patients Earlier studies comparing LR rates following BCT versus mastectomy in young women demonstrated conflicting results (extensively reviewed in Pilewskie and King) [17]; however, more recent studies demonstrate excellent LR rates after BCT in young patients with comparable LR rates and survival between BCT and mastectomy Using data from two large tumor registries in Utah, Frandsen et al. 24 demonstrated significant improvements in LR over time for young women (<40 years) treated with BCT or mastectomy and among patients treated in the modern era (after 2000), 5- and 10-year LR rates, relapse free survival (RFS) and overall survival (OS) were equivalent following BCT or mastectomy. Five-year LR rates for patients <40 years of age treated with BCT before and after 2000, were reported as 10.9% and 3.9% respectively (p<0.05) and similar 5-year LR rates were observed for patients who underwent mastectomy (9.4% and 4.8% for patients treated before and after 2000 p<0.05, respectively). In addition, there was no significant difference in 10-year LR, RFS or OS between BCT and mastectomy in patients treated after 2000 (10-year LR 6.1% vs. 7.9% p 0.57; RFS 85.1% vs.74.8% p 0.14; OS 85.1% vs. 79.7% p 0.52, respectively). 24 The impact of local therapy on breast cancer specific survival (BCSS) was also evaluated in a report from the Surveillance, Epidemiology and End Results (SEER) database which included 7665 women aged <40 years diagnosed with stage I or II invasive breast cancer treated between At a median followup of 111 months there was no difference in BCSS observed between patients treated with BCT or mastectomy, 10-year BCSS 87.7% vs. 85.2%, p 0.01, respectively. In a recent multicenter prospective cohort study including 3000 women years of age treated between 2000 and 2008 in the United Kingdom 26, LR risk varied over time with similar LR rates for BCT and mastectomy at 18 months (1.0% vs. 1.0%, p 0.35) but higher rates for BCT at 5 and 10-years (5.3% vs. 2.6% p<0.01 and 11.7% vs. 4.9% p <0.01, respectively). After adjustment for potential confounders BCT remained associated with increased risk for LR (5-year HR %CI , p <0.01 and 10-year HR %CI , p 0.02 for BCT vs. mastectomy); however, there was no significant difference in 10-year distant disease free survival (DDFS) or OS by surgery type (HR %CI , p and HR %CI p 0.081, for DDFS and OS, respectively). 26 Although the increased risk of LR overtime following BCT in this study cannot be ignored it should also be noted that BRCA mutation status in this cohort remains to be analyzed. Lastly, in a recent meta-analysis, which included data on 22,598 patients 40 years with stage I or II breast cancer from five population-based studies and a pooled study of two clinical trials, mastectomy was not associated with an improved OS or DDFS compared to BCT (BCT vs. mastectomy; summary HR = %CI )

229 Age, molecular subtypes and local therapy decision-making Young age, molecular subtypes and survival As detailed above for risk of LRR, molecular subtype is also associated with survival in young women with breast cancer. Recent data suggest that young age is associated with worse survival outcomes in Luminal type tumors but not in the HER2+ or triplenegative subtypes. Sheridan et al. conducted a population-based analysis of 3046 women <50 years of age treated in two time periods ( and ) and reported an improved RFS and OS over time for women <40 years of age and for those years of age among all tumor subtypes. 13 Importantly, although a significant association between younger age (<40) and RFS was observed in HER2-subtype patients treated in the first time cohort (5-year RFS 49% vs. 66% for patients years, p 0.02) this was no longer evident in later years following the introduction of taxanes and trastuzumab in clinical practice (5-year RFS 81% vs. 84% p 0.879). In HR+ patients RFS also improved over time (5-year RFS 65% vs. 79% p<0.01 for patients <40 years and 72% vs. 92% p<0.01 for patients years of age treated between and , respectively) coinciding with the increased use of endocrine therapy, but remained inferior in patients <40 years of age as compared to those years of age in the later time period (5-year RFS: 79% vs. 92%, p <0.01). Whereas there was no effect of age observed on either RFS or OS in TNBC in either time period ( year RFS <40 vs years, 60 % vs. 63% p 0.87 and RFS <40 vs years, 78% vs 77% p 0.93). Similar results were reported in a recent retrospective study among 4315 women treated for stage I-IV invasive breast cancer at a single institution in Mexico between with a median follow-up time of 40 months. 28 Five-year OS was lower in young women ( 40) with HR+/Her2-disease (82% vs. 87.1%, p 0.03) as compared to their older counterparts but not for HER2+ (5-year OS 77.3% vs. 83.3% p 0.17) or TNBC (72.9% vs. 69.2%, p = 0.44). When HR+/Her2- tumors were further subdivided in Luminal A or B type tumors by using the H-score (a method to determine the intensity of receptor expression), the age difference in OS was limited to the Luminal B type (5- year OS 79.1% vs. 85.2%, p 0.03). 28 These findings were recently reproduced in a large cohort study, examining data on women with newly diagnosed stage I-III breast cancer presenting to one of eight National Comprehensive Cancer Network centers between IHC surrogates were used to define molecular subtypes and the median follow-up time was 6.4 years. In multivariable analysis, controlling for demographic characteristics only, young women ( 40 years) were 90% more likely to die of their breast cancer as compared to those aged years at diagnosis (HR 1.9% 95%CI ). Additional adjustment for treatment, stage, grade and year of diagnosis attenuated this association (HR %CI Chapter

230 Chapter ). Controlling for tumor molecular subtype and method of detection, young women remained 30% more likely to die of their breast cancer when compared to their older counterparts (HR %CI ). In stratified analysis age 40 years was associated with statistically significant increases in risk of breast cancer death among women with Luminal A (HR %CI ) and B tumors (HR %CI ). In contrast, among women with TNBC, young age was associated with a borderline increased risk (HR %CI ) and among women with HER2+ disease there was no significant association between age and BCSS (HR %CI ). A population-based cohort study from Sweden 29, which included 22,017 patients (of which 471 were <40 years of age), diagnosed between also demonstrated worse 10-year BCSS for younger patients (10-year BCSS 69%, 76%, 84% and 89% for patients <35 years, years, years or years of age, respectively). However, BCSS rates in young women (<40 years) varied between tumor subtypes (10- year BCSS; Luminal A 92%, Luminal B 75%, HER2+ 68% and TNBC 67%) and young age was only significantly associated with worse 10-year BCSS in patients with Luminal B type tumors (75% vs. 80% for patients <40 vs. 40 years; HR %CI , p 0.01). 29 Several explanations for the disparities in survival outcomes among younger and older women with Luminal breast cancer have been proposed, including less benefit of endocrine therapy or chemo-endocrine effect of chemotherapy in younger patients. Some studies revealed that young age is a predictor for decreased adherence to endocrine therapy often caused by fertility concerns. 30 Furthermore, chemotherapy-related amenorrhea, which is associated with improved outcome in HR+ breast cancer, is less likely to occur in young women. 31 Others suggest that age-related differences in geneexpression, which go beyond tumor subtypes, underlie the age-associated disparities in luminal breast cancer patients. 32 Lastly, a recent study, using the PAM50 algorithm to define molecular subtypes, showed that although the proportion of patients diagnosed with the Luminal subtype increases with age, there are proportionally more Luminal B tumors in younger patients with breast cancer as compared to older patients. 3 It is possible that the relative overrepresentation of the molecular Luminal B type tumors in younger patients underlies the age-related disparity observed in Luminal patients in studies using IHC surrogates for molecular sub typing. Thus, while one may postulate that more extensive surgery would mitigate risk factors for local recurrence leading to improved outcomes in young patients, the accumulating evidence suggests that breast cancer biology and appropriate use of systemic therapy, rather than the extent of surgery, is the major determinant of survival. 230

