Doctoral dissertation Tove Filtenborg Tvedskov. The evolution of the sentinel node procedure in the treatment of breast cancer

Transcription

1 Doctoral dissertation Tove Filtenborg Tvedskov The evolution of the sentinel node procedure in the treatment of breast cancer From Section of Breast Surgery Department of Plastic Surgery, Breast Surgery and Burns Copenhagen University Hospital 2017

2 The Faculty of Health and Medical Sciences at the University of Copenhagen has accepted this dissertation for public defence for the doctoral degree in Medicine. Copenhagen, 16 January 2017 Ulla Wewer Head of Faculty The defence takes place 9 June 2017 in Henrik Dam Auditorium, Panum Institute, Blegdamsvej 3, 2200 København N - 1 -

3 In memory of my mother Signe Holst Filtenborg - 2 -

4 - 3 -

5 Preface: The work that forms the basis of this thesis has been carried out at the department of Breast surgery, Rigshospitalet. The majority of the studies are based on data from the Danish Breast Cancer Cooperative Group Database. Careful and accurate registration of breast cancer patients in the database is carried out in all breast cancer centers in Denmark, which secure the completeness of the database. Without this unique long-lasting and ongoing registration many of the studies included in this thesis would not have been possible. I am grateful that I had the opportunity to make use of the valuable information from the database. Especially I would like to thank Chief statistician Majbritt Jensen for her statistical work, advice and useful ideas, points and comments. Without collaboration nationally and internationally I would never have been able to obtain results based on a sample sizes as large as what is described in this thesis. I would especially like to thank participants from the breast surgery department in Vejle, Herlev, Ringsted and Ålborg for their comprehensive work to obtain data on patients treated by sentinel node procedure at recurrence and Consultant Dorte Teilum for her idea of investigating the use of sentinel node procedure at recurrence. I must thank Consultant Eva Balslev from the department of Pathology Herlev Hospital for her advice, supervision and invaluable pathology expertise and I would like to thank researchers from the Sino-Danish Breast Cancer Research center for doing analysis on new prognostic markers for prediction of non-sentinel node metastases. I am grateful to the Danish Quality Database of Mammography screening, and especially chairman consultant Ilse Vejborg, that made a comparison of non-sentinel node metastases between clinically and screen detected cancers possible. I thank all co-author for their contribution with new points of views, advice and careful manuscript readings. I am especially grateful for my collaboration with the highly estimated researchers from Helsinki Professor Marjut Leidenius and consultant Tuomo Meretoja that made a unique collaboration with external validation of the predictive model possible. I would like to thank my younger colleges Charlotte Uth and Olav Damsgaard for their inspiring questions and views on the research topics, making the task of supervising a source for new inspiration. I must also thank my research colleges Kenneth Greving Andersen, Linnea Langhans and Birgitte Mertz for talks, discussions and help. Last but not least, a deep-felt and special thank goes to my mentor and college Professor Niels Kroman for his supervision, good advice and widespread sharing of his network. The research would not have been possible without financial support that allowed time to immerse deeper into the research field in certain periods. Especially, I am gratefull to IMK almene fond who supported my three year PhD study as well as 4 months as a Post doc. I thank Rigshospitalet for a 3 months senior scholarship that made it possible to finalize some of the recent studies in this thesis. I would like to thank Overlæge Dr.Med.Edgar Schnohr og hustru Gulberte Schnohr s for two months grant for writing this thesis. Finally, a deepfelt thank goes to my lovely husband Jesper and my four children Ellen Signe, Hans Christian, Jens Arvid og Carl Villads for their patience and support. Frederiksberg, January 2017 Tove Filtenborg Tvedskov - 4 -

6 - 5 -

7 Contents Contents List of included papers Introduction Aim Incidence and therapeutic consequences of micrometastatic disease in the sentinel node Definition of micrometastases and isolated tumor cells Stage migration Iatrogenic displacement Axillary staging as a two-stage procedure Influence on adjuvant treatment Risk of non-sentinel node metastases Prediction of non-sentinel node metastases Model development Model validation Molecular biological risk factors Method of detection as a risk factor Prediction of axillary recurrence Prognostic significance of axillary lymph node dissection Changing guidelines Axillary staging in locally recurrent breast cancer Use of SLND in locally recurrent breast cancer Proposed changes in guidelines Conclusion and perspectives Conclusion Future reduction in axillary surgery English summary Danish Summary/sammenfatning Reference List Papers included

8 List of included papers This thesis is based on the following 10 papers I. Tvedskov TF, Jensen MB, Balslev E, Ejlertsen B, Kroman N. Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: A nationwide study. Eur J Cancer 2011 Apr; 47(6):872-8 II. III. IV. Tvedskov TF, Jensen MB, Kroman N, Balslev E. Iatrogenic displacement of tumor cells to the sentinel node after surgical excision in primary breast cancer. Breast Cancer Res Treat Jan;131(1):223-9 Damgaard OE, Jensen MB, Kroman N, Tvedskov TF. Quantifying the number of lymph nodes identified in one-stage versus two-stage axillary dissection in breast cancer. The Breast 2013, Feb 22(1):44-46 Tvedskov TF, Jensen MB, Lisse IM, Ejlertsen B, Balslev E, Kroman N. High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the sentinel node. Int J Cancer 2012, Nov 15;131(10): V. Tvedskov TF, Jensen MB, Balslev E, Kroman N. Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Acta Oncol. 2014, Feb;53(2): VI. VII. VIII. IX. Tvedskov TF, Meretoja TJ, Jensen MB, Leidenius M, Kroman N. Cross-validation of three predictive tools for non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Eur J Surg Oncol. 2014, Apr;40(4): Tvedskov TF, Bartels A, Jensen MB, Paaschburg B, Kroman N, Balslev E, Brünner N. Evaluating TIMP-1, Ki67 and HER2 as markers for non-sentinel lymph node metastases in breast cancer patients with micrometastases to the sentinel lymph node. APMIS 2011, Dec;119(12): Tvedskov TF, Jensen MJ, Balslev E, Garne JP, Christiansen P, Ejlertsen B, Kroman N. Risk of non-sentinel node metastases in patients with symptomatic cancers compared to screen-detected breast cancers. Acta Oncol. 2015, Oct 9:1-5 Tvedskov TF, Jensen MJ, Ejlertsen B, Christiansen P, Balslev E, Kroman N. Prognostisk significance of axillary dissection in breast cancer patients with micrometastases or isolated tumor cells in sentinel nodes: A nationwide study. Brest Cancer Res Treat. 2015, Oct;153(3): X. Uth CC, Christensen MH, Oldenbourg MH, Kjær C, Garne JP, Teilum D, Kroman N, Tvedskov TF: Sentinel lymph node dissection in locally recurrent breast cancer. Ann Surg Oncol. 2015, Aug;22(8): Study I, IV and VII were included in my PhD thesis entitled: Staging of women with breast cancer after introduction of the sentinel node guided axillary dissection defended at Copenhagen University, May

9 Introduction The surgical treatment of breast cancer has changed dramatically over the latest centuries and has become increasingly less aggressive. Treatment of the breast tumor has changed from Halsted s mastectomy, where breast, pectoral muscles and regional lymph nodes were removed (1), over modified radical mastectomy to breast conserving surgery supplemented by oncoplastic techniques to improve the cosmetic results (2). Likewise, the treatment and staging of the axilla for lymphatic spread has changed. Axillary lymph node dissection (ALND) was previously the standard procedure for staging of the axilla. This procedure is associated with considerable morbidity (3;4) and is redundant in women without lymph node metastases. In 1994 the sentinel lymph node dissection (SLND) was introduced in the treatment of breast cancer (5) as a procedure to identify patients who could be spared an ALND. In the following years, the use of the procedure in breast cancer rapidly increased and in 1997 the first breast cancer patient in Denmark was treated by SLND (6). In 2004 the sentinel node procedure was fully implemented in all departments of breast surgery in Denmark (6), and has now replaced ALND as standard procedure for staging of the axilla in clinically node negative primary breast cancer. Under the sentinel node procedure, the first lymph nodes receiving lymph drainage from the breast is identified, removed and examined (7). Examination is done intraoperatively on frozen sections in the majority of cases, and a supplementary final conventional histological examination is done postoperatively. Only in case of metastatic spread to these first lymph nodes, a completion ALND is performed (8). SLND can accurately stage the axilla by removing in average only two lymph nodes (8), thereby causing less arm morbidity than ALND (4). Today, more than 3000 sentinel node procedures are performed in Denmark every year, as a part of the surgical treatment of breast cancer. This procedure spares thousands of Danish breast cancer patients the risk of arm morbidity each year. Worldwide, the SLND is becoming more and more widespread. In a European Work Package run by epidemiologists in Heidelberg, the recent trends in ALND and SLND practices among breast cancer patients in different countries and centers in Europe have been compared (9). Denmark has delivered data for this comparison. The study showed a widespread and increasing use of the sentinel node procedure from 2003 to 2010, but with large differences between countries and a potential for extended use. In 2010, the SLND was offered to 79 96% of patients with pt1 tumors and 49 92% of patients with pt2 tumors. A decrease in the use of ALND in the same period was seen, but again large differences between countries, in offering ALND to sentinel node positive patients existed. This could probably be explained by important differences in guidelines between countries. In Denmark, data from all Danish women with breast cancer have been prospectively collected and registered in the Danish Breast Cancer Group (DBCG) database since 1977, including information on several patient, tumor and sentinel node characteristics, as well as adjuvant systemic treatment and radiotherapy, and follow up status (10). Furthermore, the DBCG describes guidelines for the treatment of breast cancer in Denmark (11). The DBCG recommends SLND as standard procedure for patients with primary breast cancer without lymph node metastases verified by ultrasound of the axillary (11). In addition the SLND is recommended to selected patients with ductal carcinoma in situ (DCIS). Today, the DBCG database contains information on more than breast cancer patients. Thus, the DBCG database holds data of a unique size for investigating the use of SLND in breast cancer patients. The majority of studies included in this thesis are based on data from the DBCG database

10 Aim Overtreatment is a major problem in breast cancer. The trend towards less invasive surgical treatment should continue to minimize this overtreatment. Efforts should continue towards an additional reduction in ALND in patients without prognostic benefit from the procedure. Since the introduction of the SLND in the treatment of breast cancer, the focus has changed from identifying patients who can safely be spared an ALND, to identifying patients who will benefit from, and therefore do need, an ALND. The selection of patients for SLND and subsequent ALND is of utmost importance to secure a tailor-made treatment of the axilla. ALND should be avoided in sentinel node positive patients without prognostic benefit from ALND and the use of the SLND should be extended to groups of breast cancer patients, where SLND is not yet considered as standard procedure, despite some of these patients being without axillary lymph node metastases. The work included in the present thesis was carried out to optimize the use of the sentinel node procedure in the treatment of breast cancer, to tailor the treatment in a way, so that overtreatment and unnecessary side effects is kept to a minimum. The aim of this thesis was to: investigate the incidence and therapeutic consequences of identification of micrometastatic disease in the sentinel node identify patients with metastases in the sentinel node who do not need a completion ALND extend the use of SLND beyond patients with primary breast cancer - 9 -

11 Incidence and therapeutic consequences of micrometastatic disease in the sentinel node Definition of micrometastases and isolated tumor cells In average only two lymph nodes are initially removed by the SLND, compared to more than ten lymph nodes by ALND (8). Removing fewer nodes has made more extensive histopathological examinations possible (12) and as a result more metastases are found (13), especially micrometastases and isolated tumor cells (ITC)(Fig 1). The definition and staging of these small metastases has changed along with the increasing evidence of their significance. When introducing the SLND in the treatment of breast cancer, micrometastases and ITC were classified together as metastatic lesions no larger than 2 mm, according to the fifth edition of the American Joint Committee on Cancer (AJCC) staging manual from With the increasing identification of small metastases after introduction of the SLND, this definition was changed in the sixth edition of the AJCC staging manual in Micrometastases were subsequently defined as tumor cell deposits no larger than 2 mm, and ITC were defined as tumor cell deposits no larger than 0.2 mm (14). Patients with micrometastases in the sentinel node were considered node positive (pn1mi) and patients with ITC were considered node negative (pn0(i+)). In Denmark, in addition to size, cell count was used to classify metastases. Metastases between 10 and 100 tumor cells were defined as micrometastases, and single cells or cell clusters of less than 10 cells were defined as ITC (8). The change in definition over years has made comparison of case material from different periods difficult and in some studies re-evaluation of specimens from the early SLND era was necessary (15). In 2010, cell count was added to the AJCC definition (16). Along with an increased understanding of the prognostic significance of small metastatic deposits, the staging of these patients is still under debate (17). Figure 1 (From (18)) a) Micrometastases in the sentinel node b) Isolated tumor cells in the sentinel node

12 Stage migration The use of SLND, with more extensive examination of the removed lymph nodes, has led to identification of more metastases in the axillary lymph nodes. The identification of low volume metastases has increased further with the recent introduction of the one-step nucleic acid amplification (OSNA) in some international centres for examination of the sentinel nodes (19;20). The increased identification of metastases causes stage migration, where patients that were previously regarded as node negative today are staged as node positive, not because of a more advanced stage of the disease but solely due to introduction of a new and more detailed method for lymph node examination (21). Few studies have investigated the magnitude of stage migration after introduction of SLND and only four have been population-based (22-25). Two studies have investigated the magnitude of stage migration in different parts of the Netherlands. Both studies showed an increase in the proportion of node positive patients mainly caused by an increase in patients with micrometastases (23;24). A previous smaller study including Danish patients from three different counties indicated a similar stage migration in Denmark, but patients with micrometastases and ITC were not addressed separately (22). We have determined the magnitude of stage migration after introduction of the SLND on a national basis in Denmark (Paper I) (25), which is the largest study to date on the subject. This nationwide study, based on data from more than 24,000 breast cancer patients from the DBCG database, the proportion of breast cancer patients diagnosed as node positive in , before introduction of the sentinel node procedure, has been compared to the proportion of node positive patients in , after introduction of the sentinel node procedure (25). The study showed that the introduction of the SLND has resulted in an increase from 5% to 9%, in the proportion of breast cancer patients identified with micrometastases in the lymph nodes, and the identification of patients with ITC has increased from basically non-existing to around 2 % (25). There was no difference in the proportion of patients diagnosed with macrometastases (Fig 2). Figur 2: Stage migration after introduction of SLND in breast cancer treatment in Denmark. (From (18))

13 Iatrogenic displacement It could be questioned whether all these small tumor cell deposits, like micrometastases or ITC, identified in the sentinel nodes are of clinical significance and are likely to disseminate to other axillary lymph nodes (non-sentinel nodes) (26). During the last decades the theory of cancer stem cells has been introduced (27). According to this theory, breast cancer cells represent a heterogeneous group of cells (28) with different metastatic potential. It has been proposed that some tumor cell deposits in the sentinel node represent insignificant tumor cells, spread from the breast by iatrogenic displacement, and not by cancerogenious spread (29;30). These cells might not have any metastatic potential. The existence of iatrogenically displaced tumor cells in the needle track after core needle biopsy is well described, and is not related to increased risk of local recurrence (31). The discussion on the existence of a similar iatrogenic displacement of tumor cells to the sentinel node has been going on for years but without concluding evidence (29;32). Studies have shown an increased incidence of ITC in the sentinel node after needle biopsy of the primary tumor. These tumor cells have a different morphology than metastatic tumor cells and are often placed in the subcapsular sinus of the sentinel node. The studies are however small and the clinical significance of this observation is unclear (29;30;32). In addition to needle biopsy, some breast cancer patients are offered a surgical excision to confirm the cancer diagnosis before final cancer surgery. Studies indicate that this procedure, like a needle biopsy, can cause iatrogenic displacement, but the existence and the magnitude is still under debate. One study including 663 patients with either fine needle biopsy, core needle biopsy, or surgical excisional biopsy has shown a significantly lower risk of sentinel node metastases after previous surgical excision compared to previous needle biopsy in an adjusted analysis (P=0.04) (33). The difference was however only seen in the group of patients with macrometastases and could represent residual confounding. In another study including 4016 patients, a significantly increasing risk of sentinel node metastases detected by immunohistochemistry (IHC) was seen after previous intervention on the breast; 1.2% after no previous biopsy, 3.0% after previous fine needle biopsy, 3.8% after core needle biopsy, and 4.6% after surgical excisional biopsy (P=0.002)(34). In this study adjustment for tumor size, lymphovascular invasion and location of tumor in the breast was done. In 2012, we conducted a Danish nationwide study including 17,374 patients from the DBCG database, to investigate whether a surgical excision could lead to iatrogenic displacement of tumor cells to the sentinel nodes (Paper II). The study compared the risk of sentinel node metastases in 414 breast cancer patients with previous surgical excision with 16,960 patients without previous surgical excision. We found that a recent Table 1: Sentinel node status of 17,374 Danish breast cancer patients operated with SLND according to recent surgical tumor excision. (From Paper II (35)) Recent surgical tumor excision Yes No Crude OR 95% CI P- value Adjusted OR 95% CI P- value Sentinel node status No. % No. % Negative , ITC ;4.51 < ;5.43 <.0001 Micrometastases , ; ; Macrometastases , ;0.54 < ; Total , Abbreviations: SLND: Sentinel Lymph Node Dissection, ITC: Isolated Tumor Cells; OR: Odds Ratio; CI: Confidence Interval

14 surgical excision resulted in a nearly four-fold increased risk of having ITC in the sentinel node (Odds Ratio (OR)=3.73, 95% Confidence Interval (CI)= ; P<0.0001)(35). The results supported the existence of iatrogenic displacement of tumor cells from the breast to the sentinel node after surgical excision biopsy (Table 1). Patients with diagnostic surgical excision before cancer surgery is a highly selective group of patients where the cancer diagnosis is not evident, and therefore adjustments were made for tumor size, lymphovascular invasion, histological type, and malignancy grade. Despite adjustment for these known confounders, we found an unexplained significantly lower risk of macrometastases in the sentinel node (OR=0.67; 95% CI ; P=0.001) after previous surgical excision, compared to patients without surgical excision, and some degree of residual confounding might exist. Like others, we did not find an association between time interval from the previous intervention of the breast to the sentinel node procedure and the degree of iatrogenic displacement (36). Despite ITC generally being more common in the sentinel node of patients with lobular carcinomas (37), the displaced tumor cells were especially seen in patients with ductal carcinomas. None of the 23 patients with ITC in the sentinel node after previous surgical excision had non-sentinel node metastases (Table 2). Despite the small sample size this may indicate that these iatrogenically displaced cells are clinically insignificant and should not lead to a completion ALND. Table 2: Number of patients with NSN metastases out of 369 Danish breast cancer patients with ITC and 1,793 with micrometastases in the SN, according to recent surgical tumor excision. (From Paper II (35)) Recent surgical tumor excision Yes No NSN metastases Total NSN metastases Total P-value Yes No Yes No ITC 0 0% % Micrometastases % % 1,411 1, Abbreviations: SN: sentinel node; NSN: non-sentinel node; ITC: isolated tumor cells It is possible that a similar iatrogenic displacement exists in patients with DCIS (29;38). Some of these patients are exposed to several biopsies including stereotactic biopsies and excisional biopsies before the diagnosis is settled. It has been shown that DCIS patients with several interventions before surgery have an increased risk of tumor cells in the sentinel node (39). If tumor cell deposits are found in the sentinel node of patients with DCIS, they are up-staged to invasive carcinomas, and some are offered an ALND. Further investigation on the existence and significance of iatrogenic displacement of tumor cells in patients with DCIS is warranted. Axillary staging as a two-stage procedure In % of patients with micrometastases or ITC, metastases are not identified on the intraoperative frozen sections of the sentinel node, but found by the final postoperative histopathological examination, even if IHC are used on frozen sections (40;41). These patients were until 2012 offered an additional ALND as a second procedure. This two-stage surgical procedure may potentially further increase the risk of arm

15 morbidity, due to increased formation of scar tissue and risk of nerve damage due to difficult surgical dissection in the previously operated axilla, in addition to prolonged hospital stay and increased economic expenses (42;43). Furthermore, it has been hypothesized that fibrosis caused by the primary surgical procedure may hamper surgery and subsequent histopathological examination at the second surgical procedure (42-46), thereby reducing the number of lymph nodes removed and identified. This will impair locoregional control if metastatic lymph nodes are left in the axilla. In a large nationwide study, including 8257 Danish breast cancer patients treated by SLND and ALND in either a one-stage (7393 patients) or two-stage (864 patients) procedure, we found that the average number of lymph nodes removed was 15.6 in the two-stage procedure compared to 16.6 in the one-stage procedure (Paper III) (47) (Fig 3). Due to the large sample size this slight difference was in fact statistically significant (p<0.0001), but hardly clinically relevant. This result was in line with the result from an American study including 1003 patients from the ACOSOG Z0010 and Z0011 trials, where in average 14 lymph nodes were removed in patients treated by a one-stage procedure as well as a two-stage procedure (45). Similar results of only a slight difference in the numbers of lymph nodes removed by a one-stage vs. two-stage procedure have been shown in two minor studies (42;48). These results indicate that impaired surgical treatment after a two-stage procedure cannot be supported, and intraoperative identification of metastases in the sentinel node at the primary surgical procedure is not of crucial importance for precise staging and local control. Figur 3: Number of lymph nodes removed in breast cancer patients according to age and one- or two-stage procedure 1: one-stage procedure; 2: two-stage procedure. (Based on data from Paper III (47) by permission from MD Olaf Damgaard) At the same time a large Danish cohort study, including 2847 patients from the same period and using a detailed questionnaire, showed that there was no statistically significant difference in the morbidity between the two groups treated by one- or two-stage procedures, respectively (49). This further supports the safety of a two-stage axillary procedure

16 Metastases not diagnosed at the intraoperative examination on frozen sections are in particular micrometastases and ITC with a low risk of non-sentinel node metastases. Identification of these small metastases on frozen sections could be increased by the use of IHC analyses. In 2012 a Danish study showed that significantly more micrometastases and ITC were identified in the sentinel nodes intraoperatively in a center using IHC analysis on frozen sections compared to a center of similar size using only conventional haematoxylin-eosin staining (40). Accordingly, it was concluded that intraoperative use of IHC on frozen sections would reduce the proportion of patients undergoing a two-stage axillary procedure. Since then, the Danish guidelines for ALND have change and ALND is no longer recommended in patients with only micrometastases or ITC in less than three sentinel nodes (11). Due to this change in guidelines, it is no longer important to identify micrometastases or ITC intraoperatively, and the use of IHC on frozen sections has become redundant. In 2011 the St. Gallen International Expert Consensus Guidelines stated that routine use of IHC analysis on sentinel node specimens was no longer indicated (50). Influence on adjuvant treatment In Denmark adjuvant systemic treatment is offered to patients with primary breast cancer and at least one of the following characteristics: tumor size >1 cm, age < 60 years, malignancy grade II or III, negative hormone receptor status, or positive HER2 status (Table 3). Table 3: Prognostic groups of breast cancer patients in Denmark according to DBCG guidelines. Only patients allocated to group I are not offered adjuvant systemic treatment ( Age Tumor size Positive lymph nodes >/=60 years </= 10 mm 0 Type and malignancy grade Ductal grade I/unknown Lobular grade I-II /unknown Other type Ductal II III, Lobular III ER status (% positive) </=10%/unknown Medullar (neg) HER2 status Negative/unknown Positive DBCG prognostic group 0-9% II >/=1 II >10 mm II < 60 years II I II II In addition adjuvant systemic treatment is offered in case of axillary metastases, including micrometastases (10). Thus, the increased identification of micrometastases in the sentinel node could potentially result in an increase in the proportion of patients offered adjuvant systemic treatment. However, in the Danish study on stage migration after introduction of the SLND, the increased identification of small metastases did not seem to affect the proportion of patients offered adjuvant systemic treatment by more than 1%, due to the use of several other independent risk factors in stratification for adjuvant treatment (Paper I) (25). The increasing tendency towards including new prognostic markers, including genetic subtypes, as

17 high risk criteria (51), will diminish the consequences of stage migration even further, because nodal status is gradually losing its importance as a prognostic marker. Risk of non-sentinel node metastases Far from all patients with only micrometastases or ITC in the sentinel node have metastatic spread to nonsentinel nodes. In general, only about 15-20% of patients with micrometastases (52) and 10-15% of patients with ITC (53) in the sentinel node have macrometastatic spread to non-sentinel nodes. Thus, the majority of patients with small metastases in the sentinel node does not benefit from the additional ALND but still run the risk of arm morbidity (3;4). In a Danish study, based on information from more than 1800 patients with micrometastases or ITC in the sentinel node registered in the DBCG database between 2002 and 2008, it was established that 18 % of patients with micrometastases and 9% of patients with ITC had metastases in non-sentinel nodes (Paper IV) (15). In other words, ALND was redundant in 82 % and 91 % of these patients, respectively. These results questioned the benefit form ALND in patients with micrometastases or ITC in the sentinel node. In patients with macrometastases, the risk of non-sentinel node metastases is generally considerably higher; around 40 % (8;54;55). In conclusion, the introduction of the SLND has increased the identification of micrometastases and ITC. These small metastases are often not identified until the final postoperative histopathological examination, resulting in completion ALND as a two-stage procedure. However, this two-stage procedure does not seem to impair the quality of surgery, and is redundant in a considerable proportion of patients because of a low risk of non-sentinel node metastases, e.g. patients with iatrogenically displaced tumor cells in the sentinel node. Furthermore, the increased identification of small metastases does not seem to affect the subsequent adjuvant treatment, because other prognostic markers including genetic subtypes, is gradually replacing nodal status as the most important marker

18 Prediction of non-sentinel node metastases Model development Not all non-sentinel node metastases will become clinical relevant. Despite a false negative rate of the sentinel node procedure at about 5%, the local recurrence rate is less than 1% after eight years of follow-up in patients with a negative sentinel node (56;57). This means that even some patients with non-sentinel node metastases can be spared an ALND without impairment of prognosis. Due to the low proportion of patients with non-sentinel node metastases when only micrometastases or ITC are found in the sentinel node, it could be questioned whether ALND will improve prognosis in these patients. Still a group of patients with a high risk of non-sentinel node metastases, which will benefit from an ALND, might exist. It would be advantageous to identify this group of patients. This could spare the majority of patients with micrometastass or ITC in the sentinel node a large, unnecessary and often twostaged axillary procedure. Several studies have tried to identify sentinel node positive patients with no further metastatic spread to non-sentinel nodes. When macrometastases are found in the sentinel node several risk factors have been identified (58-60) and scoring systems have been developed (61-63). Unfortunately these scoring systems are not very well adapted for small metastases (64-66). Only about 12% of patients with primary breast cancer have micrometastases or ITC in the sentinel node (25) and accordingly a large cohort of patients are needed to get a sufficient sample size for an investigation of risk factors for non-sentinel nodes in these patients. In Denmark, the first SLND was registered in the DBCG database in Between 2002 and 2008 a total number of 2293 breast cancer patients with micrometastases or ITC in the sentinel node have been registered in the database. This means that the DBCG database can provide a unique sample size for investigating risk factors for non-sentinel node metastases in patients with micrometastases or ITC of the patients from the DBCG database had micrometastases in the sentinel node and underwent an ALND, and 304 patients had ITC in the sentinel Table 4: Risk factors for NSN metastases in a multivariate analysis of 1521 Danish breast cancer patients with micrometastases and 299 patients with ITC in the sentinel node operated between (From Paper IV (15)) Variable OR 95% CI P-Value Micrometastases Tumor size, cm, trend Proportion of pos SN, 100% vs. <100% Lymphovascular invasion Hormone receptor status, neg vs. pos Location of tumor in upper lateral quadrant Isolated tumor cells Tumor size, >2 vs. 2 cm Age, <40 vs. 40 years Proportion of pos SN, 100% vs. <100% Abbreviations: SN: Sentinel Node, NSN: Non-Sentinel Node, ITC: Isolated Tumor Cells, neg: negative, pos: positive, OR: Odds Ratio, CI: Confidence Interval

19 node as well as an ALND. Based on these two groups of patients, risk factors for non-sentinel node metastases have been identified in a logistic regression model (Paper IV) (15). Based on these factors, two models have been developed for prediction of the risk of non-sentinel node metastases; one for patients with micrometastases and one for patients with ITC. The models stratified patients into risk groups of non-sentinel node metastases according to the number of risk factors present. In the model for patients with micrometastases, tumor size, hormone receptor status, lymphovascular invasion, location of tumor in the breast, and proportion of positive sentinel nodes were identified as risk factors for non-sentinel node metastases (Table 4). A group of patients (5%) was identified with 4 or more risk factors present, which had a high risk of non-sentinel node metastases on nearly 40%, comparable to the risk of non-sentinel node metastases in patients with macrometastases in the sentinel node (55). These patients might still benefit from an ALND. A group of patients with a low risk of non-sentinel node metastases less than 10% could not be identified. In the model for patients with ITC, age at diagnosis, tumor size, and proportion of positive sentinel nodes were identified as risk factors for non-sentinel node metastases (Table 4). Only four patients had all three risk factors present. These patients had a 75% risk of non-sentinel node metastases, but this risk estimate was based on a very small sample size. 32% of patients with ITC had none of the risk factors present and a risk of non-sentinel node metastases at only 2%. Model validation Before a predictive model can be taken into clinical use, validation in a new dataset is necessary. Since the development of the two models, registration of patients with micrometastases and ITC in the DBCG database has continued and between 2009 and 2010 another 900 patients have been registered. These patients were used for validation of the two models. The accuracy of the model for patients with micrometastases changed only slightly from an AUC=0.64 in the original cohort to 0.63 in the validation cohort, while the accuracy of the ITC model, based on a minor number of patients and fewer risk factors, dropped from 0.73 in the original cohort to 0.60 in the validation cohort (Paper V) (67)(Table 5). It has been shown, that validation of previously developed predictive models often fails, when tested in a foreign population (61;68). In example, the Tenon score was developed in a French population and worked well in the French validation study (69), but did not perform very well in a Swedish population (70). Likewise, the model from the Memorial Sloan-Kettering Cancer Center (58) worked well in other American populations (71), but was not very precise in a Hungarian population (72). Finally, only two out of twelve tested models worked well in a Chinese population (73). Therefore validation of the models in a Danish dataset could not secure that the models would work well in centers outside of Denmark. By the same time as the two models were developed in Denmark, development of a predictive model for non-sentinel node metastases in patients with micrometastases or ITC in the sentinel node was done in Helsinki, based on information from 484 Finnish breast cancer patients. In this model the risk of nonsentinel node metastases was associated with tumor size and multifocality, and the AUC was 0.68 (68). The Finnish model had been internally validated on 51 patients with an AUC=0.79. An agreement was made to test the Danish and Finnish models in the opposite dataset (Paper VI). The results showed that the Danish model for micrometatases was accurate when tested in the Finnish cohort, with a slight change in AUC from 0.64 to This model was developed based on the largest sample size and included the largest number of risk factors. Thus, the accuracy of the Danish micrometastatic model did not change, under neither internal nor external validation

20 The AUC of the Finnish model decreased from 0.68 to 0.58 when tested in the Danish cohort, and the AUC of the Danish model for ITC decreased from 0.73 to 0.52, when tested in the Finnish cohort (74)(Table 5). The lower performance of these two models under cross-validation could be due the development based on smaller sample sizes where fewer risk factors were identified. In addition, demographic differences and differences in diagnosis, surgical techniques, and pathology methods might exist between countries. Apart from the Danish and Finnish models, a French model exists for the prediction of non-sentinel node metastases in patients with micrometastases in the sentinel node. This model was developed from information on 909 breast cancer patients and uses four different risk factors; tumor size, detection by haematoxylin-eosin vs. IHC, lymphovascular invasion and histological type (75). The AUC for the French model was 0.66 in the initial series. The model was validated in a series of 484 patients but the AUC was not reported (76). The Finnish and the French micrometastatic model have previously been validated on 313 breast cancer patients from 5 different institutions in Europe. The two models did not perform well in this multi-institutional cohort, especially not in predicting patients with high risk of non-sentinel node metastases. The AUC for the two models was 0.58 and 0.56, respectively, at this external validation (77). Both the Finnish and the French model focused on prediction of patients with a low risk of non-sentinel node metastases. Along with the increasing evidence suggesting that ALND can be omitted in patients with only micrometastases or ITC in the sentinel node without impairment of prognosis (78), prediction of patients with low risk of non-sentinel node metastases is no longer important. The Danish micrometastatic model can be considered as a robust model for prediction of patients with micrometastases in the sentinel node with high risk of non-sentinel node metastases, which might benefit from ALND (74)

21 Table 5: Performance of models for predicting low and high risk of non-sentinel node metastases in original, internal validation and external validation cohorts. (From Paper VI (74)) Model Cohort No. No. with NSN metastases (%) Sensitivity (Low- risk) Specificity (Low-risk) Low-risk patients (%) Sensitivity (High-risk) Specificity (High-risk) High-risk patients (%) AUC (95% CI) Helsinki model Original (7.4) (84.1) Internal validation 51 5 (9.8) (70.6) (0.4) 0.68 ( ) ( ) DBCG model (MIC) DBCG model (ITC) External validation (16.6) (91.0) (0.2) 0.58 ( ) Original (17.9) (4.3) (14.4) 0.64 ( ) Internal validation (16.8) (4.3) (22.0) 0.63 ( ) External validation (8.6) (5.0) (18.7) 0.63 (0.49;0.76) Original (9.4) (32.0) (17.0) 0.73 ( ) Internal validation (12.8) (39.0) (17.0) 0.60 ( ) External validation (5.8) (49.5) (9.7) 0.52 (0.32;0.71) Abbreviations: DBCG: Danish Breast Cancer Cooperative Group, CI: Confidence Interval, AUC: Area Under the Curve, MIC: Micrometastases, ITC: Isolated Tumor Cells Cut point of 10% risk of non-sentinel node metastases or (for DBCG ITC model) no risk factors present Cut point of 30% predicted risk of non-sentinel node metastases or (for DBCG ITC model) more than one risk factor present

22 Molecular biological risk factors Despite the extensive validation, the two models do not give a perfect distinction between patients with and without non-sentinel node metastases. Therefore efforts have been made to optimize the models by including additional new risk factors for non-sentinel node metastases. It is possible that new molecular markers can be used in the prediction of non-sentinel node metastases and be included in and improve existing models. Only few markers have been tested and the results have been disappointing (64;79). In 2011, a Korean study tested 14 different biological markers in 205 breast cancer patients with sentinel node metastases but none of the markers were significantly associated with non-sentinel node metastases (80). In Denmark, we have tested Tissue Inhibitor of Metalloproteinase 1 (TIMP1) IHC, Ki67 and HER2 for the prediction of non-sentinel node metastases (Paper VII). All three factors are involved in cancer cell dissemination. TIMP1 is a protease inhibitor found to be associated with lymph node metastases and risk of recurrence in breast cancer patients (81;82). KI67 is a nuclear antigen used as a marker for cell proliferation and associated with poor prognosis (83). Finally, HER2 is a tyrosine kinase receptor involved in regulation of breast cell growth and associated with poor prognosis (84). The three markers were tested in a matched case-control study including 75 patients from Herlev Hospital operated between 2001 and 2007 with micrometastases in the sentinel node and a completion ALND. None of the three markers were significantly associated with non-sentinel node metastases (85). At present, testing of additional markers is going on in the same patient material; TIMP1 FISH, Plasminogen activator inhibitor 1 (PAI1), BMI1 and Mel-18. Translational research has indicated that these markers may serve as prognostic markers in breast cancer (86-88). Testing of BMI-1 and Mel-18 is performed at the Klinikk for kreft, kirurgi og transplantasjon in Oslo, Norway. In 2000 Perou et al. described four different breast cancer subtypes based on gene expression patterns (89). The use of these genetic subtypes as prognostic markers in breast cancer is increasing (89). It is possible that the subtypes could be used in the selection of sentinel node positive patients for completion ALND. The genetic subtypes have a high concordance with subtypes based on traditional IHC analyses of estrogen receptor status, progesterone receptor status, HER2 status, and KI67 index. A recent nationwide Danish study based on more than 20,000 patients has shown that the risk of axillary lymph node metastases is associated with breast cancer subtypes based on receptor status (90). A similar association between subtypes and non-sentinel node metastases has not yet been found (91-93), and so far genetic subtypes have not been used to select sentinel node positive patients for ALND. Method of detection as a risk factor Another way to identify patients with sentinel node metastases that may be more likely to spread beyond the sentinel node is to look at the method of cancer detection. After introduction of mammographic screening programs in many countries in the western world a peak in the incidence of breast cancers was seen (94). This is thought to be due to lead time bias, where breast cancers are identified before they become clinically evident. It is well known, that these screen-detected cancers are smaller and have a lower rate of lymph node metastases compared to symptomatic cancers (95;96). It has also been stated, that these cancers represent a group of slow growing cancers with a lower malignancy grade, lower Ki67 index and higher proportion of positive hormone receptor status (96;97). These characteristics are related to a less aggressive disease and better prognosis and it could be hypothesized, that these patients have a lower risk of non-sentinel node metastases. If this is true, the method of detection could be included in the

