Adjuvant Hormonal Therapy for Premenopausal Women Otto Metzger, MD Dana-Farber Cancer Institute Harvard Medical School Boston, USA

Chemotherapy-induced amenorrhea OFR ~10% Menopausal status should be defined prior to the initiation of adjuvant chemotherapy

Plan of the Talk Tamoxifen versus AI: Is there room for additional discussion in 2017? What is the real gain with extended endocrine therapy? Do we need molecular tests to identify patients at risk of recurrence beyond year 5?

Plan of the Talk Tamoxifen versus AI: Is there room for additional discussion in 2017? What is the real gain with extended endocrine therapy? Do we need molecular tests to identify patients at risk of recurrence beyond year 5?

Aromatase Inhibitor Overview Group Tamoxifen versus AI Recurrence Breast Cancer Mortality The Lancet 2015 386, 1341-1352DOI: (10.1016/S0140-6736(15)61074-1)

Aromatase Inhibitor Overview Group Tamoxifen AI versus AI Recurrence Breast Cancer Mortality The Lancet 2015 386, 1341-1352DOI: (10.1016/S0140-6736(15)61074-1)

Aromatase Inhibitor Overview Group Subgroup Analysis The Lancet 2015 386, 1341-1352DOI: (10.1016/S0140-6736(15)61074-1)

So Why Bother? Should we just give upfront AI or sequential strategy to all patients? My personal consideration: AI is not tolerated for a significant proportion of patients with discontinuation rates as high as 40% reported in clinical trials We can safely select a subgroup of patients that can be treated with adjuvant tamoxifen So, how can we identify a subset of patients that can be treated with upfront tamoxifen?

BIG 1-98 Analytic Cohort 12-year update (Lancet Oncol 2011) 2-Arm Option Tamoxifen Letrozole 2 5 YEARS A B C D 4-Arm Option N=1,828 Enrolled 1998-2000 N=3,094 1999-2003 N=4,922 N=911 N=917 N=1548 N=1546 Is letrozole superior than TAM in the subsets of LA and LB disease? Is there a preferred regimen for patients diagnosed with invasive lobular carcinoma 10

Disease-free survival

Disease-free survival

Multivariate Analysis Cox Model for DFS (IPCW)1 0.5 1.0 1.5 Lobular LB Lobular LA Ductal LB Ductal LA Hazard Ratio (95% CI) 0.95 (0.76-1.20) 0.64 (0.53-0.79) 0.49 (0.32-0.78) 0.33 (0.21-0.55) Favors Letrozole Favors Tamoxifen 1. Included variables: age, tumor size, nodal status, histological grade, histology, local therapy, subtype (LA/LB) and treatment Treatment by histology (ductal/lobular), p=0.006 Treatment by subtype (LA/LB), p=0.01 Interactions

What be learned from the sequential arms of BIG-198? What is the impact of tumor cell proliferation on the benefit of AI versus TAM? What be learned from the sequential arms of BIG-198? 2-Arm Option Tamoxifen Letrozole 2 5 YEARS A B C D 4-Arm Option N=6,182 Enrolled 1999-2003 N=1548 N=1540 N=1548* N=1546 *612 patients (40%) received letrozole after the tamoxifen arm was unblinded. This presentation is the intellectual property of the author/presenter. Contact ottto_metzger@dfci.harvard.edu for permission to reprint

Invasive Ductal Carcinoma Disease-free survival by LA and LB subtypes LA-like Ductal LB-like Ductal This presentation is the intellectual property of the author/presenter. Contact ottto_metzger@dfci.harvard.edu for permission to reprint

How can we use this information in our clinical practice? Points for consideration: Classic clinico-pathologic features can be used to identify indolent tumors that could be treated with upfront tamoxifen There might be a potential for the use of genomic tests in the identify “indolent” tumors

Very high and very low risk levels are informative Ultralow Threshold

Stockholm STO 3 phase III postmenopausal N0 <3 cm validates ultralow risk (IDLE, indolent lesions of epithelial origin) SABCS 2016 PD7-01  

