original articles C. G. Rusthoven 1 *, R. A. Rabinovitch 1, B. L. Jones 1, M. Koshy 2,3, A. Amini 1,N.Yeh 1, M. W. Jackson 1 & C. M.

Transcription

1 16. Smith IE, Dowsett M. Aromatase inhibitors in breast cancer. N Engl J Med 2003; 348: Parker JS, Peterson AC, Tudor IC et al. A novel biomarker to predict sensitivity to enzalutamide in triple-negative breast cancer. ASCO 2015; Abstr Robinson JL, Macarthur S, Ross-Innes CS et al. Androgen receptor driven transcription in molecular apocrine breast cancer is mediated by FoxA1. EMBO J 2011; 30: Ni M, Chen Y, Lim E et al. Targeting androgen receptor in estrogen receptornegative breast cancer. Cancer Cell 2011; 20: Lehmann BD, Bauer JA, Schafer JM et al. PIK3CA mutations in androgen receptor-positive triple negative breast cancer confer sensitivity to the combination of PI3K and androgen receptor inhibitors. Breast Cancer Res 2014; 16: : , 2016 doi: /annonc/mdw046 Published online 9 February 2016 The impact of postmastectomy and regional nodal radiation after neoadjuvant chemotherapy for clinically lymph node-positive breast cancer: a National Cancer Database (NCDB) analysis C. G. Rusthoven 1 *, R. A. Rabinovitch 1, B. L. Jones 1, M. Koshy 2,3, A. Amini 1,N.Yeh 1, M. W. Jackson 1 & C. M. Fisher 1 1 Department of Radiation Oncology, The University of Colorado School of Medicine, Aurora; Departments of 2 Radiation Oncology; 3 Radiation and Cellular Oncology, The University of Chicago School of Medicine, Chicago, USA Received 23 October 2015; revised 14 January 2016; accepted 26 January 2016 Background: Following neoadjuvant chemotherapy (NAC), the optimal strategies for postmastectomy radiotherapy (PMRT) and regional nodal irradiation (RNI) after breast-conserving surgery (BCS) are controversial. In this analysis, we evaluate the impact of these radiotherapy (RT) approaches for women with clinically node-positive breast cancer treated with NAC in the National Cancer Database (NCDB). Patients and methods: Women with ct1 3 cn1 M0 breast cancer treated with NAC were divided into four cohorts by surgery [Mastectomy (Mast) versus BCS] and post-chemotherapy pathologic nodal status (ypn0 versus ypn+). Overall survival (OS) was estimated using the Kaplan Meier method and RT approaches were analyzed using the log-rank test, multivariate Cox models, and propensity score-matched analyses. Results: From 2003 to 2011, cases were identified including 3040 Mast-ypN0, 7243 Mast-ypN+, 2070 BCSypN0, and 2962 BCS-ypN+ patients. On univariate analysis, PMRT was associated with improved OS for both MastypN0 (P = 0.019) and Mast-ypN+ (P < 0.001) patients. On multivariate analyses adjusted for factors including age, comorbidity score, ct stage, in-breast pathologic complete response, axillary surgery, ypn stage, estrogen receptor status and hormone therapy, PMRT remained independently associated with improved OS among Mast-ypN0 [hazard ratio (HR) = 0.729, 95% confidence interval (CI) , P = 0.015] and Mast-ypN+ patients (HR = 0.772, 95% CI , P < 0.001). No differences in OS were observed with the addition of RNI to breast RT for BCS-ypN0 or BCS-ypN+ patients. Propensity score-matched analyses demonstrated identical patterns of significance. On subset analysis, OS was improved with PMRT in each pathologic nodal subgroup (ypn0, ypn1, and ypn2-3) (all P < 0.05). Conclusions: In the largest reported analysis of RT for cn1 patients treated with NAC, PMRT was associated with improved OS for all pathologic nodal subgroups. No OS differences were observed with the addition of RNI to breast RT. Key words: National Cancer Database (NCDB), neoadjuvant chemotherapy, radiotherapy, postmastectomy radiation therapy (PMRT), regional lymph node irradiation (RNI), pathologic complete response ( path CR) *Correspondence to: Dr Chad Rusthoven, Department of Radiation Oncology, University of Colorado School of Medicine, 1665 N. Aurora Court, Suite 1032, Mail Stop F706, Aurora, CO 80045, USA. Tel: ; Fax: ; chad. rusthoven@ucdenver.edu The Author Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please journals.permissions@oup.com.