23 predictive models and patients with cancers detected by screening should be offered less extensive axillary surgery. Only few studies have looked at the risk of non-sentinel node metastases in patients with screen-detected breast cancers. The studies are small and only limited adjustment for confounders have been made (98-100). The most remarkable results are found in a Swedish study including 143 breast cancer patients with micrometastases or ITC in the sentinel node. The study found a five-fold increased risk of non-sentinel node metastases in symptomatic cancers compared to screen-detected cancers after adjustments for tumor size and malignancy grade (OR=5.1; 95% CI=1.4-19) (99). These results pointed in the direction of a modified treatment of the axilla in patients with screen-detected cancers. Patients with screen-detected breast cancers are however a highly selected group of patients. The Swedish study included patients between the age of 30 and 88 in the group with symptomatic cancers. Studies have shown that tumor characteristics vary by age, and that young age is associated with poor prognosis (101). In the Swedish study no adjustment for age was performed, but only adjustments for malignancy grade and tumor size, and residual confounding cannot be excluded. We tried to verify the Swedish results in a Danish dataset of 955 patients from the screening population between the age of 50 and 70, registered in the DBCG database with micrometastases or ITC in the sentinel node, after introduction of the national mammographic screening program (Paper VIII). 481 of these patients were identified in the nationwide Danish Quality Database of Mammography Screening with screen-detected cancers. The remaining 474 patients were considered having symptomatic cancers. We found no difference in the risk of non-sentinel node metastases between the two groups (OR=1.07; 95% CI= ; P=0.69), neither in patients with micrometastases nor in patients with ITC (102). In contrast to the Swedish study, adjustments for tumor size, proportion of positive sentinel nodes, lymphovascular invasion, hormone receptor status, and location of tumor in the breast were made in the Danish study; all risk factors identified in the Danish models for predicting non-sentinel node metastases developed in 2012 (15). 181 patients with micrometastases and 756 patients with ITC were included in the multivariate analyses (Table 6). Still, no significant difference was found in the risk of non-sentinel node metastases between patients with screen-detected cancers and symptomatic cancers. Table 6: Risk factors for NSN metastases in a multivariate analysis of 756 Danish breast cancer patients with micrometastases and 181 patients with ITC in the sentinel node from the screening population operated between (From Paper VIII (102)) Variable OR 95 % CI P-value Micrometastases Tumor size, cm, trend Proportion of pos SN, 100% vs. <100% Lymphovascular invasion Hormone receptor status, neg vs. pos Location of tumor in upper lateral quadrant Symptomatic vs. screen-detected Isolated tumor cells Tumor size, >2 vs. 2 cm Proportion of pos SN, 100% vs. <100% Symptomatic vs. screen-detected Abbreviations: SN: Sentinel Node, NSN: Non-Sentinel Node, ITC: Isolated Tumor Cells, neg: negative, pos; positive, OR: Odds Ratio, CI: Confidence Interval

24 Based on the Danish results, the method of detection cannot be used in the prediction of non-sentinel node metastases and the data does not support a less aggressive treatment of the axilla in patients with screendetected breast cancer. Prediction of axillary recurrence The largest concern for breast cancer patients is not the risk of occult non-sentinel node metastases but the risk of recurrence. Accordingly, the predictive models for non-sentinel node metastases ought to be tested for their ability to predict axillary recurrences. Only one study exists on such a testing (103). In this study, four different models developed for the prediction of non-sentinel node metastases in patients with macrometastases in the sentinel node was tested for the prediction of axillary recurrence in 486 Dutch patients with micrometastases in the sentinel node without completion ALND. Only one of the four models identified a group of patients with a risk of axillary recurrence just above 10%. It is possible that a model developed for the prediction of a high risk of non-sentinel node metastases in patients with micrometastases in the sentinel node could more accurately predict axillary recurrence in this patient series. Hence, testing of the Danish micrometastatic model (15) for prediction of axillary recurrences in the Dutch data material has been proposed and initiated in collaboration with researchers in Maastricht

25 Prognostic significance of axillary lymph node dissection Studies on the the impact of ALND on axillary recurrence and survival in patients with micrometastases or ITC s in the sentinel node are few and limited by short follow-up, small sample sizes, lack of multivariate analyses, and information on adjuvant treatment ( ) and the results are conflicting. Some studies show that patients with ITC or micrometastases have a worse outcome if ALND is omitted (107;109) while others can not show any difference (106;110;111). Two large register studies exist, including 6,838 and 10,259 patients with micrometastases in the sentinel node, respectively, from the American SEER database and the American National Cancer Database (NCDB). No significant difference in overall survival or axillary recurrence was found between patients treated by SLND + ALND or SLND alone (112;113), indicating that ALND could safely be omitted in these patients. These studies were however limited by missing registration of recurrences, imprecise number of removed sentinel nodes, and missing information on adjuvant treatment. The results from two recent randomized trials have further questioned the benefit from ALND in patients with micrometastases in the sentinel node. In the European IBSCSG study, where 934 breast cancer patients with micrometastases or ITC in the sentinel node were randomized to either ALND or no ALND, no difference was found in axillary recurrence or survival between groups (78). Similar results were found in the Spanish AATRN 048/13/2000 trial, including 233 patients (114). It should however be noted that both randomized trials closed down before they met the planned accrual and might be underpowered. In contrast to these studies, a Dutch research group has published the results of a large register study, including 795 patients with ITC and 1028 patients with micrometastases in the sentinel node. They found a non-significant adjusted hazard ratio of 2.4 (95% CI= ) for regional recurrence if ALND was omitted in patients with ITC in the sentinel node, and a significantly increased hazard ratio on 4.4 (95% CI= ) for regional recurrence if ALND was omitted in patients with micrometastases in the sentinel node after 5 years follow-up (115). Due to these results, the authors recommended ALND in patients with micrometastases in the sentinel node and unfavorable tumor characteristics. To investigate the safety of omitting ALND in breast cancer patients with micrometastases or ITC in the sentinel node we initiated a retrospective study in Denmark based on national data from the DBCG database (Paper IX). Until 2012 the standard treatment of Danish breast cancer patients with micrometastses or ITC in the sentinel node was a completion ALND. Still, some patients did not undergo ALND. The reason for not choosing ALND in these patients is basically unknown, but is probably due to age, comorbidity and patient preference. Accordingly, patients without ALND is a highly selected group of patients. In total, 256 patients with ITC or micrometastases in the sentinel node but without completion ALND have been registered in the DBCG database from the start of the sentinel node era in 2002 and until Table 7: Axillary recurrence and overall survival in 2074 Danish breast cancer patients with micrometastases or ITC in the sentinel node treated between 2002 and (From Paper IX (116)) 5 years cumulated incidence Micrometastases Isolated tumor cells ALND ALND Yes No Yes No Axillary recurrence OS (95% CI) 91.8 ( ) 79.4 ( ) 93.3 ( ) 87.3 ( ) Abbreviations: ITC: Isolated Tumor Cells, ALND: Axillary Lymph Node Dissection, OS: Overall Survival

26 The axillary recurrence rate in this group of patients was very low (Table 7); 1.6% after 6 years of follow-up (116), despite between 9 18 % of these patients were expected to have non-sentinel node metastases (15). After adjustment for age, no significant difference was seen in axillary recurrence between patients with and without ALND; neither for patients with micrometastases in the sentinel node (HR=1.79; 95% CI= ; P=0.44) nor for patients with ITC (HR=2.21; 95% CI= ; P=0.27) (Table 8)(116). Development of axillary recurrence from minimal metastatic disease left in the axilla might take longer than what is expected for macrometastatic disease. In a study including patients with haematoxylin-eosin negative sentinel node, where patients with metastases detected by IHC were not offered an ALND, the medium time to recurrence was 4.8 years (117). Sufficiently long follow-up for at least 5 years, like in the Danish study, is important for patients with minimal metastatic disease, to give time for an axillary recurrence to develop. As expected, the group without ALND had a significantly worse overall survival compared to patients with ALND, because one of the main reasons for omitting ALND in these patients could be age or comorbidity (Table 7). When adjusting for known risk factors for non-sentinel node metastases, age, comorbidity, and adjuvant systemic treatment and radiotherapy, there was no significant difference in overall survival between patients with and without ALND (HR=1.13; 95% CI= ; P=0.41)((Table 8) (116). These results support the safety of omitting ALND in patients with only micrometastases or ITC in the sentinel node. Table 8: Adjusted Cox proportional hazard ratios for axillary recurrence and death if ALND is omitted compared to patients with ALND in 2074 Danish breast cancer patients with micrometastases or ITC in the sentinel node treated between 2002 and (From Paper IX (116)) Axillary recurrence Death HR* 95% CI P-value HR** 95% CI P-value Micrometastases n=1673 ITC n=401 Micrometastases or ITC n= Abbreviations: ITC: Isolated Tumor Cells, ALND: Axillary Lymph Node Dissection, HR: Hazard Ratio, CI: Confidence Interval * Adjustment for age ** Adjustment for age, tumor size, histology type, malignancy grade, lymphovascular invasion, hormone receptor status, nodal status, comorbidity, adjuvant systemic treatment and adjuvant radiotherapy The remarkable difference between these results and the result from the Dutch study could be explained by the differences in the adjuvant treatment guidelines. In contrast to Danish treatment guidelines, only half of the patients included in the Dutch study received any kind of adjuvant treatment (115). In addition, it is possible that differences in adjuvant radiotherapy between studies play an important role. In fact, in the recent European AMAROS trial, where 4823 sentinel node positive patients were randomized to either ALND or axillary radiotherapy, no difference was found in the risk of axillary recurrence between groups after 6.1 years of follow-up (118). It is possible that adjuvant systemic treatment and adjuvant radiotherapy can offset the impaired prognosis after omitting ALND in patients with micrometastases or ITC in the sentinel node. In the AMAROS trial, 60 % of the included patients had macrometastases in the sentinel node. Still no significant difference in 5 years axillary recurrence, disease free survival or overall survival was seen

27 between groups (118). These results are in line with the results from the American ACOSOG Z0011 trial, where 891 patients with clinically node negative primary breast cancer treated by breast conserving surgery and maximum 2 positive sentinel nodes were randomized to ALND or no ALND. No significant difference in axillary recurrence and survival was shown after 6.3 years of follow up (110;111). Like in the AMAROS trial 60% of patients had macrometastases in the sentinel node. This study had however several methodological weaknesses (119), and both the ACOSOG Z0011 trial and the AMAROS trial were underpowered due to a low number of events. Still these studies indicate that it could be safe to omit ALND even in patients with macrometastases in the sentinel node. Recently, a Swedish randomized trial, the SENOMAC trial, has been initiated where breast cancer patients with macrometastases in the sentinel node are randomized to ALND or no ALND. Differences in recurrence and survival will be estimated. Danish breast surgery centers are going to participate in the trial. It is possible that the results from this trial will further reduce the proportion of sentinel node positive patients offered a completion ALND

28 Changing guidelines Studies on the prognostic significance of ALND in patients with micrometastases or ITC in the sentinel node have resulted in a trend towards omission of ALND in these patients. In 2009 the St. Gallen International Expert Consensus Conference stated, based on the low risk of non-sentinel node metastases, that ALND can be avoided in selected breast cancer patients with micrometastases or ITC in the sentinel node (120) (Fig 4a).Two years later, in 2011, after the publication of the results from the ACOSOG Z0011 trial, the option of omitting ALND was extended to patients with macrometastases in the sentinel node. It was however underlined that this option was only accepted for patients fulfilling the criteria for the ACOSOG Z0011 trial, e.g. clinically node negative patients, with 1 2 positive sentinel nodes, undergoing breast conserving treatment with adjuvant radiotherapy (50). In line with these recommendations, Denmark changed the DBCG treatment guidelines for patients with micrometastases or ITC in the sentinel node. Since 2012 ALND is no longer recommended in patients with less than 3 micrometastatic sentinel nodes, regardless of type of surgery in the breast (Fig 4b)(11) SLND is encouraged, and omission of ALND is accepted in sentinel node negative patients ALND is no longer recommended in patients with ITC/MIC in sentinel node ALND can be omittet in patients with MAC in sentinel node fulfilling the ACOSOG Z0011 criteria First SLND in Denmark First DBCG protokol for SLND All Danish departments of breast surgery are offering SLND to patients with tumor size < 4 cm SLND is offered regardles of tumor size Patients with ITC/MIC in sentinel node are not offered ALND 2016 Start inclusion in SENOMAC trial, where patients with MAC in sentinel node are randomised to ALND or ART Figur 4: a) Changes in the St. Gallen Guidelines for treatment of the axilla b) Changes in the DBCG guidelines for treatment of the axilla Abbreviations: SLND: sentinel lymph node dissection; ALND: axillary lymph node dissection, MAC: macrometastases; MIC: micrometastases; ITC: isolated tumor cells; DBCG: Danish Breast Cancer Group; ART: axillary radiotherapy It is possible that the predictive models for non-sentinel node metastases, based on patients from the DBCG database with micrometastases in the sentinel node, could give a better prediction of a high risk of non-sentinel node metastases than the presence of metastases in 3 or more sentinel nodes (15). In contrast to the majority of models, the Danish micrometastatic model focused on patients with high risk of nonsentinel node metastases. In the data material for model development and validation only 32 patients had 3 or more positive sentinel nodes, and only 3 (9.4%) of these patients had non-sentinel node metastases. By using the developed model on the common development and validation cohort a group of patients was identified with a particularly high risk of non-sentinel node metastases on more than 35%. ALND might be

29 more beneficial in this group of patients than in patients with 3 or more micrometastatic sentinel nodes (Table 9). Table 9: Risk of NSN metastases in 2300 Danish breast cancer patients with micrometastases in the sentinel node according to risk groups defined by number of positive sentinel nodes or predictive model No. of positive SN; 1-2 vs. >2 Predictive model: No. of risk factors; 0-3 vs. 4-5 *59 patients had missing values for predictive model Low risk group High risk group Total Total NSN metatastases (%) Total NSN metatastases (%) (17.7) 32 3 (9.4) (16.8) (35.4) 2241* Abbreviations: No.: Number, SN: Sentinel nnode, NSN: Non-Sentinel Nodes When looking at Danish patients with axillary recurrence after omitting ALND, only 5 patients without ALND, included in the study of prognostic significance of ALND in patients with micrometastases or ITC in the sentinel node (116), had axillary recurrence as first event; 2 with micrometastases and 3 with ITC in the sentinel node. One of the two patients with micrometastases had 4 or more risk factors present and was identified as high risk patient according to the predictive model developed for patients with micrometastases (Unpublished data). However the number of patients was too small to test if the predictive model for non-sentinel node metastases could predict axillary recurrence as well. Results from testing of the models to predict axillary recurrence in the Dutch data material might shed further light on this issue. In Denmark ALND is still recommended to breast cancer patients with macrometastases in the sentinel node but will be offered participation in the SENOMAC trial

30 Axillary staging in locally recurrent breast cancer Use of SLND in locally recurrent breast cancer In Denmark, treatment guidelines for patients with locally recurrent breast cancer describes that patients with previous SLND should be offered an ALND and patients with previous ALND should not receive additional axillary surgery (11). However, less than 10% of patients with local recurrence has lymphatic spread at time of recurrence (121). This means that more than 90% of patients with local recurrence and a previous SLND could again be spared an ALND. In patients with previous ALND, a new lymphatic pathway to a de novo sentinel node might have been formed. If this new sentinel node contains metastases, identification by SLND might improve locoregional control. In Denmark, there has been a growing interest to extend the potential benefits from the SLND to patients with local recurrence and some departments have started to use the procedure in selected patients. It is however questionable whether the previous axillary surgery and/or previous radiotherapy have changed the lymph drainage from the breast in such a way that SLND is hampered. Only few and smaller studies exist on the subject ( ). In 2011 and 2015 respectively, two studies were published with a reasonable number of patients; a Dutch study from 2011 including 150 patients and our Danish study from 2015 including 147 patients from 12 different departments of breast surgery in Denmark; 773 patients with a previous SLND and 74 patients with a previous ALND (Paper X). The results from both studies points in the same direction: The SLND seems to be feasible in patients with locally recurrent breast cancer. The Dutch study (the SNARB study) found a detection rate on 52%; higher in patients with previous SLND compared to previous ALND, and in patients with previous mastectomies compared to breast conserving surgery (126). In the Danish study we found a detection rate on 50%. Again the highest detection rate was found in patients with previous SLND (66%) compared to previous ALND (34%)(P=0.0001), and in patients with previous mastectomies (64%) compared to previous breast conserving surgery (48%)(121)(Table 10). However, the difference in detection rate between previous mastectomy and previous breast conserving surgery was not significant. It is possible that the trend towards an impaired detection rate in patients with previous breast conserving surgery is caused, not by the surgery itself but by the subsequent external radiation towards the residual breast. A recent Italian study included 212 patients with local recurrence after breast conserving surgery and a previously negative sentinel node. In this study they found a much higher detection rate on 92.5%. 36% of the patients had been treated with intraoperative radiotherapy or no radiotherapy at all (127). In the published studies, only few sentinel node negative patients underwent completion ALND. No additional metastases were found in these patients corresponding to a false negative rate on zero. In the Danish study, no axillary recurrences were seen after more than 3 years of follow-up (121), whereas a cumulative incidence of axillary recurrence on 3.9% was seen in the Italian study (127). An increased risk of aberrant drainage was seen in patients with previous ALND. In patients with primary breast cancer aberrant lymphatic drainage is uncommon. Especially drainage to the contralateral axilla is rare (0 2 %) (128). However, if SLND is performed at locally recurrent breast cancer after previous ALND, aberrant drainage to contralateral axilla is seen in 11-71% of the patients at lymphoscintigraphy (121;126;128). In the Danish study, 25% of patients with previous ALND had complete or partly aberrant drainage outside the ipsilateral axilla (6 patients)(121). In half of these patients sentinel node was located in the contralateral axilla and in one patient (17%) the aberrant sentinel node contained metastases (Table 10). This is in line with international results where metastases are found in 20 % of the cases if a sentinel

31 node in the contralateral axilla is removed (128). This emphasizes the need for lymphoscintigraphy in these patients, to be able to identify the location of additional metastatic lymph nodes. In the Danish study, 37 out of 73 patients (51%) with a previous SLND had a sentinel node identified without metastases. These patients could again be spared an ALND and the risk of arm morbidity. Another 11 patients did only have micrometastases or ITC in the sentinel node, and it could be argued if ALND could be omitted in these patients. In 6 out of 74 patients (8%) with a previous ALND, a positive sentinel node was identified and removed. This could potentially leading to a better local control(121). Table 10: Results of SLND at recurrence according to primary axillary operation in 147 Danish breast cancer patients. (From Paper X (121)) SLND ALND No. % No. % Total Previous operation in breast Mastectomy BCS Scintigraphy Not performed Performed SLNDAR Non-detection of SN Detection of SN Ipsilateral drainage Aberrant drainage Not performed Performed Non-detection of SN Detection of SN Ipsilateral drainage Npos Nneg Aberrant drainage Npos Nneg Abbreviations: SLNDAR: Sentinel Lymph Node Dissection After Recurrence, SLND: Sentinel Lymph Node Dissection, BCS: Breast Conserving Surgery, SN: Sentinel Node Traditionally, patients with lymph node metastases in the contralateral axilla are considered as having disseminated disease and the treatment is regarded as palliative (128). However, due to the change of lymphatic drainage after previous axillary surgery and radiotherapy the prognostic significance of lymph node metastases outside the ipsilateral axilla might have changed. Patients with lymph node metastases in

32 aberrantly draining lymph nodes at local recurrence might have the same prognosis as patients with metastases in the ipsilateral axilla (129). The prognostic significance of lymph node metastases in the contralateral axilla at recurrence is difficult to investigate because this condition is extremely rare, and even by using national register it would be difficult to obtain a sufficient sample size. Studies built on multinational cooperation are therefore of great importance. A multinational collaboration on the prognostic significance of axillary recurrence in the contralateral axilla has recently been initiated by researchers in Maastricht. Denmark are participating in this study and we have delivered data from the DBCG database on patients with recurrence in contralateral axilla. If these patients have the same prognosis as patients with ipsilateral recurrence the results can change the treatment towards a more aggressive approach to obtain locoregional control. Proposed changes in guidelines The DBCG guidelines do not yet recommend SLND at recurrence (11). Based on the current evidence, it could be suggested that SLND is used at local recurrence, including preoperative lymphoscintigraphy. If a sentinel node can be identified, frozen sections should be performed. If no metastases are found, patients with a previous SLND should not undergo ALND. In case of metastases in the sentinel node, ALND should be performed after previous SLND. If a new sentinel node is found in the ipsilateral axilla after previous ALND no further surgery should be performed unless previous ALND was incomplete. Attempts to remove sentinel nodes at extra-axillary sites should be considered, especially in the contralateral axilla. Depending on the results of the Dutch study on the prognostic significance of recurrence in contralateral axilla a contralateral ALND could be recommended. Removing internal mammary sentinel nodes is related to substantial morbidity and a prognostic advantage in primary breast cancer is still discussed (130;131). The significance of removing these sentinel nodes at recurrence is unknown. A change in radiation fields in patients with sentinel nodes at extra-axillary sites should be considered as well. In case of non-detection, treatment should follow the current guidelines

33 Conclusion and perspectives Conclusion The introduction of the SLND in breast cancer treatment spares each year hundreds of thousands of women an ALND and the following risk of arm morbidity. The more extensive examination of the lymph nodes after introduction of the SLND has however resulted in identification of more micrometastases and ITC. Only a small proportion of patients with these metastases in the sentinel node have metastases in other axillary lymph nodes and some are caused by iatrogenic displacement. Removal of additional lymph nodes by ALND in these patients do not improve prognosis. As a result ALND is no longer recommended in the majority of these patients and the use of IHC on frozen sections has become redundant. Still a group might exist, with a higher risk of axillary recurrence where ALND should still be recommended. The developed predictive model can be used to identify this group. If final histology identifies such a high risk patient where ALND is still recommended, a two-stage procedure can be offered without impairment of surgical results or increased risk of side effects compared to a one-stage procedure. This thesis adds to the increasing evidence of lack of prognostic advantage of ALND in sentinel node positive breast cancer patients. This evidence has resulted in dramatic changes in surgical treatment guidelines in Europe and Denmark, with a reduction in the proportion of patients offered ALND. It is expected that in the following years additional subgroups, where ALND can be omitted, will be identified. In addition, the thesis suggests an extension of the use of SLND to patients with locally recurrent breast cancer, and a change in treatment guidelines for this group of patients is proposed. Future reduction in axillary surgery Further extension of the SLND to other patient groups could spare an additional proportion of patients an ALND. Especially, a change in the use of SLND in patients treated by neoadjuvant treatment could reduce the use of ALND. The use of neoadjuvant treatment is increasing, to reduce tumor size and improve the possibilities of breast conserving surgery. In Denmark these patients are offered SLND before neoadjuvant treatment if they are clinically node negative. If metastases are found in the sentinel node, an ALND is offered at final surgery after neoadjuvant treatment. This does not take advantage of the neoadjuvant treatment that might reduce metastatic load in the axilla and convert patients from node positive to node negative and make further axillary surgery redundant. If the SLND is performed after neoadjuvant treatment, international studies have shown a detection rate between %, and a false negative rate (FNR) between 0 and 33% ( ). Two prospective studies have recently been published on the subject. The American ACOSOC Z1071 trial included 687 breast cancer patients with metastases in the axilla verified by biopsy before neoadjuvant treatment and a clinically node negative axilla after end of treatment. The detection rate at the subsequent SLND was 92.9% and the FNR was 12.6% at completion ALND (135). In the German SENTINA trial 592 patients with a positive sentinel node before neoadjuvant treatment was offered another SLND and ALND after end of treatment. Here the detection rate was 60% and the FNR was 52%. In the same

34 study 360 patients with lymph node metastases at biopsy before neoadjuvant treatment and a subsequent clinically negative axilla had a detection rate on 80% and a FNR on 14% at the following SLND and ALND (136). The FNR are in both studies higher than accepted by SLND at primary breast cancer. However a decreasing FNR was seen when using more than one tracer and by removing more sentinel nodes. In addition it is possible that marking of positive nodes before neoadjuvant treatment (Targeted axillary dissection) with a radioactive seed could further reduce the FNR (137), thereby making a SLND after neoadjuvant treatment acceptable. Despite the advantages of introducing the SLND, the procedure is not without side effects. It has been shown that 25-50% of patients experience pain after the SLND and 31 56% experience sensory disturbances (3;4). Thus, caution should be taken to avoid overused of the procedure. Today, all patients with primary invasive breast cancer are offered axillary surgery, either SLND or ALND. Sometimes axillary surgery is even offered to patients with only DCIS, because in 13 40% of these patients an unsuspected small area with invasive carcinoma is found at histopathological examination after final surgery (138;139). International studies have shown that metastases are found in the sentinel node in only 1-2% of patients diagnosed with only DCIS or microinvasive disease <1 mm. The SLND is therefore redundant in the vast majority of these patients (38;140;141). It is possible that some of these sentinel node metastases are caused by iatrogenic displacement of tumor cells from the breast. The risk of metastases to other lymph nodes and the need for a completion ALND in these cases is basically unknown. The number of patients diagnosed with DCIS or microinvasive disease <1 mm has increased after introduction of the National Mammographic screening program, thereby increasing the risk of overtreatment by unnecessary SLND (94). Identification of risk factors for sentinel and non-sentinel node metastases in these patients could enable a tailor-made treatment of the axilla, and avoiding overtreatment by redundant SLND. Traditionally, the advantage of axillary surgery has been based on the need for staging information for adjuvant treatment decisions and therapeutic removal of positive lymph nodes. Adjuvant treatment is now increasingly based on biological tumor characteristics and a survival advantage of removing axillary lymph nodes has never been shown. As early as in 1971 the NSABP B-04 trial was initiated where patients were randomized to radical mastectomy, total mastectomy with radiation, or total mastectomy with ALND only if nodes became positive (142). No difference in survival was shown between groups. With the increasingly precise axillary ultrasound (143) it is possible that axillary surgery with time will be reserved for patients with substantial tumor load in the axilla, whereas patients with fewer metastatic lymph nodes might be offered axillary radiotherapy which is connected to minor risk of morbidity than ALND, as shown in the AMAROS trial (118). The role for SLND will then be questioned

35 English summary This thesis is based on 10 original articles, of which 3 were previously included in the PhD thesis Staging of women with breast cancer after introduction of sentinel node guided axillary dissection (18). In the PhD thesis is was shown that the introduction of sentinel lymph node dissection (SLND) in the treatment of breast cancer in Denmark has resulted in an increased identification of patients with micrometastases or isolated tumor cells (ITC) in the lymph nodes (25). Not all these small metastases are likely to disseminate to non-sentinel nodes. This thesis provides evidence that a previous surgical excision of a breast tumor is likely to lead to iatrogenic displacement of tumor cells resulting in a nearly four-fold increased risk of ITC in the sentinel node (35). These tumor cells are not associated with non-sentinel node metastases. Especially ITC, but also micrometastases and some macrometastases, are not identified on perioperative frozen sections, but found postoperatively at the conventional histopathological examination. These patients are offered an axillary lymph node dissection (ALND) as a second procedure. It has been suggested that this two-stage procedure reduces the number of lymph nodes removed, because of fibroses from previous surgery. In this thesis it was shown that a two-stage procedure does not result in a clinically relevant impairment of the number of lymph nodes removed by ALND (47). Based on patient, tumor, and sentinel node characteristics from the Danish Breast Cancer Group database, two predictive models for non-sentinel node metastases, when only micrometastases or ITC are found in the sentinel node, were developed, as a part of the PhD thesis (15). These two models have now been internally validated (67), and a cross-validation in a Finnish patient material has been performed in cooperation with researchers from Helsinki (74). The model for patients with micrometastases proved to be robust under internal as well as external validation and could be used to identify patients with micrometastases that might still benefit from an ALND. Efforts should continue to improve the model. As a part of the PhD thesis, new molecular markers were tested for prediction of non-sentinel node metastases (85). In addition, method of detection of the breast cancer could be a possible predictor of non-sentinel node metastases. It has been hypothesized that breast cancers detected by screening represent a clinical indolent group of cancers with lower risk of non-sentinel node metastases compared to symptomatic cancers. This hypothesis was tested in this thesis in a large Danish dataset. No significant difference in the risk of non-sentinel node metastases was found between patients with screen-detected and symptomatic breast cancers (102), and a less aggressive treatment of the axilla in patients with screen-detected breast cancers cannot be supported. Likewise, the method of detection is not expected to be able to improve the predictive models. Until 2012, the standard treatment of Danish patients with micrometastases or ITC in the sentinel node was ALND. Still, in selected patients ALND was not performed. This thesis includes a comparison of the risk of axillary recurrence and survival between patients with and without ALND. The overall axillary recurrence rate was only 1.6% after 6 years of follow-up, despite between 9 18 % of these patients are expected to have non-sentinel node metastases. No significant difference was seen in axillary recurrence and overall survival between patients with and without ALND (116). These results support the safety of omitting ALND in patients with only micrometastases or ITC in the sentinel node, and since 2012 Danish breast cancer patients with ITC or up to two micrometastatic sentinel nodes are no longer offered an ALND. In Denmark the standard surgical treatment of the axilla in locally recurrent breast cancer is no further treatment of the axilla in case of previous ALND, and ALND in case of previous SLND. To investigate whether SLND can be extended to this patient group, a Danish multicenter study was performed. Despite a

36 reduced detection rate, especially after previous ALND, SLND seemed to be a feasible procedure in locally recurrent breast cancer. The procedure can spare a clinically significant number of patients an unnecessary ALND and improve staging and local control after previous ALND. The increased number of patients with aberrant drainage underlines the importance of preoperative lymphoscintigraphy at local recurrence (121)

37 Danish Summary/sammenfatning Denne afhandling er baseret på 10 original artikler, hvoraf de 3 indgik i Ph.d. afhandlingen Stadieinddeling af kvinder med mammacancer efter indførelsen af sentinel node vejledt aksildissektion (18). I Ph.d. afhandlingen blev det vist, at indførelsen af sentinel node proceduren I behandling af brystkræft I Danmark, har ført til identifikation af flere patienter, der kun har mikrometastaser eller isolerede tumorceller (ITC) i aksillens lymfeknuder(25). Langt fra alle disse patienter har spredning til andre lymfeknuder end sentinel node (non-sentinel nodes). I denne afhandling vises det, at tidligere kirurgisk eksisionsbiopsi kan føre til iatrogen spredning af tumorceller med en næsten fire gange forøget risiko for ITC i sentinel node (35). Disse tumorceller er ikke associeret til non-sentinel node metastaser. Især ITC, men også mikrometastaser og nogle makrometastaser, findes ikke ved den peroperative frysemikroskopi, men først ved den postoperative konventionelle histologiske undersøgelse. Disse patienter tilbydes en ny operation med aksilrømning. Der har tidligere været tvivl om, hvorvidt der ved denne to-stadie operation blev fjernet færre lymfeknuder på grund af fibrose efter den tidligere operation. I afhandlingen vises det at en to-stadie operation ikke forringer antallet af fjernede lymfeknuder væsentligt (47). På baggrund af patient, tumor og sentinel node karakteristika fra Dansk Brystkræft Gruppes (DBCG) database, blev der under Ph.d. studiet udviklet to modeller til at forudsige non-sentinel node metastaser hos patienter med mikrometastaser eller ITC i sentinel nodes (15). Disse to modeller er nu blevet valideret både internt i et dansk patientmateriale (67), og en krydsvalidering I et finsk patientmateriale er udført I samarbejde med forskere fra Helsinki (74). Specielt modellen for patienter med mikrometastaser viste sig at være meget robust under validering og vil kunne anvendes til at udpege patienter med mikrometastaser i sentinel node med en særlig høj risiko for non-sentinel node metastaser, som kunne have gavn af en aksilrømning. I et forsøg på at forbedre nøjagtigheden af disse modeller, blev nye molekylærbiologiske markørers evne til at forudsige non-sentinel node metastaser testet under Ph.d. studiet (85). Der blev ikke fundet nogen sammenhæng mellem hhv. TIMP-1, KI67 og HER2 og risikoen for non-sentinel node metastaser. Efterfølgende er nu også detektionsmetoden for brystkræft testet som prædiktor for non-sentinel node metastaser. Der er tidligere fremsat teorier om at brystkræft opdaget ved screening repræsenterer en gruppe tilfældigt opdagede kræftknuder med meget langsomt vækstmønster, som derfor muligvis også har mindre risiko for spredning til non-sentinel nodes. Denne teori er i denne afhandling blevet testet i et større dansk patientmateriale. Der blev ikke fundet nogen signifikant forskel i risikoen for non-sentinel node metastaser mellem patienter med screeningsdetekterede cancere og symptomatiske cancere (102). Der er derfor ikke holdepunkter for at give en mere aggressiv behandling af aksillen til patienter med symptomatiske cancere i forhold til screenings detekterede cancere, og detektionsmetoden ser ikke ud til at kunne forbedre modellerne til forudsigelse af non-sentinel node metastaser. Indtil 2012 var standardbehandlingen af danske patienter med mikrometastaser eller ITC I sentinel node en aksilrømning. Udvalgte patienter fik dog ikke foretaget aksilrømning. Denne afhandling inkluderer en sammenligning af risikoen for aksilrecidiv og overlevelse hos patienter med og uden aksilrømning. Den samlede aksilrecidiv rate for de to grupper var kun 1,6 % efter 6 års follow-up, til trods for at mellem 9 og 18 % af disse patienter forventes at have spredning til non-sentinel nodes. Der kunne ikke vises nogen forskel I aksilrecidiv og overlevelse mellem patienter med og uden aksilrømning (116). Disse resultater støtter sikkerheden I at undlade aksilrømning hos patienter med mikrometastaser eller ITC I sentinel node

38 Studierne har medvirket til at danske brystkræftpatienter siden 2012 ikke længere tilbydes en aksilrømning, hvis der kun finder ITC eller mikrometastaser i mindre end tre sentinel nodes. I Danmark er den kirurgiske standardbehandling af aksillen ved lokalrecidiv af brystkræft, at der ikke udføres yderligere aksilkirurgi, hvis der tidligere er foretaget aksilrømning, og aksilrømning, hvis patienten tidligere har fået foretaget sentinel node procedure. For at undersøge om sentinel node proceduren kan udvides til brystkræftpatienter med lokalrecidiv, blev der udført et dansk multicenterstudie. Dette studie fandt sentinel node proceduren anvendelig ved lokalrecidiv af brystkræft, omend med en reduceret detektionsrate, især efter tidligere aksilrømning. Proceduren kan spare et klinisk signifikant antal patienter en unødvendig aksilrømning og forbedre statuering og lokal kontrol efter tidligere aksilrømning. Det øgede antal patienter med aberrant drænage understreger behovet for præoperativ lymfescintigrafi hos disse patienter (121)

39 Reference List (1) Halsted WS. I. The Results of Operations for the Cure of Cancer of the Breast Performed at the Johns Hopkins Hospital from June, 1889, to January, Ann Surg 1894 November;20(5): (2) Klit A, Henriksen TF, Siersen HE, Elberg JJ, Christiansen P, Kroman N. [Oncoplastic breast surgery in Denmark.]. Ugeskr Laeger 2014 October 27;176(44). (3) Gartner R, Jensen MB, Kronborg L, Ewertz M, Kehlet H, Kroman N. Self-reported arm-lymphedema and functional impairment after breast cancer treatment--a nationwide study of prevalence and associated factors. Breast 2010 December;19(6): (4) Gartner R, Jensen MB, Nielsen J, Ewertz M, Kroman N, Kehlet H. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA 2009 November 11;302(18): (5) Giuliano AE, Kirgan DM, Guenther JM, Morton DL. Lymphatic mapping and sentinel lymphadenectomy for breast cancer. Ann Surg 1994 September;220(3): (6) Friis E, Galatius H, Garne JP. Organized nation-wide implementation of sentinel lymph node biopsy in Denmark. Acta Oncol 2008;47(4): (7) Lyman GH, Temin S, Edge SB, Newman LA, Turner RR, Weaver DL et al. Sentinel lymph node biopsy for patients with early-stage breast cancer: American Society of Clinical Oncology clinical practice guideline update. J Clin Oncol 2014 May 1;32(13): (8) Christiansen P, Friis E, Balslev E, Jensen D, Moller S. Sentinel node biopsy in breast cancer: five years experience from Denmark. Acta Oncol 2008;47(4): (9) Gondos A, Jansen L, Heil J, Schneeweiss A, Voogd AC, Frisell J et al. Time trends in axilla management among early breast cancer patients: Persisting major variation in clinical practice across European centers. Acta Oncol 2016 February 15;1-8. (10) Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47(4): (11) DBCG guidelines. [ (12) Cserni G, Amendoeira I, Apostolikas N, Bellocq JP, Bianchi S, Bussolati G et al. Pathological work-up of sentinel lymph nodes in breast cancer. Review of current data to be considered for the formulation of guidelines. Eur J Cancer 2003 August;39(12): (13) Chen SL, Hoehne FM, Giuliano AE. The prognostic significance of micrometastases in breast cancer: a SEER population-based analysis. Ann Surg Oncol 2007 December;14(12): (14) Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 2002 September 1;20(17): (15) Tvedskov TF, Jensen MB, Lisse IM, Ejlertsen B, Balslev E, Kroman N. High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the sentinel node. Int J Cancer 2012 November 15;131(10):