231 Age, molecular subtypes and local therapy decision-making SPECIAL CONSIDERATIONS IN YOUNG WOMEN Triple-Negative Breast Cancer Younger patients are more often diagnosed with the prognostic unfavorable triplenegative subtype and adjuvant systemic treatment options are limited in these patients. But as in age, it appears that the intrinsic biology, which imparts inferior outcome in TNBC is not overcome with more extensive surgery. As previously discussed, TNBC is associated with increased rates of LR after both BCT and mastectomy as compared to tumors of other biology. Nevertheless, three retrospective studies comparing outcome in patients undergoing BCT or mastectomy found no difference in LR rates or survival by type of surgery. In addition, as discussed, in more recent studies LR risk in younger patients undergoing BCT for invasive breast cancer are acceptably low even for patients with TNBC (Table 2). 12,33 Contralateral prophylactic mastectomy Although earlier retrospective reports examining a potential survival benefit after contralateral prophylactic mastectomy (CPM) generated conflicting results more recent data have led to the understanding that even among young women with breast cancer, CPM is not associated with improved survival. 37,38 With improvements in adjuvant systemic treatment modalities, included the increased use of targeted therapies, contralateral breast cancer (CBC) risk in younger patients continues to decline. 39 Tumor biology also seems to play an important role in the risk of developing a CBC: women with HR- disease face an increased risk of developing particularly HR- CBC as compared to patients presenting with a HR+ primary breast cancer, likely explained by use of endocrine therapy. 40,41 Nichols et al. 39 examined temporal trends in CBC incidence using the SEER database and reported declining incidence rates of CBC risk over time with an annual CBC incidence rate <1% (per 100/year) for women of all age categories diagnosed with an ER+ index tumor between 2001 and Slightly higher CBC incidence rates were observed among women diagnosed with an ER- index tumor between 2001 and 2005 but the annual incidence of CBC only exceeded 1% per year among women years of age (1.26% (per 100/year). 39 Neoadjuvant therapy In the setting of neoadjuvant chemotherapy (NAC) the impact of molecular subtypes on initial treatment response and risk of LRR are increasingly being recognized. High grade, HR- tumors are more likely to respond to NAC in terms of achieving a pathological complete response (pcr), which is associated with improved local-regional control; however, in the absence of achieving a pcr, the HER2 and triple-negative subtypes are associated with higher risk of LRR after both BCT and mastectomy. 42,43 Chapter

232 Chapter 12 Table 2. Local and regional recurrence rate in triple negative breast cancer Study Design No. Pts Period Stage Type of Surgery Follow-Up (median) Local Recurrence Regional Recurrence Crude risk MV analyses Sign. Factors Crude risk MV analyses HR (95% CI) HR (95% CI) Sign. factors Gangi et al. (2014) 33 Retrosp I-III BCT 60 m Luminal A (n=1341) 1.70% Luminal B (n=212) 1.90% Her2 (n=64) 12.50% TNBC vs. Luminal A TNBC vs. Luminal B TNBC vs. HER2 1.4 ( ) Tumor size 1 (T3 vs. T1 HR 4.7) 0.70% 1.6 ( ) 2.40% 1.1 ( ) 6.30% TNBC vs. Luminal A TNBC vs. Luminal B TNBC vs. HER2 1.3 ( ) 0.4 ( ) 0.2 ( ) Stage (II vs. I HR 5.2; III vs. I HR 8.3) TNBC (n=234) 4.70% n.a. n.a. 1.30% n.a. n.a. Radosa et al. (2017) 12 Retrosp I-III, TNBC only 39% BCT /61% MT 74 m Age <40 y (n=289) 6% 1 (ref) Age 40 y (n=1641) 5% None 2 2% n.a. n.a. 0.9 ( ) 2% n.a. n.a. Abbreviations: BCT = breast conserving therapy; MT = mastectomy; TNBC = triple negative breast cancer; MV = multivariable; HR = Hazard Ratio; 95%CI = 95% Confidence Interval; m = months; y = years 1 MV analyses adjusted for age, (<50, or 80y), tumor size, stage and grade. 2 MV analyses adjusted for age, type of surgery, tumor size, grade, nuclear grade, LVI, chemotherapy, radiotherapy and lymph-node status. Note. MT vs. BCT HR %CI