2 introduction The survival benefit of postmastectomy radiotherapy (PMRT) for women with pathologic lymph node involvement after upfront mastectomy has been established by multiple randomized trials [1, 2] and meta-analyses [3, 4]. Recently, reported randomized trials have also demonstrated improved disease-free survival with the addition of regional nodal irradiation (RNI) for women treated with upfront breast-conserving surgery (BCS) or mastectomy [5, 6], with one trial demonstrating a reduction in breast cancer mortality [5]. However, no randomized data are currently available for women treated with neoadjuvant chemotherapy (NAC) before surgery to clarify the role of PMRT or the addition of RNI to breast radiotherapy (RT) in this setting. As a result, the optimal RT approach for the growing number of women treated with NAC is a matter of considerable debate, particularly for those presenting with clinically positive lymph nodes who achieve a favorable pathologic response to NAC [7]. In this analysis, using data from the National Cancer Database (NCDB), we evaluated the survival impact of PMRT and the addition of RNI to breast RT for women with clinically positive lymph nodes treated with NAC and surgery. materials and methods cohort The NCDB is a hospital-based cancer registry sponsored by the American College of Surgeons and American Cancer Society including 70% of malignant cancers diagnosed in the United States. Demographics, comorbidities, tumor characteristics, and overall survival (OS) are recorded, as well as therapies delivered during the first course of treatment including surgery, RT, chemotherapy, and hormone therapy. The NCDB was queried for women 18 years with ct1 3, cn1, M0 breast cancer treated with multiagent NAC followed by mastectomy or BCS, with complete datasets for American Joint Committee on Cancer staging, treatment, and Charlson Deyo comorbidity scores. The search was limited to years where reporting facilities were accountable for the completeness of data, which included cases diagnosed from 2003 to NAC was defined by an interval from chemotherapy initiation to surgery of 84 and 270 days [8]. RT field design data in the NCDB include breast-only, chest wallonly, breast plus RNI, and chest wall plus RNI. No data are available regarding the specific types of chemotherapy or hormonal therapy agents administered. statistical analysis Women in the NCDB with ct1 3 cn1 M0 breast cancer Receiving Neoadjuvant Chemotherapy (NAC) and definitive surgery from Mastectomy (Mast) patients Exclude Patients were separated into four parallel cohorts by surgery [Mastectomy (Mast) or BCS] and post-nac pathologic lymph node stage (ypn) into those with (ypn+) and without (ypn0) residual nodal disease (Figure 1). Therefore, the final cohorts for analysis included Mast-ypN0, Mast-ypN+, BCS-ypN0, BCS-ypN+, representing four distinct clinical scenarios for treatment decisions regarding adjuvant RT. For Mast patients, the primary end point was OS from diagnosis with PMRT (with or without RNI) versus no PMRT. OS by RNI was evaluated as a secondary end point for PMRT patients. For BCS patients, the primary end point was OS with breast RT alone versus breast RT plus RNI. BCS patients who did not undergo RT were excluded. To be included in an RT cohort, a minimum of 15 RT treatments were required [9] and <15 were excluded. Survival estimates were generated using the Kaplan Meier method and compared using log-rank and unadjusted Cox models. Multivariate Cox models were adjusted for factors chosen a priori, including RT, age, race, year, comorbidity score, grade, ct stage, in-breast pathologic complete response (path CR), extent of axillary surgery, ypn stage, and estrogen 384 BCS patients treated without radiation (RT) 203 patients treated with <15 RT fractions 5032 Breast-Conserving Surgery (BCS) 3040 Mast-ypN Mast-ypN BCS-ypN BCS-ypN+ No radiation 1078 PMRT 1962 No radiation 1819 PMRT 5424 Breast only RT 1154 Breast and RNI 916 Breast only RT 1337 Breast and RNI 1625 Figure 1. Study design. NCDB, National Cancer Database; RT, radiotherapy; PMRT, postmastectomy radiotherapy; RNI, regional nodal irradiation; ypn, post-chemotherapy pathologic lymph node stage; ypn+, pathologically lymph node-positive; ypn0, pathologically lymph node-negative; Mast, Mastectomy; BCS, breast-conserving surgery. Volume 27 No. 5 May 2016 doi: /annonc/mdw