40 (16) Kumar S, Bramlage M, Jacks LM, Goldberg JI, Patil SM, Giri DD et al. Minimal disease in the sentinel lymph node: how to best measure sentinel node micrometastases to predict risk of additional nonsentinel lymph node disease. Ann Surg Oncol 2010 November;17(11): (17) Mittendorf EA, Ballman KV, McCall LM, Yi M, Sahin AA, Bedrosian I et al. Evaluation of the Stage IB Designation of the American Joint Committee on Cancer Staging System in Breast Cancer. J Clin Oncol 2015 April 1;33(10): (18) Tvedskov TF. Staging of women with breast cancer after introduction of sentinel node guided axillary dissection. Dan Med J 2012 July;59(7):B4475. (19) Cserni G. Intraoperative analysis of sentinel lymph nodes in breast cancer by one-step nucleic acid amplification. J Clin Pathol 2012 March;65(3): (20) Castellano I, Macri L, Deambrogio C, Balmativola D, Bussone R, Ala A et al. Reliability of whole sentinel lymph node analysis by one-step nucleic acid amplification for intraoperative diagnosis of breast cancer metastases. Ann Surg 2012 February;255(2): (21) Feinstein AR, Sosin DA, Wells CK. The Will Rogers phenomenon: improved technologic diagnosis and stage migration as a source of nontherapeutic improvement in cancer prognosis. Trans Assoc Am Physicians 1984;97: (22) Madsen AH, Jensen AR, Christiansen P, Garne JP, Cold S, Ewertz M et al. Does the introduction of sentinel node biopsy increase the number of node positive patients with early breast cancer? A population based study form the Danish Breast Cancer Cooperative Group. Acta Oncol 2008;47(2): (23) van der Heiden-van der Loo, Bezemer PD, Hennipman A, Siesling S, van Diest PJ, Bongers V et al. Introduction of sentinel node biopsy and stage migration of breast cancer. Eur J Surg Oncol 2006 September;32(7): (24) Maaskant AJ, van de Poll-Franse LV, Voogd AC, Coebergh JW, Tutein Nolthenius-Puylaert MC, Nieuwenhuijzen GA. Stage migration due to introduction of the sentinel node procedure: a population-based study. Breast Cancer Res Treat 2009 January;113(1): (25) Tvedskov TF, Jensen MB, Balslev E, Ejlertsen B, Kroman N. Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: A nationwide study. Eur J Cancer 2011 April;47(6): (26) van Deurzen CH, Bult P, de BM, Koelemij R, van HR, Tjan-Heijnen VC et al. Morphometry of isolated tumor cells in breast cancer sentinel lymph nodes: metastases or displacement? Am J Surg Pathol 2009 January;33(1): (27) Magee JA, Piskounova E, Morrison SJ. Cancer stem cells: impact, heterogeneity, and uncertainty. Cancer Cell 2012 March 20;21(3): (28) Klein CA, Blankenstein TJ, Schmidt-Kittler O, Petronio M, Polzer B, Stoecklein NH et al. Genetic heterogeneity of single disseminated tumour cells in minimal residual cancer. Lancet 2002 August 31;360(9334): (29) Bleiweiss IJ, Nagi CS, Jaffer S. Axillary sentinel lymph nodes can be falsely positive due to iatrogenic displacement and transport of benign epithelial cells in patients with breast carcinoma. J Clin Oncol 2006 May 1;24(13):

41 (30) Carter BA, Jensen RA, Simpson JF, Page DL. Benign transport of breast epithelium into axillary lymph nodes after biopsy. Am J Clin Pathol 2000 February;113(2): (31) Diaz LK, Wiley EL, Venta LA. Are malignant cells displaced by large-gauge needle core biopsy of the breast? AJR Am J Roentgenol 1999 November;173(5): (32) Diaz NM, Cox CE, Ebert M, Clark JD, Vrcel V, Stowell N et al. Benign mechanical transport of breast epithelial cells to sentinel lymph nodes. Am J Surg Pathol 2004 December;28(12): (33) Hansen NM, Ye X, Grube BJ, Giuliano AE. Manipulation of the primary breast tumor and the incidence of sentinel node metastases from invasive breast cancer. Arch Surg 2004 June;139(6): (34) Moore KH, Thaler HT, Tan LK, Borgen PI, Cody HS, III. Immunohistochemically detected tumor cells in the sentinel lymph nodes of patients with breast carcinoma: biologic metastasis or procedural artifact? Cancer 2004 March 1;100(5): (35) Tvedskov TF, Jensen MB, Kroman N, Balslev E. Iatrogenic displacement of tumor cells to the sentinel node after surgical excision in primary breast cancer. Breast Cancer Res Treat 2012 January;131(1): (36) Tille JC, Loubeyre P, Bodmer A, Jannot Berthier AS, Rozenholc A, Tabouret-Viaud C et al. Isolated tumor cells in sentinel lymph nodes of invasive breast cancer: cell displacement or metastasis? Breast J 2014 September;20(5): (37) Mittendorf EA, Sahin AA, Tucker SL, Meric-Bernstam F, Yi M, Nayeemuddin KM et al. Lymphovascular invasion and lobular histology are associated with increased incidence of isolated tumor cells in sentinel lymph nodes from early-stage breast cancer patients. Ann Surg Oncol 2008 December;15(12): (38) Veronesi P, Intra M, Vento AR, Naninato P, Caldarella P, Paganelli G et al. Sentinel lymph node biopsy for localised ductal carcinoma in situ? Breast 2005 December;14(6): (39) Francis AM, Haugen CE, Grimes LM, Crow JR, Yi M, Mittendorf EA et al. Is Sentinel Lymph Node Dissection Warranted for Patients with a Diagnosis of Ductal Carcinoma In Situ? Ann Surg Oncol 2015 April 24. (40) Stovgaard ES, Tvedskov TF, Laenkholm AV, Balslev E. Cytokeratin on frozen sections of sentinel node may spare breast cancer patients secondary axillary surgery. Patholog Res Int 2012;2012: (41) Holm M, Paaschburg B, Balslev E, Axelsson CK, Willemoe GL, Flyger HL. Intraoperative immunohistochemistry staining of sentinel nodes in breast cancer: clinical and economical implications. Breast 2008 August;17(4): (42) Husen M, Paaschburg B, Flyger HL. Two-step axillary operation increases risk of arm morbidity in breast cancer patients. Breast 2006 October;15(5): (43) Goyal A, Newcombe RG, Chhabra A, Mansel RE. Morbidity in breast cancer patients with sentinel node metastases undergoing delayed axillary lymph node dissection (ALND) compared with immediate ALND. Ann Surg Oncol 2008 January;15(1): (44) Husted MA, Haugaard K, Soerensen J, Bokmand S, Friis E, Holtveg H et al. Arm morbidity following sentinel lymph node biopsy or axillary lymph node dissection: a study from the Danish Breast Cancer Cooperative Group. Breast 2008 April;17(2):

42 (45) Olson JA, Jr., McCall LM, Beitsch P, Whitworth PW, Reintgen DS, Blumencranz PW et al. Impact of immediate versus delayed axillary node dissection on surgical outcomes in breast cancer patients with positive sentinel nodes: results from American College of Surgeons Oncology Group Trials Z0010 and Z0011. J Clin Oncol 2008 July 20;26(21): (46) Braems G, Denys H, Cocquyt V, Van den BR. Preceding sentinel node biopsy in early breast cancer: does it affect the number of axillary lymph nodes? Acta Chir Belg 2008 November;108(6): (47) Damgaard OE, Jensen MB, Kroman N, Tvedskov TF. Quantifying the number of lymph nodes identified in one-stage versus two-stage axillary dissection in breast cancer. Breast 2013 February;22(1):44-6. (48) Chakravorty A, Sanmugalingam N, Shrestha A, Thomee E, Rusby J, Roche N et al. Axillary nodal yields: a comparison between primary clearance and completion clearance after sentinel lymph node biopsy in the management of breast cancer. Eur J Surg Oncol 2011 February;37(2): (49) Andersen KG, Jensen MB, Tvedskov TF, Kehlet H, Gartner R, Kroman N. Persistent pain, sensory disturbances and functional impairment after immediate or delayed axillary lymph node dissection. Eur J Surg Oncol 2013 January;39(1):31-5. (50) Goldhirsch A, Wood WC, Coates AS, Gelber RD, Thurlimann B, Senn HJ. 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 August;22(8): (51) Coates AS, Winer EP, Goldhirsch A, Gelber RD, Gnant M, Piccart-Gebhart M et al. -Tailoring therapies-improving the management of early breast cancer: St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer Ann Oncol 2015 August;26(8): (52) Cserni G, Gregori D, Merletti F, Sapino A, Mano MP, Ponti A et al. Meta-analysis of non-sentinel node metastases associated with micrometastatic sentinel nodes in breast cancer. Br J Surg 2004 October;91(10): (53) van Deurzen CH, de BM, Monninkhof EM, Bult P, van der WE, Tjan-Heijnen VC et al. Non-sentinel lymph node metastases associated with isolated breast cancer cells in the sentinel node. J Natl Cancer Inst 2008 November 19;100(22): (54) Menes TS, Tartter PI, Mizrachi H, Constantino J, Estabrook A, Smith SR. Breast cancer patients with pn0(i+) and pn1(mi) sentinel nodes have high rate of nonsentinel node metastases. J Am Coll Surg 2005 March;200(3): (55) Bolster MJ, Peer PG, Bult P, Thunnissen FB, Schapers RF, Meijer JW et al. Risk factors for nonsentinel lymph node metastases in patients with breast cancer. The outcome of a multiinstitutional study. Ann Surg Oncol 2007 January;14(1): (56) Naik AM, Fey J, Gemignani M, Heerdt A, Montgomery L, Petrek J et al. The risk of axillary relapse after sentinel lymph node biopsy for breast cancer is comparable with that of axillary lymph node dissection: a follow-up study of 4008 procedures. Ann Surg 2004 September;240(3): (57) Krag DN, Anderson SJ, Julian TB, Brown AM, Harlow SP, Costantino JP et al. Sentinel-lymph-node resection compared with conventional axillary-lymph-node dissection in clinically node-negative patients with breast cancer: overall survival findings from the NSABP B-32 randomised phase 3 trial. Lancet Oncol 2010 October;11(10):

43 (58) Van Zee KJ, Manasseh DM, Bevilacqua JL, Boolbol SK, Fey JV, Tan LK et al. A nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy. Ann Surg Oncol 2003 December;10(10): (59) Barranger E, Coutant C, Flahault A, Delpech Y, Darai E, Uzan S. An axilla scoring system to predict non-sentinel lymph node status in breast cancer patients with sentinel lymph node involvement. Breast Cancer Res Treat 2005 May;91(2): (60) Kohrt HE, Olshen RA, Bermas HR, Goodson WH, Wood DJ, Henry S et al. New models and online calculator for predicting non-sentinel lymph node status in sentinel lymph node positive breast cancer patients. BMC Cancer 2008;8:66. (61) van dh, I, Kuijt G, Roumen R, Voogd A, Steyerberg EW, Vergouwe Y. A head to head comparison of nine tools predicting non-sentinel lymph node status in sentinel node positive breast cancer women. J Surg Oncol 2015 August;112(2): (62) Alran S, De RY, Fourchotte V, Charitansky H, Laki F, Falcou MC et al. Validation and limitations of use of a breast cancer nomogram predicting the likelihood of non-sentinel node involvement after positive sentinel node biopsy. Ann Surg Oncol 2007 August;14(8): (63) Cserni G, Bianchi S, Vezzosi V, Arisio R, Peterse JL, Sapino A et al. Validation of clinical prediction rules for a low probability of nonsentinel and extensive lymph node involvement in breast cancer patients. Am J Surg 2007 September;194(3): (64) Carcoforo P, Maestroni U, Querzoli P, Lanzara S, Maravegias K, Feggi L et al. Primary breast cancer features can predict additional lymph node involvement in patients with sentinel node micrometastases. World J Surg 2006 September;30(9): (65) Houvenaeghel G, Nos C, Mignotte H, Classe JM, Giard S, Rouanet P et al. Micrometastases in sentinel lymph node in a multicentric study: predictive factors of nonsentinel lymph node involvement--groupe des Chirurgiens de la Federation des Centres de Lutte Contre le Cancer. J Clin Oncol 2006 April 20;24(12): (66) Leidenius MH, Vironen JH, Riihela MS, Krogerus LA, Toivonen TS, von Smitten KA et al. The prevalence of non-sentinel node metastases in breast cancer patients with sentinel node micrometastases. Eur J Surg Oncol 2005 February;31(1):13-8. (67) Tvedskov TF, Jensen MB, Balslev E, Kroman N. Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Acta Oncol 2014 February;53(2): (68) Meretoja TJ, Strien L, Heikkila PS, Leidenius MH. A simple nomogram to evaluate the risk of nonsentinel node metastases in breast cancer patients with minimal sentinel node involvement. Ann Surg Oncol 2012 February;19(2): (69) Coutant C, Olivier C, Lambaudie E, Fondrinier E, Marchal F, Guillemin F et al. Comparison of models to predict nonsentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: a prospective multicenter study. J Clin Oncol 2009 June 10;27(17): (70) Andersson Y, Frisell J, de BJ, Bergkvist L. Prediction of non-sentinel lymph node status in breast cancer patients with sentinel lymph node metastases: evaluation of the tenon score. Breast Cancer (Auckl ) 2012;6:

44 (71) Hessman CJ, Naik AM, Kearney NM, Jensen AJ, Diggs BS, Troxell ML et al. Comparative validation of online nomograms for predicting nonsentinel lymph node status in sentinel lymph node-positive breast cancer. Arch Surg 2011 September;146(9): (72) Kocsis L, Svebis M, Boross G, Sinko M, Maraz R, Rajtar M et al. Use and limitations of a nomogram predicting the likelihood of non-sentinel node involvement after a positive sentinel node biopsy in breast cancer patients. Am Surg 2004 November;70(11): (73) Chen K, Zhu L, Jia W, Rao N, Fan M, Huang H et al. Validation and comparison of models to predict non-sentinel lymph node metastasis in breast cancer patients. Cancer Sci 2012 February;103(2): (74) Tvedskov TF, Meretoja TJ, Jensen MB, Leidenius M, Kroman N. Cross-validation of three predictive tools for non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Eur J Surg Oncol 2014 April;40(4): (75) Houvenaeghel G, Nos C, Giard S, Mignotte H, Esterni B, Jacquemier J et al. A nomogram predictive of non-sentinel lymph node involvement in breast cancer patients with a sentinel lymph node micrometastasis. Eur J Surg Oncol 2009 July;35(7): (76) Houvenaeghel G, Bannier M, Nos C, Giard S, Mignotte H, Jacquemier J et al. Non sentinel node involvement prediction for sentinel node micrometastases in breast cancer: nomogram validation and comparison with other models. Breast 2012 April;21(2): (77) Cserni G, Bori R, Maraz R, Leidenius MH, Meretoja TJ, Heikkila PS et al. Multi-institutional comparison of non-sentinel lymph node predictive tools in breast cancer patients with high predicted risk of further axillary metastasis. Pathol Oncol Res 2013 January;19(1): (78) Galimberti V, Cole BF, Zurrida S, Viale G, Luini A, Veronesi P et al. Axillary dissection versus no axillary dissection in patients with sentinel-node micrometastases (IBCSG 23-01): a phase 3 randomised controlled trial. Lancet Oncol 2013 April;14(4): (79) Singh M, Parnes MB, Spoelstra N, Bleile MJ, Robinson WA. p16 expression in sentinel nodes with metastatic breast carcinoma: evaluation of its role in developing triaging strategies for axillary node dissection and a marker of poor prognosis. Hum Pathol 2004 December;35(12): (80) Kwon Y, Ro J, Kang HS, Kim SK, Hong EK, Khang SK et al. Clinicopathological parameters and biological markers predicting non-sentinel node metastasis in sentinel node-positive breast cancer patients. Oncol Rep 2011 April;25(4): (81) Schrohl AS, Holten-Andersen MN, Peters HA, Look MP, Meijer-van Gelder ME, Klijn JG et al. Tumor tissue levels of tissue inhibitor of metalloproteinase-1 as a prognostic marker in primary breast cancer. Clin Cancer Res 2004 April 1;10(7): (82) Talvensaari-Mattila A, Turpeenniemi-Hujanen T. High preoperative serum TIMP-1 is a prognostic indicator for survival in breast carcinoma. Breast Cancer Res Treat 2005 January;89(1): (83) Yerushalmi R, Woods R, Ravdin PM, Hayes MM, Gelmon KA. Ki67 in breast cancer: prognostic and predictive potential. Lancet Oncol 2010 February;11(2): (84) Menard S, Fortis S, Castiglioni F, Agresti R, Balsari A. HER2 as a prognostic factor in breast cancer. Oncology 2001;61 Suppl 2:

45 (85) Tvedskov TF, Bartels A, Jensen MB, Paaschburg B, Kroman N, Balslev E et al. Evaluating TIMP-1, Ki67, and HER2 as markers for non-sentinel node metastases in breast cancer patients with micrometastases to the sentinel node. APMIS 2011 December;119(12): (86) Wurtz SO, Schrohl AS, Mouridsen H, Brunner N. TIMP-1 as a tumor marker in breast cancer--an update. Acta Oncol 2008;47(4): (87) Duffy MJ. Urokinase plasminogen activator and its inhibitor, PAI-1, as prognostic markers in breast cancer: from pilot to level 1 evidence studies. Clin Chem 2002 August;48(8): (88) Sparmann A, van LM. Polycomb silencers control cell fate, development and cancer. Nat Rev Cancer 2006 November;6(11): (89) Perou CM, Sorlie T, Eisen MB, van de RM, Jeffrey SS, Rees CA et al. Molecular portraits of human breast tumours. Nature 2000 August 17;406(6797): (90) Holm-Rasmussen EV, Jensen MB, Balslev E, Kroman N, Tvedskov TF. Reduced risk of axillary lymphatic spread in triple-negative breast cancer. Breast Cancer Res Treat 2015 January;149(1): (91) Boler DE, Uras C, Ince U, Cabioglu N. Factors predicting the non-sentinel lymph node involvement in breast cancer patients with sentinel lymph node metastases. Breast 2012 August;21(4): (92) Zhou W, He Z, Xue J, Wang M, Zha X, Ling L et al. Molecular subtype classification is a determinant of non-sentinel lymph node metastasis in breast cancer patients with positive sentinel lymph nodes. PLoS One 2012;7(4):e (93) Reyal F, Belichard C, Rouzier R, de GE, Senechal C, Bidard FC et al. Non-sentinel lymph node metastasis prediction in breast cancer with metastatic sentinel lymph node: impact of molecular subtypes classification. PLoS One 2012;7(10):e (94) Christiansen P, Vejborg I, Kroman N, Holten I, Garne JP, Vedsted P et al. Position paper: breast cancer screening, diagnosis, and treatment in Denmark. Acta Oncol 2014 April;53(4): (95) Nagtegaal ID, Duffy SW. Reduction in rate of node metastases with breast screening: consistency of association with tumor size. Breast Cancer Res Treat 2013 February;137(3): (96) Chiarelli AM, Edwards SA, Sheppard AJ, Mirea L, Chong N, Paszat L et al. Favourable prognostic factors of subsequent screen-detected breast cancers among women aged Eur J Cancer Prev 2012 November;21(6): (97) Meshkat B, Prichard RS, Al-Hilli Z, Bass GA, Quinn C, O'Doherty A et al. A comparison of clinicalpathological characteristics between symptomatic and interval breast cancer. Breast 2015 June;24(3): (98) Farshid G, Pradhan M, Kollias J, Gill PG. A decision aid for predicting non-sentinel node involvement in women with breast cancer and at least one positive sentinel node. Breast 2004 December;13(6): (99) Grabau D, Dihge L, Ferno M, Ingvar C, Ryden L. Completion axillary dissection can safely be omitted in screen detected breast cancer patients with micrometastases. A decade's experience from a single institution. Eur J Surg Oncol 2013 June;39(6): (100) Barry M, Kell MR. Re-evaluating the role of axillary lymph node dissection in screen-detected breast cancer patients. Breast 2012 February;21(1):

46 (101) Kroman N, Jensen MB, Wohlfahrt J, Mouridsen HT, Andersen PK, Melbye M. Factors influencing the effect of age on prognosis in breast cancer: population based study. BMJ 2000 February 19;320(7233): (102) Tvedskov TF, Jensen MB, Balslev E, Garne JP, Vejborg I, Christiansen P et al. Risk of non-sentinel node metastases in patients with symptomatic cancers compared to screen-detected breast cancers. Acta Oncol 2015 October 9;1-5. (103) Pepels MJ, Vestjens JH, de BM, Bult P, van Dijck JA, Menke-Pluijmers M et al. Models predicting non-sentinel node involvement also predict for regional recurrence in breast cancer patients without axillary treatment. Eur J Surg Oncol 2013 December;39(12): (104) Degnim AC, Zakaria S, Boughey JC, Sookhan N, Reynolds C, Donohue JH et al. Axillary recurrence in breast cancer patients with isolated tumor cells in the sentinel lymph node [AJCC N0(i+)]. Ann Surg Oncol 2010 October;17(10): (105) Onishi T, Jinno H, Takahashi M, Hayashida T, Sakata M, Nakahara T et al. Non-Sentinel Lymph Node Status and Prognosis of Breast Cancer Patients with Micrometastatic Sentinel Lymph Nodes. Eur Surg Res 2010 November 22;45(3-4): (106) Wasif N, Maggard MA, Ko CY, Giuliano AE. Underuse of axillary dissection for the management of sentinel node micrometastases in breast cancer. Arch Surg 2010 February;145(2): (107) Cox CE, Kiluk JV, Riker AI, Cox JM, Allred N, Ramos DC et al. Significance of sentinel lymph node micrometastases in human breast cancer. J Am Coll Surg 2008 February;206(2): (108) Pernas S, Gil M, Benitez A, Bajen MT, Climent F, Pla MJ et al. Avoiding axillary treatment in sentinel lymph node micrometastases of breast cancer: a prospective analysis of axillary or distant recurrence. Ann Surg Oncol 2010 March;17(3): (109) Weaver DL, Ashikaga T, Krag DN, Skelly JM, Anderson SJ, Harlow SP et al. Effect of occult metastases on survival in node-negative breast cancer. N Engl J Med 2011 February 3;364(5): (110) Giuliano AE, Hunt KK, Ballman KV, Beitsch PD, Whitworth PW, Blumencranz PW et al. Axillary dissection vs no axillary dissection in women with invasive breast cancer and sentinel node metastasis: a randomized clinical trial. JAMA 2011 February 9;305(6): (111) Giuliano AE, McCall L, Beitsch P, Whitworth PW, Blumencranz P, Leitch AM et al. Locoregional recurrence after sentinel lymph node dissection with or without axillary dissection in patients with sentinel lymph node metastases: the American College of Surgeons Oncology Group Z0011 randomized trial. Ann Surg 2010 September;252(3): (112) Bilimoria KY, Bentrem DJ, Hansen NM, Bethke KP, Rademaker AW, Ko CY et al. Comparison of sentinel lymph node biopsy alone and completion axillary lymph node dissection for node-positive breast cancer. J Clin Oncol 2009 June 20;27(18): (113) Yi M, Giordano SH, Meric-Bernstam F, Mittendorf EA, Kuerer HM, Hwang RF et al. Trends in and outcomes from sentinel lymph node biopsy (SLNB) alone vs. SLNB with axillary lymph node dissection for node-positive breast cancer patients: experience from the SEER database. Ann Surg Oncol 2010 October;17(Suppl 3): (114) Sola M, Alberro JA, Fraile M, Santesteban P, Ramos M, Fabregas R et al. Complete axillary lymph node dissection versus clinical follow-up in breast cancer patients with sentinel node

47 micrometastasis: final results from the multicenter clinical trial AATRM 048/13/2000. Ann Surg Oncol 2013 January;20(1): (115) Pepels MJ, de BM, Bult P, van Dijck JA, van Deurzen CH, Menke-Pluymers MB et al. Regional recurrence in breast cancer patients with sentinel node micrometastases and isolated tumor cells. Ann Surg 2012 January;255(1): (116) Tvedskov TF, Jensen MB, Ejlertsen B, Christiansen P, Balslev E, Kroman N. Prognostic significance of axillary dissection in breast cancer patients with micrometastases or isolated tumor cells in sentinel nodes: a nationwide study. Breast Cancer Res Treat 2015 October;153(3): (117) Kapoor NS, Sim MS, Lin J, Giuliano AE. Long-term Outcome of Patients Managed With Sentinel Lymph Node Biopsy Alone for Node-Negative Invasive Breast Cancer. Arch Surg 2012 July 16;1-7. (118) Donker M, van TG, Straver ME, Meijnen P, van de Velde CJ, Mansel RE et al. Radiotherapy or surgery of the axilla after a positive sentinel node in breast cancer (EORTC AMAROS): a randomised, multicentre, open-label, phase 3 non-inferiority trial. Lancet Oncol 2014 November;15(12): (119) Goyal A, Dodwell D, Reed MW, Coleman RE. Axillary treatment in women with one or two sentinel nodes with macrometastases: more evidence is needed to inform practice. J Clin Oncol 2014 December 1;32(34):3902. (120) Goldhirsch A, Ingle JN, Gelber RD, Coates AS, Thurlimann 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 August;20(8): (121) Uth CC, Christensen MH, Oldenbourg MH, Kjaer C, Garne JP, Teilum D et al. Sentinel Lymph Node Dissection in Locally Recurrent Breast Cancer. Ann Surg Oncol 2015 August;22(8): (122) Intra M, Trifiro G, Galimberti V, Gentilini O, Rotmensz N, Veronesi P. Second axillary sentinel node biopsy for ipsilateral breast tumour recurrence. Br J Surg 2007 October;94(10): (123) Cox CE, Furman BT, Kiluk JV, Jara J, Koeppel W, Meade T et al. Use of reoperative sentinel lymph node biopsy in breast cancer patients. J Am Coll Surg 2008 July;207(1): (124) Derkx F, Maaskant-Braat AJ, van der Sangen MJ, Nieuwenhuijzen GA, van de Poll-Franse LV, Roumen RM et al. Staging and management of axillary lymph nodes in patients with local recurrence in the breast or chest wall after a previous negative sentinel node procedure. Eur J Surg Oncol 2010 July;36(7): (125) Teilum D, Kroman N, Friis E. [Sentinel node technique in local relapse after breast-conserving surgery]. Ugeskr Laeger 2008 September 29;170(40): (126) Maaskant-Braat AJ, Roumen RM, Voogd AC, Pijpers R, Luiten EJ, Rutgers EJ et al. Sentinel Node and Recurrent Breast Cancer (SNARB): results of a nationwide registration study. Ann Surg Oncol 2013 February;20(2): (127) Intra M, Viale G, Vila J, Grana CM, Toesca A, Gentilini O et al. Second Axillary Sentinel Lymph Node Biopsy for Breast Tumor Recurrence: Experience of the European Institute of Oncology. Ann Surg Oncol 2015 July;22(7): (128) Lizarraga IM, Scott-Conner CE, Muzahir S, Weigel RJ, Graham MM, Sugg SL. Management of contralateral axillary sentinel lymph nodes detected on lymphoscintigraphy for breast cancer. Ann Surg Oncol 2013 October;20(10):

48 (129) Moossdorff M, Vugts G, Maaskant-Braat AJ, Strobbe LJ, Voogd AC, Smidt ML et al. Contralateral lymph node recurrence in breast cancer: Regional event rather than distant metastatic disease. A systematic review of the literature. Eur J Surg Oncol 2015 September;41(9): (130) Chen RC, Lin NU, Golshan M, Harris JR, Bellon JR. Internal mammary nodes in breast cancer: diagnosis and implications for patient management -- a systematic review. J Clin Oncol 2008 October 20;26(30): (131) Ozmen V, Ozcinar B, Bozdogan A, Eralp Y, Yavuz E, Dincer M. The effect of internal mammary lymph node biopsy on the therapeutic decision and survival of patients with breast cancer. Eur J Surg Oncol 2015 October;41(10): (132) van Deurzen CH, Vriens BE, Tjan-Heijnen VC, van der WE, Albregts M, van HR et al. Accuracy of sentinel node biopsy after neoadjuvant chemotherapy in breast cancer patients: a systematic review. Eur J Cancer 2009 December;45(18): (133) Tan VK, Goh BK, Fook-Chong S, Khin LW, Wong WK, Yong WS. The feasibility and accuracy of sentinel lymph node biopsy in clinically node-negative patients after neoadjuvant chemotherapy for breast cancer--a systematic review and meta-analysis. J Surg Oncol 2011 July 1;104(1): (134) Xing Y, Foy M, Cox DD, Kuerer HM, Hunt KK, Cormier JN. Meta-analysis of sentinel lymph node biopsy after preoperative chemotherapy in patients with breast cancer. Br J Surg 2006 May;93(5): (135) Boughey JC, Suman VJ, Mittendorf EA, Ahrendt GM, Wilke LG, Taback B et al. Sentinel lymph node surgery after neoadjuvant chemotherapy in patients with node-positive breast cancer: the ACOSOG Z1071 (Alliance) clinical trial. JAMA 2013 October 9;310(14): (136) Kuehn T, Bauerfeind I, Fehm T, Fleige B, Hausschild M, Helms G et al. Sentinel-lymph-node biopsy in patients with breast cancer before and after neoadjuvant chemotherapy (SENTINA): a prospective, multicentre cohort study. Lancet Oncol 2013 June;14(7): (137) Donker M, Straver ME, Wesseling J, Loo CE, Schot M, Drukker CA et al. Marking axillary lymph nodes with radioactive iodine seeds for axillary staging after neoadjuvant systemic treatment in breast cancer patients: the MARI procedure. Ann Surg 2015 February;261(2): (138) Doyle B, Al-Mudhaffer M, Kennedy MM, O'Doherty A, Flanagan F, McDermott EW et al. Sentinel lymph node biopsy in patients with a needle core biopsy diagnosis of ductal carcinoma in situ: is it justified? J Clin Pathol 2009 June;62(6): (139) Polom K, Murawa D, Wasiewicz J, Nowakowski W, Murawa P. The role of sentinel node biopsy in ductal carcinoma in situ of the breast. Eur J Surg Oncol 2009 January;35(1):43-7. (140) Gojon H, Fawunmi D, Valachis A. Sentinel lymph node biopsy in patients with microinvasive breast cancer: A systematic review and meta-analysis. Eur J Surg Oncol 2014 January;40(1):5-11. (141) Intra M, Rotmensz N, Veronesi P, Colleoni M, Iodice S, Paganelli G et al. Sentinel node biopsy is not a standard procedure in ductal carcinoma in situ of the breast: the experience of the European institute of oncology on 854 patients in 10 years. Ann Surg 2008 February;247(2): (142) Fisher B, Anderson S, Bryant J, Margolese RG, Deutsch M, Fisher ER et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2002 October 17;347(16):

49 (143) Houssami N, Turner RM. Staging the axilla in women with breast cancer: the utility of preoperative ultrasound-guided needle biopsy. Cancer Biol Med 2014 June;11(2):

50 Papers included

51 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) available at journal homepage: Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: A nationwide study Tove Filtenborg Tvedskov a, *, Maj-Britt Jensen b, Eva Balslev c, Bent Ejlertsen d, Niels Kroman a a Department of Breast Surgery, Copenhagen University Hospital, Blegdamsvej 9, 2100 København Ø, Denmark b Danish Breast Cancer Cooperative Group, Blegdamsvej 9, 2100 København Ø, Denmark c Department of Pathology, Herlev Hospital, Herlev Ringgade 75, 2730 Herlev, Denmark d Department of Oncology, Copenhagen University Hospital, Blegdamsvej 9, 2100 København Ø, Denmark ARTICLE INFO ABSTRACT Article history: Received 26 August 2010 Received in revised form 19 November 2010 Accepted 24 November 2010 Available online 29 December 2010 Keywords: Breast neoplasms Sentinel lymph node biopsy Neoplasm metastasis Neoplasm staging Purpose: To estimate the size and therapeutic consequences of stage migration after introduction of sentinel lymph node dissection (SLND) in breast cancer treatment in Denmark. Patients and methods: We compared the distribution of lymph node metastases in breast cancer patients operated in and ; before and after introducing SLND. The study was based on data from the national Danish Breast Cancer Cooperative Group (DBCG) database. Results: We included 24,051 patients in the study; 10,231 patients from the first period and 13,820 from the second period. The proportion of patients having macrometastases was not significantly different in the two periods, whereas the proportion of patients with micrometastases increased from 5.1% to 9.0% (P < ). However, this only resulted in an estimated change, from 7.8% to 8.8%, in the proportion of patients offered adjuvant systemic treatment due to positive nodal status as the only high-risk criterion, when using today s criteria for risk-allocation. In addition, we found that negative hormone receptor status was associated to negative nodal status when adjusted for confounders (odds ratios (OR) 0.83, P < ). Conclusion: Introduction of SLND in breast cancer treatment in Denmark has resulted in a stage migration on 4% due to identification of more micrometastases. However, this stage migration has only minor impact on patients offered adjuvant systemic treatment because nodal status today is less important in risk-allocation. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction Axillary nodal status is the most important prognostic factor in breast cancer. Axillary lymph node dissection (ALND) has until recently been the standard procedure for staging of the axilla, but this procedure is associated with considerable morbidity 1 and is redundant for women without lymph node metastases. Sentinel lymph node dissection (SLND) has gradually replaced ALND as standard procedure, because it can accurately stage the axilla by removing in average only two lymph nodes. 2 Following SLND, patients are only recommended ALND in case of verified sentinel node metastases. * Corresponding author: Tel.: ; fax: address: tft@dadlnet.dk (T.F. Tvedskov) /$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi: /j.ejca