233 Age, molecular subtypes and local therapy decision-making As discussed, a higher incidence of HER2 and TNBC is seen in younger women, highlighting the need to understand the interplay between young age at diagnosis, tumor biology and outcome in the neo-adjuvant setting. HER2 and triple-negative subtypes are associated with higher risk of LRR after both BCT and mastectomy. 42,43 As discussed, a higher incidence of HER2 and TNBC is seen in younger women, highlighting the need to understand the interplay between young age at diagnosis, tumor biology and outcome in the neo-adjuvant setting. In a series of 595, predominantly stage I and II, breast cancer patients who received NAC followed by BCT between the highest LRR-free survival rates were observed in patients with HR+/HER2- and HR+/HER2+ disease (5-year LRR-free survival of 97.0% and 95.9%, respectively). The lowest rates of LRR-free survival were observed in HR-/HER2+ and HR-/HER2- patients (86.5% and 89.5%, respectively) with the difference in LRR-free survival being highly significant across subgroups (p 0.001). In addition, the impact of experiencing a pcr on LRR-free survival was dependent on subtype: patients with HR+ disease had excellent LRR-free survival regardless of response to NAC; whereas patients with HR-/HER2+ disease who did not achieve a pcr experienced higher rates of LRR (5-year LRR-free survival with pcr 95% vs. 83% without pcr) as did patients with HR-/HER2- disease (5-year LRR-free survival 99% with pcr vs. 84% without pcr). 44 As in the upfront surgical setting, two recent studies indicate that with the introduction of trastuzumab and the use of taxanes in addition to anthracyclines in the neo-adjuvant setting, LRR-free survival among the HR- and HER2+ subtypes have improved and LRR is now comparable across subtypes. In a series of 751 predominantly stage II disease patients undergoing NAC, which included trastuzumab in case of HER2+ disease, followed by BCT, 5-year LRR-free survival rates were 97.2%, 96.1%, 94.4% and 93.5% for patients with HR+/HER2-, HR+/HER2+, HR-/HER2+ and HR-/ HER2- disease, respectively (p 0.44). Among patients with HR+ disease who achieved a pcr 5-year LRR-free survival was 100% and similarly excellent locoregional outcomes were also observed among patients with HR-/HER2+ or HR-/HER2- disease who achieved a pcr (97.4% and 98.6%, respectively) vs. 86.6% and 89.9% in patients who did not experience a pcr. In multivariable analysis HR-/HER2- disease, stage II disease and lack of experiencing a pcr were associated with worse LRR-free survival. 45 In a smaller study Jwa et al. evaluated 335 consecutive patients treated with NAC and BCT and reported 5-year LRR free survival of 92.9% for HER2+ patients who received trastuzumab as compared to 78.3% among those who did not. 42 Chapter

234 Chapter 12 In patients treated with mastectomy after NAC, HR+ or HER2+ disease is also associated with favorable rates of LRR regardless of NAC response; whereas in TNBC achieving a pcr seems particularly prognostic for locoregional outcome. In a single-institution cohort study including 233 patients with stage II-III breast cancer who received NAC followed by mastectomy and RT, TNBC patients had the highest 5-year LRR rate as compared to HER2+ or HR+ disease (20% vs. 6% vs.4% p 0.01, respectively). None of the patients who experienced a pcr suffered a LRR at 5 years. Among patients who did not have a pcr the 5-year LRR risk was 26%, 7% and 4% for patients with TNBC, HER2+ or HR+ disease, respectively (p <0.01). 43 None of these studies focus on the interplay between age, molecular subtypes and locoregional outcome in the neo-adjuvant setting. OLDER AGE, MOLECULAR SUBTYPES AND LOCOREGIONAL RECURRENCE Breast cancer incidence has risen steadily in women over the age of 70 and, as the mean age of the global population increases, breast cancer in older patients will be increasingly encountered in clinical practice. 46 Compared to younger women, older women are more likely to present with breast cancer of favorable tumor biology. Over 80% of breast cancers encountered in women years of age are ER+, compared to 60% in patients between years. 47 In addition, the incidence of HER2+ tumors decreases from 22% in women <40 years to 10% in women 70 years. 48 Resulting in a lower incidence of HER2 and triple-negative subtypes and a higher incidence of the favorable Luminal tumor subtype in elderly patients. 14 Two smaller series, using ER/ HER2 and Ki67 or the PAM50 to determine tumor subtypes, also report an increasing incidence of Luminal A type and decreasing incidence of Luminal B, Her2 or TNBC with age. As in younger patients, HER2 and TNBC are also associated with worse distant metastasis-free survival (DMFS) and BCSS as compared to Luminal type tumors in elderly patients. 3,49 In contrast, older patients tend to present with larger tumors and some have reported an increased incidence of nodal involvement, partially explained by delay in diagnosis. 50,51 Recent studies report declining rates of LR in elderly patients diagnosed with breast cancers with favorable tumor biology. Van der Leij et al. 52 demonstrated a 5- and 10-year LR rate of 2% and 3%, respectively, for patients >65 years of age with pt1-2/pn0-2 breast cancer who underwent BCT followed by RT between Seventy-eight percent (n = 409) of these patients, who could be divided into subtypes based on IHC assessment of ER and HER2, had ER+/Her2- disease and only 40% of patients received endocrine therapy. 52 Other studies, in whom all patients received endocrine therapy, reveal even lower LR rates, ranging between 1 2.5% at 5 years in elderly patients with 234

235 Age, molecular subtypes and local therapy decision-making low-risk ER+ breast cancer treated with BCT As the favorable Luminal subtype is overrepresented in older women, there is a growing awareness that de-escalation of locoregional treatment may not significantly compromise BCSS. 46 Omission of RT after BCS in elderly patients An important pillar in the locoregional management of breast cancer is the administration of whole-breast RT after BCT. However, the premise that women at low-risk of recurrence might benefit less from RT led to several randomized controlled trials (RCT) which all suggest that there is a favorable subgroup including but not limited to of older women with ER+ tumors treated with endocrine therapy, in whom RT may not provide meaningful benefit. Three RCTs revealed that RT after BCS reduces first recurrences but the absolute benefit in the older lower-risk group receiving endocrine therapy is very small and does not impact BCSS[56]. An overview of these three RCTs is given in Table 3. 53,54,57 In the ABCSG trial the addition of RT resulted in an absolute decrease in 5-year LR of 2% (2% vs. 6% p<0.01) in post-menopausal patients (median age 66 years) with tumors 3 cm undergoing BCT followed by tamoxifen or an aromatase inhibitor (AI). 57 The most recent trial, the PRIME II trial, demonstrated similar results with low 5-year LR rates for patients 65 years of age undergoing BCT for tumors 3 cm followed by tamoxifen or an AI without RT (1% vs. 4% p<0.01). 54 Further, there was no significant difference in DFS or OS with the use of RT at 5-years of follow-up in either the ABCSG or the PRIME II trial. In the RCT with longest follow-up (12.5 years), the CALGB9343 trial, an absolute decrease in LR of 8% (2% vs. 10% p <0.01) was observed after BCT with RT and tamoxifen compared to surgery and tamoxifen alone in patients 70 years of age with tumors 2 cm; yet there was no difference in BCSS or OS. 53 Chapter