3 receptor (ER) status/hormone therapy (HT) groups. Because HT decisions and efficacy are dependent on ER status, these variables were combined into four groups for analysis (ER+/HT+, ER+/HT,ER, and unknown ER and/ or HT). HER2 data were inadequate for analyses secondary to unknown status in 60% of patients and the absence of data for HER2-directed therapy. Additional variables with unknown values ( 6% overall by covariate) were analyzed categorically. Sensitivity analyses excluding unknown variables were performed and demonstrated identical patterns of significance to the primary analyses. Propensity score-matched analyses were also carried out in each of the four cohorts evaluating the impact of RT approaches. One-to-one matching without replacement was completed using the nearest neighbor match on the logit of the propensity score for RT approach (derived from age, race, comorbidities, year, ct stage, in-breast response, axillary surgery, ypn stage, grade, ER/HT groupings). The caliper width was 0.05 the standard deviation of the logit of the propensity score. Subgroup analyses evaluated the impact of RT approach by five variables chosen a priori based on prognostic factors for locoregional recurrence (LRR) in National Surgical Adjuvant Breast and Bowel Project (NSABP) clinical trials of NAC [10], including age, in-breast path CR, ct, and ypn stage. The impact of RT by the extent of axillary dissection was also evaluated. Given that the increased NAC utilization in the latter years of this study contributed to a limited follow-up overall [median 39 months (range months); median 41 months among survivors], a sensitivity analysis was carried out restricted to diagnostic years allowing for >5 years of median follow-up. To control for potential immortal time bias in the PMRT cohorts [11], a landmark analysis was carried out restricted to women surviving a minimum of 12 months from diagnosis. results Our search returned cases, including 3040 Mast-ypN0 (19.8%), 7243 Mast-ypN+ (47.3%), 2070 BCS-ypN0 (13.5%), and 2962 BCS-ypN+ (19.3%). Patient characteristic are presented in Table 1. On univariate analysis (Figure 2), OS was improved with PMRT for both the Mast-ypN0 cohort [hazard ratio (HR) = 0.743, 95% confidence interval (CI) , P = 0.019: 5-year OS, 88.3% versus 84.8%) and the Mast-ypN+ cohort (HR = 0.795, 95% CI , P < 0.001: 5-year OS, 74.1% versus 70.9%). No OS differences were observed with the addition of RNI to breast RT in the BCSypN0 cohort (HR = 0.944, 95% CI , P = 0.720: 5-year OS, 91.0% versus 90.0%) or the BCS-ypN+ cohort (HR = 1.152, 95% CI , P = 0.133; 5-year OS, 79.7% versus 82.3%). On multivariate analyses (Table 2), PMRT remained independently associated with improved survival in both mastectomy cohorts (Mast-ypN0: HR = 0.729, 95% CI , P = 0.015; Mast-ypN+: HR = 0.772, 95% CI , P < 0.001). For BCS patients, the addition of RNI to breast RT was not associated with survival in either cohort (BCS-ypN0: HR = 0.969, 95% CI , P = 0.851; BCS-ypN+: HR = 1.037, 95% CI , P = 0.700). Similarly, on subgroup analysis of Mast patients receiving PMRT, no OS differences were observed based on the addition of RNI to chest wall RT (supplementary material, available at online). On propensity score-matched analyses, OS remained improved with PMRT in both the Mast-ypN0 cohort (1039 PMRT versus 1039 no-rt; HR = 0.695, 95% CI , P = 0.014) and the Mast-ypN+ cohort (1787 PMRT versus 1787 no-rt; HR = 0.845, 95% CI , P = 0.015). On propensity score-matched analyses evaluating BCS patients, no significant differences were observed with RNI in the BCS-ypN0 cohort (860 RNI versus 860 no-rni; HR = 1.028, 95% CI , P = 0.880) or the BCS-ypN+ cohort (1244 RNI versus 1244 no-rni; HR = 0.962, 95% CI , P = 0.704). subgroup and secondary analyses Subgroup analyses (Table 3) demonstrated no significant interactions between the survival impact of PMRT or RNI based on age, axillary surgery, ypn stage, or in-breast pathologic response. In the Mast-ypN+ cohort, the magnitude of PMRT benefit was greater for ct3 tumors (interaction P = 0.020); although, patients with ct1 2 tumors also significantly benefited. Forest plots for the survival impact of PMRT for all patients undergoing mastectomy by ypn stage are shown in Figure 3, demonstrating improved OS with PMRT in each pathologic nodal subgroup (ypn0, ypn1, and ypn2 3). On sensitivity analyses restricted to diagnostic years allowing for >5 years of median follow-up ( ) (supplementary material, available at online), an OS advantage on multivariate analyses was redemonstrated with PMRT for Mast-ypN0 (HR = 0.714, 95% CI , P = 0.037) and