52 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) After ALND, the removed lymph nodes are traditionally examined by standard microscopy of a few haematoxylin eosin stained histological sections. More extensive histopathological examinations using immunohistochemical staining on multi-sections of all removed lymph nodes will identify metastasis in 9 31% of the cases considered negative by standard examination. 3 These time-consuming methods are not used routinely in ALND, but are feasible following SLND. As a result more metastases, including micrometastases, are found. 4 Hence, more patients are diagnosed as being node positive and classified as having a more advanced stage of the disease, resulting in stage migration or Will Rogers phenomenon as described by Feinstein. 5 The magnitude of this stage migration has been investigated previously, 6 14 but only few studies have been population-based 7 9 and none have been nationwide. The consequences of stage migration on patients offered adjuvant treatment have not been systematically investigated and are basically unknown. In Denmark, the Danish Breast Cancer Cooperative Group (DBCG) database gives us a unique opportunity for a nationwide study on stage migration. The purpose of the present study was to estimate the size and therapeutic consequences of stage migration after introduction of SLND in breast cancer treatment in Denmark. 2. Patients and methods In Denmark, clinical and histopathological data as well as information on treatment and follow-up status of women with breast cancer have been prospectively registered in a national database managed by DBCG since Furthermore, DBCG describes guidelines for breast cancer treatment in Denmark. Today the database contains information on more than 80,000 breast cancer patients. 15 SLND was introduced in Denmark in 1997 and completely implemented in all Danish surgical departments by the end of Today, SLND is a standard procedure in Denmark for axillary staging of women with unifocal breast cancers without verified lymph node metastases and without history of operation in the upper lateral quadrant of the breast. 2 All removed sentinel nodes are examined according to the national DBCG guidelines with at least two step sections 500 lm apart of the bivalved sentinel node. In case of no metastases found by haematoxylin eosin staining, immunohistochemical cytokeratin staining is performed. Lymph nodes removed by ALND are examined by bisectioning and haematoxylin eosin staining. Metastases are classified as described by the American Joint Committee on Cancer (AJCC) 17 in combination with cell counts defining metastases with between 10 and 100 tumour cells as micrometastases and with less than 10 cells as isolated tumour cells. 2 In the primary tumour, hormone receptor status is defined from percentage of stained tumour cells where patients with staining for either oestrogen or progesterone receptors in P10% of the cells are considered as being hormone receptor positive. We estimated the size of stage migration by comparing the distribution of lymph node metastases in breast cancer patients operated in two different periods of 4 years: from 1993 to 1996 and from 2005 to Data on lymph node metastases, age at diagnosis, hormone receptor status, tumour size, histological type and malignancy grade were retrieved from the DBCG database. All registered patients, regardless of inclusion in specific treatment protocols, were included in the study to avoid selection bias. We collected missing information on nodal status manually from the original pathology file when possible. Altogether, 1617 patients were excluded: 53 patients from the first period due to missing information on nodal status and 1038 patients from the first and 526 patients from the second period due to missing information of the number of lymph nodes removed or less than four lymph nodes removed by ALND. To investigate if the introduction of SLND had changed the proportion of patients offered adjuvant systemic treatment, we divided patients from the two periods into risk groups according to the risk criteria described at the 10th St. Gallen International Expert Consensus Meeting 2007 (Table 2). 18 Accordingly, negative nodal status, tumour size 6 2 cm, positive or unknown hormone receptor status, age P 35 years and ductal carcinoma malignancy grade I or unknown grade were considered as low risk criteria. Furthermore, non-ductal carcinoma was considered as a low risk criterion, because only ductal carcinomas were graded histologically in the first period. HER2 status and peritumoural vascular invasion were not included as risk criteria when comparing the two periods because these parameters were only measured in the last period. The study was approved by the Ethical Committees of the Capital Region, protocol no. H , and by the Danish Data Protection Agency Statistical analysis The DBCG Data Centre was responsible for data collection and data analysis. Associations between pairs of variables were analysed by the v 2 -test (excluding unknowns). Univariate and multivariate logistic regression models were applied to examine the effect of age at diagnosis, tumour size, histological type and grade, hormone receptor status and period on nodal status. Odds ratios (OR) and 95% confidence intervals (CI) were calculated, and the Wald test was used to test the overall significance of each parameter. For departments of pathology involved in both periods, a multivariate model including interaction terms of departments and period was set up to test heterogeneity using the Wald test. Two-tailed P-values were applied and level of significance was set to 5%. All statistical analyses were done using SAS 9.1 (SAS Institute, Cary, NC, USA). 3. Results We included a total of 24,051 patients in the study; 10,231 patients were operated between 1993 and 1996, and 13,820 patients were operated between 2005 and Patient and tumour characteristics are described in Table 1. In , we identified 307 patients having only isolated tumour cells in their lymph nodes. These patients are considered as node negative when staged according to the AJCC 17 and were

53 874 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) Table 1 Patient and tumour characteristics by period of diagnosis among 24,051 breast cancer patients in Denmark. Period of diagnosis No. % No. % Number of patients 10, , Removed LN(lymph nodes) by ALND (axillary lymph node dissection) 4 9 removed LN P10 removed LN Nodal status Node negative a 50.3 Node positive: macrometastases Node positive: micrometastases Age (years) P Tumour size (mm) P Unknown Histological type and grade Ductal grade I Ductal grade II Ductal grade III Ductal grade unknown Lobular grade I III Other Hormone receptor status Positive , Negative Unknown Risk allocation High-risk , Low risk Allocation not possible a The number includes 307 patients with only isolated tumour cells in the lymph nodes. Table 2 Criteria for risk allocation. Danish high-risk criteria 2007 St. Gallen intermediateor high-risk criteria 2007 High-risk criteria used in the present study Nodal status Positive Positive Positive Tumour size >2 cm >2 cm >2 cm Age at diagnosis <35 years <35 years <35 years Grade Ductal grade 2 3 Grade 2 3 Ductal grade 2 3 Lobular grade 3 Hormone receptor status Negative Negative Negative HER2 status Positive Positive Not used Peritumoural vascular invasion Not used Present Not used included in the group of node negative patients. No patients with only isolated tumour cells were registered in the first period. The distribution of nodal status, age at diagnosis, tumour size, hormone receptor status, histological type and grade changed significantly over time. From the first to the second period we found an increasing age at diagnosis (P < ), increasing grade (P < ), increasing proportion of patients having ductal carcinomas (P < ),

54 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) increasing proportion of patients having hormone receptor positive tumours (P < ) and a decreasing tumour size (P < ) (Table 1). The overall number of node positive patients increased significantly from 45.6% before to 49.7% after introduction of SLND; the proportion of patients with micrometastases increased from 5.1% to 9.0% (P < ), whereas the proportion of patients having macrometastases was unchanged. In a univariate analysis the risk of being node positive was significantly increased after introduction of SLND compared to before (OR 1.18; CI , P < ). Furthermore, the risk of being node positive was significantly related to histological type and grade, increasing tumour size and younger age. There was no significant difference in the risk of having lymph node metastases between patients with positive or negative hormone receptor status (Table 3). In a multivariate analysis, adjusting for changes in tumour size, age at diagnosis, hormone receptor status, histological type and grade, the risk of being node positive when operated in the last period compared to the first remained significantly increased (OR 1.20; CI , P < ) (Table 3). When specifying this analysis according to the risk of having either macrometastases or micrometastases, we found an even more increased risk for having micrometastases after introduction of SLND compared to before (OR 1.85; CI , P < ) while the risk of having macrometastases was unchanged (OR 1.01; CI , P = 0.77). In the multivariate analysis younger age, increasing tumour size and histological grade remained significantly related to node positive disease (P < ), but in contrast to the results of the univariate analysis, negative hormone receptor status turned out to be significantly related to negative nodal status (OR 0.83; CI , P < ). Patients with unknown hormone receptor status were found to be significantly related to negative nodal status as well (OR 0.81; CI ; P < ). However, this group represented only 5% of all patients and negative hormone receptor status remained significantly related to negative nodal status when compared to the common group of patients with either positive or unknown hormone receptor status (OR 0.85; CI ; P < ). To examine whether different departments of pathology in Denmark contributed equally to the increase in the amount of node positive patients after introduction of SLND, sub-analyses were made for single departments of pathology. Nine departments were no longer part of a breast unit in the last period because of centralisation of breast cancer treatment in Denmark. For the remaining 16 departments multivariate analyses adjusting for changes in tumour size, age at diagnosis, hormone receptor status, histological type and grade were made to investigate interactions between department and period. A total of 21,276 patients were included in these sub-analyses; 7478 operated in , and 13,798 in Table 3 Probability of positive axillary lymph nodes (macro- or micrometastases) among 24,051 breast cancer patients in Denmark treated in or Univariate analysis Multivariate analysis OR (odds ratio) 95% CI (confidence interval) P-value OR 95% CI P-value Period of diagnosis < < Age at diagnosis (years) < < P Tumour size (mm) < < P Unknown Type and grade < < Ductal grade I Ductal grade II 1 1 Ductal grade III Ductal grade unknown Lobular grade I III Other Hormone receptor status < < Positive 1 1 Negative Unknown

55 876 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) Odds ratios for being node positive in compared to did not vary significantly between the single departments of pathology (P = 0.11). Finally, we estimated the impact of the increased proportion of node positive patients on the proportion offered adjuvant systemic treatment. Patients from the two periods were divided into risk groups according to the modified St. Gallen risk criteria as described (Table 2). By doing this, we estimated that 71% of the patients in the first period and 73% of the patients in the second period would have been high-risk patients according to the risk-criteria of today (Table 1), and out of those only 788 patients (150 with micrometastases, 638 with macrometastases) in the first period, corresponding to 7.8% of the patients, and 1217 patients (361 with micrometastases, 856 with macrometastases) in the last period, corresponding to 8.8% of the patients, became high-risk patients because of positive nodal status as the only high-risk criterion. The majority became high-risk patients regardless of nodal status but due to the existence of other high-risk criteria. The minor increase in high-risk patients caused by nodal status, from 7.8% to 8.8%, was, however, significant (P = 0.006). In the last period, 75% of the included patients (10,433 patients) underwent SLND. If we used all available risk-criteria defined at the 10th St. Gallen International Expert Consensus Meeting 2007, 18 including HER2 status, peritumoural vascular invasion and histological grading of lobular carcinomas, the proportion of high-risk patients increased even further to 80% of the patients (8334 patients) and still only a minor proportion, 7.9% (820 patients), had nodal status as the only high-risk criterion. 4. Discussion In this nationwide study, we examined the influence of introducing SLND on the proportion of lymph node positive patients in 24,051 breast cancer patients. Our results provide support for an absolute increase in the proportion of node positive patients on nearly 4% points, exclusively due to more women diagnosed with micrometastases. However, this stage migration resulted in only 1% absolute increase in the proportion of patients who, according to current guidelines, would be offered adjuvant systemic treatment. This study has some potential limitations. First, misclassification could arise if the pathologists were not consistent in how they evaluated nodal metastases. 19 In this study, the Danish departments of pathology reported the results of nodal examination prospectively during the study period using a standardised form. To examine whether heterogeneity between departments had any impact on our results we calculated odds ratios for being node positive for every single department of pathology. This revealed no significant differences in odds ratios between the departments. Thus, the DBCG data on nodal status, used in this study, are not significantly affected by minor local differences in pathology procedures and can be considered as quite uniform. Second, we found a significant drift in risk factors for having lymph node metastases over time which made it difficult to compare different periods. A decrease in average tumour size over time as well as an increase in oestrogen receptor positive tumours and age at diagnosis have been shown earlier, and could be explained by the introduction of mammography screening in some Danish counties. 21,22 It may also represent a trend towards an overall biologically less aggressive disease, but at the same time, an increase in malignancy grade was seen, pointing in the opposite direction. These opposite trends resulted in a basically unchanged proportion of high-risk patients (Table 1). Malignancy grading is to some degree a subjective measurement and changes in registrational practice could theoretically explain the trend towards increasing grade. Still, there have been no change in Danish grading guidelines between the two periods 22 and furthermore, a large intraobserver agreements in malignancy grading have been shown. 23 Further studies are needed to evaluate the observed increase in grade. After adjustment for the described changes, when estimating the size of stage migration, the risk of being node positive after introduction of SLND remained significantly increased compared to the period before. The risk was still only increased in the group of patients with micrometastases. That implies that the increase in micrometastases after introduction of SLND is actually the result of a more extensive lymph node examination and cannot be explained by changes in age at diagnosis, tumour size, hormone receptor status and malignancy grade over time. The strengths of our study include its large sample size of more than 24,000 breast cancer patients and its populationbased approach including 94% of all breast cancer patients registered in the national DBCG database in the two periods. Data in the DBCG database have been prospectively collected from all Danish women with breast cancer and more than 95% concordance with the Danish Cancer Register and the National Pathology Register, which are considered close to complete, has been shown. 15 Accordingly, existing discrepancies in the proportion of node positive patients and in tumour size between the DBCG database and national databases from other contries 24 are not likely to be caused by incomplete registration but could rather be explained by differences in population and breast cancer screening policies. Among other studies investigating the magnitude of stage migration after introduction of SLND, 6 14 only three studies have been population-based. 7 9 In The Netherlands van der Heiden-van der Loo and colleagues have investigated the magnitude of stage migration in 3665 patients in the central part of the country during the introduction of SLND. 7 They found that the proportion of node positive patients increased significantly from 28% in 1997 to 38% in 2002 and this increase was mainly caused by micrometastases. In contrast, Maaskant and colleagues made a similar investigation on 17,100 patients in the south-eastern part of The Netherlands, but included the entire population in the investigated area and the complete period for the introduction of SLND. They found a much lower increase in percentage of patients having micrometastases from 1% in 1994 to 4.3% in In Denmark, stage migration after introduction of SLND has been investigated in a smaller study including 2116 patients from two different counties. An increase in the proportion of node positive patients of 7.3% and 13.3%, respectively, was found from to These variations in the size of stage migration between different studies may reflect local differences in lymph node examinations but may also be a

56 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) result of different study sizes and study periods. Our estimated size of stage migration on nearly 4% is similar to the findings of Maaskant and colleagues which is the only previous study excluding the complete period for implementing SLND in the entire population. 9 Our findings of an increase in node positive patients exclusively represented by micrometastases confirm the results of several previous studies. 7 9 In our study, the number of lymph nodes excised increased from the first to the second period (Table 1). It has previously been shown that the proportion of node positive patients will increase with increasing number of lymph nodes examined. 25 In Denmark, removal of at least 10 lymph nodes is today required for sufficient surgery when ALND is performed, 15 but this was not the case in the first period of our study where many patients had only between 4 and 9 lymph nodes removed. It should be noted that this will tend to underestimate the number of node positive patients in the first period and consequently the magnitude of stage migration caused by the introduction of SLND alone will be lower than the estimate found in this study. In addition to the main results, we showed that positive nodal status was related to younger age and increasing tumour size, which is in accordance to the literature. 26,27 Moreover, we showed that negative nodal status was significantly related to negative hormone receptor status, as an independent factor, despite the fact that hormone receptor negative tumours generally are considered being biologically more aggressive. 20 Reports on hormone receptor status in relation to lymph node metastases are controversial. 27,28 One of the largest studies on the subject included 18,025 patients and showed the same relation between node negative disease and negative hormone receptor status as we do, indicating that this is a true observation. 26 Still, our results are based on patients from two different time periods and a study including patients between the two periods should be performed to confirm the results. The prognostic significance of micrometastases in the sentinel node remains unclear and identification of more micrometastases after introduction of SLND, 29 leading to stage migration, might cause overtreatment of these patients. 30 We, therefore, investigated the therapeutic consequences of this stage migration. The criteria for offering adjuvant systemic treatment have changed over years. 15,18 By allocating patients from the two periods into risk groups according to the risk criteria of today (Table 2), we showed that introduction of SLND had only minor effect on the number of patients offered adjuvant systemic treatment, with only 1% absolute increase in the proportion of patients with positive nodal status as the only high-risk criterion. The national criteria for risk allocation used in Denmark are slightly more conservative than the modified St. Gallen criteria used in this study (Table 2). 15 Using more conservative high-risk criteria will tend to increase the effect of stage migration on patients offered adjuvant treatment. Conversely, the trend towards inclusion of several new high-risk criteria in the decision for adjuvant systemic treatment will diminish the therapeutic consequences of stage migration. In our study, several high-risk criteria were available in the last period. When using all available criteria we found that nodal status was important in risk-allocation of only 10% of high-risk patients offered SLND (820/8334 patients). This indicates that axillary nodal status is losing its significance in the decision for adjuvant systemic treatment. In conclusion, we found a 4% absolute increase in the proportion of node positive patients after introduction of SLND. The increase was exclusively caused by identification of more micrometastases. However, this stage migration had only minor impact on the proportion of patients offered adjuvant systemic treatment, because nodal status is gradually losing its significance in risk allocation, due to introduction of other risk factors. Conflict of interest statement None declared. Funding The study was supported by IMK common foundation, HOC research initiative foundation and the Danish cancer research foundation. Acknowledgements The study was supported financially by IMK common foundation, HOC research initiative foundation and the Danish cancer research foundation. REFERENCES 1. Gartner R, Jensen MB, Nielsen J, et al. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA 2009;302: Christiansen P, Friis E, Balslev E, Jensen D, Moller S. Sentinel node biopsy in breast cancer: five years experience from Denmark. Acta Oncol 2008;47: Tjan-Heijnen VC, Buit P, de Widt-Evert LM, Ruers TJ, Beex LV. Micro-metastases in axillary lymph nodes: an increasing classification and treatment dilemma in breast cancer due to the introduction of the sentinel lymph node procedure. Breast Cancer Res Treat 2001;70: Turner RR, Giuliano AE, Hoon DS, Glass EC, Krasne DL. Pathologic examination of sentinel lymph node for breast carcinoma. World J Surg 2001;25: Feinstein AR, Sosin DA, Wells CK. The Will Rogers phenomenon: improved technologic diagnosis and stage migration as a source of nontherapeutic improvement in cancer prognosis. Trans Assoc Am Physicians 1984;97: Vanderveen KA, Schneider PD, Khatri VP, Goodnight JE, Bold RJ. Upstaging and improved survival of early breast cancer patients after implementation of sentinel node biopsy for axillary staging. Ann Surg Oncol 2006;13: van der Heiden-van der Loo M, Bezemer PD, Hennipman A, et al. Introduction of sentinel node biopsy and stage migration of breast cancer. Eur J Surg Oncol 2006;32: Madsen AH, Jensen AR, Christiansen P, et al. Does the introduction of sentinel node biopsy increase the number of node positive patients with early breast cancer? A population based study form the Danish Breast Cancer Cooperative Group. Acta Oncol 2008;47:

57 878 E U R O P E A N J O U R NA L O F CA N C E R47 (2011) Madsen AH, Jensen AR, Christiansen P, et al. Stage migration due to introduction of the sentinel node procedure: a population-based study.. Breast Cancer Res Treat 2009;113: Giuliano AE, Dale PS, Turner RR, et al. Improved axillary staging of breast cancer with sentinel lymphadenectomy. Ann Surg 1995;222: Groen RS, Oosterhuis AW, Boers JE. Pathologic examination of sentinel lymph nodes in breast cancer by a single haematoxylin eosin slide versus serial sectioning and immunocytokeratin staining: clinical implications. Breast Cancer Res Treat 2007;105: van Rijk MC, Peterse JL, Nieweg OE, et al. Additional axillary metastases and stage migration in breast cancer patients with micrometastases or submicrometastases in sentinel lymph nodes. Cancer 2006;107: de Widt-Levert L, Tjan-Heijnen V, Bult P, Ruers T, Wobbes T. Stage migration in breast cancer: surgical decisions concerning isolated tumour cells and micro-metastases in the sentinel lymph node. Eur J Surg Oncol 2003;29: Bolster MJ, Bult P, Wauters CA, et al. More tumor-affected lymph nodes because of the sentinel lymph node procedure but no stage migration, because the 2002 TNM classifies small tumor deposits as pathologic N0 breast cancer. Cancer 2009;115: Moller S, Jensen MB, Ejlertsen B, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47: Friis E, Galatius H, Garne JP. Organized nation-wide implementation of sentinel lymph node biopsy in Denmark. Acta Oncol 2008;47: Singletary SE, Allred C, Ashley P, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 2002;20: Goldhirsch A, Wood WC, Gelber RD, et al. Progress and promise: highlights of the international expert consensus on the primary therapy of early breast cancer Ann Oncol 2007;18: Cserni G, Amendoeira I, Apostolikas N, et al. Discrepancies in current practice of pathological evaluation of sentinel lymph nodes in breast cancer. Results of a questionnaire based survey by the European Working Group for Breast Screening Pathology. J Clin Pathol 2004;57: Bentzon N, During M, Rasmussen BB, Mouridsen H, Kroman N. Prognostic effect of estrogen receptor status across age in primary breast cancer. Int J Cancer 2008;122: Mouridsen HT, Bjerre KD, Christiansen P, Jensen MB, Moller S. Improvement of prognosis in breast cancer in Denmark , based on the nationwide reporting to the DBCG Registry. Acta Oncol 2008;47: Kiaer HW, Laenkholm AV, Nielsen BB, Bjerre KD. Classical pathological variables recorded in the Danish Breast Cancer Cooperative Group s register Acta Oncol 2008;47: Robbins P, Pinder S, de Klark N, et al. Histological grading of breast carcinomas: a study of interobserver agreement. Hum Pathol 1995;26: Lundin J, Lehtimaki T, Lundin M, et al. Generalisability of survival estimates for patients with breast cancer a comparison across two population-based series. Eur J Cancer 2006;42: Axelsson CK, Rank F, Blichert-Toft M, Mouridsen HT, Jensen MB. Impact of axillary dissection on staging and regional control in breast tumors < or = 10 mm the DBCG experience. The Danish Breast Cancer Cooperative Group (DBCG), Rigshospitalet, Copenhagen, Denmark. Acta Oncol 2000;39: Gann PH, Colilla SA, Gapstur SM, Winchester DJ, Winchester DP. Factors associated with axillary lymph node metastasis from breast carcinoma: descriptive and predictive analyses. Cancer 1999;86: Patani NR, Dwek MV, Douek M. Predictors of axillary lymph node metastasis in breast cancer: a systematic review. Eur J Surg Oncol 2007;33: Colleoni M, Rotmensz N, Peruzzotti G, et al. Size of breast cancer metastases in axillary lymph nodes: clinical relevance of minimal lymph node involvement. J Clin Oncol 2005;23: Patani N, Mokbel K. The clinical significance of sentinel lymph node micrometastasis in breast cancer. Breast Cancer Res Treat 2009;114: Hansen NM, Grube B, Ye X, et al. Impact of micrometastases in the sentinel node of patients with invasive breast cancer. J Clin Oncol 2009;27:

58 Breast Cancer Res Treat (2012) 131: DOI /s y EPIDEMIOLOGY Iatrogenic displacement of tumor cells to the sentinel node after surgical excision in primary breast cancer Tove F. Tvedskov Maj-Britt Jensen Niels Kroman Eva Balslev Received: 26 May 2011 / Accepted: 3 August 2011 / Published online: 18 August 2011 Ó Springer Science+Business Media, LLC Abstract Isolated tumor cells (ITC) are more common in the sentinel node (SN) after needle biopsy of a breast cancer, indicating iatrogenic displacement of tumor cells. We here investigate whether similar iatrogenic displacement occurs after surgical excision of a breast tumor. We compared the incidence of ITC in the SN of 414 breast cancer patients with recent surgical excision to a group of 16,960 patients without recent surgical procedure in a multivariate analysis by linking data from the Danish Breast Cancer Cooperative Group database and the Danish National Health Register. Moreover, the incidence of spread to non-sns in patients with ITC in the SN after recent surgical excision was analyzed. We found an adjusted odds ratio on 3.73 (95% CI ; P \ ) for having ITC in the SN after surgical excision. The increase in ITC after surgical excision was especially seen in patients with ductal carcinomas (OR 4.66; 95% CI ). None of the patients with ITC in SN after surgical excision had further spread to non-sns compared to 12% in the group without recent surgical excision (P = 0.09). The nearly fourfold increase in ITC in the SN after surgical excision indicates that this procedure induces iatrogenic displacement of tumor cells. This T. F. Tvedskov (&) N. Kroman Department of Breast Surgery, 3104, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark tft@dadlnet.dk M.-B. Jensen Danish Breast Cancer Cooperative Group, Strandboulevarden 47, Bygn. C, St, 2100 Copenhagen, Denmark E. Balslev Department of Pathology, Herlev Hospital, Herlev Ringvej 75, 2730 Herlev, Denmark displacement was more common in ductal carcinomas. We found no further dissemination to non-sns in patients with ITC in the SN after recent surgical excision, and it is questioned whether these patients benefit from an axillary lymph node dissection. Keywords Breast cancer Sentinel lymph node Isolated tumor cells Iatrogenic displacement Surgical excision Introduction Sentinel lymph node dissection (SLND) has become the standard procedure for axillary staging of woman with early primary breast cancer. This has made more extensive examinations of the lymph nodes possible. As a result more micrometastases and isolated tumor cells (ITC) are found in the sentinel node (SN) [1]. It is shown that the presence of micrometastases is of prognostic importance [2 4] while the significance of ITC is uncertain. ITC represent a heterogeneous group of cells [5] and not all are likely to develop into metastases [6]. Studies have shown that only 10 15% of patients with ITC in the SN have further spread to non-sns [7]. Nevertheless, patients with ITC in the SN will often be recommended an axillary lymph node dissection (ALND) due to the risk of non-sn metastases [8], and as a consequence, run the risk of arm morbidity [9]. It is shown that ITC are more common in the SN after needle biopsy or manipulation of a breast tumor [10 13]. This is probably due to iatrogenic displacement of tumor cells. Sometimes it is necessary to perform a surgical excision biopsy of a tumor to confirm the cancer diagnosis before a breast cancer operation, or an unexpected breast cancer can be found in relation to a supposed benign breast operation. Such a surgical excision may lead to a similar or 123

59 224 Breast Cancer Res Treat (2012) 131: even higher rate of ITC in the SN and it is not evident whether further ALND is indicated in these patients. So far, only few studies have addressed this subject and the results are controversial [12, 13]. One study shows a lower incidence of SN metastases after surgical excision biopsy compared to needle biopsy [13], while another study shows that a surgical excision biopsy increases the incidence of SN metastases [12]. Furthermore, the incidence of non-sn metastases in case of iatrogenically displaced tumor cells in the SN is unknown. The aim of this population-based study was to investigate the incidence of ITC and micrometastases in the SN after recent surgical excision of a breast tumor, to clarify if this procedure can lead to iatrogenic displacement of tumor cells to the SN. Furthermore, we investigated the incidence of non-sn metastases among these patients, to clarify whether an ALND is indicated in these patients. Materials and methods In Denmark, clinical and histopathological data and information on treatment of women with breast cancer have been prospectively registered in a national database managed by the Danish Breast Cancer Cooperative Group (DBCG) since Today the database contains information on more than 95,000 breast cancer patients [14]. We retrieved information from this database on age at diagnosis, tumor size, histological type and grade, hormone receptor status, lymphovascular invasion and SN and non- SN status of breast cancer patients with SLND performed between January 2002 and the end of In the database, metastases are classified according to the American Joint Committee on Cancer staging manual [15]. Micrometastases are defined as tumor cell deposits between 0.2 and 2 mm or cell count less than 100 tumor cells. ITC are defined as tumor cell deposits less than 0.2 mm or cell count less than 10 [8]. In case of missing information on nodal status, data were collected manually from the original pathology file when possible. Information on recent surgical excision was retrieved from the Danish National Health Register (DNHR) which was founded in 1977 and contains information on hospital admission, diagnosis, and treatment of all hospitalized patients in Denmark. Recent surgical excision was defined as surgical excision biopsy or removal of a supposed benign tumor within 2 months before an ipsilateral SLND. We allowed up to 1 week discordance in dates of SLND between the DNHR and the DBCG databases, because this was considered as registration errors. A total of 29,174 Danish women with primary breast cancer with surgery from January 2002 to the end of 2009 were identified using the DBCG database. Out of those, 17,945 were registered with a SLND. 571 patients had incomplete information on SN status and were excluded. Another 65 patients were excluded because of major discordance in registered dates or laterality between databases (Fig. 1). Patient and tumor characteristics of the remaining 17,374 patients are shown in Table 1. When analyzing the risk of non-sn metastases we excluded patients without an adequate ALND (less than 10 lymph nodes removed). The study was approved by the Danish Data Protection Agency (J.nr ). Statistical analyses The DBCG Data Centre was responsible for data collection and data analysis. Associations between pairs of variables were analyzed by the v 2 -test or Fishers exact test. Univariate and multivariate logistic regression models were applied to examine the effect of recent surgical excision on nodal status, adjusted for age at diagnosis, tumor size, lymphovascular invasion, histological type and grade, and hormone receptor status. Also time interval from recent surgical excision was investigated. Odds ratios (OR) and 95% confidence intervals (CI) were calculated, and the Wald test was used to test the overall significance of each parameter. For histological type a multivariate model including interaction terms of type and recent surgical excision was set up to test heterogeneity using the Wald test. Two-tailed P values were applied and the level of significance was set to 5%. All statistical analyses were done using SAS 9.1 (SAS Institute, Cary, NC, USA). Results A total number of 17,374 patients were included in the study; 414 patients with and 16,960 patients without a recent surgical excision. We found a significant difference in age at diagnosis, tumor size, histological type (P \ ), and lymphovascular invasion (P = 0.001) between the two groups of patients (Table 1). Patients with recent surgical excision were younger, had smaller tumors, often without lymphovascular invasion, and were more likely to have non-ductal, non-lobular carcinomas. Furthermore, a significantly different distribution of SN metastases was seen between the two groups (P \ ) (Table 2). About 9% of patients with recent surgical excision had ITC in the SN compared to only 3% of patients without recent surgical excision. In a univariate analysis the crude odds ratio was 3.17 (95% CI ; P \ ) for having ITC in the SN after surgical excision compared to patients without recent surgical procedure. In a multivariate analysis including tumor size, 123

60 Breast Cancer Res Treat (2012) 131: Fig. 1 Flowchart for inclusion of patients. DBCG indicated Danish Breast cancer Cooperative Group, SN sentinel node, SLND sentinel lymph node dissection, ALND axillary lymph node dissection, ITC isolated tumor cells, NSN non-sentinel node lymphovascular invasion, histological type, and malignancy grade the corresponding adjusted odds ratio was 3.73 (95% CI ; P \ ). Age at diagnosis and hormone receptor status was not significantly related to the risk of having ITC in the SN and was not included in the multivariate analysis. The crude odds ratio for having micrometastases in the SN after recent surgical excision compared to patients without recent surgical procedure was not significant (OR 1.27; 95% CI ; P = 0.09). In a multivariate analysis including age at diagnosis, lymphovascular invasion, tumor size, histological type, malignancy grade, and hormone receptor status we found an adjusted odds ratio on 1.49 (95% CI ; P = 0.006) for having micrometastases in the SN after recent surgical excision compared to patients without a recent surgical procedure. Finally, a difference was seen in the proportion of patients having macrometastases between the two groups; only 12% of patients in the group with recent surgical excision had macrometastases compared to 26% of patients in the control group. When adjusting for differences in age at diagnosis, lymphovascular invasion, tumor size, histological type, malignancy grade, and hormone receptor status between the two groups we found an odds ratio on 0.67 (95% CI ; P = 0.01) for having macrometastases in the group with recent surgical excision compared to the control group. To investigate whether the risk of having ITC in the SN after recent surgical excision depended on the time interval between the surgical procedure and the final SLND, we calculated the crude OR for having ITC in the SN after recent surgical excision for different time intervals. We found no relation between time interval and the risk of having ITC in the SN, with crude OR on 3.78 (95% CI ), 1.64 (95% CI ), 3.33 (95% CI ), and 5.27 (95% CI ) for patients with less than 2 weeks, 2 3 weeks, 3 4 weeks, and more than 123

61 226 Breast Cancer Res Treat (2012) 131: Table 1 Patient and tumor characteristics of 17,374 Danish breast cancer patients operated with SLND in according to recent surgical tumor excision and sentinel node status Recent surgical tumor excision Sentinel node status Total Yes No Negative ITC Micrometastases Macrometastases No. % No. % No. % No. % No. % No. % No. % Patients , , , , Age (years) B , , , , , , , , , , , C , , , Tumor size (mm) , , , , , , , , , , , , , , C Unknown WHO type Ductal , , , , , Lobular , , Other , , , Unknown Grade Grade I , , , , Grade II , , , , Grade III , , , Ungradable , , , Hormone receptor status Positive , , , , , Negative , , , Unknown Lymphovascular invasion Present , , Absent , , Unknown SLND sentinel lymph node dissection, ITC isolated tumor cells 4 weeks, respectively, between the two procedures, compared to patients with no recent surgical procedure. In general, ITC were significantly more common in the SN in patients with lobular carcinomas compared to ductal carcinomas; 7.4% of patients with lobular carcinomas had ITC in the SN compared to 2.5% of patients with ductal carcinomas and other cancer types. The adjusted OR for having ITC in the SN in patients with lobular carcinomas was 2.94 (95% CI ; P \ ) compared to ductal carcinomas, in a multivariate analysis including recent surgical excision, tumor size, lymphovascular invasion, and malignancy grade. Conversely, the increase in ITC in the SN after surgical excision was especially seen in patients with ductal carcinomas. In an analysis of interaction according to histological type the adjusted OR for having ITC in the SN after recent surgical excision in patients with ductal carcinomas was 4.66 (95% CI ) compared to patients without recent surgical excision whereas the OR for lobular carcinomas was only 1.18 (95% CI ). This difference in OR was significant (P = 0.03 for heterogeneity). 123

62 Breast Cancer Res Treat (2012) 131: Table 2 Sentinel node status of 17,374 Danish breast cancer patients operated with SLND according to recent surgical tumor excision Sentinel node status Recent surgical tumor excision Crude OR 95% CI P value Adjusted OR 95% CI P value Yes No No. % No. % Negative , ITC \ \ Micrometastases , Macrometastases , \ Total , SLND sentinel lymph node dissection, ITC isolated tumor cells, OR odds ratio, CI confidence interval Table 3 Number of patients with NSN metastases out of 369 Danish breast cancer patients with ITC and 1,793 with micrometastases in the SN, according to recent surgical tumor excision Recent surgical tumor excision P value Yes No NSN metastases Total NSN metastases Total Yes No Yes No ITC 0 0% % Micrometastases % % 1,411 1, SN sentinel node, NSN non-sentinel node, ITC isolated tumor cells In contrast to patients with ITC, micrometastases were most often seen in the SN in patients with ductal carcinomas with a crude odds ratio on 0.81 (95% CI ; P = 0.009) for having micrometastases in the SN for patients with lobular carcinomas compared to ductal carcinomas. In a multivariate analysis including recent surgical excision, age at diagnosis, tumor size, lymphovascular invasion, malignancy grade, and hormone receptor status the corresponding adjusted OR was 0.76 (95% CI ; P = ). There was no significant difference in the risk of having micrometastases in the SN after recent surgical excision comparing patients with lobular and ductal carcinomas (P = 0.49 for heterogeneity). An adequate ALND was performed in 369 out of 531 patients with ITC in the SN; 23 patients with and 346 patients without recent surgical excision (Fig. 1). None of the 23 patients with ITC in the SN after recent surgical excision had dissemination to non-sns, neither as ITC, micrometastases nor as macrometastases. In contrast, 12% of patients with ITC in the SN but without recent surgical procedure had dissemination to non-sns. However, these differences in the risk of having non-sn metastases did not reach statistical significance (P = 0.09) (Table 3). Out of 2,130 patients with micrometastases in the SN 1,793 patients had an adequate ALND; 52 with and 1,741 without recent surgical excision (Fig. 1). 12% of patients with recent surgical excision had dissemination to non-sns compared to 19% of patients without recent surgical excision (Table 3). This difference was not significant (P = 0.18). Discussion We found that surgical excision of a breast tumor in the weeks before SLND conferred a nearly fourfold increased risk of having ITC in the SN. This increase was mainly seen in ductal carcinomas, despite the fact that lobular carcinomas, in general, were more likely to present with ITC in the SN. None of the patients with ITC in the SN following recent surgical excision had non-sn metastases. This study is a register based study using data from two different databases with the risk of registration bias. Nevertheless, both databases have been extensively validated. Data in the DBCG database have been prospectively collected and registered using standardized forms. Linking the DBCG database and the Danish Cancer Register, which is considered close to complete, has revealed over 95% concordance [16]. Furthermore, validation of the information on date of surgery and laterality registered in the DBCG database has shown a more than 90 and 95% concordance, respectively, and in case of discrepancy in date of surgery the difference was found small [17]. Data in the DNHR have been reported electronically from all Danish 123

63 228 Breast Cancer Res Treat (2012) 131: hospitals. Validation of the registered treatment has shown a 66 99% concordance [18]. Both databases are nationwide which provides us with a large and population-based data material. However, despite the population-based design the study lacks statistical power. The unexpected finding of a breast cancer or the need for a surgical excision biopsy is rare when using a modern diagnostic set-up. Moreover, the overall risk of having ITC in the SN is reported to be only about 3 4% [19, 20], and the reported risk of non-sn metastases in these patients is only 10 15% [7]. Even in a population-based study like this, based on more than 17,000 women, we end up with a small study group when assessing the risk of non-sn metastases. The fact that the differences in dissemination of ITC to non-sns in patients with and without recent surgical excision did not reach statistical significance is possibly related to study size. Patients with surgical excision in the weeks before SLND represents a highly selected group compared to the cohort of breast cancer patients without recent surgical procedure, with younger age and smaller tumor size. We adjusted for these differences by using a multivariate design including all known confounders, but still residual confounding may exist. The increased incidence of macrometastases in the control group, even after adjustment for confounders, indicate that some degree of residual confounding exists. Furthermore, information on recent needle biopsy of the breast tumor was not included in the study. It has earlier been shown that such a procedure is related to a higher risk of ITC in the SN [10 13]. In Denmark, almost all breast cancer patients are offered either fine needle aspiration or core needle biopsy in order to obtain a valid cancer diagnosis before surgery, but registration validity of this information in the DBCG database tends to be poor. We assume that basically all patients included in this study underwent needle biopsy, but it is possible that some patients in the group with surgical excision did not have a needle procedure before excision. This would tend to underestimate the effect of surgical excision on displacement of tumor cells. The nearly fourfold increased risk of having ITC in the SN after surgical excision indicates that this procedure may induce iatrogenic displacement of tumor cells. This phenomenon has been described earlier [10, 21]. Studies have shown a higher incidence of ITC in the SN after needle biopsy or manipulation of a breast cancer [11 13]. However, the amount of iatrogenic displacement after surgical excision is controversial and not very well investigated. To our knowledge, only two studies have addressed the subject [12, 13]. Hansen et al. [13] have shown a significantly lower incidence of SN metastases after surgical excision compared to needle biopsy, while Moore et al. [12] have shown an increased number of SN metastases identified by immunohistochemistry if a surgical excision biopsy is performed before SLND. Our results support the existence of iatrogenic displacement of tumor cells to the SN after surgical excision. However, our data could not show a time dependent relationship between the degree of displacement and the interval since recent surgical excision. Nevertheless, it must be taken into account that the number of patients with ITC in the SN in each time interval was small and a time dependent relation cannot be excluded. Furthermore, we found a smaller, but significant, increase in the risk of having micrometastases in the SN after surgical excision. This has not been shown earlier. The increase could theoretically be explained by growth of displaced ITC into micrometastases, but a more simple explanation is misclassification of ITC as micrometastases or residual confounding. It has been shown that lobular carcinomas are more likely to present with ITC in the SN than ductal carcinomas [22, 23]. Our results confirm this observation. In contrast, the increased risk of having ITC in the SN after recent surgical excision was especially seen in patients with ductal carcinomas. A similar tendency of displacement of cells from ductal carcinomas is seen in the surrounding breast tissue after core needle biopsy [24], but the phenomenon has never been shown in the SN before. Not all ITC in the SN are likely to disseminate to non-sns [6]. Especially, the clinical significance of iatrogenically displaced ITC is questioned but research on the subject is scarce. Only one study by Hansen et al. [13] has investigated the presence of non-sn metastases in relation to iatrogenic displacement. They found that 39% of patients with positive SN after fine needle aspiration had non-sn metastases, 37% had non-sn metastases after core needle biopsy and 31% after surgical excision biopsy. No significant difference between groups was found. No control group was included in the study and the incidence of non-sn metastases was not stratified according to size of the SN metastases. We investigated the incidence of non-sn metastases after recent surgical excision in patients with either ITC or micrometastases in the SN. We found no dissemination to non-sns if only ITC were found in the SN, in contrast to patients with micrometastases where 12% had spread to non-sns. The absence of non-sn metastases in patients with ITC in the SN after recent surgical excision suggests that these ITC are not an indicator of further axillary spread. However, the number of patients was low, and the results did not reach statistical significance. Conclusion Breast cancer patients with surgical excision of the breast tumor in the weeks before SLND have a nearly fourfold 123