TAILORx trial first report: N0, ER+, HER2-negative 21 gene Recurrence Score <11, endocrine therapy alone Sparano J et al. NEJM 2015

9-Years Risk of Distant Recurrence (%) 10 20 30 40 50 60 70 80 90 100 9-Years Risk of Distant Recurrence (%) 5 15 25 35 45 Recurrence Score Node Negative 1-3 Positive Nodes 4+ Positive 95%CI Mean TransATAC Low incidence of distant recurrence at 9 years for those with very low RS levels even in the subset of 1-3 positive nodes

Node positivity does not necessarily imply the need for upfront AI Metzger et al. Unpublished data

Plan of the Talk Tamoxifen versus AI: Is there room for additional discussion in 2017? What is the real gain with extended endocrine therapy? Do we need molecular tests to identify patients at risk of recurrence beyond year 5?

Extended Endocrine Therapy 1. Goss PE et al, J Natl Cancer Inst 2005;97:1262–71. 2. Mamounas EP et al., J Clin Oncol 2008;26:1965-1971. 3. Jakesz et al., J Natl Cancer Inst. 2007 Dec 19;99(24):1845-53. 4. Davies C et al., Lancet. 2013 ;381(9869):805-16. 5. Gray et al., J Clin Oncol 31, 2013 (suppl; abstr 5). 6. Goss PE et al., N Engl J Med. 2016

(1.9% absolute benefit, P=0.03) New Data From SABCS 2016 Results from 3 extended AI randomized studies presented at SABCS 2016 In all 3 studies, primary analyses demonstrated no statistically significant benefit in DFS from extending AI therapy in post-menopausal patients. Of note, results from B-42 and DATA were generally similar to previous extended endocrine therapy trials (~3- 4% absolute benefit) NSABP B-42 DATA IDEAL Population 3966 patients who completed 5 years of AI or up to 3 years TAM followed by AI (for a total of 5 years) 1912 patients who completed 2-3 years of Tamoxifen 1824 patients who completed 5 years of Tamoxifen Treatment 5y AI vs Placebo 3y AI vs 6y AI 2.5y AI vs 5y AI HR 0.85 (0.73 – 0.999) 0.79 (0.62-1.02) 0.96 (0.76-1.20) DFS 84.7% (5 years AI) vs 81.3% (Placebo) 83.1% (6 years AI) vs 79.4% (3 years AI) 87.9% (5 years AI) vs 88.4% (2.5 years AI) P value P=0.048 (n.s.) P=0.07 (n.s.) P=0.7 (n.s.) Note, there was a statistically significant benefit in terms of prevention of distant recurrence (1.9% absolute benefit, P=0.03) Discontinuation of AIs in all these trials was ~40% in the extended period 1. Mamounas E et al, 2016 SABCS. 2. Tjan-Heijnen VC et al, 2016 SABCS. 3. Block EJ et al, 2016 SABCS

NSABP B-42 or Stratification: Postmenopausal Pts with ER+ or PR+ Breast Cancer Stage I, II, or IIIa invasive BC at diagnosis Disease-free After 5 Years of Endocrine Therapy or AI X 5 yrs TAM X < 3 yrs  AI to Complete 5 yrs Stratification: Pathological nodal status (Negative, Positive) Prior adjuvant TAM (Yes, No) Lowest BMD T score: spine, hip, femur (>-2.0, ≤ -2.0 SD) R Letrozole X 5 yrs Placebo X 5 yrs

NSABP B-42: Disease-Free Survival Letrozole Placebo 100 80 60 40 20 0 2 4 6 7 8 Years After Randomization Disease-Free Survival Letrozole 1959 1813 1644 1225 216 Placebo 1964 1814 1639 1208 210 HR=0.85 (0.73-0.999) P = 0.048 84.7% 81.3% # Events 292 339 *P-value did not reach statistical significance level of 0.0418