Mast-ypN+ patients (HR = 0.746, 95% CI , P < 0.001). No OS differences were observed with RNI for either BCS cohort. Among mastectomy patients, a landmark analysis restricted to patients surviving a minimum of 12 months from diagnosis also demonstrated improved OS on multivariate analysis in the Mast-ypN0 (HR = 0.748, 95% CI , P = 0.028) and Mast-ypN+ cohorts (HR = 0.809, 95% CI , P = 0.001). Additionally, to account for the potential impact of unmeasured cardiac effects from RT [12], a subgroup analysis evaluating the impact of RT for right- versus left-sided breast cancers was carried out, which demonstrated no significant interactions between PMRT or RNI and tumor laterality (supplementary material, available at online). discussion For women treated with NAC, the optimal adjuvant RT approach is controversial due, in great extent, to the absence of randomized data evaluating RT in this setting. After NAC, the current National Comprehensive Cancer Network (NCCN) guidelines recommend adjuvant RT decisions based on the pre-nac staging and risk factors [9]. However, the post-nac pathology adds an important layer of prognostic information by allowing for an in vivo response assessment to pre-operative systemic therapy, potentially allowing for de-escalation of therapy in selected patients. Responses to NAC and, in particular, pathologic complete responses in the axillary lymph nodes and breast represent consistent prognostic factors for improved locoregional control and survival [10, 13]. Whether a favorable pathologic response to NAC also represents a predictive factor influencing the relative benefits of PMRT or the addition of RNI to breast RT remains an open question and is the subject of an ongoing cooperative randomized trial (NSABP B-51/RTOG 1304) for women with cn1 disease who convert to ypn0 after NAC (clinicaltrials.gov: NCT ). 820 Rusthoven et al. Volume 27 No. 5 May 2016

4 Volume 27 No. 5 May 2016 doi: /annonc/mdw Table 1. Patient characteristics Mastectomy/ypN0 (3040) Mastectomy/ypN-positive (7243) Breast conservation/ypn0 (2070) Breast conservation/ypn-positive (2962) No PMRT PMRT P No PMRT PMRT P Breast RT Breast and RNI P Breast RT Breast and RNI P ALL Age (years) < (53%) 1127 (57%) (48%) 2818 (52%) (48%) 426 (47%) (44%) 703 (43%) (47%) 835 (43%) 940 (52%) 2606 (48%) 597 (52%) 490 (54%) 746 (56%) 922 (57%) Race White 818 (76%) 1501 (77%) (76%) 4272 (79%) (70%) 665 (73%) (77%) 1194 (74%) Black 192 (17%) 349 (18%) 336 (19%) 847 (16%) 284 (25%) 199 (22%) 243 (18%) 352 (22%) Other 60 (6%) 96 (5%) 74 (4%) 243 (5%) 50 (4%) 42 (5%) 54 (4%) 69 (4%) Unknown 8 (1%) 16 (1%) 23 (1%) 62 (1%) 11 (1%) 10 (1%) 15 (1%) 10 (1%) Year of diagnosis (2%) 41 (2%) (4%) 219 (4%) (4%) 25 (3%) (5%) 73 (5%) (4%) 57 (3%) 101 (6%) 239 (4%) 41 (4%) 33 (4%) 82 (6%) 69 (4%) (5%) 88 (5%) 111 (6%) 293 (5%) 67 (6%) 39 (4%) 101 (8%) 99 (6%) (5%) 106 (5%) 148 (8%) 345 (6%) 79 (7%) 47 (5%) 99 (7%) 102 (6%) (8%) 166 (9%) 184 (10%) 524 (10%) 107 (9%) 88 (10%) 126 (9%) 152 (9%) (13%) 244 (12%) 264 (15%) 702 (13%) 158 (14%) 96 (11%) 167 (13%) 203 (13%) (18%) 313 (16%) 285 (16%) 882 (16%) 186 (16%) 128 (14%) 219 (16%) 284 (18%) (22%) 425 (22%) 332 (18%) 1053 (19%) 217 (19%) 195 (21%) 236 (18%) 260 (16%) (24%) 522 (27%) 323 (18%) 1167 (22%) 254 (22%) 265 (29%) 245 (18%) 383 (24%) Comorbidity score (92%) 1791 (91%) (90%) 4889 (90%) (92%) 818 (89%) (91%) 1464 (90%) (6%) 158 (8%) 157 (9%) 469 (9%) 92 (8%) 86 (9%) 103 (8%) 141 (9%) 2 16 (12%) 13 (1%) 34 (2%) 66 (1%) 6 (1%) 12 (1%) 14 (1%) 20 (1%) Tumor grade 1 43 (4%) 66 (3%) (6%) 391 (7%) < (2%) 21 (2%) (6%) 89 (6%) (26%) 492 (25%) 640 (35%) 2105 (39%) 295 (26%) 220 (24%) 498 (37%) 563 (35%) (62%) 1251 (64%) 958 (53%) 2546 (47%) 767 (67%) 628 (67%) 672 (50%) 883 (54%) Unknown 84 (8%) 153 (8%) 113 (6%) 382 (7%) 67 (6%) 47 (5%) 84 (6%) 90 (6%) ct stage (16%) 188 (9.6%) < (19%) 595 (11%) < (15%) 127 (14%) (19%) 307 (19%) (55%) 882 (45%) 884 (49%) 2492 (46%) 735 (64%) 588 (64%) 876 (66%) 1009 (62%) (29%) 892 (46%) 583 (32%) 2337 (43%) 248 (22%) 201 (22%) 206 (15%) 309 (19%) In-breast response Path CR 440 (41%) 834 (43%) (7%) 273 (5%) < (40%) 447 (49%) < (8%) 113 (7%) Residual 638 (59%) 1128 (58%) 1686 (93%) 5151 (95%) 694 (60%) 469 (51%) 1236 (92%) 1512 (93%) Axillary LN surgery 4 LNs 313 (29%) 519 (27%) (13%) 539 (10%) (39%) 321 (35%) (15%) 226 (14%) LNs 202 (19%) 368 (19%) 314 (17%) 920 (17%) 186 (16%) 153 (17%) 232 (17%) 255 (16%) (50%) 1027 (52%) 1254 (69%) 3871 (71%) 478 (41%) 416 (45%) 885 (66%) 1111 (68%) Unknown 20 (2%) 48 (2%) 20 (1%) 94 (2%) 45 (4%) 26 (3%) 20 (15%) 33 (2%) Continued