64 Breast Cancer Res Treat (2012) 131: increased risk of being diagnosed with ITC in the SN compared with patients treated in a one-step surgical procedure, indicating that the surgical excision induces iatrogenic displacement of tumor cells. The increase was especially seen in patients with ductal carcinomas. We found no further dissemination to non-sns in patients with ITC after recent surgical excision. In Denmark, all patients with ITC in the SN are offered a subsequent ALND because of the risk of non-sn metastases [8]. With a modern diagnostic set-up, surgical excision biopsy of a breast cancer is supposed to be reduced to a minimum. In case of recent surgical tumor excision, ITC in the SN should be interpreted with caution and omission of ALND should be considered, especially in patients with ductal carcinomas. Acknowledgments The study was supported by IMK common foundation, HOC research initiative foundation, the Danish cancer research foundation and A.P. Møller Foundation for the Advancement of Medical Science. Conflicts of interest References None. 1. Chen SL, Hoehne FM, Giuliano AE (2007) The prognostic significance of micrometastases in breast cancer: a SEER population-based analysis. Ann Surg Oncol 14: Dowlatshahi K, Fan M, Snider HC, Habib FA (1997) Lymph node micrometastases from breast carcinoma: reviewing the dilemma. Cancer 80: Kuijt GP, Voogd AC, van de Poll-Franse LV, Scheijmans LJ, van Beek MW, Roumen RM (2005) The prognostic significance of axillary lymph-node micrometastases in breast cancer patients. Eur J Surg Oncol 31: Grabau D, Jensen MB, Rank F, Blichert-Toft M (2007) Axillary lymph node micrometastases in invasive breast cancer: national figures on incidence and overall survival. APMIS 115: Klein CA, Blankenstein TJ, Schmidt-Kittler O, Petronio M, Polzer B, Stoecklein NH, Riethmuller G (2002) Genetic heterogeneity of single disseminated tumour cells in minimal residual cancer. Lancet 360: van Deurzen CH, Bult P, de Boer M et al (2009) Morphometry of isolated tumor cells in breast cancer sentinel lymph nodes: metastases or displacement? Am J Surg Pathol 33: van Deurzen CH, de Boer M, Monninkhof EM, Bult P, van der Wall E, Tjan-Heijnen VC, van Diest PJ (2008) Non-sentinel lymph node metastases associated with isolated breast cancer cells in the sentinel node. J Natl Cancer Inst 100: Christiansen P, Friis E, Balslev E, Jensen D, Moller S (2008) Sentinel node biopsy in breast cancer: five years experience from Denmark. Acta Oncol 47: Husen M, Paaschburg B, Flyger HL (2006) Two-step axillary operation increases risk of arm morbidity in breast cancer patients. Breast 15: Bleiweiss IJ, Nagi CS, Jaffer S (2006) Axillary sentinel lymph nodes can be falsely positive due to iatrogenic displacement and transport of benign epithelial cells in patients with breast carcinoma. J Clin Oncol 24: Diaz NM, Cox CE, Ebert M et al (2004) Benign mechanical transport of breast epithelial cells to sentinel lymph nodes. Am J Surg Pathol 28: Moore KH, Thaler HT, Tan LK, Borgen PI, Cody HS III (2004) Immunohistochemically detected tumor cells in the sentinel lymph nodes of patients with breast carcinoma: biologic metastasis or procedural artifact? Cancer 100: Hansen NM, Ye X, Grube BJ, Giuliano AE (2004) Manipulation of the primary breast tumor and the incidence of sentinel node metastases from invasive breast cancer. Arch Surg 139: Moller S, Jensen MB, Ejlertsen B et al (2008) The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 47: Singletary SE, Allred C, Ashley P et al (2002) Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 20: Rostgaard K, Holst H, Mouridsen HT, Lynge E (2000) Do clinical databases render population-based cancer registers obsolete? The example of breast cancer in Denmark. Cancer Causes Control 11: Jensen AR, Storm HH, Moller S, Overgaard J (2003) Validity and representativity in the Danish Breast Cancer Cooperative Group a study on protocol allocation and data validity from one county to a multi-centre database. Acta Oncol 42: Nickelsen TN (2001) Data validity and coverage in the Danish National Health Registry. A literature review. Ugeskr Laeger 164: Kiaer HW, Laenkholm AV, Nielsen BB, Bjerre KD (2008) Classical pathological variables recorded in the Danish Breast Cancer Cooperative Group s register Acta Oncol 47: Straver ME, Meijnen P, van Tienhoven G et al (2010) Sentinel Node Identification Rate and Nodal Involvement in the EORTC AMAROS Trial. Ann Surg Oncol 17: Carter BA, Jensen RA, Simpson JF, Page DL (2000) Benign transport of breast epithelium into axillary lymph nodes after biopsy. Am J Clin Pathol 113: Mittendorf EA, Sahin AA, Tucker SL et al (2008) Lymphovascular invasion and lobular histology are associated with increased incidence of isolated tumor cells in sentinel lymph nodes from early-stage breast cancer patients. Ann Surg Oncol 15: Vincent-Salomon A, Caly M, De RY et al (2009) Lobular phenotype related to occult-metastatic spread in axillary sentinel node and/or bone marrow in breast carcinoma. Eur J Cancer 45: Diaz LK, Wiley EL, Venta LA (1999) Are malignant cells displaced by large-gauge needle core biopsy of the breast? AJR Am J Roentgenol 173:

65 The Breast 22 (2013) 44e46 Contents lists available at SciVerse ScienceDirect The Breast journal homepage: Original article Quantifying the number of lymph nodes identified in one-stage versus two-stage axillary dissection in breast cancer Olaf E. Damgaard a, Maj-Britt Jensen b, Niels Kroman a, Tove F. Tvedskov a, * a Dept of Breast Surgery, 4124, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark b Danish Breast Cancer Cooperative Group, 2501, Copenhagen University Hospital, Blegdamsvej 9, Copenhagen, Denmark article info abstract Article history: Received 9 February 2012 Received in revised form 14 March 2012 Accepted 18 March 2012 Keywords: Sentinel node Axillary lymph node dissection Breast cancer Purpose: To establish whether a different number of lymph nodes is identified in a delayed versus an immediate axillary lymph node dissection (ALND) in breast cancer patients. Methods: Using data from the Danish National Patient Register and the Danish Breast Cancer Cooperative Group Database we identified 864 breast cancer patients with sentinel lymph node dissection (SLND) and delayed ALND and 7393 breast cancer patients with SLND and immediate ALND operated between 2002 and We compared the number of lymph nodes identified in the two groups by a student s t-test. Results: The mean number of lymph nodes identified in patients with immediate and delayed ALND was and 15.59, respectively. This difference was statistically significant (P < ). Conclusion: The number of lymph nodes identified in breast cancer patients is slightly reduced if delayed ALND is performed. However, the difference is small and considered to be without clinical significance. Ó 2012 Elsevier Ltd. All rights reserved. Introduction Sentinel lymph node dissection (SLND) is today a standard procedure for staging of the axilla in breast cancer patients. In Denmark, removed sentinel nodes are generally examined peroperatively on frozen sections. If metastases are detected, the surgeon will immediately proceed to an axillary lymph node dissection (ALND) during the same operation. The detection rate for metastases on frozen sections is 85e90%, 1,2 which means that about 10e15% of sentinel node positive patients are initially found negative, but will during the postoperative conventional histological examination be found positive. Those patients will be assigned to a second operation where an ALND is performed. The smaller the metastasis, the higher the likelihood of being overlooked on frozen sections. 1 Since the detection of isolated tumor cells (ITC) and micrometastases are improving, 3,4 the number of patients who end up receiving a two-stage procedure will increase, as long as these patients are offered an ALND. ALND as a second procedure may be more challenging due to inflammatory changes, scarring and fibrosis in the axilla after * Corresponding author. Tel.: þ ; fax: þ address: tft@dadlnet.dk (T.F. Tvedskov). a recent SLND. We hypothesize that the initial SLND can hamper the number of lymph nodes detected by the second procedure, thereby reducing the total number of lymph nodes identified. This will make a two-stage procedure less precise and emphasize the importance of a high detection rate of metastases on frozen sections. The literature on the subject is limited and results are diverging. 5e8 By using data from the Danish National Patient Register (NPR) and the Danish Breast Cancer Cooperative Group (DBCG) database we investigated the number of lymph nodes identified in a onestage versus a two-stage axillary procedure in a large population of breast cancer patients. Materials and methods In Denmark, clinical and histopathological data as well as information on treatment and follow-up status on women with breast cancer are prospectively registered in a national database managed by the DBCG. The DBCG was founded in 1977 and today the database contains information on more than 80,000 breast cancer patients. 9 Linking the DBCG database and the Danish Cancer Register, which is considered close to complete, has revealed over 95% concordance. 10 Validation of the information on date of surgery and laterality registered in the DBCG database has shown a more /$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi: /j.breast

66 O.E. Damgaard et al. / The Breast 22 (2013) 44e46 45 than 90% and 95% accuracy, respectively. In case of discrepancy in date of surgery, the difference was found to be small. 11 The NPR was founded in 1977 and contains information on hospital admission, diagnosis and treatment of all hospitalized patients in Denmark. Data is registered electronically from all Danish hospitals. Validation of the registered treatment has shown a66e99% accuracy. 12 From the DBCG database we retrieved information on the total number of lymph nodes identified in patients operated by SLND and a subsequent ALND between 2002 and 2010 as well as date and laterality of the procedure and patients age at diagnosis. In the DBCG database only the date of the first surgical procedure is registered, regardless of a later ALND or re-excision. In the NPR, exact dates for every single procedure are registered. Accordingly, we used the NPR to identify patients registered with an ALND more than 7 days and a maximum of 60 days after the SLND, making it a two-stage procedure. Patients identified in the NPR as having a two-stage procedure were combined with information retrieved from the DBCG database by using the social security number. Patients from the DBCG database, not identified in the NPR with a two-stage procedure, were considered as having a one-stage procedure. We excluded patients with discrepancies in date or laterality of the SLND between the DBCG and the NPR databases. We allowed however up to one week discordance in registered dates between databases, because this was considered registration errors. No lower limit of the number of lymph nodes identified by ALND was used. However, patients with no difference between the number of sentinel nodes and the total number of lymph nodes identified were excluded. The mean total number of lymph nodes identified in each group was compared by a student s t-test. The association between the two groups, patients age at diagnosis and the number of lymph nodes identified was tested in a linear regression model. The study has been approved by the Danish Data Protection Agency (J.nr ), and by the National Board of Health (J.nr /1). Results A total number of 8257 patients were included in the study, 7393 patients with a one-stage procedure and 864 patients with a two-stage procedure. An equal proportion of patients in the two groups (93%) had a total number of 10 or more lymph nodes identified by SLND and ALND. In the one-stage group we found a mean number of lymph nodes identified, and in the twostage group the mean number was 15.59, leaving a difference of Table 1 Mean number of lymph nodes identified according to age in a one- and two-stage axillary procedures in 8257 Danish breast cancer patients. Age, years One-stage procedure Two-stage procedure No. Patients Mean No. LN removed No. Patients Mean No. LN removed 20e >30e >40e >50e >60e >70e >80e > e Total LN ¼ lymph nodes, ALND ¼ axillary lymph node dissection lymph nodes. This difference was statistically significant (P < ). A small but significant decrease in the number of identified lymph nodes was seen by increasing age (P < ) (Table 1). Mean age at diagnosis in the two groups differed by less than half a year, being years and years, respectively, in the oneand two-stage group. To exclude an age-dependent difference in the number of lymph nodes identified in the two groups we made an age-adjusted analysis. After adjusting for patients age at diagnosis we still found a significantly lower number of lymph nodes identified by a two-stage procedure, with a difference between groups on 0.97 lymph nodes (P < ). We furthermore made a subanalysis excluding patients above the age of 70. This reduced the number of patients to 5994 in the one-stage group, and 762 in the two-stage group. The mean number of lymph nodes identified increased only slightly to for a one-stage, and for a two-stage procedure, resulting in a difference of 1.00 lymph node less identified by a two-stage procedure. The difference remained statistically significant (P < ). The number of lymph nodes identified according to age at diagnosis for the two groups is shown in Table 1. Discussion We found approximately only one lymph node less identified in breast cancer patients operated by a two-stage axillary procedure compared to a one-stage procedure. Earlier studies have investigated differences between one- and two-stage axillary procedures. A Danish study found on average one lymph node less removed by a two-stage procedure compared to a one-stage procedure. 5 The same trend was found in a minor study by Chakravorty et al. 8 In contrast, Goyal et al. found a higher number of in average two more lymph nodes more removed by a two-stage procedure. 6 None of the differences were found to be statistically significant, but the studies were generally small including only about 100 patients in each group. Olson et al. investigated the difference between one-stage and two-stage procedures in a cohort 4e5 times larger than the former three studies and they found no difference in the mean number of lymph nodes removed. 7 However, in a smaller Danish study comparing one- and two-stage procedures 10% of patients in the two-stage group had less than 10 lymph nodes identified compared to only 1% in the one-stage group. 13 The majority of the above mentioned studies were not aiming directly to establish whether there was a difference in the number of lymph nodes identified in one- versus two-stage procedures. This was the primary aim of our study. Furthermore, our results were based on a large and nation-wide cohort of more than 8000 breast cancer patients, which provided statistical power to show even a small difference. However, basing a study on register data gives us some limitations. Some patients in the database were registered as having an ALND, but still only one or two lymph nodes had been identified. It is not likely, that a complete ALND was intended in these patients. Still, a very low number of lymph nodes registered could not be used as exclusion criteria when the aim of the study was to examine the number of lymph nodes identified. Excluding patients with less than 10 lymph nodes identified would not reveal if patients in a two-stage procedure were more likely to have an insufficient number of lymph nodes removed, e.g. only nine or even only eight lymph nodes. Age has been shown to influence the number of lymph nodes removed by ALND. 14e16 We confirmed this association of a decreasing number of lymph nodes identified by increasing age.

67 46 O.E. Damgaard et al. / The Breast 22 (2013) 44e46 However, this age-dependent association could not explain the difference between the two groups. In our study, the mean age difference between groups was less than half a year and neither an age-adjusted analysis nor a subanalysis of patients under the age of 70 did change the results. The surgical approach to both one- and two-stage operations is very similar, but the second operation in a two-stage procedure may prove more difficult due to fibrosis and scar tissue making it possible that lymph nodes are left behind. It is furthermore possible that lymph nodes are undetected by pathologists due to inflammatory changes or destroyed or lost in the procedure due to surgical difficulties around the scarred areas. This can explain the slightly reduced number of lymph nodes identified. Above all, the reduction in the number of lymph nodes identified by a two-stage procedure was small. Moreover, in average over 15 lymph nodes were identified in both groups. In general, removal of at least 10 lymph nodes is considered adequate, 17 indicating that a satisfactory number of lymph nodes were removed even at the two-stage procedure. Furthermore, there was no difference in the proportion of patients with less than 10 lymph nodes identified in the two groups. If frozen sections are omitted in selected patients with a low risk of sentinel node metastases, to spare operating time 1 and avoid loss of tissue, 18 some of these patients will undergo a two-stage procedure. This will, according to our results, not deteriorate their possibilities of a precise axillary staging at final histopathological examination. In conclusion, our findings show that a two-stage procedure does influence the number of lymph nodes identified. However, the difference was small and is considered to be without clinical significance. Conflict of interest statement The authors have no conflicts of interest to declare. Acknowledgments The study was supported by IMK common foundation, Danish Cancer Research Foundation and A.P. Møller Foundation for the Advancement of Medical Science. Balder Kjerulff was helpful with data handling. References 1. Holm M, Paaschburg B, Balslev E, Axelsson CK, Willemoe GL, Flyger HL. Intraoperative immunohistochemistry staining of sentinel nodes in breast cancer: clinical and economical implications. Breast 2008;17(4):372e5. 2. Kroman N, Christiansen P, Hussain ZB, Kehlet H. Breast cancer surgery in Denmark. Ugeskr Laeger 2010;172(41):2836e9. 3. Tvedskov TF, Jensen MB, Balslev E, Ejlertsen B, Kroman N. Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: a nationwide study. Eur J Cancer 2011;47(6):872e8. 4. Chen SL, Hoehne FM, Giuliano AE. The prognostic significance of micrometastases in breast cancer: a SEER population-based analysis. Ann Surg Oncol 2007;14(12):3378e Husen M, Paaschburg B, Flyger HL. Two-step axillary operation increases risk of arm morbidity in breast cancer patients. Breast 2006;15(5):620e8. 6. Goyal A, Newcombe RG, Chhabra A, Mansel RE. Morbidity in breast cancer patients with sentinel node metastases undergoing delayed axillary lymph node dissection (ALND) compared with immediate ALND. Ann Surg Oncol 2008;15(1):262e7. 7. Olson Jr JA, McCall LM, Beitsch P, Wietworth PW, Reintgen DS, Blumencranz PW, et al. Impact of immediate versus delayed axillary node dissection on surgical outcomes in breast cancer patients with positive sentinel nodes: results from American College of Surgeons Oncology Group Trials Z0010 and Z0011. J Clin Oncol 2008;26(21):3530e5. 8. Chakravorty A, Sanmugalingam N, Shrestha A, Thomee E, Rusby J, Roche N, et al. Axillary nodal yields: a comparison between primary clearance and completion clearance after sentinel lymph node biopsy in the management of breast cancer. Eur J Surg Oncol 2011;37(2):122e6. 9. Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47(4):506e Rostgaard K, Holst H, Mouridsen HT, Lynge E. Do clinical databases render population-based cancer registers obsolete? The example of breast cancer in Denmark. Cancer Causes Control 2000;11(7):669e Jensen AR, Storm HH, Moller S, Overgaard J. Validity and representativity in the Danish Breast Cancer Cooperative Groupea study on protocol allocation and data validity from one county to a multi-centre database. Acta Oncol 2003;42(3):179e Nickelsen TN. Data validity and coverage in the Danish national Health Registry. A literature review. Ugeskr Laeger 2001;164(1):33e Madsen AH, Jensen AR, Christiansen P, Garne JP, Cold S, Ewertz M, et al. Does the introduction of sentinel node biopsy increase the number of node positive patients with early breast cancer? A population based study form the Danish Breast Cancer Cooperative Group. Acta Oncol 2008;47(2):239e Braems G, Denys H, Cocquyt V, Van den BR. Preceding sentinel node biopsy in early breast cancer: does it affect the number of axillary lymph nodes? Acta Chir Belg 2008;108(6):691e Chagpar AB, Scoggins CR, Martin RC, Sahoo S, Carlson DJ, Laidley AL, et al. Factors determining adequacy of axillary node dissection in breast cancer patients. Breast J 2007;13(3):233e Petrik DW, McCready DR, Sawka CA, Goel V. Association between extent of axillary lymph node dissection and patient, tumor, surgeon, and hospital factors in patients with early breast cancer. J Surg Oncol 2003;82(2):84e Axelsson CK, Rank F, Blichert-Toft M, Mouridsen HT, Jensen MB. Impact of axillary dissection on staging and regional control in breast tumors < or ¼ 10 mmethe DBCG experience. The Danish Breast Cancer Cooperative Group (DBCG), Rigshospitalet, Copenhagen, Denmark. Acta Oncol 2000;39(3):283e Fritzsche FR, Reineke T, Morawietz L, Kristiansen G, Dietel M, Fink D, et al. Pathological processing techniques and final diagnosis of breast cancer sentinel lymph nodes. Ann Surg Oncol 2010;17(11):2892e8.

68 IJC International Journal of Cancer High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the sentinel node Tove Filtenborg Tvedskov 1, Maj-Britt Jensen 2, Ida Marie Lisse 3, Bent Ejlertsen 4, Eva Balslev 3 and Niels Kroman 1 1 Department of breast surgery, Copenhagen University Hospital, Copenhagen, Denmark 2 Danish Breast Cancer Cooperative Group, Copenhagen, Denmark 3 Department of pathology, Herlev Hospital, Copenhagen, Denmark 4 Department of oncology, Copenhagen University Hospital, Copenhagen, Denmark Axillary lymph node dissection (ALND) in breast cancer patients with positive sentinel nodes is under debate. We aimed to establish two models to predict non-sentinel node (NSN) metastases in patients with micrometastases or isolated tumor cells (ITC) in sentinel nodes, to guide the decision for ALND. A total of 1,577 breast cancer patients with micrometastases and 304 with ITC in sentinel nodes, treated by sentinel lymph node dissection and ALND in were identified in the Danish Breast Cancer Cooperative Group database. Risk of NSN metastases was calculated according to clinicopathological variables in a logistic regression analysis. We identified tumor size, proportion of positive sentinel nodes, lymphovascular invasion, hormone receptor status and location of tumor in upper lateral quadrant of the breast as risk factors for NSN metastases in patients with micrometastases. A model based on these risk factors identified 5% of patients with a risk of NSN metastases on nearly 40%. The model was however unable to identify a subgroup of patients with a very low risk of NSN metastases. Among patients with ITC, we identified tumor size, age and proportion of positive sentinel nodes as risk factors. A model based on these risk factors identified 32% of patients with risk of NSN metastases on only 2%. Omission of ALND would be acceptable in this group of patients. In contrast, ALND may still be beneficial in the subgroup of patients with micrometastases and a high risk of NSN metastases. Axillary nodal status is the most important prognostic factor in breast cancer. Axillary lymph node dissection (ALND) has previously been standard procedure for staging of the axilla, but ALND is associated with considerable arm morbidity 1,2 and is redundant for women without lymph node metastases. Sentinel lymph node dissection (SLND) causes limited arm morbidity compared to ALND 3,2 and has, therefore, gradually replaced ALND as standard procedure for staging of the Key words: breast carcinoma, sentinel lymph node biopsy, micrometastases, isolated tumor cells, non-sentinel node metastases Abbreviations: AJCC: American Joint Committee on Cancer; ALND: Axillary lymph node dissection; AUC: area under curve; CI: confidence interval; DBCG: Danish Breast Cancer Cooperative Group; ITC: isolated tumor cells; NSN: non-sentinel node; OR: odds ratio; ROC: receiver operating characteristic; SLND: sentinel lymph node dissection Grant sponsors: IMK common foundation, HOC research initiative foundation, Danish Cancer Research Foundation, A.P. Møller Foundation for the Advancement of Medical Science DOI: /ijc History: Received 11 Nov 2011; Accepted 20 Jan 2012; Online 18 Feb 2012 Correspondence to: Tove Filtenborg Tvedskov, Department of breast surgery, Copenhagen University Hospital, Blegdamsvej 9, 2100 København Ø, Denmark, Tel.: þ , Fax: þ , tft@dadlnet.dk axilla. SLND allows more extensive histopathological examinations of the lymph nodes and as a result more metastases, especially more micrometastases and isolated tumor cells (ITC), are found. 4,5 Micrometastases are in Denmark defined as metastases measuring between 0.2 and 2 mm or between 10 and 100 tumor cells and ITC are defined as single cells or cell clusters <0.2 mm or <10 cells. 6 Sentinel node positive patients have generally been recommended additional ALND even if only micrometastases or ITC are found in the sentinel node. However, metaanalyses have shown that only about 20% of patients with micrometastases 7 and 12% of patients with ITC 8 in the sentinel node have metastatic spread to non-sentinel node (NSN) and the benefit from ALND in these patients is now under debate. 9,10 A recent randomized trial from the American College of Surgeons Oncology Group (Z0011) could not show any difference in axillary recurrence and survival between sentinel node positive patients with or without ALND. 10 However, the study was closed early, including less than half the planned number of patients and included only patients with breast conserving surgery. Thus, the results cannot be transformed into any breast cancer patient with a positive sentinel node. Accordingly, a model is needed to identify patients who will need an ALND in the future because of a high risk of relapse. The presence of NSN metastases can be considered as a surrogate endpoint for axillary recurrence. Several risk factors for metastatic spread to NSN have been identified in patients with macrometastases in the Epidemiology Int. J. Cancer: 131, (2012) VC 2012 UICC

69 2368 Risk of non-sentinel node metastases sentinel node 11 and scoring systems have been developed. 12,13 Unfortunately, these scoring systems are not very precise in patients with only small metastases Studies indicate that tumor size and lymphovascular invasion are associated with further metastatic spread to NSN in patients with minimal metastatic disease in the sentinel node, 8,19 24 but most studies have been too small to develop a useful risk-model to support the decision for omitting ALND in selected patients. In Denmark, clinical and histopathological data on women with breast cancer have been registered in a national database managed by the Danish Breast Cancer Cooperative Group (DBCG) since Today, the database contains information on >80,000 breast cancer patients. 25 This gave us data material of a unique size for developing models to predict NSN metastases. The aim of our study was to establish a model to identify a group of patients with micrometastases or ITC in the sentinel node where ALND may still be recommended because of high risk of NSN metastases and furthermore to identify a group of patients where ALND can safely be omitted because of minimal risk of NSN metastases. Epidemiology Material and Methods Patients In Denmark, all departments of breast surgery must be certified by DBCG to perform SLND. 26 Since 2002, all SLND have been performed according to national guidelines described by DBCG. A combination of radioactive tracer and blue dye is recommended, while lymphoscintigraphy is not used as a routine. 6 All removed sentinel nodes are examined with at least two step sections 500 lm apart of the bivalved sentinel node. In case of no metastases found by hematoxylin-eosin staining, immunohistochemical cytokeratin staining is performed. Lymph nodes removed by ALND are examined by bisectioning and hematoxylin-eosin staining. From the introduction of SLND in 2002 to the end of 2008, a total number of 2,293 breast cancer patients have been registered in the DBCG database with micrometastases or ITC in the sentinel node. Since 2005, metastases have been classified according to the American Joint Committee on Cancer (AJCC) staging manual 27 in combination with cell count, where metastases between 0.2 and 2 mm or between 10 and 100 tumor cells are defined as micrometastases, and single cells or cell clusters <0.2 mm or <10 cells are defined as ITC. 6 From 2005 to 2008, a total number of 368 patients were registered with ITC and 1,474 patients were registered with micrometastasis in the sentinel node. This information was validated using original pathology files. Fifteen patients had macrometastases; and in two patients, the pathology file could not be found. These patients were excluded. Eleven patients registered with ITC in the sentinel node were identified with micrometastases and one patient registered with micrometastases was identified with ITC. In the first years after introduction of SLND in breast cancer treatment in Denmark, all tumor cell deposits under 2 Figure 1. Flowchart for inclusion of patients from the DBCG database with micrometastases or isolated tumor cells in the sentinel node. Abbreviations: DBCG: Danish breast cancer cooperative group; ALND: axillary lymph node dissection. mm or <100 tumor cells were registered as micrometastases. From 2002 to 2004, a total number of 451 patients were registered with micrometastases in the DBCG database. A reevaluation of specimens from these patients identified 278 patients with micrometastases and 68 patients with ITC when using the staging manual 27 from AJCC in combination with cell count as described. Of the remaining patients, 74 patients had macrometastases, eight patients were node negative and in 23 patients the specimen was missing. These 105 patients were excluded. The re-evaluation was performed by two breast-trained pathologists from the Department of Pathology, Herlev Hospital. Missing information on additional ALND was collected from original pathology files. In total, 279 patients did not undergo an adequate ALND with at least seven lymph nodes removed and were excluded. Another 11 patients were excluded because of neoadjuvant treatment. A total number of 1,577 patients with micrometastases and 304 patients with ITC were eligible for final analysis (Fig. 1). Int. J. Cancer: 131, (2012) VC 2012 UICC

70 Tvedskov et al Variables From the DBCG database, we retrieved information on age at diagnosis, tumor size, hormone receptor status, histological type, malignancy grade, number of removed sentinel nodes, number of positive sentinel nodes, lymphovascular invasion, multifocality, HER2 status, location of tumor in the breast and presence of NSN metastases. Location of tumor in the breast was divided into location in upper lateral quadrant versus located in other quadrants, centrally or on the edge of the upper lateral quadrant. Original pathology files were searched for information on HER2 status of the primary tumor not reported to the database, and for information on location of metastasis in the sentinel node. For micrometastases, we furthermore searched the files for information on extracapsular extension of the sentinel node metastasis. Statistical analyses Associations between presence of NSN metastases and the characteristics listed in Table 1 were analyzed by v 2 test and Fischer s exact test, excluding unknowns, for patients with ITC and patients with micrometastases, respectively, in the sentinel node. Univariate and multivariate logistic regression models were applied to examine the influence of age at diagnosis, tumor size, histological type, malignancy grade, lymphovascular invasion, hormone receptor status, HER2 status, location of tumor in the breast, focality of tumor, location of metastasis in the sentinel node, extracapsular extension (for micrometastases only), number of removed sentinel nodes as well as the proportion of positive sentinel nodes among removed sentinel nodes on the risk of NSN metastases in patients with ITC and patients with micrometastases in the sentinel node. Fifty-six patients with micrometastases and five patients with ITC in the sentinel node were excluded from the multivariate analyses because of missing information on variables. Adjusted odds ratio (OR) and 95% confidence intervals (CI) were calculated and the Wald test was used to test the significance of each variable. Test for interaction between covariates were performed pairwise. For the multivariate model for micrometastases, a receiver operating characteristic (ROC) curve including the c-statistic [area under curve (AUC)] was produced. Also a score was assigned to each patient by adding the relevant b-coefficients from the multivariate logistic regression model, which was supplemented by a simplified score by adding number of risk factors present. SAS version 9.1 (SAS Institute, Cary, NC) was used for all statistical analyses. The study was approved by the Danish Data Protection Agency (J.nr ). Results Distributions of variables between patients with and without NSN metastases are shown in Table 1. NSN metastases were found in 28 of 304 patients with ITC in the sentinel node, corresponding to 9% of the patients, and 283 of 1,577 patients with micrometastases in the sentinel node, corresponding to 18% of the patients. An average number of 16.4 lymph nodes (range 7 40) were removed. Micrometastases In patients with micrometastases, the risk of NSN metastases was in the univariate analyses, significantly associated with increasing tumor size, lymphovascular invasion, negative hormone receptor status, multifocality, location of tumor in the upper lateral quadrant of the breast, number of removed sentinel nodes and increasing proportion of positive sentinel nodes (Table 1). In the multivariate analysis, the risk of NSN metastases remained significantly associated with increasing tumor size (trend cm), lymphovascular invasion, negative hormone receptor status, location of tumor in the upper lateral quadrant of the breast and increasing proportion of positive sentinel nodes [100% vs. (0 25%), (>25 33%), (>33 to <100%)]. Number of removed sentinel nodes (p ¼ 0.26) and multifocality (p ¼ 0.18) were no longer significantly associated with NSN metastases. A significant interaction between tumor size and lymphovascular invasion was found with lymphovascular invasion as a stronger risk factor in larger tumors. OR for the significant variables are shown in Table 2. The five significant variables from the multivariate logistic regression analysis were included in a model to predict NSN metastases. A score was assigned to each patient by adding the relevant b- coefficients. In the model, 5% of patients were identified with a high risk of NSN metastases on 37.4%. However, the model was unable to identify a subgroup of patients with a very low risk of NSN metastases; patients with the lowest risk score had a 12% risk for NSN metastasis. The AUC for the constructed model was To make the model suitable for daily clinical use, the risk score was simplified according to the five risk factors from Table 2, counting one risk score per present risk factor. Fifty-seven patients had at least four risk factors present. 40% of these patients had NSN metastases. Fifty-eight patients did not have any of the five identified risk factors, but still 10% of these patients had NSN metastases (Table 3). In a subanalysis excluding patients with multifocal carcinomas, the estimated OR in the multivariate analysis changed only slightly. In this analysis, the association between negative hormone receptor status and NSN metastases was no longer significant. ITC In patients with ITC in the sentinel node younger age at diagnosis (<40 vs. 40þ), increasing tumor size (>2 cm vs. 2 cm) and increasing proportion of positive sentinel nodes (100% vs. <100%) were significantly associated with NSN metastases in the univariate analyses (Table 1). In the multivariate analysis, all three variables remained significantly associated with NSN metastases. OR for the significant variables are shown in Table 2. A simple model was constructed Epidemiology Int. J. Cancer: 131, (2012) VC 2012 UICC

71 2370 Risk of non-sentinel node metastases Epidemiology Table 1. Patient, tumor and sentinel node characteristics according to the risk of NSN metastases in 304 Danish breast cancer patients with ITC and 1,577 patients with micrometastases in the sentinel node operated between Isolated tumor cells Micrometastases NSN metastases NSN metastases Variables No Yes % p-value No Yes % p-value Total , Age, years 0.06* 0.41 < Tumor size, cm # > > > Unknown WHO type Ductal , Lobular Other Unknown Grade Grade I Grade II Grade III Ungradable LVI Present Absent , Unknown Hormone receptor status Positive , Negative Unknown HER2 status Positive Negative Unknown Location of tumor in breast Upper lateral Borderline upper lateral Not upper lateral Unknown Int. J. Cancer: 131, (2012) VC 2012 UICC

72 Tvedskov et al Table 1. Patient, tumor and sentinel node characteristics according to the risk of NSN metastases in 304 Danish breast cancer patients with ITC and 1,577 patients with micrometastases in the sentinel node operated between (Continued) Isolated tumor cells Micrometastases NSN metastases NSN metastases Variables No Yes % p-value No Yes % p-value Focality of tumor in breast Multifocal Unifocal , Unknown Location of metastasis in SN Capsula Parenchyma Vessels Multifocal Unknown Extracapsular extension 0.58 Present Absent 1, Number of removed SN Unknown Positive SN/Removed SN % >33%, <100% >25; % Unknown *p ¼ 0.03 <40 vs. 40, # p ¼ cmvs. >2 cm, p ¼ % vs. <100%. Abbreviations: ITC: isolated tumor cells; NSN: non-sentinel nodes; SN: sentinel node. based on the three risk factors from Table 2 (Table 4). In the model, patients without any of the three risk factors had a very low risk, on only 2%, of having NSN metastases. This group represented 32% of patients with ITC. Only four patients had all three risk factors present. Three of these four patients (75%) had NSN metastases. In the subanalysis excluding patients with multifocal carcinomas age at diagnosis was no longer significantly associated with NSN metastases in the multivariate analysis. However, in our data material, age was only decisive in a few patients, and tumor size and proportion of positive sentinel nodes still identified 1/3 of patients with ITC with a risk of NSN metastases <2%. Discussion According to our results, omission of ALND in breast cancer patients with ITC in the sentinel node may be reasonable if age is 40 years at diagnosis and tumor size is 2 cm in the presence of at least one additional negative sentinel node, because only 2% of these patients had NSN metastases. We were unable to identify a subgroup of patients with micrometastases in the sentinel node with a risk of NSN metastases below 10%. In a multivariate analysis, we identified tumor size >2cm, lymphovascular invasion, negative hormone receptor status, location of tumor in the upper lateral quadrant of the breast and absence of negative sentinel nodes as risk factors for NSN metastases. By these risk factors, patients with micrometastases in the sentinel node could be split into risk groups according to the risk of NSN metastases, where the high risk group with four or more risk factors present had a 40% risk of NSN metastases. The results were based on a large sample size from the DBCG database of nearly 2,300 patients with micrometastases Epidemiology Int. J. Cancer: 131, (2012) VC 2012 UICC