236 Chapter 12 Table 3. Overview of randomized controlled trials evaluating the role of radiotherapy after breast conserving therapy (BCT) followed by adjuvant endocrine therapy treatment. Study No. Pts Period Follow- Up ABSCG tria l5 7 CALGB9343 tria l Age (y) 4.5 y Postmenopause Tumor Size Arms No. of Local Recurrence (%) Sign. 3 cm Tam or AI 19 (6%) <0.001 (median 66) Tam or AI 2 (2%) + RT 12.6 y 70 2 cm Tam 27 (10%) <0.001 Tam + RT 6 (2%) PRIME II tria l y 65 3 cm Tam or AI 26 (4%) <0.001 Tam or AI + RT 5 (1%) Abbreviations: Tam=Tamoxifen; AI= Aromatase Inhibitor; RT= radiotherapy; y = years The importance of tumor biology in defining the subgroup of patients in whom RT could be safely omitted is illustrated by a recent study by Daugherty et al. who queried the SEER database from for patients 70 years of age receiving BCT for small ER- tumors and demonstrated that RT was associated with improved 5-year BCSS (93.1% vs. 85.0% p <0.01) and OS (81.0% vs. 61.7%, p <0.01). 58 These results highlight that tumor biology and not age alone should be considered in treatment decisionmaking. Current ongoing initiatives focusing on the use of molecular subtyping to select patients in whom RT could be omitted include the PRECISION trial (NCT ) and the Canadian LUMINA study (NCT ). Both studies aim to identify a subgroup of younger patients (> 50 years and >55 years of age respectively) with ER+ early stage breast cancer in whom RT can be safely withheld after BCT based on molecular subtype as determined by the gene-expression Prosigna test or IHC combined with Ki67 assessment, respectively. 236

237 Age, molecular subtypes and local therapy decision-making De-escalating axillary surgery in elderly breast cancer patients Standard of care for evaluation of the axilla in clinically node-negative (cn0) breast cancer patients is sentinel lymph-node biopsy (SLNB) with completion axillary lymphnode dissection (ALND) applied selectively to node positive populations depending on the procedure performed on the breast (BCT vs mastectomy) and the volume of axillary nodal disease (micro metastases vs. macro metastases and number of involved nodes). For patients with early stage breast cancer undergoing BCT, the results of the ACOSOG Z011 trial support omission of ALND in patients with 1-2 positive SLN. 59 Although there was initial controversy over the widespread applicability of these results to all breast cancer patients, there was limited concern for implementing these findings in postmenopausal patients with ER+ breast cancer as these patients comprised the majority of the Z011 cohort. At 10 years of follow-up, rates of RR in patients treated with SLN biopsy alone in Z011 were 1.5% despite the fact that 27% of patients in the ALND arm were found to have additional nodal involvement. 60 The recognition that not all axillary disease requires ALND has also been demonstrated in RCT of omission of ALND in elderly patients with low-risk ER+ breast cancer treated with tamoxifen. Martelli et al. compared ALND versus no-alnd in women aged years with primary ct1n0 (<2 cm) breast cancer treated with BCS followed by RT and tamoxifen and reported a 6% (95%CI 0%-12.6%) crude cumulative incidence of axillary recurrence in the no-alnd arm compared to zero in the ALND arm with no difference in 15-year BCSS, DMFS or OS between groups. 61 These findings are supported by the results of the IBCSG trial where no BCSS benefit was observed after ALND in patients 60 years of age surgically treated for ER+, cn0 breast cancer followed by tamoxifen. Furthermore, transiently improved quality of life (mainly during the first year after treatment) was observed in patients who did not receive ALND. 62 Notably, in the CALGB 9343 trial described above only 60% of patients had any type of axillary procedure. The long-term follow-up (12.5 years) from this trial reported an overall axillary recurrence rate of 0% and 1.5% in the RT and no-rt arms respectively. Among 392 patients who did not undergo ALND, the axillary recurrence rate was 0% and 3% in the RT and no-rt arms, again with no difference in DFS or OS; suggesting that omission of axillary staging does not translate into inferior outcomes among elderly women with ER+ breast cancer treated with endocrine therapy. 53 In aggregate these data suggest that axillary staging may have little value in women >70 years with early stage ER+ breast cancer who will receive endocrine therapy. This is highlighted by the Choosing Wisely campaign, an initiative of the American Board of Internal Medicine, which aims to reduce wasteful medical tests and avoid risks associated with unnecessary procedures[63]. As part of this campaign, the Society of Chapter

238 Chapter 12 Surgical Oncology (SSO) released a list of procedures that are commonly ordered but not always necessary in surgical oncology which included the recommendation to not routinely perform SLNB in older patients (>70 years) with ER+ breast cancer. 64 Some argue that omission of axillary staging leads to a lack of staging information however information provided from resected axillary nodes rarely influences postoperative treatment options, which are now mainly determined by tumor biology and in particular the endocrine responsiveness of the tumor. 65 With the increasing use of gene-expression profiles as an adjunct prognosticator the influence of axillary staging information on adjuvant treatment recommendations will probably decline even further. In addition, as the ALND vs no-alnd trials showed no survival differences between groups, it can be inferred that the administered adjuvant systemic treatment regimens were adequate. CONCLUSIONS Accepting the variations among tumor subtypes, improvements in adjuvant systemic therapy, likely combined with attention to margins and the use of boost RT, have resulted in acceptably low rates of LR among young women with breast cancer and young age by itself does not seem to justify aggressive local treatment. Similar improvements in local recurrence rates are seen in patients at the other end of the age spectrum, with excellent breast cancer specific outcomes for older breast cancer patients with favorable tumor biology subtypes, even after de-intensified local-regional treatment. Thus, tumor biology, and not age, should be the driver in local therapy decision-making. 238

239 Age, molecular subtypes and local therapy decision-making ACKNOWLEDGMENTS We thank Lior Braunstein and Jay Harris for providing the additional numbers on 5-year and 10-year local recurrence risk stratified for molecular subtype and age in their study. We thank Kaitlyn Bifolck for her expert editorial assistance. Chapter

240 Chapter 12 REFERENCE LIST 1. Copson E, Eccles B, Maishman T et al. Prospective observational study of breast cancer treatment outcomes for UK women aged years at diagnosis: the POSH study. J Natl Cancer Inst 2013; 105: Collins LC, Marotti JD, Gelber S et al. Pathologic features and molecular phenotype by patient age in a large cohort of young women with breast cancer. Breast Cancer Res Treat 2012; 131: Jenkins EO, Deal AM, Anders CK et al. Age-specific changes in intrinsic breast cancer subtypes: a focus on older women. Oncologist 2014; 19: van de Vijver MJ, He YD, van t Veer LJ et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002; 347: Carey LA, Perou CM, Livasy CA et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. JAMA 2006; 295: Cancello G, Maisonneuve P, Rotmensz N et al. Prognosis and adjuvant treatment effects in selected breast cancer subtypes of very young women (<35 years) with operable breast cancer. Ann Oncol 2010; 21: van Laar C, van der Sangen MJ, Poortmans PM et al. Local recurrence following breast-conserving treatment in women aged 40 years or younger: trends in risk and the impact on prognosis in a population-based cohort of 1143 patients. Eur J Cancer 2013; 49: Aalders KC, Postma EL, Strobbe LJ et al. Contemporary Locoregional Recurrence Rates in Young Patients With Early-Stage Breast Cancer. J Clin Oncol 2016; 34: Arvold ND, Taghian AG, Niemierko A et al. Age, breast cancer subtype approximation, and local recurrence after breast-conserving therapy. J Clin Oncol 2011; 29: Ahn SH, Son BH, Kim SW et al. Poor outcome of hormone receptor-positive breast cancer at very young age is due to tamoxifen resistance: nationwide survival data in Korea--a report from the Korean Breast Cancer Society. J Clin Oncol 2007; 25: Kim EK, Noh WC, Han W, Noh DY. Prognostic significance of young age (<35 years) by subtype based on ER, PR, and HER2 status in breast cancer: a nationwide registry-based study. World J Surg 2011; 35: Radosa JC, Eaton A, Stempel M et al. Evaluation of Local and Distant Recurrence Patterns in Patients with Triple-Negative Breast Cancer According to Age. Ann Surg Oncol 2017; 24: Sheridan W, Scott T, Caroline S et al. Breast cancer in young women: have the prognostic implications of breast cancer subtypes changed over time? Breast Cancer Res Treat 2014; 147:

241 Age, molecular subtypes and local therapy decision-making 14. Partridge AH, Hughes ME, Warner ET et al. Subtype-Dependent Relationship Between Young Age at Diagnosis and Breast Cancer Survival. J Clin Oncol 2016; 34: Braunstein LZ, Taghian AG, Niemierko A et al. Breast-cancer subtype, age, and lymph node status as predictors of local recurrence following breast-conserving therapy. Breast Cancer Res Treat 2017; 161: Truong PT, Sadek BT, Lesperance MF et al. Is biological subtype prognostic of locoregional recurrence risk in women with pt1-2n0 breast cancer treated with mastectomy? Int J Radiat Oncol Biol Phys 2014; 88: Pilewskie M, King TA. Age and molecular subtypes: impact on surgical decisions. J Surg Oncol 2014; 110: Coulombe G, Tyldesley S, Speers C et al. Is mastectomy superior to breast-conserving treatment for young women? Int J Radiat Oncol Biol Phys 2007; 67: Cao JQ, Truong PT, Olivotto IA et al. Should women younger than 40 years of age with invasive breast cancer have a mastectomy? 15-year outcomes in a population-based cohort. Int J Radiat Oncol Biol Phys 2014; 90: Kroman N, Holtveg H, Wohlfahrt J et al. Effect of breast-conserving therapy versus radical mastectomy on prognosis for young women with breast carcinoma. Cancer 2004; 100: van der Sangen MJ, van de Wiel FM, Poortmans PM et al. Are breast conservation and mastectomy equally effective in the treatment of young women with early breast cancer? Long-term results of a population-based cohort of 1,451 patients aged </= 40 years. Breast Cancer Res Treat 2011; 127: Mahmood U, Morris C, Neuner G et al. Similar survival with breast conservation therapy or mastectomy in the management of young women with early-stage breast cancer. Int J Radiat Oncol Biol Phys 2012; 83: Bantema-Joppe EJ, de Munck L, Visser O et al. Early-stage young breast cancer patients: impact of local treatment on survival. Int J Radiat Oncol Biol Phys 2011; 81: e Frandsen J, Ly D, Cannon G et al. In the Modern Treatment Era, Is Breast Conservation Equivalent to Mastectomy in Women Younger Than 40 Years of Age? A Multi-Institution Study. Int J Radiat Oncol Biol Phys 2015; 93: Ye JC, Yan W, Christos PJ et al. Equivalent Survival With Mastectomy or Breast-conserving Surgery Plus Radiation in Young Women Aged < 40 Years With Early-Stage Breast Cancer: A National Registry-based Stage-by-Stage Comparison. Clin Breast Cancer 2015; 15: Maishman T, Cutress RI, Hernandez A et al. Local Recurrence and Breast Oncological Surgery in Young Women With Breast Cancer: The POSH Observational Cohort Study. Ann Surg 2016 Jul 26 [Epub ahead of print]. doi: /SLA Vila J, Gandini S, Gentilini O. Overall survival according to type of surgery in young (</=40 years) early breast cancer patients: A systematic meta-analysis comparing breast-conserving surgery versus mastectomy. Breast 2015; 24: Chapter

242 Chapter Villarreal-Garza C, Mohar A, Bargallo-Rocha JE et al. Molecular Subtypes and Prognosis in Young Mexican Women With Breast Cancer. Clin Breast Cancer 2016 Nov 23 [Epub ahead of print]. doi: /j.clbc Fredholm H, Magnusson K, Lindstrom LS et al. Long-term outcome in young women with breast cancer: a population-based study. Breast Cancer Res Treat 2016; 160: Partridge AH, Wang PS, Winer EP, Avorn J. Nonadherence to adjuvant tamoxifen therapy in women with primary breast cancer. J Clin Oncol 2003; 21: Walshe JM, Denduluri N, Swain SM. Amenorrhea in premenopausal women after adjuvant chemotherapy for breast cancer. J Clin Oncol 2006; 24: Azim HA, Jr., Michiels S, Bedard PL et al. Elucidating prognosis and biology of breast cancer arising in young women using gene expression profiling. Clin Cancer Res 2012; 18: Gangi A, Chung A, Mirocha J et al. Breast-conserving therapy for triple-negative breast cancer. JAMA Surg 2014; 149: King TA, Sakr R, Patil S et al. Clinical management factors contribute to the decision for contralateral prophylactic mastectomy. J Clin Oncol 2011; 29: Boughey JC, Hoskin TL, Degnim AC et al. Contralateral prophylactic mastectomy is associated with a survival advantage in high-risk women with a personal history of breast cancer. Ann Surg Oncol 2010; 17: Peralta EA, Ellenhorn JD, Wagman LD et al. Contralateral prophylactic mastectomy improves the outcome of selected patients undergoing mastectomy for breast cancer. Am J Surg 2000; 180: Bedrosian I, Hu CY, Chang GJ. Population-based study of contralateral prophylactic mastectomy and survival outcomes of breast cancer patients. J Natl Cancer Inst 2010; 102: Pesce C, Liederbach E, Wang C et al. Contralateral prophylactic mastectomy provides no survival benefit in young women with estrogen receptor-negative breast cancer. Ann Surg Oncol 2014; 21: Nichols HB, Berrington de Gonzalez A, Lacey JV, Jr. et al. Declining incidence of contralateral breast cancer in the United States from 1975 to J Clin Oncol 2011; 29: Bouchardy C, Benhamou S, Fioretta G et al. Risk of second breast cancer according to estrogen receptor status and family history. Breast Cancer Res Treat 2011; 127: Rusner C, Wolf K, Bandemer-Greulich U et al. Risk of contralateral second primary breast cancer according to hormone receptor status in Germany. Breast Cancer Res 2014; 16:

243 Age, molecular subtypes and local therapy decision-making 42. Jwa E, Shin KH, Kim JY et al. Locoregional Recurrence by Tumor Biology in Breast Cancer Patients after Preoperative Chemotherapy and Breast Conservation Treatment. Cancer Res Treat 2016; 48: Yang TJ, Morrow M, Modi S et al. The Effect of Molecular Subtype and Residual Disease on Locoregional Recurrence in Breast Cancer Patients Treated with Neoadjuvant Chemotherapy and Postmastectomy Radiation. Ann Surg Oncol 2015; 22 Suppl 3: S Caudle AS, Yu TK, Tucker SL et al. Local-regional control according to surrogate markers of breast cancer subtypes and response to neoadjuvant chemotherapy in breast cancer patients undergoing breast conserving therapy. Breast Cancer Res 2012; 14: R Swisher SK, Vila J, Tucker SL et al. Locoregional Control According to Breast Cancer Subtype and Response to Neoadjuvant Chemotherapy in Breast Cancer Patients Undergoing Breast-conserving Therapy. Ann Surg Oncol 2016; 23: Biganzoli L, Wildiers H, Oakman C et al. Management of elderly patients with breast cancer: updated recommendations of the International Society of Geriatric Oncology (SIOG) and European Society of Breast Cancer Specialists (EUSOMA). Lancet Oncol 2012; 13: e Anderson WF, Katki HA, Rosenberg PS. Incidence of breast cancer in the United States: current and future trends. J Natl Cancer Inst 2011; 103: de Munck L, Schaapveld M, Siesling S et al. Implementation of trastuzumab in conjunction with adjuvant chemotherapy in the treatment of non-metastatic breast cancer in the Netherlands. Breast Cancer Res Treat 2011; 129: Konigsberg R, Pfeiler G, Hammerschmid N et al. Breast Cancer Subtypes in Patients Aged 70 Years and Older. Cancer Invest 2016; 34: Schonberg MA, Marcantonio ER, Li D et al. Breast cancer among the oldest old: tumor characteristics, treatment choices, and survival. J Clin Oncol 2010; 28: Wildiers H, Van Calster B, van de Poll-Franse LV et al. Relationship between age and axillary lymph node involvement in women with breast cancer. J Clin Oncol 2009; 27: van der Leij F, van Werkhoven E, Bosma S et al. Low risk of recurrence in elderly patients treated with breast conserving therapy in a single institute. Breast 2016; 30: Hughes KS, Schnaper LA, Bellon JR et al. Lumpectomy plus tamoxifen with or without irradiation in women age 70 years or older with early breast cancer: long-term follow-up of CALGB J Clin Oncol 2013; 31: Kunkler IH, Williams LJ, Jack WJ et al. Breast-conserving surgery with or without irradiation in women aged 65 years or older with early breast cancer (PRIME II): a randomised controlled trial. Lancet Oncol 2015; 16: van de Water W, Bastiaannet E, Scholten AN et al. Breast-conserving surgery with or without radiotherapy in older breast patients with early stage breast cancer: a systematic review and meta-analysis. Ann Surg Oncol 2014; 21: Chapter

244 Chapter Early Breast Cancer Trialists Collaborative Group, Darby S, McGale P et al. Effect of radiotherapy after breast-conserving surgery on 10-year recurrence and 15-year breast cancer death: meta-analysis of individual patient data for 10,801 women in 17 randomised trials. Lancet 2011; 378: Potter R, Gnant M, Kwasny W et al. Lumpectomy plus tamoxifen or anastrozole with or without whole breast irradiation in women with favorable early breast cancer. Int J Radiat Oncol Biol Phys 2007; 68: Daugherty EC, Daugherty MR, Bogart JA, Shapiro A. Adjuvant Radiation Improves Survival in Older Women Following Breast-Conserving Surgery for Estrogen Receptor-Negative Breast Cancer. Clin Breast Cancer 2016; 16: e Giuliano AE, Hunt KK, Ballman KV et al. Axillary dissection vs no axillary dissection in women with invasive breast cancer and sentinel node metastasis: a randomized clinical trial. JAMA 2011; 305: Giuliano AE, Ballman K, McCall L et al. Locoregional Recurrence After Sentinel Lymph Node Dissection With or Without Axillary Dissection in Patients With Sentinel Lymph Node Metastases: Long-term Follow-up From the American College of Surgeons Oncology Group (Alliance) ACOSOG Z0011 Randomized Trial. Ann Surg 2016; 264: Martelli G, Boracchi P, Ardoino I et al. Axillary dissection versus no axillary dissection in older patients with T1N0 breast cancer: 15-year results of a randomized controlled trial. Ann Surg 2012; 256: International Breast Cancer Study Group, Rudenstam CM, Zahrieh D et al. Randomized trial comparing axillary clearance versus no axillary clearance in older patients with breast cancer: first results of International Breast Cancer Study Group Trial J Clin Oncol 2006; 24: Choosing Wisely: An initiative of the ABIM Foundation. org. Accessed January The Society of Surgical Oncology Encourages Doctors, Patients to Question Specific Commonly-used Tests and Treatments as part of choosing wisely campaign. surgonc.org/news-publications/for-the-press/2016/07/12/the-society-of-surgical-oncology-encourages-doctors-patients-to-question-specific-commonly-used-tests-and-treatments-as-part-of-choosing-wisely-campaign. Accessed January Sanghani M, Balk EM, Cady B. Impact of axillary lymph node dissection on breast cancer outcome in clinically node negative patients: a systematic review and meta-analysis. Cancer 2009; 115:

245 Age, molecular subtypes and local therapy decision-making Chapter

246

247 CHAPTER 13 Summary in Dutch - Nederlandse samenvatting

248 Chapter 13 De behandeling van borstkanker kan bestaan uit de volgende drie pijlers: chirurgie, aanvullende radiotherapie en/of aanvullende systeem therapie behandeling. Dit proefschrift gaat over de besluitvorming omtrent aanvullende systeem therapie behandeling in patiënten met vroeg stadium borstkanker. Chemotherapie, hormonale therapie en targeted therapie zijn vormen van aanvullende systeem therapie.. Sinds 2001 is er een landelijke richtlijn opgesteld waarin aangegeven wordt welke patiënten in aanmerking komen voor aanvullende behandeling. Sinds de eerste versie van deze richtlijn is deze regelmatig aangepast en werd het indicatiegebied voor het gebruik van systeem therapie steeds verder uitgebreid. Niet iedere borstkanker patiënt heeft baat bij systeem therapie omdat bij sommige patiënten de te behalen overlevingswinst niet opweegt tegen de bijwerkingen. In de afgelopen jaren is steeds meer bekend geworden over de rol van tumor biologie in relatie tot borstkanker uitkomst. Er zijn moleculaire subtypes van borstkanker geïdentificeerd die een directe associatie hebben met de vooruitzichten op langere termijn en tevens zijn er genexpressie profielen ontwikkeld die kunnen discrimineren tussen patiënten met een hoog of laag risico op terugkeer van ziekte of uitzaaiingen. De inzichten in de moleculaire subtypes en de genexpressieprofielen dragen bij aan een betere selectie van patiënten die met aanvullende systeem therapie behandeld dienen te worden. In de meest recente borstkanker richtlijn wordt het gebruik van een genexpressie profiel aangeraden voor patiënten bij wie twijfel bestaat over het nut van chemotherapie behandeling. In het eerste deel van dit proefschrift laten we zien dat het gebruik van aanvullende systeemtherapie bij borstkanker patiënten in Nederland over de afgelopen twintig jaar aanzienlijk is toegenomen, maar dat sinds 2010 het gebruik van systeem therapie afvlakt. Daarnaast bestaat er aanzienlijke variatie in Nederland in het navolgen van de landelijke richtlijnen voor het gebruik van systeem therapie: een groot aantal patiënten ontving geen chemotherapie of hormoon therapie terwijl ze daar volgens de richtlijn wel voor in aanmerking kwamen en vice versa (hoofdstuk 2). In hoofdstuk 3 laten we zien dat het al dan niet navolgen van deze landelijke richtlijnen voor systeemtherapie niet samenhangt met de sociaaleconomische status of etniciteit van borstkanker patiënten. In het tweede deel van dit proefschrift onderzoeken we wat de invloed van genexpressie profielen is op het gebruik van chemotherapie in Nederland. In hoofdstuk 4 laten we zien dat de inzet van genexpressie profielen in Nederland over de afgelopen jaren is toegenomen. Toch ontving slechts een beperkt deel van de patiënten die volgens de richtlijn kandidaten waren voor een genexpressie profiel ook daadwerkelijk deze test. Het viel op dat genexpressie profielen meer werden ingezet bij jongere patiënten of patiënten met een hoge sociaaleconomische status. Patiënten die, volgens de richtlijn, in aanmerking kwamen voor het gebruik van een genexpressie profiel en die deze 248

249 Summary in Dutch - Nederlandse samenvatting test ook daadwerkelijk ontvingen kregen minder vaak aanvullende chemotherapie in vergelijking met patiënten bij wie geen genexpressieprofiel was gebruikt (hoofdstuk 5). Het daadwerkelijk toedienen van aanvullende chemotherapie varieerde in sommige subgroepen patiënten aanzienlijk tussen twee tijdsperiodes waarin verschillende borstkanker richtlijnen van kracht waren. Gebruik van een genexpressie profiel was geassocieerd met een meer consistent chemotherapie beleid in dezelfde tijdspanne (hoofdstuk 6). In hoofdstuk 7 worden de resultaten gepresenteerd van een prospectieve observationele studie naar de invloed van een genexpressie profiel (het 70-genen profiel, MammaPrint ) op de besluitvorming omtrent chemotherapie. Borstkanker patiënten bij wie, op basis van conventionele prognostische factoren, twijfel bestond over de indicatie voor chemotherapie, kwamen in aanmerking voor deelname aan deze studie. De arts werd gevraagd een advies te formuleren over het al dan niet geven van aanvullende chemotherapie voorafgaand aan bekendmaking van het genexpressieprofiel resultaat en achteraf. In ongeveer de helft van de in totaal 660 geïncludeerde patiënten leidde het gebruik van een genexpressieprofiel tot een verandering in het chemotherapie behandeladvies. Wat opviel was dat bij patiënten met vergelijkbare borst tumoren, het chemotherapie behandeladvies voorafgaand aan bekendmaking van het genexpressie profiel resultaat aanzienlijk varieerde tussen artsen: in de helft van de patiënten werd wel chemotherapie aangeraden, en in de andere helft niet. Ondanks de richtlijn aanbeveling om genexpressie profielen alleen in te zetten voor patiënten bij wie twijfel bestaat over de chemotherapie indicatie, worden genexpressie profielen ook vaak ingezet bij patiënten voor wie er op basis van conventionele prognostische factoren al een duidelijke aanbeveling over chemotherapie gegeven kan worden. Bij deze patiënten bestond een aanzienlijk discordantie tussen de richtlijn aanbeveling en het genexpressie profiel resultaat; d.w.z. dat op basis van de richtlijn wel chemotherapie werd geadviseerd maar volgens het genexpressie profiel niet of andersom. Bij patiënten die in lijn met de richtlijn een duidelijke aanbeveling kregen om behandeld te worden met chemotherapie was het gebruik van een genexpressie profiel geassocieerd met minder chemotherapie gebruik. Wanneer er volgens de richtlijn geen indicatie bestond voor nabehandeling met chemotherapie was de inzet van een genexpressieprofiel geassocieerd met een toename in het gebruik van chemotherapie (hoofdstuk 8). Recent zijn de resultaten gepubliceerd van een grote gerandomiseerde studie waarin aangetoond werd dat in patiënten die volgens de conventionele richtlijnen wel in aanmerking komen voor chemotherapie (klinisch hoog-risico) maar die volgens 70-genen profiel niet in aanmerking komen voor chemotherapie, chemotherapie veilig Chapter