5 822 Rusthoven et al. Volume 27 No. 5 May 2016 Table 1. Continued Mastectomy/ypN0 (3040) Mastectomy/ypN-positive (7243) Breast conservation/ypn0 (2070) Breast conservation/ypn-positive (2962) No PMRT PMRT P No PMRT PMRT P Breast RT Breast and RNI P Breast RT Breast and RNI P ypn stage 1 NA NA 1318 (73%) 3186 (59%) <0.001 NA NA 1073 (80%) 1187 (73%) NA NA 338 (19%) 1537 (28%) NA NA 212 (16%) 327 (20%) 3 NA NA 163 (9%) 701 (13%) NA NA 52 (4%) 111 (7%) ER status Negative 565 (52%) 943 (48%) (34%) 1350 (25%) < (51%) 487 (53%) (33%) 521 (32%) Positive 468 (43%) 955 (49%) 1098 (60%) 3800 (70%) 499 (43%) 397 (43%) 822 (62%) 1015 (63%) Unknown 45 (4%) 64 (3%) 111 (6%) 274 (5%) 62 (5%) 32 (4%) 79 (6%) 89 (6%) Hormone therapy (HT) Negative 632 (59%) 991 (51%) < (46%) 1571 (29%) < (55%) 501 (55%) (37%) 566 (35%) Positive 417 (39%) 934 (48%) 917 (50%) 3757 (69%) 494 (43%) 399 (44%) 820 (61%) 1024 (63%) Unknown 29 (3%) 37 (2%) 60 (3%) 96 (2%) 26 (2%) 16 (2%) 22 (2%) 35 (2%) ER/HT groups ER+/HT+ 374 (35%) 850 (43%) < (47%) 3510 (65%) < (38%) 355 (39%) (56%) 944 (58%) ER+/HT 81 (8%) 88 (5%) 200 (11%) 233 (4%) 51 (4%) 35 (4%) 65 (5%) 59 (4%) ER 565 (52%) 943 (48%) 610 (34%) 1350 (25%) 593 (51%) 487 (53%) 436 (33%) 521 (32%) Unknown 58 (5%) 81 (4%) 154 (9%) 331 (6%) 68 (5.9%) 39 (4%) 91 (7%) 101 (6%) Status at last FU Alive 970 (90%) 1811 (92%) 1379 (76%) 4320 (80%) 1062 (92.0%) 854 (93%) 1131 (85%) 1362 (84%) Deceased 108 (10%) 151 (8%) 440 (24%) 1104 (20%) 92 (8.0%) 62 (7%) 206 (15%) 263 (16%) PMRT, postmastectomy radiotherapy; RT, radiotherapy; RNI, regional nodal irradiation; ct stage, clinical T stage; ypn stage, post-therapy pathologic nodal stage; ypn0, pathologically lymph node-negative; ypn+, pathologically lymph node positive; path CR, pathologic complete response; residual, residual invasive disease; Comorbidity, Charlson Deyo Comorbidity Score; ER, estrogen receptor; HT, hormone therapy; LN, lymph node; FU, follow-up.

6 Percent survival Mastectomy (mast) ypn PMRT 50 PMRT 40 No radiation 40 No radiation HR = 0.743, Cl , P = HR = 0.795, Cl , P < Months Months No radiation PMRT Percent survival Breast & RNI Breast conservation (BCS) ypn Breast and RNI 40 Breast RT HR = 0.944, Cl , P = Months Breast RT In this analysis, we report the outcomes for > women in the United States with ct1 3, cn1, M0 breast cancer treated with NAC, surgery, and differing adjuvant RT approaches in the modern era. A consistent OS advantage was observed for women with cn1 disease treated with PMRT, irrespective of the pathologic lymph node response to NAC. Conversely, no significant differences in OS were observed after BCS with the addition of RNI to breast RT. These patterns of significance were also maintained in multivariate and propensity score-matched analyses controlling for important clinical variables such as age, comorbidity score, pre- and post-nac staging, axillary surgery, ER status, and hormone therapy. Clinically, the most significant findings of this analysis are related to the observed survival benefits of PMRT among women with both ypn1 and ypn0 disease. An influential series from MD Anderson evaluated >700 patients treated with NAC and reported improved survival with PMRT limited to patients with 4 involved nodes at the time of surgery [14]. More recently, a pooled analysis of two prospective trials of NAC (NSABP B-18/B-27) provided further insights into recurrence patterns Percent survival Percent survival Mastectomy (mast) ypn+ Breast conservation (BCS) ypn Breast and RNI 40 Breast RT HR = 1.152, Cl , P = Months Figure 2. Kaplan Meier survival curves. RT, radiotherapy; PMRT, postmastectomy radiotherapy; RNI, regional nodal irradiation; ypn, post-chemotherapy pathologic lymph node stage; ypn+, pathologically lymph node-positive; ypn0, pathologically lymph node-negative; Mast, Mastectomy; BCS, breast-conserving surgery. without PMRT after mastectomy and without RNI after BCS [10]. In the NSABP analysis, LRRs decreased with favorable responses to NAC, with 10-year rates for ypn2-3, ypn1, ypn0 (with residual in-breast disease), and ypn0 (with in-breast path CR) of 11% 27%, 6% 21%, 9% 12%, and 0% 9%, respectively. In aggregate, the findings of retrospectives studies have led some authors to individualize PMRT decisions for ypn1 disease in the context of additional risk factors [7], whereas others have recommended PMRT for all women with persistent pathologic nodal disease after NAC [15]. For women with cn+ disease who convert to ypn0, the value of PMRT has been even more uncertain. A separate report from MD Anderson specific to women achieving a path CR after NAC reported a survival benefit of PMRT restricted to those with clinical stage III or higher ( T3N1 or N2 3) [16], whereas other small institutional series have reported no benefit of PMRT for ypn0 patients [17, 18]. The NSABP B-51/RTOG 1304 trial is designed to clarify the role of RT approaches for the ypn0 subgroup and ongoing enrollment should be actively encouraged. While prospective results are awaited, the present analysis offers meaningful Volume 27 No. 5 May 2016 doi: /annonc/mdw