73 2372 Risk of non-sentinel node metastases Epidemiology Table 2. Risk factors for NSN metastases in a multivariate analysis of 299 Danish breast cancer patients with ITC and 1,521 patients with micrometastases in the sentinel node operated between Isolated tumor cells Variable OR 95% CI p-value Tumor size, >2 vs. 2 cm Age, <40 vs. 40 years Proportion of pos SN, % vs. <100% Micrometastases Tumor size, cm, trend Proportion of pos SN, % vs. <100% Lymphovascular invasion Hormone receptor status, neg vs. pos Location of tumor in upper lateral quadrant Abbreviations: CI: confidence interval; ITC: isolated tumor cells; neg: negative; NSN: non-sentinel node; OR: Odds ratio; pos: positive; SN: sentinel node. Table 3. Model predicting the risk of NSN metastases in 1,521 Danish breast cancer patients with micrometastases in the sentinel node operated in based on number of risk factors present Patients with Patients NSN metastases Risk factors No. % No. % Total 1, Abbreviation: NSN: non-sentinel node. or ITC in the sentinel node, which makes this study far the largest study to date on the subject. We chose not to use patients from the current data material for validation of the models but included all available patients in the primary analysis. That enabled us to test all possible risk factors for NSN metastases and at the same time get the most robust estimates. Patients operated after end of inclusion for this study will be used for internal validation. We furthermore plan a subsequent validation on an external dataset from another institution outside Denmark. Data used in the study have been prospectively collected from all Danish women with breast cancer and registered on standardized forms in the DBCG database. Earlier studies have shown a >95% concordance between the DBCG database and the Danish Cancer Register as well as the National Table 4. Model predicting the risk of NSN metastases in 299 Danish breast cancer patients with ITC in the sentinel node operated in based on the number of risk factors present Patients with Patients NSN metastases Risk factors No. % No. % Total Abbreviations: ITC: isolated tumor cells; NSN: non-sentinel node. Pathology Register, which are considered close to complete. 25 Furthermore, data have been thoroughly validated using original pathology files and specimens. However, misclassification of metastases could arise if the pathologists were not consistent in how they evaluated and registered nodal metastases. 28 Nevertheless, a recent study investigating the heterogeneity in registration practice between Danish departments of pathology revealed no significant differences. 5 Thus, the DBCG data on nodal status can be considered as quite uniform. The benefit from ALND in breast cancer patients with minimal metastatic disease in the sentinel node is now under debate. 29 The recent randomized trial from the American College of Surgeons Oncology Group (Z0011) indicates that ALND can be safely omitted in sentinel node positive patients without impairment of prognosis. However, the study was closed early after including less than half the planned number of patients and the observed recurrence rate was low. Accordingly, it cannot be ruled out, that the similar recurrence rates for patients with and without ALND may reflect lack of power. Moreover, the study included only patients with small tumors, 1 2 positive sentinel nodes and breast conserving surgery with subsequent radiotherapy 10 which to some extend would result in irradiation of the lower part of the axilla and this can have affected the outcome. Furthermore, the majority received adjuvant systemic treatment. Accordingly, the results cannot be transformed into any breast cancer patient with a positive sentinel node. Hence, a tool is needed for a more detailed evaluation of the risk of NSN metastases to substantiate a safe omission of ALND. Several authors have tried to predict the risk of NSN metastases in sentinel node positive patients. In case of macrometastases in the sentinel node, several risk factors have been identified 11 and scoring systems have been developed. 12,13 However, these models are not very precise in patients with micrometastases in the sentinel node where they tend to overestimate the risk of NSN metastases ,22 Furthermore, none of the models have been tested in a population of patients with only ITC in the sentinel node. The unique size of our study population enabled us to test several possible risk factors for NSN metastases in breast Int. J. Cancer: 131, (2012) VC 2012 UICC

74 Tvedskov et al cancer patients with micrometastases or ITC in the sentinel node. We confirmed that increasing tumor size and lymphovascular invasion are predictors of NSN metastases in these patients. 14,19 21,23 We furthermore identified increased proportion of positive sentinel nodes, negative hormone receptor status and younger age at diagnosis as risk factors. Neither hormone receptor status nor age at diagnosis has earlier been found associated with NSN metastases in a population of patients with micrometastases or ITC in the sentinel node. However, an increasing proportion of positive lymph nodes has earlier been found associated with NSN metastases 30,31 as well as decreasing overall survival in patients with nodal micrometastases. 32 In contrast to patients with macrometastases, 11 we did not find an increased risk of NSN metastases in patients with extracapsular extension. This is in accordance with the results of a previous smaller study on patients with micrometastases. 33 In addition to these traditional prognostic factors, we tested HER2 status, location of tumor in the breast and location of metastases in the sentinel node. We confirmed a significant association between NSN metastases and location of tumor in the upper lateral quadrant, as shown previously. 34 This may be explained by different lymph draining patterns from different parts of the breast, where tumors in the upper lateral quadrant may be more likely to drain directly to additional lymph nodes in the more apical part of the axilla. 35,36 Similar to other studies including only patients with micrometastases, we could not show an association between HER2 status 19,14 and NSN status. Location of metastases in the parenchyma of the sentinel node has in previous studies been associated with NSN metastases. 37,38 We could not confirm this association in our study. Furthermore, it has earlier been shown, that the risk of NSN metastases is associated with the size of the micrometastases. 39 We did not have sufficient information on the exact size of the micrometastases to investigate this association. In a recent study on patients with micrometastases or ITC in the sentinel node multifocality was, besides tumor size, identified as the only predictor of NSN metastases in a multivariate analysis. 14 We could not confirm this association. The use of SLND in multifocal breast cancer remains in question, 40 and SLND is not yet standard treatment in multifocal breast cancer in all Danish departments. 6 Accordingly, a subanalysis excluding patients with multifocal breast carcinoma was performed. The estimated OR for NSN metastases showed robustness and changed only slightly in the subanalysis. Ki67 was not tested as a risk factor for NSN metastases in this study. It has been shown that large interlaboratory variability exists in the measurements of Ki67, making it difficult to compare results from several different departments. 41 In an earlier minor study, we have tested the ability of Ki67 to predict NSN metastases in breast cancer patients with micrometastases in the sentinel node from a single institution. 42 We found no association between Ki67 and NSN metastases. Ki67 was not measured as a routine in patients included in our study and these earlier results did not justify new analyses of Ki67. Based on all significant risk factors, we have constructed two models for the prediction of NSN metastases in patients with minimal metastatic disease in the sentinel node. In patients with micrometastases, the model showed no clearly defined low-risk group and the AUC was only Only two earlier studies have tried to construct a predictive model based on a population of patients with micrometastases or ITC. 14,24 These studies, including large sample sizes of 909 and 484 patients, respectively, could like us, not identify a substantial subgroup of patients with a very low risk of NSN metastases and the AUC for the constructed models were only 0.66 and The proportion of patients diagnosed with micrometastases or ITC in the sentinel node is generally low, making a further increase in sample size limited. In our opinion, collaboration between institutions is now needed to improve and validate the existing models. In addition, it cannot be excluded that the search for new biochemical markers might reveal new clinically significant predictors for NSN metastases that can improve the models. 43 It is possible that the presence of NSN metastases are associated with the genetic subtype of the tumor and it still needs to be elucidated whether gene expression profiles 44 can be useful in the models. Such an analysis will however require fresh frozen tissue and was for that reason not possible in our study. For patients with ITC in the sentinel node, we identified increasing tumor size, younger age at diagnosis and increasing proportion of positive sentinel nodes as risk factors for NSN metastases. Using these risk factors, we could separate patients into different risk groups. One third of the patients did not have any of the three risk factors. If ALND had been omitted in this group, NSN metastases would have been missed in only two of the 299 patients with ITC. This is, in our opinion acceptable, bearing in mind that a false negative rate on 5% is accepted for SLND in general. Furthermore, the results of a large clinical trial have recently shown that <1% of sentinel node negative patients will have regional recurrence after 8 years of follow-up, despite about 4% being false negative. 45 This indicates that not all lymph node metastasis will become clinically relevant. Accordingly, omission of ALND in patients with ITC and none of the three identified risk factors might only result in a minimal increase in axillary recurrences, and would not deteriorate the safety of the procedure. To date, this is the first model for the prediction of NSN metastases based on a population with only ITC in the sentinel node. However, the model needs to be validated in another dataset. If future studies, together with recent published data, 9,10 indicate that ALND may safely be omitted in sentinel node positive patients despite the risk of leaving NSN metastases in the axilla, a very low risk of NSN metastases will become irrelevant. On the contrary, a model will be needed to Epidemiology Int. J. Cancer: 131, (2012) VC 2012 UICC

75 2374 Risk of non-sentinel node metastases identify patients with high risk of relapse. 46 In contrast to earlier models predicting the risk of NSN metastases, 24,14 we also focused our models on patient at the high end of the risk scale. We identified a group of patients with micrometastases in the sentinel node with risk of NSN metastases at a level comparable to patients with macrometastases. 31,47,48 This group of patients, having at least four of the five identified risk factors, may still benefit from an ALND. In patients with ITC, only four patients had all risk factors present, hereof 75% with NSN metastases. However, the number of patients in this subgroup was too small to draw any conclusions. References Epidemiology 1. Gartner R, Jensen MB, Nielsen J, Ewertz M, Kroman N, Kehlet H. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA 2009;302: Gartner R, Jensen MB, Kronborg L, Ewertz M, Kehlet H, Kroman N. Self-reported armlymphedema and functional impairment after breast cancer treatment a nationwide study of prevalence and associated factors. Breast 2010;19: Veronesi U, Paganelli G, Viale G, Luini A, Zurrida S, Galimberti V, Intra M, Veronesi P, Robertson C, Maisonneuve P, Renne G, de Cicco C, et al. A randomized comparison of sentinelnode biopsy with routine axillary dissection in breast cancer. N Engl J Med 2003;349: Cserni G, Amendoeira I, Apostolikas N, Bellocq JP, Bianchi S, Bussolati G, Boecker W, Borisch B, Connolly CE, Decker T, Dervan P, Drijkoningen M, et al. Pathological work-up of sentinel lymph nodes in breast cancer. Review of current data to be considered for the formulation of guidelines. Eur J Cancer 2003;39: Tvedskov TF, Jensen MB, Balslev E, Ejlertsen B, Kroman N. Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: A nationwide study. Eur J Cancer 2011;47: Christiansen P, Friis E, Balslev E, Jensen D, Moller S. Sentinel node biopsy in breast cancer: five years experience from Denmark. Acta Oncol 2008;47: Cserni G, Gregori D, Merletti F, Sapino A, Mano MP, Ponti A, Sandrucci S, Baltas B, Bussolati G. Meta-analysis of non-sentinel node metastases associated with micrometastatic sentinel nodes in breast cancer. Br J Surg 2004;91: van Deurzen CH, de BM, Monninkhof EM, Bult P, van der WE, Tjan-Heijnen VC, van Diest PJ. Non-sentinel lymph node metastases associated with isolated breast cancer cells in the sentinel node. J Natl Cancer Inst 2008;100: Giuliano AE, McCall L, Beitsch P, Whitworth PW, Blumencranz P, Leitch AM, Saha S, Hunt KK, Morrow M, Ballman K. Locoregional recurrence after sentinel lymph node dissection with or without axillary dissection in patients with sentinel lymph node metastases: the American College of Surgeons Oncology Group Z0011 randomized trial. Ann Surg 2010;252: Giuliano AE, Hunt KK, Ballman KV, Beitsch PD, Whitworth PW, Blumencranz PW, Leitch AM, Saha S, McCall LM, Morrow M. Axillary dissection vs no axillary dissection in women with invasive breast cancer and sentinel node metastasis: a randomized clinical trial. JAMA 2011;305: Degnim AC, Griffith KA, Sabel MS, Hayes DF, Cimmino VM, Diehl KM, Lucas PC, Snyder ML, Chang AE, Newman LA. Clinicopathologic features of metastasis in nonsentinel lymph nodes of breast carcinoma patients. Cancer 2003;98: Van Zee KJ, Manasseh DM, Bevilacqua JL, Boolbol SK, Fey JV, Tan LK, Borgen PI, Cody HS, III, Kattan MW. A nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy. Ann Surg Oncol 2003;10: Coutant C, Olivier C, Lambaudie E, Fondrinier E, Marchal F, Guillemin F, Seince N, Thomas V, Leveque J, Barranger E, Darai E, Uzan S, et al. Comparison of models to predict nonsentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: a prospective multicenter study. J Clin Oncol 2009;27: Meretoja TJ, Strien L, Heikkila PS, Leidenius MH. A simple nomogram to evaluate the risk of nonsentinel node metastases in breast cancer patients with minimal sentinel node involvement. Ann Surg Oncol 2012;19: Cyr A, Gillanders WE, Aft RL, Eberlein TJ, Gao F, Margenthaler JA. Micrometastatic disease and isolated tumor cells as a predictor for additional breast cancer axillary metastatic burden. Ann Surg Oncol 2010;17: Gur AS, Unal B, Johnson R, Ahrendt G, Bonaventura M, Gordon P, Soran A. Predictive probability of four different breast cancer nomograms for nonsentinel axillary lymph node metastasis in positive sentinel node biopsy. JAm Coll Surg 2009;208: Rayhanabad J, Yegiyants S, Putchakayala K, Haig P, Romero L, DiFronzo LA. Axillary recurrence is low in patients with breast cancer who do not undergo completion axillary lymph node dissection for micrometastases in sentinel lymph nodes. Am Surg 2010;76: Alran S, De RY, Fourchotte V, Charitansky H, Laki F, Falcou MC, Benamor M, Freneaux P, Salmon RJ, Sigal-Zafrani B. Validation and limitations of use of a breast cancer nomogram predicting the likelihood of non-sentinel node involvement after positive sentinel node biopsy. Ann Surg Oncol 2007;14: Carcoforo P, Maestroni U, Querzoli P, Lanzara S, Maravegias K, Feggi L, Soliani G, Basaglia E. Primary breast cancer features can predict additional lymph node involvement in patients with sentinel node micrometastases. World J Surg 2006;30: Schrenk P, Konstantiniuk P, Wolfl S, Bogner S, Haid A, Nemes C, Jagoutz-Herzlinger M, Redtenbacher S. Prediction of non-sentinel lymph node status in breast cancer with a micrometastatic sentinel node. Br J Surg 2005;92: Houvenaeghel G, Nos C, Mignotte H, Classe JM, Giard S, Rouanet P, Lorca FP, Jacquemier J, Bardou VJ. Micrometastases in sentinel lymph node in a multicentric study: predictive factors of nonsentinel lymph node involvement Groupe des Chirurgiens de la Federation des Centres de Lutte Contre le Cancer. J Clin Oncol 2006;24: Kumar S, Bramlage M, Jacks LM, Goldberg JI, Patil SM, Giri DD, Van Zee KJ. Minimal disease in the sentinel lymph node: how to best measure sentinel node micrometastases to predict risk of additional non-sentinel lymph node disease. Ann Surg Oncol 2010;17: Gipponi M, Canavese G, Lionetto R, Catturich A, Vecchio C, Sapino A, Friedman D, Cafiero F. The role of axillary lymph node dissection in breast cancer patients with sentinel lymph node micrometastases. Eur J Surg Oncol 2006;32: Houvenaeghel G, Nos C, Giard S, Mignotte H, Esterni B, Jacquemier J, Buttarelli M, Classe JM, Cohen M, Rouanet P, Penault LF, Bonnier P, et al. A nomogram predictive of non-sentinel lymph node involvement in breast cancer patients with a sentinel lymph node micrometastasis. Eur J Surg Oncol 2009;35: Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB, Christiansen P, Mouridsen HT. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG): its 30-years experience and future promise. Acta Oncol 2008; 47: Friis E, Galatius H, Garne JP. Organized nationwide implementation of sentinel lymph node biopsy in Denmark. Acta Oncol 2008;47: Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI, Borgen PI, Clark G, Edge SB, Hayes DF, Hughes LL, Hutter RV, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 2002;20: Cserni G, Amendoeira I, Apostolikas N, Bellocq JP, Bianchi S, Boecker W, Borisch B, Connolly CE, Decker T, Dervan P, Drijkoningen M, Ellis IO, et al. Discrepancies in current practice of pathological evaluation of sentinel lymph nodes in breast cancer. Results of a questionnaire based survey by the European Working Group for Breast Screening Pathology. J Clin Pathol 2004;57: Pepels MJ, Vestjens JH, de BM, Smidt M, van Diest PJ, Borm GF, Tjan-Heijnen VC. Safety of avoiding routine use of axillary dissection in early stage breast cancer: a systematic review. Breast Cancer Res Treat 2011;125: Leidenius MH, Vironen JH, Riihela MS, Krogerus LA, Toivonen TS, von Smitten KA, Heikkila PS. The prevalence of non-sentinel node metastases in breast cancer patients with sentinel node micrometastases. Eur J Surg Oncol 2005;31: Menes TS, Tartter PI, Mizrachi H, Constantino J, Estabrook A, Smith SR. Breast cancer patients with pn0(iþ) and pn1(mi) sentinel nodes have Int. J. Cancer: 131, (2012) VC 2012 UICC

76 Tvedskov et al high rate of nonsentinel node metastases. JAm Coll Surg 2005;200: Truong PT, Vinh-Hung V, Cserni G, Woodward WA, Tai P, Vlastos G. The number of positive nodes and the ratio of positive to excised nodes are significant predictors of survival in women with micrometastatic node-positive breast cancer. Eur J Cancer 2008;44: Cserni G. Axillary sentinel lymph node micrometastases with extracapsular extension: a distinct pattern of breast cancer metastasis? J Clin Pathol 2008;61: Carvalho MJ, Dias MF, Silva TS, Custodio S, de Oliveira CF. Breast cancer patients with micrometastases in sentinel lymph nodes: differences considering additional metastatic lymph nodes. Eur J Gynaecol Oncol 2009;30: Estourgie SH, Nieweg OE, Olmos RA, Rutgers EJ, Kroon BB. Lymphatic drainage patterns from the breast. Ann Surg 2004;239: Kroman N, Wohlfahrt J, Mouridsen HT, Melbye M. Influence of tumor location on breast cancer prognosis. Int J Cancer 2003;105: Li J, Rudas M, Kemmner W, Warnick P, Fischer J, Gnant M, Schlag PM, Bembenek A. The location of small tumor deposits in the SLN predicts Non-SLN macrometastases in breast cancer patients. Eur J Surg Oncol 2008;34: Di TL, Arizzi C, Rahal D, Destro A, Roncalli M, Alloisio M, Orefice S, Rubino A, Morenghi E, Masci G, Santoro A, Del P, et al. Anatomic location of breast cancer micrometastasis in sentinel lymph node predicts axillary status. Ann Surg 2006;243: Viale G, Maiorano E, Mazzarol G, Zurrida S, Galimberti V, Luini A, Renne G, Pruneri G, Maisonneuve P, Veronesi U. Histologic detection and clinical implications of micrometastases in axillary sentinel lymph nodes for patients with breast carcinoma. Cancer 2001;92: Bendifallah S, Werkoff G, Borie-Moutafoff C, Antoine M, Chopier J, Gligorov J, Uzan S, Coutant C, Rouzier R. Multiple synchronous (multifocal and multicentric) breast cancer: clinical implications. Surg Oncol 2010;19:e Dowsett M, Nielsen TO, A Hern R, Bartlett J, Coombes RC, Cuzick J, Ellis M, Henry NL, Hugh JC, Lively T, McShane L, Paik S, et al. Assessment of Ki67 in breast cancer: recommendations from the International Ki67 in Breast Cancer working group. J Natl Cancer Inst 2011;103: Tvedskov TF, Bartels A., Jensen MB, Paachburg B, Kroman N, Balslev E, Brunner N. Evaluating TIMP-1, Ki67 and HER2 as markers for nonsentinel lymph node metastases in breast cancer patients with micrometastases to the sentinel lymph node. APMIS 2011;119: Kwon Y, Ro J, Kang HS, Kim SK, Hong EK, Khang SK, Gong G, Ro JY. Clinicopathological parameters and biological markers predicting non-sentinel node metastasis in sentinel nodepositive breast cancer patients. Oncol Rep 2011; 25: Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, Hastie T, Eisen MB, van de RM, Jeffrey SS, Thorsen T, Quist H, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98: Krag DN, Anderson SJ, Julian TB, Brown AM, Harlow SP, Costantino JP, Ashikaga T, Weaver DL, Mamounas EP, Jalovec LM, Frazier TG, Noyes RD, et al. Sentinel-lymph-node resection compared with conventional axillary-lymph-node dissection in clinically node-negative patients with breast cancer: overall survival findings from the NSABP B-32 randomised phase 3 trial. Lancet Oncol 2010;11: Galimberti V, Botteri E, Chifu C, Gentilini O, Luini A, Intra M, Baratella P, Sargenti M, Zurrida S, Veronesi P, Rotmensz N, Viale G, et al. Can we avoid axillary dissection in the micrometastatic sentinel node in breast cancer? Breast Cancer Res Treat 2012;131: Bolster MJ, Peer PG, Bult P, Thunnissen FB, Schapers RF, Meijer JW, Strobbe LJ, van Berlo CL, Klinkenbijl JH, Beex LV, Wobbes T, Tjan- Heijnen VC. Risk factors for non-sentinel lymph node metastases in patients with breast cancer. The outcome of a multi-institutional study. Ann Surg Oncol 2007;14: Noguchi M. Avoidance of axillary lymph node dissection in selected patients with node-positive breast cancer. Eur J Surg Oncol 2008;34: Epidemiology Int. J. Cancer: 131, (2012) VC 2012 UICC

77 Acta Oncologica ISSN: X (Print) X (Online) Journal homepage: Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node Tove F. Tvedskov, Maj-Britt Jensen, Eva Balslev & Niels Kroman To cite this article: Tove F. Tvedskov, Maj-Britt Jensen, Eva Balslev & Niels Kroman (2014) Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node, Acta Oncologica, 53:2, , DOI: / X To link to this article: Published online: 17 Jun Submit your article to this journal Article views: 202 View related articles View Crossmark data Citing articles: 3 View citing articles Full Terms & Conditions of access and use can be found at Download by: [Copenhagen University Library] Date: 24 February 2016, At: 05:02

78 Acta Oncologica, 2014; 53: ORIGINAL ARTICLE Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node TOVE F. TVEDSKOV 1, MAJ-BRITT JENSEN 2, EVA BALSLEV 3 & NIELS KROMAN 1 Downloaded by [Copenhagen University Library] at 05:02 24 February Department of Breast Surgery, Copenhagen University Hospital, Copenhagen, Denmark, 2 Danish Breast Cancer Cooperative Group, Copenhagen, Denmark and 3 Department of Pathology, Herlev Hospital, Herlev, Denmark Abstract Background. Benefit from axillary lymph node dissection in sentinel node positive breast cancer patients is under debate. Based on data from 1820 Danish breast cancer patients operated in , we have developed two models to predict high risk of non-sentinel node metastases when micrometastases or isolated tumor cells are found in sentinel node. The aim of this study was to validate these models in an independent Danish dataset. Material and methods. We included 720 breast cancer patients with micrometastases and 180 with isolated tumor cells in sentinel node operated in from the Danish Breast Cancer Cooperative Group database. Accuracy of the models was tested in this cohort by calculating area under the receiver operating characteristic curve (AUC) as well as sensitivity and specificity. Results. AUC for the model for patients with micrometastases was comparable to AUC in the original cohort: 0.63 and 0.64, respectively. The sensitivity and specificity for predicting risk of non-sentinel node metastases over 30% was 0.36 and 0.81, respectively, in the validation cohort. AUC for the model for patients with isolated tumor cells decreased from 0.73 in the original cohort to 0.60 in the validation cohort. When dividing patients with isolated tumor cells into high and low risk of non-sentinel node metastases according to number of risk factors present, 37% in the high-risk group had non-sentinel node metastases. Specificity and sensitivity was 0.48 and 0.88, respectively, in the validation cohort when using this cut-point. Conclusion. In this independent dataset, the model for patients with micrometastases was robust with accuracy similar to the original cohort, while the model for patients with isolated tumor cells was less accurate. The models may be used to identify patients where axillary lymph node dissection should still be considered. Today, sentinel lymph node dissection (SLND) has replaced axillary lymph node dissection (ALND) as standard procedure for staging of the axilla in clinically node negative breast cancer. SLND can accurately stage the axilla by removing on average only two lymph nodes. In case of metastatic spread to sentinel nodes an ALND is still recommended. Removing fewer nodes has made more extensive histopathological examinations of the lymph nodes possible and as a consequence more metastases are found, especially more micrometastases and isolated tumor cells (ITC) [1]. The overall benefit from ALND in patients with minor spread to axillary lymph nodes is now under debate [2]. Cohort studies have not been able to show an increased risk of axillary recurrence if ALND is omitted in patients with micrometastases or ITC in the sentinel node [3,4]. Furthermore, the newly published randomized IBCSG trial could not show a difference in disease-free survival after five years of follow-up, between patients with sentinel node micrometastases, with and without ALND [5]. This has, together with the results from the randomized trial from American College of Surgeons Oncology Group (ACOSOC) (Z0011), where no difference was found in axillary recurrence rate or survival between sentinel node positive patients with or without ALND [6,7], started a trend towards omitting ALND in breast cancer patients with minimal metastatic disease in the sentinel node. However, a minor risk of axillary recurrence still exists if ALND is completely omitted in patients with micrometastases or ITC in Correspondence: T. Filtenborg Tvedskov, Department of Breast Surgery, 4124, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark. Tel: Fax: tft@dadlnet.dk (Received 6 March 2013 ; accepted 15 May 2013 ) ISSN X print/issn X online 2014 Informa Healthcare DOI: / X

79 Downloaded by [Copenhagen University Library] at 05:02 24 February T. F. Tvedskov et al. the sentinel node [8]. To assure a safe omission of ALND a tool is needed to identify patients where ALND may still be offered in the future due to a high risk of recurrence. The presence of non-sentinel node metastases can be considered as a surrogate endpoint for axillary recurrence. Several models for predicting the risk of non-sentinel node metastases in patients with positive sentinel nodes have been developed [9,10], but only three of these models have been developed in a population of patients with micrometastases or ITC in the sentinel node [11,12]. Moreover, the existing models have been directed towards identifying patients with a low risk of non-sentinel node metastases. To identify patients with a potentially high risk of relapse, a model predicting patients with high risk of non-sentinel node metastases is needed. We have previously developed two models for predicting patients with high risk of non-sentinel node metastases when only micrometastases or ITC are found in the sentinel node, based on a large population-based data material [13]. The aim of this study was to validate these models in a large independent group of Danish breast cancer patients with micrometastases or ITC in the sentinel node. Material and methods In Denmark, all SLND are performed according to national guidelines described by the Danish Breast Cancer Cooperative Group (DBCG). A combination of radioactive tracer and blue dye is recommended, while lymphoscintigraphy is not used as a routine [14]. All removed sentinel nodes are examined with at least two step sections 500 μ m apart of the bivalved sentinel node. If no metastases are found by hematoxylineosin staining, immunohistochemical cytokeratin staining is performed. Lymph nodes removed by ALND are examined by bisectioning and hematoxylineosin staining. Since 2005, metastases have in Denmark been classified according to the 6th American Joint Committee on Cancer (AJCC) staging manual [15] in combination with cell count, where metastases between 0.2 and 2 mm or between 10 and 100 tumor cells are defined as micrometastases, and single cells or cell clusters less than 0.2 mm or less than 10 cells are defined as ITC [14]. Clinical and histopathological data on Danish women with breast cancer are prospectively collected and registered on standardized forms in a national database managed by the DBCG [16]. Based on data from this database we have developed two models to predict non-sentinel node metastases in breast cancer patients with micrometastases or ITC in the sentinel node. The models were based on 299 patients with ITC and 1521 patients with micrometastases in the sentinel node, operated in with SLND and a completion ALND (original cohort) [13]. In patients with ITC, from this original cohort, the risk of non-sentinel node metastases was significantly associated with younger age at diagnosis, increasing tumor size and increasing proportion of positive sentinel nodes in a multivariate analysis. In patients with micrometastases, from the original cohort, the risk of non-sentinel node metastases was significantly associated with increasing tumor size, lymphovascular invasion, negative hormone receptor status, location of tumor in the upper lateral quadrant of the breast and increasing proportion of positive sentinel nodes in a multivariate analysis. Based on these five risk factors a logistic regression model for predicting non-sentinel node metastases was developed for patients with micrometastases: Logit(p) * * 1 LVI * 1 HRneg * 1 UL * 1 proportion100% [where 1 1,2,3,4,5,6 (tumor size 10, 11 20, 21 30,31 40,41 50,51 mm )] A significant interaction between tumor size and lymphovascular invasion was found in the original cohort, with lymphovascular invasion being a stronger risk factor in larger tumors. However, this interaction did not change the accuracy of the model substantially and was subsequently not included in the model for validation analysis. In 2009 and 2010, a total number of 1072 breast cancer patients were registered in the DBCG database, with micrometastases or ITC in the sentinel node. A total of 147 patients did not have an ALND and five patients had less than seven lymph nodes removed by ALND. These 152 patients were excluded. The remaining 920 patients were eligible for the validation study. From the database we retrieved information on age at diagnosis, tumor size, hormone receptor status, number of removed sentinel nodes, number of positive sentinel nodes, lymphovascular invasion, location of tumor in the breast and presence of non-sentinel node metastases. Location of tumor in the breast was divided into location in upper lateral quadrant versus located in other quadrants, centrally or on the edge of the upper lateral quadrant. Data were validated, and missing information collected if possible, using the original pathology files. Nineteen patients with micrometastases and one with ITC in the sentinel node were excluded from validation analysis due to missing information on variables. The remaining 900 patients constituted the validation cohort.

80 Validation of model for NSN metastases 211 Statistical analyses The multivariate models of the original cohort were examined to evaluate risk factors for non-sentinel node metastases when ITC or micrometastases were found in the sentinel node. In the validation cohort associations between presence of non-sentinel node metastases and the risk factors listed in Table I were analyzed by χ 2 test and Fischer s exact test, excluding unknowns, for patients with ITC and patients with Table I. Patient, tumor and sentinel node characteristics according to risk of NSN metastases in 181 Danish breast cancer patients with ITC and 739 patients with micrometastases in the SN operated between 2009 and Isolated tumor cells NSN metastases Micrometastases NSN metastases Variables No Yes % p-value No Yes % p-value Downloaded by [Copenhagen University Library] at 05:02 24 February 2016 Total Age, years Tumor size, cm Unknown WHO type Ductal Lobular Other Unknown Grade Grade I Grade II Grade III Unknown/other LVI Present Absent Unknown Hormone receptor status Positive Negative Unknown HER2 status Positive Negative Unknown Location of tumor in breast Upper lateral Not upper lateral Unknown No. removed SN Positive SN/ Removed SN % %, 100% ; % ITC, isolated tumor cells; LVI, lymphovascular invasion; NSN, non-sentinel nodes; SN, sentinel node.