NSABP B-42: Cumulative Incidence of Distant Recurrence Letrozole Placebo Cumulative Incidence of DR 12 10 8 6 4 2 0 2 4 6 7 8 5.8% 3.9% HR=0.72 (0.53-0.97) P=0.03 Years After Randomization # Pts # Events 73 1964 102

DATA Study Design 6 years anastrozole 3 years anastrozole Postmenopausal at randomization ER+ and/or PR+ No sign of disease recurrence M0 breast cancer After 2-3 years adjuvant tamoxifen 6 years anastrozole 1 mg daily Stratification Nodal status ER/PR status HER2 status Tamoxifen duration 3 years anastrozole 1 mg daily 80% power to detect an increase in 3-year adapted Disease-Free Survival (aDFS) from 90% to 94%, i.e., a hazard ratio (HR) of 0.60 and a significance level of 0.05 Accounting for 10% drop-out: 950 patients per group to be included (n=1912 patients actually included) Minimum follow-up: ≥6 years after randomization, i.e., ≥ 3 years of adapted follow-up (last patient included in August 2009)

adapted Disease-Free Survival (aDFS) San Antonio Breast Cancer Symposium, December 6-10, 2016 adapted Disease-Free Survival (aDFS) N=1660 6-year Anastrozole (N=827) 3-year Anastrozole (N=833) 5-y aDFS (%) 83.1 79.4 HR (95% CI) 0.79 (0.62 – 1.02) P-value 0.07 3-year aDFS: 90.7% vs. 88.9%

aDFS if ER+ and PR+, HER2-, pN+, Chemo+ San Antonio Breast Cancer Symposium, December 6-10, 2016 aDFS if ER+ and PR+, HER2-, pN+, Chemo+ N=597 6-year Anastrozole (N=293) 3-year Anastrozole (N=286) 5-y aDFS (%) 86.0 75.9 HR (95% CI) 0.58 (0.39 – 0.89) P-value 0.01

IDEAL Study DFS 5 years HR: 0.96 (0.76-1.20) P-value: 0.70 2.5y: 88.4%

Plan of the Talk Tamoxifen versus AI: Is there room for additional discussion in 2017? What is the real gain with extended endocrine therapy? Do we need molecular tests to identify patients at risk of recurrence beyond year 5?

NODAL STATUS IS AN IMPORTANT PROGNOSTIC INDICATOR Saphner et al. JCO 1996

What is the risk of distant recurrence for T1 N0 and T2 N0 ? H.Pan for EBCTCG, ASCO 2106

H.Pan for EBCTCG, ASCO 2106

H.Pan for EBCTCG, ASCO 2106

Limitations of Oxford Overview Data Old studies Lack of detailed information on biomarker data Larger clinico-pathologic variables (?) Can we do better in identifying those less likely to recur using classic clinico-pathologic variables alone?

Preliminary Results from ATAC – Clinical Treatment Score Node: 0 = Negative 1 = 1 positive 2 = 2-3 positive 3 = 4-9 positive 4 = >9 positive Size: Continuous (if >50 then = 50) Grade: 0 = Grade 1 1 = Grade 2 2 = Grade 3 Endocrine Rx 0 = Tamoxifen 1 = Anastrozole Age: Low risk <1% DR/yr; intermediate risk 1-2% DR/yr; high risk >2%/yr 25 50 75 100 Distant recurrence free (%) 1157 938 853 747 648 318 High 1223 1047 987 893 803 426 Intermediate 1959 1757 1674 1581 1449 777 Low Number at risk 5 6 7 8 9 10 Follow-up time [years] 2.4% 7.5% 20.6% Hazard ratio Reference 3.40 (2.31-5.00) 9.39 (6.65-13.28) DRs 5-10yr 39 77 181 Validity being assessed with IBCSG BIG 1-98 Courtesy from Mitch Dowsett

Can we take advantage of genomic signatures to identify those at risk of disease recurrence beyond year 5? Can we use genomic tests to identify those more likely to benefit from extended endocrine therapy?