7 824 Rusthoven et al. Volume 27 No. 5 May 2016 Table 2. Multivariate survival analysis Mastectomy/ypN0 (3040) Mastectomy/ypN-positive (7243) Breast conservation/ypn0 (2070) Breast conservation/ypn-positive (2962) HR Low High P HR Low High P HR Low High P HR Low High P Radiation No RT 1 No RT 1 Breast 1 Breast 1 PMRT PMRT <0.001 Br and RNI Br and RNI Age <50 1 <50 1 <50 1 < Race White 1 White 1 White 1 White 1 Black Black <0.001 Black Black Other Other Other Other Unknown Unknown Unknown Unknown Year of Dx Per year Per Year Per Year Per Year Grade < <0.001 Unknown Unknown Unknown Unknown Ax LN Sx 4 LNs 1 4 LNs 1 4 LNs 1 4 LNs Unknown Unknown Unknown Unknown ct stage < ypn stage < < < <0.001 In-breast Path CR 1 Path CR 1 Path CR 1 Path CR 1 Residual <0.001 Residual Residual <0.001 Residual ER/HT ER+/HT+ 1 ER+/HT+ 1 ER+/HT+ 1 ER+/HT+ 1 ER+/HT ER+/HT <0.001 ER+/HT ER+/HT ER <0.001 ER <0.001 ER <0.001 ER <0.001 Unknown Unknown <0.001 Unknown Unknown <0.001 Comorbidity RT, radiotherapy; PMRT, postmastectomy radiotherapy; RNI, regional nodal irradiation; ct stage, clinical T stage; ypn stage, post-therapy pathologic nodal stage; ypn0, pathologically lymph node negative; ypn +, pathologically lymph node positive; path CR, pathologic complete response; residual, residual invasive disease; ER, estrogen receptor; HT, hormone therapy; LN, lymph node; Comorbidity, Charlson Deyo Comorbidity Score.

8 Table 3. Overall survival impact of PMRT and RNI by subgroup Mastectomy/ypN0 N (Events) N (Events) N (Events) Multivariate analysis Interaction P Total PMRT No RT HR w/pmrt Low High P Axillary surgery 4 LNs 832 (68) 519 (42) 313 (26) LNs 570 (53) 368 (33) 202 (20) LNs 1570 (135) 1027 (74) 543 (61) In-breast response Path CR 1274 (61) 834 (37) 440 (24) Residual 1766 (198) 1128 (114) 638 (84) Clinical T stage ct (124) 1070 (58) 764 (66) ct (135) 892 (93) 314 (42) Age < (115) 1127 (69) 569 (46) (144) 835 (82) 509 (62) Mastectomy/ypN+ Total PMRT No RT HR w/pmrt Low High P Interaction P Axillary surgery 4 LNs 770 (142) 539 (98) 231 (44) LNs 1234 (256) 920 (184) 314 (72) LNs 5125 (1121) 3871 (805) 1254 (316) yp Nodal stage ypn (736) 3186 (493) 1318 (243) ypn (808) 2238 (611) 501 (197) <0.001 In-breast response Path CR 406 (58) 273 (19) 133 (39) Residual 6837 (1486) 5151 (421) 1686 (1065) <0.001 Clinical T stage ct (797) 3087 (559) 1236 (238) ct (747) 2337 (545) 583 (202) <0.001 Age < (704) 2818 (517) 879 (187) (840) 2606 (587) 940 (253) <0.001 BCS/ypN0 Total Breast + RNI Breast only RT HR w/rni Low High P Interaction P Axillary surgery 4 LNs 766 (56) 321 (24) 445 (32) LNs 339 (27) 153 (11) 186 (16) LNs 894 (61) 416 (25) 478 (36) In-breast response Path CR 907 (38) 447 (20) 460 (18) Residual 1163 (1163) 469 (72) 694 (44) Clinical T stage ct (104) 715 (44) 906 (60) ct3 449 (50) 201 (18) 248 (32) Age < (74) 426 (44) 557 (30) (80) 490 (48) 597 (32) BCS/ypN+ Total Breast + RNI Breast only RT HR w/rni Low High P Interaction P Axillary surgery 4 LNs 426 (60) 226 (24) 200 (36) LNs 487 (66) 255 (41) 1117 (25) LNs 1996 (334) 1111 (191) 20 (143) yp Nodal stage ypn (285) 1187 (145) 1073 (140) ypn (184) 438 (118) 264 (66) In-breast response Path CR 214 (27) 113 (17) 101 (10) Residual 2748 (442) 1512 (246) 1236 (196) Continued Volume 27 No. 5 May 2016 doi: /annonc/mdw