81 212 T. F. Tvedskov et al. Table II. Multivariate analyses of patients and tumor characteristics according to non-sentinel node metastases in Danish breast cancer patients with ITCs or micrometastases in the sentinel node. Original cohort Validation cohort Isolated tumor cells n 299 n 180 OR 95% CI p-value OR 95% CI p-value Tumor size, 2 vs. 2 cm Age at diagnosis, 40 vs. 40 years Proportion of pos SN, 100% vs. 100% Micrometastases n 1521 n 720 Downloaded by [Copenhagen University Library] at 05:02 24 February 2016 Tumor size, cm, Trend Proportion of pos SN, 100% vs. 100% Lymphovascular invasion HR status, neg vs. pos Location of tumor in upper lateral quadrant CI, confidence interval; HR, hormone receptor; neg, negative; OR, odds ratio; pos, positive; SN, sentinel node. micrometastases, respectively, in the sentinel node. Multivariate logistic regression models were applied to examine the influence of age at diagnosis, tumor size, as well as the proportion of positive sentinel nodes among removed sentinel nodes, and for patients with micrometastasis in the sentinel node also lymphovascular invasion, hormone receptor status and location of tumor in the breast, on the risk of non-sentinel node metastases. Adjusted odds ratio (OR) and 95% confidence intervals (CI) were calculated and the Wald test was used to test the significance of each variable. Discrimination of the models was assessed by area under the receiver operating characteristic curve (AUC) for both the original and the validation cohort. For ITC number of risk factors was used to define a cut-point. A score was assigned to each patient with micrometastases by adding the relevant β -coefficients from the multivariate logistic regression model of the original cohort. A cut-off value separating patients with more than 30% observed risk of non-sentinel node metastases in the original cohort was applied to both cohorts. Sensitivity and specificity were determined for these thresholds. SAS version 9.2 (SAS Institute, Cary, NC, USA) was used for all statistical analyses. The study was approved by the Danish Data Protection Agency (J.nr ). Results Of the 920 eligible patients in the validation cohort 739 had micrometastases and 181 had ITC in the sentinel node. Patient, tumor and sentinel node characteristics of patients in the validation cohort are shown in Table I. Characteristics of patients in the original cohort have been reported elsewhere [13]. The proportion of patients with non-sentinel node metastases was not significantly different in the validation and original cohorts; 13% (23/181) and 9% (28/304), respectively, for patients with ITC, and 17% (125/739) and 18% (283/1577), respectively, for patients with micrometastases. The identified risk factors for non-sentinel node metastases in the original cohort remained associated with an increased risk of further spread beyond the sentinel node in the validation cohort (Table II) but the association only remained statistical significant for tumor size and lymphovascular invasion. In the validation cohort, no patients with ITC under 40 years at diagnosis had non-sentinel node metastases. In the original cohort, 17% of patients with ITC had more than one risk factor present. The risk of further spread beyond sentinel node in this group was 23%, compared to 6% in patients with only one or none of the risk factors present. In the validation cohort, a similar proportion (17%) had more than one risk factor present. Thirty-seven percent of these patients had non-sentinel node metastases, compared to only 8% of patients with only one or none of the risk factors present. Dividing patients into high- and low-risk groups for having non-sentinel node metastases with a cut-point of more than one risk factor present the sensitivity and specificity was 0.48 and 0.88, respectively, in the validation cohort compared to 0.43 and 0.85, respectively, in the original cohort. AUC for a model based on the three risk factors was 0.73 (95% CI ) in the original cohort, but

82 Validation of model for NSN metastases 213 Table III. Performance of models predicting non-sentinel node metastases in original and validation cohorts of Danish breast cancer patients with ITCs or micrometastases in the sentinel node. Model Cohort Operation year No. of patients Sensitivity Specificity AUC (95%CI) ITC Original cohort * 0.85 * 0.73 ( ) Validation cohort * 0.88 * 0.60 ( ) MIC Original cohort ( ) Validation cohort ( ) ITC, isolated tumor cells; MIC, micrometastases. * For cut-point of more than one risk factor present; For cut-point of 30% observed risk of non-sentinel node metastases. Downloaded by [Copenhagen University Library] at 05:02 24 February 2016 decreased to 0.60 (95% CI ) in the validation cohort (Table III). When a score was assigned to each patient with micrometastases in the original cohort by adding the relevant β -coefficients from the multivariate logistic regression model, a cut-off value on could separate patients in the original cohort into a group with less than 30% risk of non-sentinel node metastases and a high-risk group with at least 30% risk of further spread. This high-risk group represented 14% of patients in the original cohort. In the validation cohort, 22% of patients had a risk score above 1.128, and 28% of these high-risk patients had non-sentinel node metastases. Using this cutpoint the sensitivity and specificity was 0.36 and 0.81, respectively, in the validation cohort compared to 0.25 and 0.88, respectively, in the original cohort (Table III). The AUC for the model for patients with micrometastases changed only slightly from 0.64 (95% CI ) in the original cohort to 0.63 (95% CI ) in the validation cohort. Discussion We have previously developed two models for predicting breast cancer patients with high risk of nonsentinel node metastases when only minimal metastatic disease is found in the sentinel node. The models are now validated in a large independent Danish data material. Especially the model for patients with micrometastases showed to be robust with only a slight change in accuracy in the new data material. By using the DBCG database, the models could be based on a data material of a unique size of more than 1800 patients. In addition, the database allowed us to validate the models in an independent data material of 900 patients. This comprehensive and nationwide data material made the models very robust when validated. All risk factors for non-sentinel node metastases, identified in the original cohort, remained associated with an increased risk of non-sentinel node metastases in the validation cohort, although several of the factors were no longer statistical significant. This can partly be explained by the lower number of patients in the validation cohort on 900 patients compared to 1820 patients in the original cohort. Despite being a robust model, AUC was only 0.63 for the model for patients with micrometastases, and AUC for the model for patients with ITC decreased to 0.60 when validated. The possibilities of increasing the accuracy of the models are limited. It is possible that the search for new biochemical markers will reveal new clinically significant predictors for non-sentinel node metastases that can improve the models [17,18]. Only two earlier studies have tried to construct a predictive model based on a population of patients with micrometastases or ITC [11,12]. Like us, these studies included large sample sizes of 484 and 909 patients. Nevertheless, AUC was only 0.68 and 0.66, respectively, for the two models. Validation of the first model resulted in an AUC on 0.79, but only 51 patients were included in the validation series [12]. The second model has been validated in a series of 484 patients but the AUC was not reported [19]. These earlier models have focused on predicting a group of patients with low risk of non-sentinel node metastases, where ALND could be omitted without compromising locoregional disease control. Locoregional disease control is however not only based on surgical treatment but on adjuvant treatment as well. The criteria for offering adjuvant systemic treatment have changed over years [20] and there is a trend towards inclusion of several new high-risk criteria in the decision for adjuvant systemic treatment. This has led to a larger proportion of patients offered systemic treatment. It is possible that systemic treatment together with whole breast irradiation can eliminate low volume axillary metastases. This could explain the increasing evidence indicating that ALND can safely be omitted in sentinel node positive patients despite the risk of leaving non-sentinel node metastases in the axilla [3 6]. In cohort studies based on data from the National Cancer institute s surveillance Epidemiology, and

83 Downloaded by [Copenhagen University Library] at 05:02 24 February T. F. Tvedskov et al. End Results database [4] and the American National Cancer Data Base [3] there was no difference in outcome between patients with and without ALND when only micrometastases were found in the sentinel node. Results from only two randomized trials exist. In the IBCSG trial no significant difference in disease-free survival was found between patients with and without ALND after five years of follow-up, when only micrometastases was found in the sentinel node [5]. Likewise, the ACOSOC Z0011 trial found no significant difference in loco-regional recurrence and overall survival between sentinel node positive patients with and without ALND [6,7]. In the light of these studies, the prediction of a very low risk of non-sentinel node metastases is irrelevant. Still, a small group of sentinel node positive patients will experience an axillary recurrence despite having only minimal metastatic disease in the sentinel node, and these patients might benefit from an ALND. In a recent Dutch cohort study, including more than 2600 patients, the adjusted hazard ratio for regional recurrence was 4.39 if ALND was omitted in patients with micrometastases in the sentinel node. A similar but insignificant trend was seen for patients with ITC [21]. Therefore a predictive model may be needed to identify patients where adjuvant treatment might not be sufficient to eliminate residual axillary metastases resulting in a high risk of relapse. High risk of nonsentinel node metastases can be considered as a surrogate endpoint for axillary recurrence. In contrast to earlier models predicting the risk of non-sentinel node metastases [11,12] we focused our models on patient at the high end of the risk scale. In patients with micrometastases, we choose a cut-point that identified patients with more than 30% risk of nonsentinel node metastases in the original cohort, which is at a level comparable to patients with macrometastases [22]. This group of patients may still benefit from an ALND. For patients with ITC the cut-point was guided by the number of risk factors present. In the original cohort, only 23% of patients with more than one risk factor present had non-sentinel node metastases, which could argue for an even higher cutpoint on more than two risk factors present. Still, 37% of patients in the high-risk group of the validation cohort had non-sentinel node metastases when a cut-point of more than one risk factor was used, indicating that this is a reasonable cut-point. The two validated models seem to be suitable for a Danish population of breast cancer patients. However, when models are tested in a foreign population, they generally tend to work poor and a high variation in model performance between different centers has been shown [12]. For example, the Tenon score was developed in a French population and worked well in the French validation study [10], but did not perform very well in a Swedish population [23]. Likewise, the model from the Memorial Sloan- Kettering Cancer Center [9] worked well in other American populations [24], but was not very precise in a Hungarian population [25]. Finally, only two of 12 tested models worked well in a Chinese population [26]. Accordingly, validation of the models on an external data material outside Denmark is necessary to ensure international validity. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. The study was supported by IMK common foundation, HOC research initiative foundation, Danish Cancer Research Foundation and A.P. M ø ller Foundation for the Advancement of Medical Science. References [1] Tvedskov TF, Jensen MB, Balslev E, Ejlertsen B, Kroman N. Stage migration after introduction of sentinel lymph node dissection in breast cancer treatment in Denmark: A nationwide study. Eur J Cancer 2011 ; 47 : [2] Croshaw RL, Erb KM, Shapiro-Wright HM, Julian TB. The need for axillary dissection in patients with positive axillary sentinel lymph nodes. Curr Oncol Rep 2011 ; 13 : [3] Bilimoria KY, Bentrem DJ, Hansen NM, Bethke KP, Rademaker AW, Ko CY, et al. Comparison of sentinel lymph node biopsy alone and completion axillary lymph node dissection for node-positive breast cancer. J Clin Oncol 2009 ; 27 : [4] Yi M, Giordano SH, Meric-Bernstam F, Mittendorf EA, Kuerer HM, Hwang RF, et al. Trends in and outcomes from sentinel lymph node biopsy (SLNB) alone vs. SLNB with axillary lymph node dissection for node-positive breast cancer patients: Experience from the SEER database. Ann Surg Oncol 2010 ; 17 : [5] Galimberti V, Cole BF, Zurrida S, Viale G, Luini A, Veronesi P, et al. Axillary dissection versus no axillary dissection in patients with sentinel-node micrometastases (IBCSG 23-01): A phase 3 randomised controlled trial. Lancet Oncol 2013 ; 14 : [6] Giuliano AE, Hunt KK, Ballman KV, Beitsch PD, Whitworth PW, Blumencranz PW, 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 : [7] Giuliano AE, McCall L, Beitsch P, Whitworth PW, Blumencranz P, Leitch AM, et al. Locoregional recurrence after sentinel lymph node dissection with or without axillary dissection in patients with sentinel lymph node metastases: The American College of Surgeons Oncology Group Z0011 randomized trial. Ann Surg 2010 ; 252 : [8] Pepels MJ, Vestjens JH, de BM, Smidt M, van Diest PJ, Borm GF, et al. Safety of avoiding routine use of axillary dissection in early stage breast cancer: A systematic review. Breast Cancer Res Treat 2011 ; 125 : [9] Van Zee KJ, Manasseh DM, Bevilacqua JL, Boolbol SK, Fey JV, Tan LK, et al. A nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy. Ann Surg Oncol 2003 ; 10 :

84 Downloaded by [Copenhagen University Library] at 05:02 24 February 2016 [10] Coutant C, Olivier C, Lambaudie E, Fondrinier E, Marchal F, Guillemin F, et al. Comparison of models to predict nonsentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: A prospective multicenter study. J Clin Oncol 2009 ; 27 : [11] Houvenaeghel G, Nos C, Giard S, Mignotte H, Esterni B, Jacquemier J, et al. A nomogram predictive of non-sentinel lymph node involvement in breast cancer patients with a sentinel lymph node micrometastasis. Eur J Surg Oncol 2009 ; 35 : [12] Meretoja TJ, Strien L, Heikkila PS, Leidenius MH. A simple nomogram to evaluate the risk of nonsentinel node metastases in breast cancer patients with minimal sentinel node involvement. Ann Surg Oncol 2012 ; 19 : [13] Tvedskov TF, Jensen MB, Lisse IM, Ejlertsen B, Balslev E, Kroman N. High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the sentinel node. Int J Cancer 2012 ; 131 : [14] Christiansen P, Friis E, Balslev E, Jensen D, Moller S. Sentinel node biopsy in breast cancer: Five years experience from Denmark. Acta Oncol 2008 ; 47 : [15] Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 2002 ; 20 : [16] Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008 ; 47 : [17] Kwon Y, Ro J, Kang HS, Kim SK, Hong EK, Khang SK, et al. Clinicopathological parameters and biological markers predicting non-sentinel node metastasis in sentinel nodepositive breast cancer patients. Oncol Rep 2011 ; 25 : [18] Tvedskov TF, Bartels A, Jensen MB, Paaschburg B, Kroman N, Balslev E, et al. Evaluating TIMP-1, Ki67, and HER2 as markers for non-sentinel node metastases in breast Validation of model for NSN metastases 215 cancer patients with micrometastases to the sentinel node. APMIS 2011 ; 119 : [19] Houvenaeghel G, Bannier M, Nos C, Giard S, Mignotte H, Jacquemier J, et al. Non sentinel node involvement prediction for sentinel node micrometastases in breast cancer: Nomogram validation and comparison with other models. Breast 2012 ; 21 : [20] Goldhirsch A, Wood WC, Coates AS, Gelber RD, Thurlimann B, Senn HJ. 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 : [21] Pepels MJ, de BM, Bult P, van Dijck JA, van Deurzen CH, Menke-Pluymers MB, et al. Regional recurrence in breast cancer patients with sentinel node micrometastases and isolated tumor cells. Ann Surg 2012 ; 255 : [22] Noguchi M. Avoidance of axillary lymph node dissection in selected patients with node-positive breast cancer. Eur J Surg Oncol 2008 ; 34 : [23] Andersson Y, Frisell J, de BJ, Bergkvist L. Prediction of nonsentinel lymph node status in breast cancer patients with sentinel lymph node metastases: Evaluation of the tenon score. Breast Cancer (Auckl.) 2012 ; 6 : [24] Hessman CJ, Naik AM, Kearney NM, Jensen AJ, Diggs BS, Troxell ML, et al. Comparative validation of online nomograms for predicting nonsentinel lymph node status in sentinel lymph node-positive breast cancer. Arch Surg 2011 ; 146 : [25] Kocsis L, Svebis M, Boross G, Sinko M, Maraz R, Rajtar M, et al. Use and limitations of a nomogram predicting the likelihood of non-sentinel node involvement after a positive sentinel node biopsy in breast cancer patients. Am Surg 2004 ; 70 : [26] Chen K, Zhu L, Jia W, Rao N, Fan M, Huang H, et al. Validation and comparison of models to predict non-sentinel lymph node metastasis in breast cancer patients. Cancer Sci 2011 ; 103 :

85 Available online at ScienceDirect EJSO 40 (2014) 435e441 Cross-validation of three predictive tools for non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node T.F. Tvedskov a, *, T.J. Meretoja b, M.B. Jensen c, M. Leidenius b, N. Kroman a a Department of Breast Surgery, Copenhagen University Hospital, Afsnit 4124, Blegdamsvej 9, 2100 Copenhagen, Denmark b Breast Surgery Unit, Helsinki University Central Hospital, P.O. Box 140, HUS, Helsinki, Finland c Danish Breast Cancer Cooperative Group, Copenhagen University Hospital, Afsnit 2501, Blegdamsvej 9, Copenhagen, Denmark Accepted 23 January 2014 Available online 1 February 2014 Abstract Background: We cross-validated three existing models for the prediction of non-sentinel node metastases in patients with micrometastases or isolated tumor cells (ITC) in the sentinel node, developed in Danish and Finnish cohorts of breast cancer patients, to find the best model to identify patients who might benefit from further axillary treatment. Material and method: Based on 484 Finnish breast cancer patients with micrometastases or ITC in sentinel node a model has been developed for the prediction of non-sentinel node metastases. Likewise, two separate models have been developed in 1577 Danish patients with micrometastases and 304 Danish patients with ITC, respectively. The models were cross-validated in the opposite cohort. Results: The Danish model for micrometatases was accurate when tested in the Finnish cohort, with a slight change in AUC from 0.64 to The AUC of the Finnish model decreased from 0.68 to 0.58 when tested in the Danish cohort, and the AUC of the Danish model for ITC decreased from 0.73 to 0.52, when tested in the Finnish cohort. The Danish micrometastatic model identified 14e22% of the patients as high-risk patients with over 30% risk of non-sentinel node metastases while less than 1% was identified by the Finish model. In contrast, the Finish model predicted a much larger proportion of patients being in the low-risk group with less than 10% risk of non-sentinel node metastases. Conclusion: The Danish model for micrometastases worked well in predicting high risk of non-sentinel node metastases and was accurate under external validation. Ó 2014 Elsevier Ltd. All rights reserved. Keywords: Breast cancer; Sentinel node; Micrometastases; Isolated tumor cells; Non-sentinel node metastases Introduction The majority of patients with micrometastases or isolated tumor cells (ITC) in the sentinel node have no further spread beyond the sentinel node and will not benefit from an axillary lymph node dissection (ALND). 1,2 In addition, the recent International Breast Cancer Study Group (IBCSG) trial, where breast cancer patients with micrometastases or ITC in the sentinel node were randomized to either ALND or no * Corresponding author. Fax: þ address: tft@dadlnet.dk (T.F. Tvedskov). treatment of the axilla, did not find any difference in disease-free survival between patients, with and without ALND. 3 Accordingly, it is possible that adjuvant systemic treatment together with whole breast irradiation can eliminate low volume axillary metastases left in the axilla in these patients, making an ALND redundant. As a result, ALND is now generally abandoned in patients with only micrometastases or ITC in the sentinel node. Still, studies indicate that a small group of patients with micrometastases exists, with a high risk of non-sentinel node (NSN) metastases, 4,5 and some of these patients might experience an axillary recurrence if further axillary treatment is omitted. This is /$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.

86 436 T.F. Tvedskov et al. / EJSO 40 (2014) 435e441 underlined by the results from a recent Dutch study reporting a four-fold increased risk of axillary recurrence in patients with micrometastases in the sentinel node, if the axilla was left untreated. 4 Therefore a tool is needed to identify patients, in whom adjuvant treatment might not be sufficient to eliminate residual axillary metastases, resulting in a high risk of relapse. Such a tool would create a tailor-made treatment of the axilla, and spare the majority from an unnecessary ALND, but still prevent an axillary recurrence in high-risk patients. Several tools have been developed, 6,7 and validated 7 for patients with macrometastases in the sentinel node, but these tools are not very precise in patients with micrometastases. 8e13 Furthermore, they have a large degree of inter-institutional variation, 13e18 which hampers their direct implementation into clinical use in a new population. Only a few predictive tools have been developed based on patients with only micrometastases or ITC in the sentinel node, 5,12,19 and external validation of these models is sparse. 13,20 The aim of the present study was to cross-validate three existing models for predicting NSN metastases, developed in two separate populations of breast cancer patients with micrometastases or ITC in the sentinel node. We compare the performance of the models and evaluate their robustness for inter-institutional use. Material and method Between 2003 and 2011, 302 Finnish breast cancer patients with micrometastases and 235 with ITC in the sentinel node, have been operated at the Breast surgery unit of Helsinki University Central Hospital. 24 patients with micrometastases and 29 patients with ITC did not undergo a completion ALND. Based on the remaining 484 patients a model has been developed for the prediction of NSN metastases. The model includes tumor size and multifocality as risk factors. 12 The model was internally validated in a separate series of 51 Finnish breast cancer patients with micrometastases or ITC in the sentinel node. In Denmark, 2137 breast cancer patients with either micrometastases or ITC in the sentinel node were operated between 2002 and 2008 in eighteen different Danish breast surgery departments. Patients were registered prospectively in the Danish Breast Cancer Cooperative Group (DBCG) Database. The DBCG database is a national breast cancer database and has been described in details elsewhere patients, 147 with micrometastases and 109 with ITC, did not undergo a completion ALND. Based on the remaining 1881 patients, 1577 with micrometastases and 304 with ITC, two models were developed patients were excluded due to missing information on variables included in the final models. The final DBCG model for prediction of NSN metastases in patients with micrometastases in the sentinel node was based on 1521 patients and included tumor size, lymphovascular invasion, hormone receptor status, location of tumor in the breast and proportion of positive sentinel nodes as risk factors. The final model for prediction of NSN metastases in patients with ITC was based on 299 patients and included tumor size, young age (<40) and proportion of positive sentinel nodes as risk factors. The DBCG model for patients with micrometastases has been validated in an independent cohort of 720 Danish breast cancer patients with micrometastases, operated in 2009 and 2010 and prospectively registered in the DBCG database. 22 Likewise, the DBCG model for patients with ITC has been validated in 180 Danish patients with ITC, operated in 2009 and Further details on development and internal validation of the three models have been described elsewhere. 5,12 Micrometastases were, in all cohorts, defined as tumor deposits not larger than 2 mm, and ITC were defined as tumor deposits not larger than 0.2 mm. Additionally, cell count was used to classify metastases in the Danish cohorts. Metastases between 10 and 100 tumor cells were defined as micrometastases, and single cells or cell clusters of less than 10 cells were defined as ITC. 23 Histopathological examination of the sentinel nodes has been described elsewhere. 5,12 Multifocality was defined differently in the two cohorts. In the Danish cohort, multifocality was defined as more than one invasive carcinoma placed more than 2 cm apart. Invasive carcinomas within 2 cm were defined as satellite tumors and not included as multifocal carcinomas. In contrast, the Finnish cohort defined any cancer with more than one invasive focus as multifocal, regardless of distance between the tumors. Preoperative axillary ultrasound was performed in all patients and fine needle aspiration was performed in case of suspicious lymph nodes. A combination of radioactive tracer and blue dye was used to identify sentinel nodes. In the Finnish cohort lymphoscintigraphy was used as a routine, while it was optional in the Danish cohort. 23 Radioactive or stained lymph nodes were removed as sentinel nodes together with any lymph nodes considered suspicious by palpation or inspection. The cohort for external validation of the Helsinki model consisted of the original cohort of Danish breast cancer patients with micrometastases or ITC in the sentinel node. 43 patients with micrometastases and 7 patients with ITC were excluded due to missing information on either tumor size or focality, leaving 1534 patients for external validation. The cohort used for external validation of the DBCG model for patients with micrometastases consisted of the 278 Finnish patients with micrometastases. The cohort used for external validation of the DBCG model for patients with ITC consisted of the 206 Finnish patients with ITC. Patient, tumor and sentinel node characteristics of the Danish and Finnish cohorts are shown in Table 1. Statistical analysis Patient and disease characteristics for the original Danish and Finnish cohorts listed in Table 1 were analyzed by c 2 test, excluding unknowns. For the multivariate models developed from the original cohorts, discrimination

87 T.F. Tvedskov et al. / EJSO 40 (2014) 435e Table 1 Patient, tumor and lymph node characteristics of a Danish and Finnish cohort of breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Variables Danish cohort Finnish cohort P-value No. % No. % Total Age, years 0.05 < e e e Tumor size, cm >1 e >2 e > Unknown WHO type < Ductal Lobular Other Unknown Grade < Grade I Grade II Grade III Unknown LVI < Present Absent Unknown Hormone receptor status 0.09 Positive Negative Unknown Location of tumor in breast < Upper lateral Not upper lateral Unknown Focality < Multifocal Unifocal Unknown No. removed SN < Unknown SN metastases < Micrometastases Isolated tumor cells Positive SN/Removed SN < % >33%, <100% >25; % Unknown NSN metastases < Yes No Abbreviations: LVI: lymphovascular invasion, SN: sentinel node, NSN: non-sentinel node. of the models was assessed by area under the receiver operating characteristic curve (AUC) for the original, the internal and the external validation cohort. For the DBCG model for ITC, the number of risk factors was used to define a cutpoint. For the Helsinki model and the DBCG model for micrometastases, a score was assigned to each patient by adding the relevant b-coefficients from the multivariate logistic regression model of the original cohorts. In the original studies on development of the models, cut points separating patients with high and low risk of NSN metastases were set differently in the two cohorts. 12,26 To make a uniform comparison under external validation a cut point on 30% risk of NSN metastases was chosen for the DBCG micrometastatic model and for the Helsinki model, for identifying high-risk patients, and a cut point on 10% was chosen to identify low-risk patients. In the simpler DBCG model for patients with ITC it was not possible to choose cut points on exact 30% and 10% risk of NSN metastases. A cut point on two or more risk factors present was chosen for identifying high-risk patients, because the risk in this group was closest to 30%, and a cut point on no risk factors present was chosen to identify low-risk patients. Sensitivity and specificity were determined for these cut points. SAS version 9.2 (SAS Institute, Cary, NC) was used for all statistical analyses. Results The risk of NSN metastases varied from 6% to 18% in the different cohorts, being highest in the cohorts containing only patients with micrometastases (Table 2). The risk of NSN metastases was significantly lower in the Finnish cohort compared to the Danish cohort (P < ) where a larger proportion of patients had micrometastases in the sentinel node (P < )(Table 1). Patients in the Finnish cohort had smaller tumor size (P ¼ ), higher tumor grade (P < ), more often non-ductal carcinomas (P < ), lymphovascular invasion (P < ) and tumors in the upper lateral quadrant of the breast (P < ), and more patients were registered with multifocal tumors (P < ) (Table 1). In addition, more sentinel nodes were removed compared to the Danish cohort (P < ) (Table 1). AUC as well as sensitivity and specificity for cut points were calculated for the three models, to test the performance of the models in the external validation cohort. Data are shown in Table 2, together with data for performance in the original cohort and internal validation studies. The DBCG model for patients with ITC was the best performing model in the original cohort, followed by the Helsinki model, which performed even better under internal validation. When validated in the external cohort both of these models were less accurate with an AUC of 0.52 and 0.58, respectively. The DBCG model for patients with micrometastases was the best performing model in the external validation series. The AUC changed only slightly between the original, the internal and the external

88 438 T.F. Tvedskov et al. / EJSO 40 (2014) 435e441 Table 2 Performance of models for predicting low and high risk of non-sentinel node metastases in original, internal validation and external validation cohorts. Model Cohort No. No. with NSN metastases (%) Sensitivity (low-risk) a Specificity (low-risk) a Low-risk patients (%) a Sensitivity (High-risk) b Specificity (High-risk) b High-risk patients (%) b AUC (95% CI) Helsinki model Original (7.4) (84.1) (0.4) 0.68 (0.59e0.77) Internal 51 5 (9.8) (70.6) (0.64e0.95) validation External (16.6) (91.0) (0.2) 0.58 (0.55e0.62) validation DBCG model Original (17.9) (4.3) (14.4) 0.64 (0.60e0.67) (MIC) Internal (16.8) (4.3) (22.0) 0.63 (0.57e0.68) validation External (8.6) (5.0) (18.7) 0.63 (0.49; 0.76) validation DBCG model Original (9.4) (32.0) (17.0) 0.73 (0.64e0.82) (ITC) Internal (12.8) (39.0) (17.0) 0.60 (0.46e0.75) validation External validation (5.8) (49.5) (9.7) 0.52 (0.32; 0.71) Abbreviations: DBCG: Danish Breast Cancer Cooperative Group, CI: Confidence interval, AUC: area under the curve, MIC: micrometastases, ITC: isolated tumor cells. a Cut point of 10% risk of non-sentinel node metastases or (for DBCG ITC model) no risk factors present. b Cut point of 30% predicted risk of non-sentinel node metastases or (for DBCG ITC model) more than one risk factor present. validation cohorts. When using a cut point at 10% risk of NSN metastases, the Helsinki model identified 71e91% of the patients in the studied cohorts being in the low-risk group, while only 4e5% were identified as low-risk patients in the DBCG micrometastatic model. In contrast, the DBCG model identified 14e22% of the patients being at high risk of NSN metastases using a cut point at 30%, while less than 1% of the patients were identified as high-risk patients in the Helsinki model. The DBCG model for patients with ITC identified between 32 and 50% of the patients from the different cohorts with low risk of NSN metastases (none of the risk factors present), while between 10 and 17% of the patients were defined as high-risk patients with more than one risk factor present. Numbers of high- and low-risk patients as well as sensitivity and specificity for different cut points in the cohorts are shown in Table 2. Multifocality was defined differently in Finnish and Danish cohorts. In the DBCG database, 232 patients from the Danish cohort were registered as having unifocal tumors with satellite tumors, corresponding to the Finnish definition of multifocality. If these patients were included in the group of multifocal tumors, the proportion of patients with multifocality increased from 6% to 19% in the Danish cohort, which is close to the proportion of 17% in the Finnish cohort. When using this definition of multifocality, the sensitivity of the Helsinki model changed from 0.11 to 0.23 under external validation, and the specificity changed from 0.91 to 0.82, identifying low-risk patients. The AUC was basically unchanged at 0.58 (95% CI: 0.54e0.61). Discussion In this study we have tested three different models for predicting NSN metastases in breast cancer patients with micrometastases or ITC in the sentinel node. Largely different set of risk factors for NSN metastases were identified in the three models. The only risk factor that was identified in all three models was tumor size. This could indicate that tumor size is a strong risk factor for NSN metastases or could be explained by measurement of tumor size being consistent and to a minor degree influenced by possible inter-institutional variations in pathology methods. Due to the differences in identified risk factors, it could be hypothesized that models would only work well in the cohort used for their development. However, the accuracy of the DBCG model for patients with micrometastases did not change under neither internal nor external validation, with a stable AUC around In contrast, the Helsinki model was the most promising model in the original and internal validation cohort, but showed to be less accurate when validated in an external cohort. The same was seen for the DBCG model for patients with ITC. There could be various explanations for the poor performance of the Helsinki model and the DBCG model for ITC under external validation. Demographic differences exist between the cohorts, apparent in the significantly different baseline characteristics (Table 1). Both of the cohorts were population based, but while the Helsinki cohort included patients from only one highly specialized center, the DBCG cohort was nation-wide and included patients from eighteen different breast units. Furthermore, differences in diagnosis, surgical techniques and pathology methods may exist between cohorts. First, multifocality was defined differently in the two cohorts. Though, unifying the definition, by including satellite tumors in the group of multifocal tumors in the Danish cohort, did not improve the performance of the Helsinki model.

89 T.F. Tvedskov et al. / EJSO 40 (2014) 435e Second, even though the sentinel node procedure is standardized, the propensity to harvest additional lymph nodes together with the sentinel nodes, due to suspicious palpation or inspection, may vary. This may explain the significantly larger number of sentinel nodes removed in the Helsinki cohort compared to the Danish cohort. A higher number of harvested sentinel nodes and suspicious nodes may result in a lower proportion of patients with NSN metastases by completion ALND. 24 In addition, differences in screening policies might have affected the risk of NSN metastases. Screening has been performed in Helsinki since 1991 but in Denmark it was not nation-wide until Earlier studies have found a decreased risk of NSN metastases in patients with screenings detected breast cancers and micrometastases in the sentinel node. 25 Furthermore, the proportion of patients with NSN metastases could be influenced by the sensitivity of preoperative ultrasound, which may vary between centers. In example, the rate of NSN metastases in patients with micrometastases varied from 12 to 29% between the different centers in Denmark, and the rate of NSN metastases in patients with ITC varied from 0 to 30%. The difference in proportions of NSN metastases may partly account for the varying performance of the models. Indeed, we have previously shown that including the prevalence of NSN metastases in a predictive model may increase its performance and calibration. 18 Though, this is only possible if the model is developed from various populations. Another explanation for the difference in proportion of NSN metastases could be the slightly different definition of micrometastases and ITC in the two cohorts. It should be noted that the definition of micrometastases and ITC was changed in the 7th edition of the American Joint Committee on Cancer (AJCC) staging manual. This new definition was not yet implemented in the study period. This should be kept in mind when using the models in patients staged according to 7th edition of the AJCC staging manual. Third, in the study period a different practice for omitting ALND in patients with only minimal metastatic disease in the sentinel node may have existed. In Denmark around 9% of patients with micrometastases and 26% of patients with ITC did not undergo a completion ALND and were excluded from the patient groups used for model development. In Helsinki the proportion of patients not treated by ALND was 8% for patients with micrometastases and 12% for patients with ITC. Selection bias caused by the different proportion of patients with ITC not treated by ALND cannot be excluded. Finally, the performance may reflect the sample size for model development. The DBCG model for patients with micrometastases was developed in a very large population-based cohort containing 1521 patients compared to only 484 and 299 patients used for development of the two other models. This makes the model less sensitive to minor local differences. Despite being less accurate in the original setting, the cross-validation shows that this model may perform well in different populations. It must be emphasized that all predictive models, especially ones with small patient numbers, need to be validated in each center before clinical implementation. The three models performed differently in identifying high- and low-risk patients, according to the chosen cut points. The Helsinki model identified a large proportion of patients with low risk of NSN metastases while the DBCG micrometastatic model identified only 4e5% as low-risk patients. In contrast, the DBCG model identified 14e22% of patients being at high risk of NSN metastases, while the Helsinki model predicted only very few patients being at high risk. The majority of earlier model for predicting NSN metastases has aimed at identifying patients where ALND could safely be omitted because of a negligible risk of leaving metastases behind in the axilla. 7 It is now evident, that not all metastases left in the axilla will become clinically relevant. 27 As a consequence, ALND is generally no longer recommended in patients with micrometastases or ITC in the sentinel node, where the risk of spread beyond the sentinel node is low 1,2. Nevertheless, studies indicate that a small group of patients with micrometastases have a high risk of NSN metastases 5 resulting in an increased risk of axillary recurrence if axillary treatment is omitted. These patients might benefit from an ALND or axillary radiotherapy. 4 Accordingly, models are needed to identify these high-risk patients, to create a tailor-made treatment of the axilla. In models for predicting NSN metastases in patients with micrometastases or ITC in the sentinel node, a high cut point should be chosen, to identify patients with a high risk of further spread beyond the sentinel node. The Helsinki model has, together with two French micrometastatic models, earlier been validated in 5 different centers, including a Helsinki-based center, for predicting high risk of NSN metastases in patients with micrometastases in the sentinel node. 13 In that study, a cut point at 20% risk of NSN metastases was chosen. A total number of 313 patients were included in the validation series, resulting in between 34 and 74 patients from each center. The AUC for the three models varied substantially between centers. This variation could be explained by the relatively small number of patients from each center. The overall AUC for the whole data material was only 0.55e indicating that neither the French nor the Helsinki model seem to work well in predicting high risk of NSN metastases when only micrometastases are found in the sentinel node. In conclusion, the DBCG model for patients with micrometastases seems to be the best model for interinstitutional use, when predicting patients with high risk of NSN metastases, when micrometastases are found in the sentinel node. The accuracy of the model did not change under neither internal nor external validation. The AUC for the model was however only around 0.60, despite including 5 different risk factors. It has been suggested that genetic subtypes are associated with the risk of metastatic spread to sentinel nodes. 28 It is possible that the same

90 440 T.F. Tvedskov et al. / EJSO 40 (2014) 435e441 association exists for further metastatic spread to NSN, and inclusion of genetic subtypes in the models might have improved accuracy. Few earlier studies have tested the prediction of NSN metastases according to genetic subtypes and the results have been disappointing. 29,30 In addition, analysis of genetic subtypes will require fresh frozen tissue and was for that reason not possible in development of the present models. Finally, it is not evident that patients identified with a high risk of NSN metastases will experience axillary recurrence or survival disadvantage. The ultimate goal is to identify patients in whom further axillary treatment could prevent breast cancer deaths or at least axillary recurrences. Due to low incidence of axillary recurrences and generally excellent outcome of these patients, a very large patient series with long follow-up would be needed to develop such a model. It is possible that macrometastatic NSN or extent of involved NSN would be a better surrogate end point for axillary recurrence than metastatic NSN. Choosing the right end point, to identify patients where ALND can safely be omitted, should be considered in future model development. Conflict of interest statement There are no conflicts of interest. References 1. van Deurzen CH, de BM, Monninkhof EM, et al. Non-sentinel lymph node metastases associated with isolated breast cancer cells in the sentinel node. J Natl Cancer Inst 2008;100: Cserni G, Gregori D, Merletti F, et al. Meta-analysis of non-sentinel node metastases associated with micrometastatic sentinel nodes in breast cancer. Br J Surg 2004;91: Galimberti V, Cole BF, Zurrida S, et al. Axillary dissection versus no axillary dissection in patients with sentinel-node micrometastases (IBCSG 23-01): a phase 3 randomised controlled trial. Lancet Oncol 2013;14: Pepels MJ, de BM, Bult P, et al. Regional recurrence in breast cancer patients with sentinel node micrometastases and isolated tumor cells. Ann Surg 2012;255: Tvedskov TF, Jensen MB, Lisse IM, Ejlertsen B, Balslev E, Kroman N. High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the sentinel node. Int J Cancer 2012;131: Van Zee KJ, Manasseh DM, Bevilacqua JL, et al. A nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy. Ann Surg Oncol 2003;10: Coutant C, Olivier C, Lambaudie E, et al. Comparison of models to predict nonsentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: a prospective multicenter study. J Clin Oncol 2009;27: Cyr A, Gillanders WE, Aft RL, Eberlein TJ, Gao F, Margenthaler JA. Micrometastatic disease and isolated tumor cells as a predictor for additional breast cancer axillary metastatic burden. Ann Surg Oncol 2010;17: Gur AS, Unal B, Johnson R, et al. Predictive probability of four different breast cancer nomograms for nonsentinel axillary lymph node metastasis in positive sentinel node biopsy. J Am Coll Surg 2009;208: Rayhanabad J, Yegiyants S, Putchakayala K, Haig P, Romero L, DiFronzo LA. Axillary recurrence is low in patients with breast cancer who do not undergo completion axillary lymph node dissection for micrometastases in sentinel lymph nodes. Am Surg 2010;76: Alran S, De RY, Fourchotte V, et al. Validation and limitations of use of a breast cancer nomogram predicting the likelihood of non-sentinel node involvement after positive sentinel node biopsy. Ann Surg Oncol 2007;14: Meretoja TJ, Strien L, Heikkila PS, Leidenius MH. A simple nomogram to evaluate the risk of nonsentinel node metastases in breast cancer patients with minimal sentinel node involvement. Ann Surg Oncol 2012;19: Cserni G, Bori R, Maraz R, et al. Multi-institutional comparison of non-sentinel lymph node predictive tools in breast cancer patients with high predicted risk of further axillary metastasis. Pathol Oncol Res 2013;19: Andersson Y, Frisell J, de BJ, Bergkvist L. Prediction of non-sentinel lymph node status in breast cancer patients with sentinel lymph node metastases: evaluation of the tenon score. Breast Cancer (Auckl) 2012;6: Kocsis L, Svebis M, Boross G, et al. Use and limitations of a nomogram predicting the likelihood of non-sentinel node involvement after a positive sentinel node biopsy in breast cancer patients. Am Surg 2004;70: Chen K, Zhu L, Jia W, et al. Validation and comparison of models to predict non-sentinel lymph node metastasis in breast cancer patients. Cancer Sci 2012;103: Cserni G, Boross G, Maraz R, et al. Multicentre validation of different predictive tools of non-sentinel lymph node involvement in breast cancer. Surg Oncol 2012;21: Meretoja TJ, Leidenius MH, Heikkila PS, et al. International multicenter tool to predict the risk of nonsentinel node metastases in breast cancer. J Natl Cancer Inst 2012;104: Houvenaeghel G, Nos C, Giard S, et al. A nomogram predictive of nonsentinel lymph node involvement in breast cancer patients with a sentinel lymph node micrometastasis. Eur J Surg Oncol 2009;35: Ouldamer L, Arbion F, Balagny A, Fourquet F, Marret H, Body G. Validation of a breast cancer nomogram for predicting nonsentinel node metastases after minimal sentinel node involvement: validation of the Helsinki breast nomogram. Breast 2013;22: Moller S, Jensen MB, Ejlertsen B, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47: Tvedskov TF, Jensen MB, Balslev E, Kroman N. Robust and validated models to predict high risk of non-sentinel node metastases in breast cancer patients with micrometastases or isolated tumor cells in the sentinel node. Acta Oncol 2014 Feb;53(2): Christiansen P, Friis E, Balslev E, Jensen D, Moller S. Sentinel node biopsy in breast cancer: five years experience from Denmark. Acta Oncol 2008;47: Hwang RF, Krishnamurthy S, Hunt KK, et al. Clinicopathologic factors predicting involvement of nonsentinel axillary nodes in women with breast cancer. Ann Surg Oncol 2003;10: Grabau D, Dihge L, Ferno M, Ingvar C, Ryden L. Completion axillary dissection can safely be omitted in screen detected breast cancer patients with micrometastases. A decade s experience from a single institution. Eur J Surg Oncol 2013;39: Tvedskov TF, Jensen MB, Balslev E, et al. High risk of non-sentinel node metastases in a group of breast cancer patients with micrometastases in the Sentinel node. Eur J Cancer 2011;47:S Krag DN, Anderson SJ, Julian TB, et al. Sentinel-lymph-node resection compared with conventional axillary-lymph-node dissection in clinically node-negative patients with breast cancer: overall survival findings from the NSABP B-32 randomised phase 3 trial. Lancet Oncol 2010;11:

91 T.F. Tvedskov et al. / EJSO 40 (2014) 435e Reyal F, Rouzier R, pont-hazelzet B, et al. The molecular subtype classification is a determinant of sentinel node positivity in early breast carcinoma. PLoS One 2011;6:e Reyal F, Belichard C, Rouzier R, et al. Non-sentinel lymph node metastasis prediction in breast cancer with metastatic sentinel lymph node: impact of molecular subtypes classification. PLoS One 2012; 7:e Zhou W, He Z, Xue J, et al. Molecular subtype classification is a determinant of non-sentinel lymph node metastasis in breast cancer patients with positive sentinel lymph nodes. PLoS One 2012;7:e35881.