Smoothed hazard rates for RS, IHC4 and ROR in TransATAC over 10 years (Node negative and positive combined) 50% high risk (or above the median) and 50% low risk as defied by each of the score. Sestak et al 2013, JNCI, 105, 1504-11

What is the additional prognostic value of signatures to clinical variables? Information included Clinical Treatment Score (CTS) Nodal status, grade, tumour size, age, treatment Immunohistochemical markers (IHC4) ER, PgR, Ki67, HER2 Oncotype Recurrence Score (RS) 21 genes (oestrogen, proliferation, invasion, HER2 genes) Breast Cancer Index (BCI) H/I and 5 proliferation genes (Molecular Grade Index) Prosigna (ROR) 46 genes, proliferation score, tumour size (EU cut-offs from transATAC for N- and N+) EndoPredict (EPclin) 12 genes (proliferation, differentiation, oestrogen); nodal status and tumour size Sestak et al SABCS 2016

Prognostic value years 5-10 – node-negative (n=591; 34 events) Likelihood Ratio χ2 EPclin IHC4 CTS ROR RS BCI Likelihood Ratio ∆χ2 % Improvement 20.0% 67.5% 11.4% 111.0% 62.0% *

Prognostic value years 5-10 – node-positive (n=227; 31 events) Likelihood Ratio χ2 EPclin IHC4 CTS ROR RS BCI Likelihood Ratio ∆χ2 % Improvement 7.5% 11.3% 6.9% 25.6% 27.5% *

Do we have a Genomic Tool able to Predict the benefit of Endocrine Therapy?

BCI Predictive (H/I) Results: MA.17 RCT Cohort In patients with High H/I, extended letrozole reduced recurrence rate significantly from 27% to 10.5% (P=0.007) No significant reduction in patients with Low H/I (P=0.35) P=0.007 16.5% absolute benefit P=0.35 No significant benefit Recurrences (%) Sgroi et al, J Natl Cancer Inst. 2013;105:1036-1042

Summary of BCI Predictive (H/I) Validation Data H/I shown to be a significant predictor of endocrine benefit in 3 randomized trial cohorts Study Cohort Treatment Predictive analysis Interaction P value Stockholm (n=600)1 Adjuvant tamoxifen vs untreated H/I High HR: 0.35 (0.19-0.65); p=0.0005 H/I Low HR: 0.67 (0.36-1.24), p=0.2 0.003 TransATAC (n=665)2 Adjuvant anastrozole vs tamoxifen H/I High HR: 0.51 (0.27-0.97); p=0.04 H/I Low HR: 1.33 (0.65-2.71), p=0.4 0.004 MA.17 (n=249)3 Extended letrozole vs placebo H/I High OR: 0.33 (0.15-0.73); p=0.006 H/I Low OR: 0.58 (0.25-1.36), p=0.21 0.03 Results suggest generalizability as an endocrine response biomarker 1. Zhang Y, et al. Clin Cancer Res. 2013;19(15):4196-205. 2. Sgroi D, et al. Lancet Oncol. 2013 Oct;14(11):1067-76. 3. Sgroi et al, J Natl Cancer Inst. 2013;105:1036-1042

Take Home Messages Tumor burden (i.e. tumor size and nodal status) reliably identify those more likely to recur between years 0-5 and 5-10 Genomic tools add prognostic value to clinical variables particularly in the node-negative subset We should be careful when evaluating the additional value of genomic signatures if the “genomic algorithms” include tumor size and nodal status BCI provides predictive information as an endocrine response biomarker

Conclusions Adjuvant Tamoxifen for low-risk patients may be as good as AI High-risk patients (defined by tumor burden) are the ones more likely to benefit from 10 years of adjuvant endocrine therapy BCI predictive value should be explored in a prospective and large phase III clinical trial focused on patients with intermediate to low tumor burden for which the value of extended endocrine therapy is not so clear