9 Table 3. Continued BCS/ypN+ N (events) N (events) N (events) Multivariate analysis Interaction P Total Breast + RNI Breast only RT HR w/rni Low High P Clinical T stage ct (359) 1316 (190) 1131 (169) ct3 515 (110) 309 (73) 206 (37) Age < (194) 703 (114) 591 (80) (275) 922 (149) 746 (126) Hazard ratios (HR) for mastectomy patients correspond to the mortality hazard with PMRT on multivariate analysis ( No RT is the reference with HR = 1). HRs for BCS patients correspond to the mortality hazard with Breast + RNI on multivariate analysis ( Breast Only RT is the reference with HR = 1). Multivariate analyses were adjusted for identical factors to the primary analysis listed in the Materials and Methods section. Interaction P < 0.05 indicates significant heterogeneity in the impact of RT approaches within subgroups of a given variable. Mastectomy-ypN0 cohort ypn0 HR = 0.739, Cl , P = Mastectomy-ypN+ cohort ypn1 ypn2 3 support to PMRT for cn1 patients after NAC irrespective of ypn stage, with the observation of improved survival in large cohorts of each pathologic nodal subgroup. The addition of RNI to breast RT or chest wall RT was not a significant prognostic factor for OS among cn1 women treated with NAC in this analysis. Our collective understanding of the impact of RNI in the setting of upfront surgery has been informed recently by two randomized trials of RNI for cn1 and high-risk cn0 disease from the National Cancer Institute of Canada (NCIC) [6] and European Organisation for Research and Treatment of Cancer (EORTC) [5]. All women in the NCIC trial were treated with BCS, whereas in the EORTC trial three fourths were treated with BCS and one-fourth with mastectomy. In the NCIC trial, RNI was associated with improved regional control, metastasis-free survival, and disease-free survival, but this did not translate into a significant OS advantage at 10 years. The EORTC demonstrated similar improvements in locoregional and distant control, as well as a significant 2% improvement in breast cancer mortality and a near-significant 1.6% (P = 0.06) improvement in OS at 10 years. With regard to OS as a primary end point, the results of the present analysis may be viewed as generally consistent with these and other randomized trials evaluating RNI strategies after upfront surgery [5, 6, 19], given the relative difficulty in demonstrating a clear OS advantage with RNI even with rigorous prospective trial designs and HR = 0.835, Cl , P =0.026 HR = 0.678, Cl , P <0.001 Interaction P = Multivariate hazard ratio Survival improved with PMRT Figure 3. Forest plot: survival impact of PMRT by ypn stage. PMRT, postmastectomy radiotherapy; ypn, post-chemotherapy pathologic lymph node stage; ypn+, pathologically lymph node-positive; ypn0, pathologically lymph node-negative extended follow-up. Importantly, data regarding more granular intermediate oncologic end points (e.g. local, regional, and distant control, and disease-free survival), representing meaningful outcomes for women treated with curative intent that were consistently improved with RNI in the NCIC and EORTC trials [5, 6], were not evaluable in this NCDB analysis. In addition to OS by RT approaches, this analysis provides valuable data regarding prognostic factors and patterns of care for women with cn1 breast cancer treated with NAC in a large registry cohort (supplementary Discussion, available at Annals of Oncology online). This retrospective analysis has several important limitations. All analyses are subject to selection bias and imbalances in unquantified variables. The median follow-up was limited owing, in large part, to increasing NAC in the latter years of this study. Although survival curve separation may be observed as early as 2 3 years after PMRT in trials involving upfront surgery [1, 2], similar differences may not be apparent until 8 10 years, if present, after RNI [5]. To address this limitation, a sensitivity analysis was carried out with >5 years of median follow-up in each cohort, which demonstrated similar findings to the primary analysis. Details regarding RNI fields and techniques, locoregional control, and disease-free survival were unavailable. Similarly, a potential dilution of the impact of RNI due to the coverage of axillary levels I and II with standard or high- 826 Rusthoven et al. Volume 27 No. 5 May 2016

10 tangent RT in the no-rni cohorts, as observed in randomized trials of axillary management [20], would also apply to this analysis. Overall, the RT outcomes in this analysis, particularly in the setting of RNI, must be interpreted in the context of absent RT quality control data. Data regarding the specific chemotherapy and hormone therapies administered were unavailable, and inadequate data on HER2 and the absence targeted therapy data precluded comment on the relative impact of RT approaches according to HER2 or triple-negative molecular status. The general nature of the available systemic therapy data and insufficient HER2 data in the NCDB represent important limitations of this analysis, as systemic therapy