92 APMIS 119: Ó 2011 The Authors APMIS Ó 2011 APMIS DOI /j x Evaluating TIMP-1, Ki67, and HER2 as markers for non-sentinel node metastases in breast cancer patients with micrometastases to the sentinel node TOVE FILTENBORG TVEDSKOV, 1 ANNETTE BARTELS, 2 MAJ-BRITT JENSEN, 3 BIRGITTE PAASCHBURG, 4 NIELS KROMAN, 1 EVA BALSLEV 5 and NILS BRU NNER 2 1 Department of Breast Surgery, Copenhagen University Hospital, Copenhagen; 2 Section of Pathobiology, Faculty of Life Sciences, University of Copenhagen, Copenhagen; 3 Danish Breast Cancer Cooperative Group, Copenhagen; 4 Department of Breast Surgery, Herlev Hospital, Copenhagen; and 5 Department of Pathology, Herlev Hospital, Copenhagen, Denmark Tvedskov TF, Bartels A, Jensen M-B, Paaschburg B, Kroman N, Balslev E, Bru nner N. Evaluating TIMP-1, Ki67, and HER2 as markers for non-sentinel node metastases in breast cancer patients with micrometastases to the sentinel node. APMIS 2011; 119: The aim was to investigate whether the biochemical prognostic markers TIMP-1, Ki67, and HER2 could predict metastatic spread to non-sentinel nodes (NSN) in breast cancer patients with micrometastases to sentinel node (SN). We included all breast cancer patients with micrometastases to SN operated between 2001 and 2007 at the Department of Breast Surgery, Herlev Hospital. The study was designed as a matched case control study with 25 cases with micrometastases to SN and, in addition, metastatic spread to NSN and 50 matched controls with micrometastases to SN, but without NSN metastases. Patient and tumor characteristics were retrieved from the Danish Breast Cancer Cooperative Group database. Immunohistochemical analyses of TIMP-1 and Ki67 and measurements of HER2 on formalin-fixed paraffin-embedded tumor tissue were performed. No significant differences in the immunoreactivity of TIMP-1 and Ki67 were found between patients with and without NSN metastases. Six of seven HER2 positive patients did not have NSN metastases, but the results did not reach statistical significance. Despite being prognostic markers in breast cancer, TIMP-1 and Ki67 could not predict NSN metastases in women with micrometastatic disease to SN. Larger studies are needed to further validate HER2 as a marker for NSN metastases in these patients. Key words: Micrometastases; sentinel node; TIMP-1; Ki67; HER2. Tove Filtenborg Tvedskov, Department of Breast Surgery, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark. tft@dadlnet.dk Sentinel lymph node dissection (SLND) has become the standard of care for axillary staging of women with primary breast cancer. This procedure has made more extensive examinations of the tumor-draining lymph nodes possible and as a result more metastases, especially more micrometastases, are found (1). In many centers, Received 7 February Accepted 2 May 2011 all patients with metastases in the sentinel lymph nodes (SN) are offered an additional axillary lymph node dissection (ALND), which is an extensive surgical procedure often resulting in severe arm morbidity (2). However, only about 15% of patients with micrometastases to the SN have metastatic spread to non-sentinel lymph nodes (NSN) (3). Thus, the majority of patients with micrometastases to the SN will 844

93 TIMP-1, KI67, AND HER2 RELATION TO NSN METASTASES have no benefit from an additional ALND, but run the risk of arm morbidity. It would be advantageous to identify patients without spread to NSN and thereby avoiding unnecessary ALND. Several attempts have been made to predict metastatic spread to NSN in patients with micrometastases in the SN. Traditional prognostic markers, such as age at diagnosis, histological type, malignancy grade, tumor size, lymphovascular invasion, and hormone receptor status, have been investigated, but only tumor size and lymphovascular invasion were shown to be associated with the risk of NSN metastases with odds ratios on 2.54 and 1.71, respectively (4 7). Thus, studies using traditional prognostic markers do not appear to provide a clinically applicable method to identify a subgroup of patients where the additional ALND can safely be omitted when only micrometastases are found in the SN. Therefore, further attempts must be made to identify additional markers for NSN involvement in breast cancer patients. In Denmark, clinical and histopathologic data as well as information on treatment and followup status of all women with breast cancer have been prospectively registered in a national database managed by the Danish Breast Cancer Cooperative Group (DBCG) since This allowed us to identify two comparable groups of patients with micrometastases to the SN with and without NSN metastases. Using primary tumor tissue from these patients, we have initiated a search for biomarkers to be used to optimize the separation of patients with micrometastases to the SN into groups with and without additional NSN metastatic spread. For the present exploratory study, we selected the proteinase inhibitor tissue inhibitor of metalloproteinase 1 (TIMP-1), the proliferation marker Ki67, and the proto-oncogene HER2. These three biomarkers are all involved in the process of cancer cell dissemination (8 10), and have been shown to be associated with poor prognosis (8, 10, 11) and positive nodal status in breast cancer patients (12, 13). Hence, it is hypothesized that measurements of these markers will provide data that can be used to identify patients with further metastatic spread from SN to NSN. MATERIALS AND METHODS Patients The study was designed as a case control study. We consecutively included all breast cancer patients with micrometastases to the SN operated at the Department of Breast Surgery, Herlev Hospital, between 2001 and Micrometastases had been identified using multisectioning and immunohistochemical (IHC) cytokeratin staining of the SN and classified according to the American Joint Committee on Cancer (14), in combination with cell counts, defining metastases with tumor cells as micrometastases. NSN had been examined by bisectioning and hematoxylin eosin staining. A total of 257 patients with micrometastases in the SN were identified retrospectively. All patients had formalin-fixed paraffinembedded (FFPE) tumor tissue stored. Pathology data on tumor size, hormone receptor status, histological type, malignancy grade, age at diagnosis, number of removed lymph nodes, and presence of NSN metastases from these patients were reviewed from the original pathology files and the DBCG database (15). The following 54 patients were excluded: male patients registered as having only carcinoma in situ, patients with bilateral tumors, patients not registered in the DBCG database, and patients without additional ALND. The remaining 203 patients were eligible for the study. On average, 20 lymph nodes had been removed in these patients, ranging from 7 to 40 lymph nodes. In 26 patients (13%), NSN metastases had been identified. These patients were considered as cases. For each case, two matched controls without NSN metastases were found among the remaining 177 patients. Patients were matched by the following criteria: tumor size ( 2 cm, >2 cm), hormone receptor status, age at diagnosis (±5 years), and malignancy grade (grade I, grade II III), if possible. One patient was excluded because no suitable match was found, leaving 25 cases and 50 matched controls for further analyses. Patient and tumor characteristics are shown in Table 1. Immunohistochemical staining for TIMP-1 and Ki67 Blinded IHC analyses of TIMP-1 and Ki67 on existing FFPE tissue blocks of the primary tumors were performed. The IHC procedures have previously been described in detail (16). In brief, 3 lm whole sections were deparaffinized in xylene and rehydrated in graded concentrations of ethanol. For antigen retrieval, the sections were microwave treated in citrate buffer ph = 6, and endogen peroxidase activity was blocked by hydrogen peroxide. Immunohistochemical staining for TIMP-1 used the mouse monoclonal antibody, clone VT7, raised Ó 2011 The Authors APMIS Ó 2011 APMIS 845

94 TVEDSKOV et al. Table 1. Patient and tumor characteristics of 75 breast cancer patients with micrometastases in the sentinel node with and without additional metastasis in NSN Characteristics NSN metastases, n=25 No NSN metastases, n=50 Total, n=75 Mean age in 62 (36 77) 61 (36 79) 61 (36 79) years (range) Mean number of 20 (9 34) 20 (7 39) 20 (7 39) LN removed (range) Mean tumor size in mm (range) 18 (4 38) 17 (7 33) 17 (4 38) Histological type (%) Ductal 23 (92) 44 (88) 67 (89) Lobular 2 (8) 6 (12) 8 (11) Tumor grade (%) Grade I 12 (48) 23 (46) 35 (46) Grade II 8 (32) 20 (40) 28 (37) Grade III 5 (20) 7 (14) 12 (16) Hormone receptor status (%) Positive 24 (96) 48 (96) 72 (96) Negative 1 (4) 2 (4) 3 (4) NSN, non-sentinel node metastases; LN, lymph nodes. against recombinant human TIMP-1 at concentration 0.25 lg ml (17). This antibody has previously been found optimal for IHC detection of TIMP-1 on FFPE tissue sections (18). Sections were stained with primary antibody overnight at 4 C. Immunohistochemical staining for Ki67 was performed using the monoclonal mouse anti-human Ki67, clone MIB-1 (Code No. M7240) (from Dako, Glostrup, Denmark) at a concentration of 1.6 mg ml. Sections were stained with primary antibody for 30 min at room temperature. Both antibodies were detected with Advance HRP (Code No. K4068), and the reaction was visualized with DAB+ (Code No. K3468). All sections were counterstained with Mayer s hematoxylin. An irrelevant monoclonal antibody (Anti-TNP), raised against tri-nitro-phenol hapten, was used as a negative control. A human mammary carcinoma known to contain the investigated antigens was included as a positive control. Two independent observers (EB and AB) assessed the sections semi-quantitatively using light microscopy. In case of discrepancies, agreement was reached by looking at the slides together. Tumor sections were considered as TIMP-1 positive if any degree of staining was seen. In addition, TIMP-1 antigen immunoreactivity in the tumor cells was graded from 0 to 3 according to intensity and extensity of cytoplasmic staining, respectively. The extensity score was graded as 0 if no tumor cells were stained, 1 if >0% and <25% were stained, 2 if 25% and 50% were stained, and 3 if more than 50% of the tumor cells were stained (Fig. 1A,B). Intensity score was based on the average intensity of staining, and graded as 0 if the staining was absent, 1 for weak staining, 2 for moderate staining, and 3 for intense staining. Finally, a common score was made for each patient by multiplying the grades. TIMP-1 antigen immunoreactivity in stromal cells of the tumors was characterized as negative if no staining was seen and as positive if any degree of staining was observed (Fig. 1C). Ki67 antigen immunoreactivity in tumor cells was determined as percentage of stained tumor cells present at the invasive front of the tumor. The Ki67 staining was considered as positive if more than 14% of the tumor cells were stained for Ki67 (Fig. 1E,F) (8). Fluorescence in situ hybridization (FISH) for HER2 Twenty seven patients had a known HER2 status registered in the DBCG database. For these patients HER2 status had been determined previously, according to international recommendations (19), using the HercepTest (DakoCytomation, Glostrup, Denmark) for IHC analysis according to the manufacturer s manual, where 1+ was considered as negative, 3+ was considered as positive, and 2+ was considered as equivocal. In case of 2+, a supplementary FISH-test for gene amplification had been performed to determine the final HER2 status. In the remaining 48 patients, the HER2 status was unknown. In these patients, HER2 status was determined retrospectively using the HER2 FISH pharmdx Kit (Dako) for gene amplification on whole sections of existing FFPE tissue blocks from the primary tumors as described elsewhere (20). The tumors were considered as HER2 positive if the ratio of gene amplification was >2.2 (21). Statistical analysis The exploratory character of this study did not allow precise power analyses, but the sample size of 25 cases and 50 controls would give sufficient power to detect a medium or large difference in the proportions between groups (22) with a power of 90% and a = 0.05 [Lenth, R. V. (2006 9), Java Applets for Power and Sample Size (Computer software), retrieved 10th of October 2010, from Associations between the presence of NSN metastases and TIMP-1 positive tumor cells, HER2 positive tumor cells, Ki67 positive tumor cells, or TIMP-1 positive stromal cells were analyzed using Cochrane Mantel Haenszel test 846 Ó 2011 The Authors APMIS Ó 2011 APMIS

95 TIMP-1, KI67, AND HER2 RELATION TO NSN METASTASES A B C D E F Fig. 1. TIMP-1 and Ki67 IHC. (A) Tumor with weak TIMP-1 staining considered as intensity grade 1, (B) Tumor with intense TIMP-1 staining considered as intensity grade 3, (C) TIMP-1 staining stromal cells, (D) TIMP-1 stained section with changing intensity, (E) Ki67 stained tumor considered as Ki67 negative, (F) Ki67 stained tumor considered as Ki67 positive. excluding unknowns. The differences in TIMP-1 common score and in percentage of cells with Ki67 staining between cases and their matched controls were found to be normally distributed and analyzed using a paired Student s t-test. For the two matched controls in each pair, an average common score of TIMP-1 and an average percentage of Ki67 stained tumor cells was calculated for this analysis. Paraffin specimens from three controls did not contain sufficient tumor tissue for analyses, and the remaining control in these three pairs was used alone. Associations between proliferation rate and HER2 status and TIMP-1 status, respectively, were analyzed using a Student s t-test. Two-tailed p-values were applied and the level of significance was set to 5%. All statistical analyses were done using SAS 9.1 (SAS Institute, Cary, NC, USA). The study was approved by the Ethical Committees of the Capital Region, protocol nr. H , and by the Danish Data Protection Agency (J.nr ). RESULTS Patients included in the study had generally small, low grade tumors (Table 1). Relations between TIMP-1, Ki67, and HER2 expression and NSN status are shown in Table 2. Seventyfive percent of the patients had TIMP-1 positive tumors, corresponding to 80% of the cases and 72% of the controls, and thus, no significant difference in the distribution of TIMP-1 positive tumors between cases and controls (p = 0.77) was observed. A large variation was seen in the number of stained tumor cells and in the intensity of staining between the different specimens (Fig. 1D). To reflect this variation, a common score including both extensity and intensity of TIMP-1 staining was calculated as described. No significant difference was seen in TIMP-1 Ó 2011 The Authors APMIS Ó 2011 APMIS 847

96 TVEDSKOV et al. Table 2. Associations between TIMP-1, Ki67, and HER2 expression and NSN status in 75 breast cancer patients with micrometastases in the sentinel node with and without additional metastasis in NSN Variables NSN metastases, No NSN metastases, Total, p-value n=25 n=50 n=75 TIMP-1 positive tumor cells (%) Positive 20 (80) 36 (72) 56 (75) 0.77 Negative 5 (20) 11 (22) 16 (21) Unknown 0 (0) 3 (6) 3 (4) TIMP-1 common score, mean (range) 2.88 (0 9) 3.81 (0 9) 3.49 (0 9) 0.21 TIMP-1 positive stromal cells (%) Positive 11 (44) 20 (40) 31 (41) 0.93 Negative 13 (52) 18 (36) 31 (41) Unknown 1 (4) 12 (24) 13 (18) Ki67 positive tumor cells (%) Positive 11 (44) 22 (44) 33 (44) 0.93 Negative 14 (52) 25 (50) 39 (52) Unknown 0 (0) 3 (6) 3 (4) Percentage of Ki67 positive 13.9 (0 60) 16.0 (0 50) 15.3 (0 60) 0.52 tumor cells, mean (range) HER2 status (%) Positive 1 (4) 6 (12) 7 (9) 0.12 Negative 24 (96) 37 (74) 61 (82) Unknown 0 (0) 7 (14) 7 (9) NSN, non-sentinel node; TIMP, tissue inhibitor of metalloproteinase; HER, human epidermal growth factor receptor. common score between patients with and without NSN metastases (p = 0.21). Stromal cells were identified in tumors from 62 patients. In half of these patients, the stromal cells stained positive for TIMP-1, with an even distribution between cases and controls (p = 0.93). Forty four percent of the patients had Ki67 positive tumors, with an even distribution between cases and controls (p = 0.93). Furthermore, there was no significant difference in the percentage of Ki67 stained tumor cells in the invasive front of the tumor between patients with and without NSN metastases (p = 0.52). In total, only seven patients (9%) were HER2 positive. Six of these seven HER2 positive patients did not have NSN metastases (12% of controls). Nevertheless, this difference in HER2 expression between cases and controls did not reach statistical significance (p = 0.12). In general, the tumor cells of HER2 positive patients showed a significantly higher level of proliferation, measured by immunoreactivity of Ki67, compared with HER2 negative patients (p = ). In contrast, TIMP-1 immunoreactivity in both tumor and stromal cells was not found significantly associated with proliferation rate (p = 0.14) or with HER2 overexpression (p = 0.17). DISCUSSION The optimal surgical treatment of patients with minimal metastatic disease in the sentinel node is under debate worldwide (23). ALND may not be necessary in the majority of these patients (3), and efforts must be made to identify subgroups of patients where ALND can be safely omitted. In this study, we selected a number of well known prognostic biomarkers to be tested for the hypothesis that specific biologic characteristics of the primary tumor can be used to identify patients without spread beyond the SN, allowing omission of ALND in these patients. The three biomarkers selected for this study, TIMP-1, Ki67, and HER2, are all involved in the process of cancer cell dissemination. TIMP- 1 is a proteinase inhibitor which, in preclinical studies, has been shown to promote tumor progression (9). Furthermore, overexpression of TIMP-1 has been found to be associated with positive nodal status (13, 24) and with high risk of relapse (11, 25 27) in breast cancer patients. The nuclear antigen Ki67 is used as marker for cell proliferation (8). In breast cancer patients, Ki67 has been shown to be associated with increased risk of lymph node metastases (28, 29) 848 Ó 2011 The Authors APMIS Ó 2011 APMIS

97 TIMP-1, KI67, AND HER2 RELATION TO NSN METASTASES and with poor prognosis (8, 30, 31). Finally, the proto-oncogene HER2 is a tyrosine kinase receptor involved in the regulation of breast cell growth (10). Overexpression of this receptor has been shown to be associated with both poor prognosis (10) and with positive axillary nodal status in breast cancer patients (29, 32). We tested the three biomarkers on consecutively included breast cancer patients with micrometastases to the SN, operated at the Department of Breast Surgery, Herlev Hospital, one of the largest centers for breast surgery in Denmark. The time period for patient inclusion was from the introduction of the SLND in 2001 to the end of This resulted in a sample size of 203 eligible patients with micrometastases to the SN, of which 13% had NSN metastases, which is in accordance to other studies (3, 33). This sample size was adequate for showing a medium or large difference between groups (22). Detection of smaller differences was not considered as clinically relevant. Studies on the expression of biomarkers in breast cancer patients with micrometastases to the SN are scarce. Especially, the expression of TIMP-1 has never been investigated in relation to micrometastatic disease before. In our study, 75% of all patients had TIMP-1 positive tumors. This level is similar to what has been found in earlier studies on TIMP-1 expression in breast cancer patients (13, 16). The presence of NSN metastases was not significantly associated with either qualitative or quantitative levels of TIMP-1 immunoreactivity in tumor cells. Furthermore, we did not find TIMP-1 immunoreactivity in tumor cells significantly associated with HER2 overexpression or with proliferation rate of the tumor. This is in contrast to previous studies where TIMP-1 has been shown to be inversely associated with proliferation rate (34, 35). As shown in both the present and previous studies, TIMP-1 may also be present in stromal cells of the tumor (34, 35), and it is possible that the microenvironment around the tumor plays an active role in tumor cell dissemination (36). We therefore tested if TIMP-1 immunoreactivity in stromal cells was associated with the presence of NSN metastases. However, we did not find any association between TIMP-1 immunoreactivity in stromal cells and the presence of NSN metastases. Furthermore, we found that TIMP-1 immunoreactivity in stromal cells was not significantly associated with proliferation rate of the tumor, which is in accordance with previous studies, where TIMP-1 immunoreactivity in stromal cells was found to be unassociated with cell proliferation and prognosis (34). Altogether, we found that TIMP-1 was not significantly associated with NSN status in breast cancer patients with micrometastases to the SN, both according to TIMP-1 immunoreactivity in tumor cells and immunoreactivity in the stromal cells of the tumor. Forty-four percent of the patients in this study had Ki67 positive tumors. This is the same level as in other studies using a similar threshold of >14% stained tumor cells as Ki67 positive (31). Two previous studies have tested if immunoreactivity of Ki67 could predict the presence of NSN metastases, but the results are conflicting. Viale et al. included 1228 patients with either macro- or micrometastases to the SN. They found no association between Ki67 immunoreactivity and NSN status (37). In contrast, Carcoforo et al. found a significantly increased Ki67 immunoreactivity in three of eight patients with NSN metastases (4), but the study size was small and no adjustment for confounders was done. In our study, where possible confounders were adjusted for by matching, we did not find any association between Ki67 immunoreactivity and the presence of NSN metastases. Only 9% of the patients in this study were HER2 positive in contrast to 20 30% of breast cancer patients in general (10). The low number of HER2 positive patients found in our study is similar to what was found in another study only including patients with micrometastases in the SN (4). As a consequence of the low number of HER2 positive patients, this study lacks statistical power to show a significant association between HER2 status and the presence of NSN metastases, despite the fact that only one of seven HER2 positive patients had NSN metastases. HER2 has earlier been tested for the ability to predict NSN metastases in studies with a mixed group of patients with micro- or macrometastases to the SN (37 43). None of these studies found HER2 useful in predicting metastatic spread to NSN. Only one study, by Carcoforo et al., exclusively included patients with micrometastases to the SN (4). Only 58 patients were included in the study, of which only eight Ó 2011 The Authors APMIS Ó 2011 APMIS 849

98 TVEDSKOV et al. had NSN metastases and no association between HER2 status and NSN status was found. Finally, as shown by others, we found that HER2 overexpression was significantly associated with proliferation rate measured by Ki67 (44). This study has some limitations. First, the biomarker analyses were made on primary tumor tissue, and not on the corresponding SN micrometastasis. Analyses on micrometastases were not technically possible because of limited amounts of metastatic tissue. Thus, we cannot be sure that the expression of the tested markers would be the same in the SN metastases as in the primary tumor. To underline this problem, previous studies have observed differences in the immunoreactivity of TIMP-1 in primary tumor tissue and the corresponding axillary lymph node metastasis (45). On the other hand, studies have shown a high degree of concordance in the immunoreactivity of Ki67 and in both immunoreactivity and gene amplification of HER2 between primary tumor tissue and corresponding axillary lymph node metastasis (46 48), which suggests that even if we could have analyzed the tumor cells in the SN, no major changes in the results would have been observed for these two biomarkers. Second, scoring of the biomarkers Ki67 and TIMP-1 is difficult and not yet standardized. The right cut-off point for Ki67 is under debate (8, 49). We chose a cut-off point of 14% as used in the majority of earlier studies (8, 31). Likewise, no generally accepted scoring system exists for TIMP-1. We used a modified H-score, similar to what has been used in earlier studies (50, 51). We found a large variation in the expression of TIMP-1 throughout the same specimen, changing from areas where all tumor cells were intensively stained to areas without any staining at all (Fig. 1D). To avoid false negative results, we therefore used analyses of whole sections and not tissue microarrays (TMA), despite the use of TMA becoming more and more widespread. Furthermore, scoring was done manually without the use of image analyzing systems. Taking these limitations into account, this study still indicates that no major difference exists in the immunoreactivity of Ki67 and TIMP-1 between patients with micrometastases to the SN with and without additional tumor cell dissemination to the remaining axillary lymph nodes. This indicates that these biomarkers, despite being prognostic markers in breast cancer, cannot be used in the decision for ALND when only micrometastatic disease is found in the SN. A non-significant trend toward differences in HER2 status between cases and controls was found, but the number of HER2 positive patients was low. Further studies based on larger populations are needed to test if HER2 status can be used to predict NSN metastases in patients with only micrometastases in the SN, either alone or in combination with traditional prognostic markers like tumor size and lymphovascular invasion. The study was supported by IMK common foundation, HOC research initiative foundation, Danish Cancer Research Foundation, Sino-Danish Breast Cancer Research Centre, Sawmill owner Jeppe Juhl and Ovita Juhl Memorial Foundation and A.P. Møller Foundation for the Advancement of Medical Science. REFERENCES 1. Chen SL, Hoehne FM, Giuliano AE. The prognostic significance of micrometastases in breast cancer: a SEER population-based analysis. Ann Surg Oncol 2007;14: Gartner R, Jensen MB, Nielsen J, Ewertz M, Kroman N, Kehlet H. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA 2009;302: Cserni G, Gregori D, Merletti F, Sapino A, Mano MP, Ponti A, et al. Meta-analysis of non-sentinel node metastases associated with micrometastatic sentinel nodes in breast cancer. Br J Surg 2004;91: Carcoforo P, Maestroni U, Querzoli P, Lanzara S, Maravegias K, Feggi L, et al. Primary breast cancer features can predict additional lymph node involvement in patients with sentinel node micrometastases. World J Surg 2006;30: Gipponi M, Canavese G, Lionetto R, Catturich A, Vecchio C, Sapino A, et al. The role of axillary lymph node dissection in breast cancer patients with sentinel lymph node micrometastases. Eur J Surg Oncol 2006;32: Houvenaeghel G, Nos C, Mignotte H, Classe JM, Giard S, Rouanet P, et al. Micrometastases in sentinel lymph node in a multicentric study: predictive factors of nonsentinel lymph node involvement Groupe des Chirurgiens de la Federation des Centres de Lutte Contre le Cancer. J Clin Oncol 2006;24: Ó 2011 The Authors APMIS Ó 2011 APMIS

99 TIMP-1, KI67, AND HER2 RELATION TO NSN METASTASES 7. Schrenk P, Konstantiniuk P, Wolfl S, Bogner S, Haid A, Nemes C, et al. Prediction of non-sentinel lymph node status in breast cancer with a micrometastatic sentinel node. Br J Surg 2005;92: Yerushalmi R, Woods R, Ravdin PM, Hayes MM, Gelmon KA. Ki67 in breast cancer: prognostic and predictive potential. Lancet Oncol 2010;11: Wurtz SO, Schrohl AS, Mouridsen H, Brunner N. TIMP-1 as a tumor marker in breast cancer an update. Acta Oncol 2008;47: Menard S, Fortis S, Castiglioni F, Agresti R, Balsari A. HER2 as a prognostic factor in breast cancer. Oncology 2001;61(Suppl 2): Schrohl AS, Holten-Andersen MN, Peters HA, Look MP, Meijer-van Gelder ME, Klijn JG, et al. Tumor tissue levels of tissue inhibitor of metalloproteinase-1 as a prognostic marker in primary breast cancer. Clin Cancer Res 2004;10: Fehm T, Becker S, Pergola-Becker G, Kramer B, Gruber I, Sotlar K, et al. Influence of tumor biological factors on tumor cell dissemination in primary breast cancer. Anticancer Res 2004;24: Wu ZS, Wu Q, Yang JH, Wang HQ, Ding XD, Yang F, et al. Prognostic significance of MMP-9 and TIMP-1 serum and tissue expression in breast cancer. Int J Cancer 2008;122: Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol 2002;20: Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47: Willemoe GL, Hertel PB, Bartels A, Jensen MB, Balslev E, Rasmussen BB, et al. Lack of TIMP-1 tumour cell immunoreactivity predicts effect of adjuvant anthracycline-based chemotherapy in patients (n = 647) with primary breast cancer. A Danish Breast Cancer Cooperative Group Study. Eur J Cancer 2009;45: Moller SN, Dowell BL, Stewart KD, Jensen V, Larsen L, Lademann U, et al. Establishment and characterization of 7 new monoclonal antibodies to tissue inhibitor of metalloproteinases-1. Tumour Biol 2005;26: Sorensen IV, Fenger C, Winther H, Foged NT, Lademann U, Brunner N, et al. Characterization of anti-timp-1 monoclonal antibodies for immunohistochemical localization in formalin-fixed, paraffin-embedded tissue. J Histochem Cytochem 2006;54: Rasmussen BB, Andersson M, Christensen IJ, Moller S. Evaluation of and quality assurance in HER2 analysis in breast carcinomas from patients registered in Danish Breast Cancer Group (DBCG) in the period of A nationwide study including correlation between HER-2 status and other prognostic variables. Acta Oncol 2008;47: Olsen KE, Knudsen H, Rasmussen BB, Balslev E, Knoop A, Ejlertsen B, et al. Amplification of HER2 and TOP2A and deletion of TOP2A genes in breast cancer investigated by new FISH probes. Acta Oncol 2004;43: Wolff AC, Hammond ME, Schwartz JN, Hagerty KL, Allred DC, Cote RJ, et al. American Society of Clinical Oncology College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. Arch Pathol Lab Med 2007;131: Cohen J. Statistical power analysis for the behavioral sciences, Routledge Academic; 2 edn, Wasif N, Maggard MA, Ko CY, Giuliano AE. Underuse of axillary dissection for the management of sentinel node micrometastases in breast cancer. Arch Surg 2010;145: Ree AH, Florenes VA, Berg JP, Maelandsmo GM, Nesland JM, Fodstad O. High levels of messenger RNAs for tissue inhibitors of metalloproteinases (TIMP-1 and TIMP-2) in primary breast carcinomas are associated with development of distant metastases. Clin Cancer Res 1997;3: Schrohl AS, Christensen IJ, Pedersen AN, Jensen V, Mouridsen H, Murphy G, et al. Tumor tissue concentrations of the proteinase inhibitors tissue inhibitor of metalloproteinases-1 (TIMP-1) and plasminogen activator inhibitor type 1 (PAI-1) are complementary in determining prognosis in primary breast cancer. Mol Cell Proteomics 2003;2: Wurtz SO, Christensen IJ, Schrohl AS, Mouridsen H, Lademann U, Jensen V, et al. Measurement of the uncomplexed fraction of tissue inhibitor of metalloproteinases-1 in the prognostic evaluation of primary breast cancer patients. Mol Cell Proteomics 2005;4: Talvensaari-Mattila A, Turpeenniemi-Hujanen T. High preoperative serum TIMP-1 is a prognostic indicator for survival in breast carcinoma. Breast Cancer Res Treat 2005;89: Marrelli D, Pinto E, Neri A, Megha T, Gioffre W, Cioppa T, et al. Mib-1 proliferation index is an independent predictor of lymph node metastasis in invasive breast cancer: a prospective study on 675 patients. Oncol Rep 2006;15: Fehm T, Maul H, Gebauer S, Scharf A, Baier P, Sohn C, et al. Prediction of axillary lymph node status of breast cancer patients by tumorbiological factors of the primary tumor. Strahlenther Onkol 2005;181: Ó 2011 The Authors APMIS Ó 2011 APMIS 851

100 TVEDSKOV et al. 30. Stuart-Harris R, Caldas C, Pinder SE, Pharoah P. Proliferation markers and survival in early breast cancer: a systematic review and meta-analysis of 85 studies in 32,825 patients. Breast 2008;17: de Azambuja E, Cardoso F, de Castro G Jr, Colozza M, Mano MS, Durbecq V, et al. Ki-67 as prognostic marker in early breast cancer: a metaanalysis of published studies involving 12,155 patients. Br J Cancer 2007;96: Van CB, Vanden BI, Drijkoningen M, Pochet N, Cheng J, Van HS, et al. Axillary lymph node status of operable breast cancers by combined steroid receptor and HER-2 status: triple positive tumours are more likely lymph node positive. Breast Cancer Res Treat 2009;113: Abdessalam SF, Zervos EE, Prasad M, Farrar WB, Yee LD, Walker MJ, et al. Predictors of positive axillary lymph nodes after sentinel lymph node biopsy in breast cancer. Am J Surg 2001;182: Nakopoulou L, Giannopoulou I, Lazaris AC, Alexandrou P, Tsirmpa I, Markaki S, et al. The favorable prognostic impact of tissue inhibitor of matrix metalloproteinases-1 protein overexpression in breast cancer cells. APMIS 2003;111: Nakopoulou L, Giannopoulou I, Stefanaki K, Panayotopoulou E, Tsirmpa I, Alexandrou P, et al. Enhanced mrna expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) in breast carcinomas is correlated with adverse prognosis. J Pathol 2002;197: Chambers AF, Groom AC, MacDonald IC. Dissemination and growth of cancer cells in metastatic sites. Nat Rev Cancer 2002;2: Viale G, Maiorano E, Pruneri G, Mastropasqua MG, Valentini S, Galimberti V, et al. Predicting the risk for additional axillary metastases in patients with breast carcinoma and positive sentinel lymph node biopsy. Ann Surg 2005;241: van Deurzen CH, van HR, Hobbelink MG, Seldenrijk CA, Koelemij R, van Diest PJ. Predictive value of tumor load in breast cancer sentinel lymph nodes for second echelon lymph node metastases. Cell Oncol 2007;29: Yu JC, Hsu GC, Hsieh CB, Sheu LF, Chao TY. Prediction of metastasis to non-sentinel nodes by sentinel node status and primary tumor characteristics in primary breast cancer in Taiwan. World J Surg 2005;29: Turner RR, Chu KU, Qi K, Botnick LE, Hansen NM, Glass EC, et al. Pathologic features associated with nonsentinel lymph node metastases in patients with metastatic breast carcinoma in a sentinel lymph node. Cancer 2000;89: Kapur U, Rubinas T, Ghai R, Sinacore J, Yao K, Rajan PB. Prediction of nonsentinel lymph node metastasis in sentinel node-positive breast carcinoma. Ann Diagn Pathol 2007;11: Zhang T, Wang H, Chen BP, Zhang HS, Wei XL, Fu Y, et al. Risk factors for non-sentinel lymph node metastases in breast cancer patients with positive sentinel lymph nodes. Chin Med J (Engl) 2008;121: Tan YY, Fan YG, Lu Y, Hwang S, Ewing C, Esserman L, et al. Ratio of positive to total number of sentinel nodes predicts nonsentinel node status in breast cancer patients. Breast J 2005;11: Offersen BV, Alsner J, Ege OK, Riisbro R, Brunner N, Sorensen FB, et al. A comparison among HER2, TP53, PAI-1, angiogenesis, and proliferation activity as prognostic variables in tumours from 408 patients diagnosed with early breast cancer. Acta Oncol 2008;47: Garcia MF, Gonzalez-Reyes S, Gonzalez LO, Junquera S, Berdize N, Del Casar JM, et al. Comparative study of the expression of metalloproteases and their inhibitors in different localizations within primary tumours and in metastatic lymph nodes of breast cancer. Int J Exp Pathol 2010;91: Cho EY, Han JJ, Choi YL, Kim KM, Oh YL. Comparison of Her-2, EGFR and cyclin D1 in primary breast cancer and paired metastatic lymph nodes: an immunohistochemical and chromogenic in situ hybridization study. J Korean Med Sci 2008;23: Van den Eynden GG, Van der Auwera I, Van Laere SJ, Trinh XB, Colpaert CG, van Dam P, et al. Comparison of molecular determinants of angiogenesis and lymphangiogenesis in lymph node metastases and in primary tumours of patients with breast cancer. J Pathol 2007;213: Strien L, Leidenius M, von Smitten K, Heikkila P. Concordance between HER-2 and steroid hormone receptor expression between primary breast cancer, sentinel node metastases, and isolated tumor cells. Pathol Res Pract 2010;206: Ahlin C, Aaltonen K, Amini RM, Nevanlinna H, Fjallskog ML, Blomqvist C. Ki67 and cyclin A as prognostic factors in early breast cancer. What are the optimal cut-off values?. Histopathology 2007;51: Del Casar JM, Carreno G, Gonzalez LO, Junquera S, Gonzalez-Reyes S, Gonzalez JM, et al. Expression of metalloproteases and their inhibitors in primary tumors and in local recurrences after mastectomy for breast cancer. J Cancer Res Clin Oncol 2010;136: Vizoso FJ, Gonzalez LO, Corte MD, Rodriguez JC, Vazquez J, Lamelas ML, et al. Study of matrix metalloproteinases and their inhibitors in breast cancer. Br J Cancer 2007;96: Ó 2011 The Authors APMIS Ó 2011 APMIS

101 Acta Oncologica ISSN: X (Print) X (Online) Journal homepage: Risk of non-sentinel node metastases in patients with symptomatic cancers compared to screendetected breast cancers Tove F. Tvedskov, Maj-Britt Jensen, Eva Balslev, Jens P. Garne, Ilse Vejborg, Peer Christiansen, Bent Ejlertsen & N. Kroman To cite this article: Tove F. Tvedskov, Maj-Britt Jensen, Eva Balslev, Jens P. Garne, Ilse Vejborg, Peer Christiansen, Bent Ejlertsen & N. Kroman (2015): Risk of non-sentinel node metastases in patients with symptomatic cancers compared to screen-detected breast cancers, Acta Oncologica, DOI: / X To link to this article: Published online: 09 Oct Submit your article to this journal Article views: 51 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at Download by: [Copenhagen University Library] Date: 24 February 2016, At: 04:57