250 Chapter 13 achterwege gelaten kan worden. Daarnaast liet deze studie zien dat het gebruik van dit genexpressieprofiel niet van toegevoegde waarde is in patiënten die volgens de conventionele richtlijn niet in aanmerking komen voor chemotherapie nabehandeling (klinisch laag risico). In hoofdstuk 9 van dit proefschrift laten we zien dat dit laatste ook geldt voor patiënten met multifocaal borstkanker. Gelijktijdig met de ontwikkeling van genexpressie profielen werden vier moleculaire subtypes van borstkanker geïdentificeerd die direct geassocieerd zijn aan borstkanker uitkomst en deze subtypes worden steeds vaker meegenomen bij de keuzes voor systeemtherapie. Als surrogaatbepaling van de moleculaire subtypes wordt vaak een combinatie van conventionele pathologie bepalingen van oestrogeen-, progesteron en Her2neu status gebruikt. De indeling in deze subtypes kan ook worden gemaakt op basis van genexpressie. In hoofdstuk 10 laten we zien dat er een forse discrepantie bestaat tussen subtypering op basis van een genexpressie test (80-genen profiel, BluePrint ) en subtypering op basis van hormoon receptoren en Her2-neu status gecombineerd met een Ki67 bepaling. Daarnaast bestaat er veel variabiliteit in de beoordeling van Ki67 tussen pathologen. Zowel genexpressie profielen als moleculaire subtypering van borstkanker worden gebruikt om beter vat te krijgen op het biologische gedrag van een tumor bij een individuele patiënt. De tumorbiologie is niet alleen van belang voor de besluitvorming omtrent systeem therapie, maar heeft ook implicaties voor de lokale behandeling. Hoofdstuk 12 van dit proefschrift is een review van de literatuur over het samenspel van moleculaire subtypes van borstkanker en de optimale strategie voor de lokale behandeling van borstkanker patiënten. 250

251 Summary in Dutch - Nederlandse samenvatting Chapter

252

253 CHAPTER 14 Review committee List of publications Acknowledgements About the author

254 Chapter 14 Review committee of this thesis Prof. dr. S.C. Linn Prof. dr. M. van Vulpen Prof. dr. H.M. Verkooijen Prof. dr. M.R. Vriens Prof. dr. T.A. King 254

255 Review committee List of publications Acknowledgements About the author LIST OF PUBLICATIONS (NOT INCLUDED IN THIS THESIS) Trends on Axillary Surgery in Nondistant Metastatic Breast Cancer Patients Treated Between 2011 and 2015: A Dutch Population-based Study in the ACOSOG-Z0011 and AMAROS Era. Poodt I., Spronk P., Vugts G., van Dalen T., Peeters M., Rots M., Kuijer A., Nieuwenhuijzen G., Schipper R. Ann Surg 2017 [Epub ahead of print] Impaired bone healing in trauma patients is associated with altered leukocyte kinetics after major trauma. Bastian O., Kuijer A., Koenderman L., Stellato R., van Solinge W., Leenen L., Blokhuis T. J Inflamm Res ,9: Coronary artery bypass grafting in Takayasu s disease - importance of the proximal anastomosis: a case report. Kuijer A., van Oosterhout M., Kloppenburg G., Morshuis W. J Med Case Rep 2015;9:283. Combined surgery for primary colorectal cancer and synchronous pulmonary metastasis: a pilot experience in two patients. Kuijer A., Furnee E., Smakman N. Eur J Gastroenterol Hepatol 2016;28:15-9. Chapter

256 Chapter

257 Review committee List of publications Acknowledgements About the author ACKNOWLEDGEMENTS The past two years have been a wonderful journey during which I have come across many special people who inspired, supported or guided me while writing this thesis. I would like to thank all of the people who contributed, in any possible way, to this thesis. Some of them I would like to thank in particular: Dr. T. van Dalen, dr. S.G. Elias, prof. dr. I.H.M. Borel Rinkes, Prof. dr. M.A.A.J. van den Bosch, prof. dr. T.A. King, prof. dr. A. Partridge, prof. S. Siesling, prof. dr. M.R. Vriens, prof. dr. E.J.Th. Rutgers, prof. dr. S.C. Linn, prof. C.H. van Gils, prof. dr. P.J. van Diest, Prof. dr. H.M. Verkooijen, prof. dr. M. van Vulpen, prof. dr. M. van den Vijver, prof. dr. R. van Hillegersberg, Marianne Deelen, deelnemende ziekenhuizen en principle investigators van de Triple A studie, alle patienten die deelgenomen hebben aan de Triple A studie, J. Verloop, dr. G.S. Sonke, dr. A. Jager, dr. M. E. Straver, Kim Aalders, Julia van Steenhoven, Annemarie Verschoor, Kay Schreuder, O. Visser, dr. C.H. Smorenburg, Annelotte van Bommel, dr. C.A. Drukker, dr. M. van der Heijder-van der Loo, dr. P.J. Westenend, dr. J. Wesseling, the Helping Ourselves Helping Others team members, Lexane van der Zee, Jelske Hondema, Famke Wouters, Maloes Mulder, Loes Rosendaal, Karlijn Schulkes, Marleen Roos, Akke Pronk, Lianne Heuthorst, Nienke Visser, Anne Sophie Kruit, Laurien Kuhry, Ellen Carbo, Anouk Ligthart, Jolijn Verbeek, Inge de Broekert, Heleen van Dinther, Irene van Hessen, Athol guys and Niles, Hans en Wiebe, Jan en Trinette, Dames 5 senior, SEOHS 2016 commissie, familie Kuijer, familie van Poppel, familie Zegers, Jos en Beppie, Margot Ringers, epidemiology en geneeskunde maatjes, Diak collega s, stafleden chirurgie Diakonessenhuis, Trois Ballon vrouwen, andere fietsmaatjes, Zoetjes. Astrid, Renée en Jos. Chapter

258 Chapter

259 Review committee List of publications Acknowledgements About the author ABOUT THE AUTHOR Anne Kuijer was born on the 14 th of August 1989 in Utrecht, The Netherlands. She grew up in Utrecht and graduated from the Herman Jordan Lyceum in Zeist in She then started her medical school at the University of Utrecht in 2008 and graduated in 2015 after completing her senior internship at General Surgery at the Diakonessen Hospital in Utrecht. During her medical school Anne started medical research, first at the department of trauma surgery at the University Medical Center Utrecht and during her last year of medical school she started conducting research on gene-expression profiling in early stage breast cancer patients under the supervision of dr. T. van Dalen at the Diakonessen Hospital Utrecht. After graduating from medical school she received a research grant from the Dutch Cancer Society (KWF), which enabled her to fill a position as clinical investigator at the Diakonessen Hospital Utrecht until December 2016, while collaborating with the University Medical Center of Utrecht. This research resulted in the current PhD thesis on the interplay between gene-expression profiling and adjuvant systemic therapy decision-making in early stage breast cancer patients. During her PhD research Anne also participated in a post-graduate Master of Science Epidemiology at the University of Utrecht, from which she is expected to graduate this year. From January until June 2017 Anne worked as a postdoctoral research fellow at the Dana Farber Cancer Institute and Harvard Medical School in Boston, under the supervision of prof. dr. T.A. King. Her research at the Dana Farber Cancer Institute focused on local therapy decision-making in young breast cancer patients. After returning to the Netherlands Anne now started her clinical career, working as a surgical resident in the Diakonessen Hospital Utrecht, and she is pursuing to start surgical training in She is motivated to continue conducting part-time medical research throughout her career. Chapter