regimens, molecular categorizations of breast cancer, and targeted therapies may directly impact pathologic down-staging, disease recurrence, and survival outcomes [13, 21, 22]. conclusions In this analysis, we present the largest reported evaluation of RT approaches for women with cn1 breast cancer treated with NAC. After mastectomy, a significant OS advantage was observed with PMRT for all pathologic nodal subgroups. No OS benefit was observed with the addition of RNI to breast RT; although, potential differences in locoregional control and disease-free survival were not evaluable. disclosure The authors have declared no conflicts of interest. references 1. Overgaard M, Hansen PS, Overgaard J et al. Postoperative radiotherapy in highrisk premenopausal women with breast cancer who receive adjuvant chemotherapy. N Engl J Med 1997; 337: Ragaz J, Jackson SM, Le N et al. Adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 1997; 337: Effect of radiotherapy after mastectomy and axillary surgery on 10-year recurrence and 20-year breast cancer mortality: meta-analysis of individual patient data for 8135 women in 22 randomised trials. Lancet 2014; 383: Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of radiotherapy and of differences in the extent of surgery for early breast cancer on local recurrence and 15-year survival: an overview of the randomised trials. Lancet 2006; 366: Poortmans PM, Collette S, Kirkove C et al. Internal mammary and medial supraclavicular irradiation in breast cancer. N Engl J Med 2015; 373: Whelan TJ, Olivotto IA, Parulekar WR et al. Regional nodal irradiation in early-stage breast cancer. N Engl J Med 2015; 373: Fowble BL, Einck JP, Kim DN et al. Role of postmastectomy radiation after neoadjuvant chemotherapy in stage II-III breast cancer. Int J Radiat Oncol Biol Phys 2012; 83: Mougalian SS, Soulos PR, Killelea BK et al. Use of neoadjuvant chemotherapy for patients with stage I to III breast cancer in the United States. Cancer 2015; 121: Gradishar WJ, Anderson BO, Balassanian R et al. Breast cancer version J Natl Compr Canc Netw 2015; 13: Mamounas EP, Anderson SJ, Dignam JJ et al. Predictors of locoregional recurrence after neoadjuvant chemotherapy: results from combined analysis of National Surgical Adjuvant Breast and Bowel Project B-18 and B-27. J Clin Oncol 2012; 30: Park HS, Gross CP, Makarov DV, Yu JB. Immortal time bias: a frequently unrecognized threat to validity in the evaluation of postoperative radiotherapy. Int J Radiat Oncol Biol Phys 2012; 83: Darby SC, Ewertz M, McGale P et al. Risk of ischemic heart disease in women after radiotherapy for breast cancer. N Engl J Med 2013; 368: Rastogi P, Anderson SJ, Bear HD et al. Preoperative chemotherapy: updates of national surgical adjuvant breast and bowel project protocols B-18 and B-27. J Clin Oncol 2008; 26: Huang EH, Tucker SL, Strom EA et al. Postmastectomy radiation improves localregional control and survival for selected patients with locally advanced breast cancer treated with neoadjuvant chemotherapy and mastectomy. J Clin Oncol 2004; 22: Mak KS, Harris JR. Radiotherapy issues after neoadjuvant chemotherapy. J Natl Cancer Inst Monogr 2015; 2015: McGuire SE, Gonzalez-Angulo AM, Huang EH et al. Postmastectomy radiation improves the outcome of patients with locally advanced breast cancer who achieve a pathologic complete response to neoadjuvant chemotherapy. Int J Radiat Oncol Biol Phys 2007; 68: Le Scodan R, Selz J, Stevens D et al. Radiotherapy for stage II and stage III breast cancer patients with negative lymph nodes after preoperative chemotherapy and mastectomy. Int J Radiat Oncol Biol Phys 2012; 82: e1 e Shim SJ, Park W, Huh SJ et al. The role of postmastectomy radiation therapy after neoadjuvant chemotherapy in clinical stage II-III breast cancer patients with pn0: a multicenter, retrospective study (KROG 12-05). Int J Radiat Oncol Biol Phys 2014; 88: Hennequin C, Bossard N, Servagi-Vernat S et al. Ten-year survival results of a randomized trial of irradiation of internal mammary nodes after mastectomy. Int J Radiat Oncol Biol Phys 2013; 86: Jagsi R, Chadha M, Moni J et al. Radiation field design in the ACOSOG Z0011 (Alliance) Trial. J Clin Oncol 2014; 32: Early Breast Cancer Trialists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005; 365: Smith I, Procter M, Gelber RD et al. 2-year follow-up of trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer: a randomised controlled trial. Lancet 2007; 369: Volume 27 No. 5 May 2016 doi: /annonc/mdw