Polley, Mei‐Yin C.

591 Background: Both Cb and Bev demonstrate activity when combined with standard chemotherapy in TNBC. CALGB 40603 is a 2x2 randomized trial that previously demonstrated that adding Cb to NACT significantly increased pathologic complete responses in the breast/axilla (pCR), while adding Bev did not (Sikov, JCO 2015). Here we report 5-year LTOs and assess factors that influenced them. Methods: 443 patients with clinical stage II-III previously untreated TNBC received 12 weeks of paclitaxel (wP) +/- Cb then dose-dense AC, +/- Bev before surgery. The primary endpoint was pCR. Analyses of LTOs (event-free survival (EFS), distant recurrence-free interval (DRFI) and overall survival (OS)), impact of residual cancer burden and other variables were secondary. Results: Median follow-up was 5.7 years (y); 5y EFS was 70.9% (95% CI; 66.7%-75.4%), DRFI 76.3% (72.3%-80.5%) and OS 76.9% (72.9%-81.2%). Pretreatment clinical stage and achieving pCR correlated with LTOs, while age, race, subtype (basal-like vs. not) and tumor grade did not. Among pCR 5y EFS was 86.4% vs. 57.5% for non-pCR (HR 0.28, 0.19-0.43), OS was 88.7% vs 66.5% (HR = 0.28, 0.17-0.44). This relationship was similar in all trial arms. Any residual disease conferred poorer outcome; compared with pCR/Residual Cancer Burden (RCB) 0, EFS HRs were 2.29 (1.32-3.97), 3.01 (1.90-4.74), and 9.67 (5.66-16.51) for RCBI, II and III, respectively. There were no improvements in LTOs with Cb (EFS HR 0.99, 0.70-1.40) or Bev (EFS HR 0.91, 0.64-1.29). In an exploratory analysis, receipt of ≥11 doses of wP was associated with better EFS (HR 1.92, 1.33-2.77); this was particularly notable in Cb-treated arms. Conclusions: As expected, regardless of treatment arm pCR was associated with markedly better LTOs, and pts with any residual disease had significantly worse outcomes. The addition of Cb or Bev to standard NACT for TNBC did not improve LTOs in this trial, although it should be noted that the trial was not powered for this endpoint. Omission of chemotherapy doses may result in poorer outcomes, especially among Cb-treated pts, which may warrant further evaluation. Support: U10CA180821; U10CA180882; Genentech; https://acknowledgments.alliancefound.org ; NCT00861705 Clinical trial information: NCT00861705.

PURPOSE Alliance/CALGB 50303 (NCT00118209), an intergroup, phase III study, compared dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (DA-EPOCH-R) with standard rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) as frontline therapy for diffuse large B-cell lymphoma. PATIENTS AND METHODS Patients received six cycles of DA-EPOCH-R or R-CHOP. The primary objective was progression-free survival (PFS); secondary clinical objectives included response rate, overall survival (OS), and safety. RESULTS Between 2005 and 2013, 524 patients were registered; 491 eligible patients were included in the final analysis. Most patients (74%) had stage III or IV disease; International Prognostic Index (IPI) risk groups included 26% IPI 0 to 1, 37% IPI 2, 25% IPI 3, and 12% IPI 4 to 5. At a median follow-up of 5 years, PFS was not statistically different between the arms (hazard ratio, 0.93; 95% CI, 0.68 to 1.27; P = .65), with a 2-year PFS rate of 78.9% (95% CI, 73.8% to 84.2%) for DA-EPOCH-R and 75.5% (95% CI, 70.2% to 81.1%) for R-CHOP. OS was not different (hazard ratio, 1.09; 95% CI, 0.75 to 1.59; P = .64), with a 2-year OS rate of 86.5% (95% CI, 82.3% to 91%) for DA-EPOCH-R and 85.7% (95% CI, 81.4% to 90.2%) for R-CHOP. Grade 3 and 4 adverse events were more common ( P < .001) in the DA-EPOCH-R arm than the R-CHOP arm, including infection (16.9% v 10.7%, respectively), febrile neutropenia (35.0% v 17.7%, respectively), mucositis (8.4% v 2.1%, respectively), and neuropathy (18.6% v 3.3%, respectively). Five treatment-related deaths (2.1%) occurred in each arm. CONCLUSION In the 50303 study population, the more intensive, infusional DA-EPOCH-R was more toxic and did not improve PFS or OS compared with R-CHOP. The more favorable results with R-CHOP compared with historical controls suggest a potential patient selection bias and may preclude generalizability of results to specific risk subgroups.

533 Background: Current guidelines define TNBC as complete absence of estrogen (ER) and progesterone receptor (PR), without HER2 amplification. However, the prognostic impact of clinical and histopathological factors, sTILs, and AdjCT in TNBC meeting these strict criteria is unknown. Methods: From a cohort of 9985 women who underwent upfront surgery for M0 breast cancer (BC) at Mayo Clinic Rochester from 1985-2012, 1159 pts with ER negative or low (≤10%) BC were identified for central ER/PR/HER2 staining and HER2 FISH (IHC2+ only) to select those with TNBC by modern definitions. Cox proportional hazards models were used to assess the impact of clinicopathological variables on invasive disease-free (IDFS) and overall survival (OS). Results: Tumors from 605 pts (median age 56.3 yrs) met criteria for TNBC (ER < 1%, PR < 1% and HER2 0, 1 or 2+ and FISH negative). 51% were T1, 65% N0, 88% grade 3, and 75% had Ki67 > 15%. Histologically, 39% were anaplastic, 26% invasive ductal, 16% medullary, 8% metaplastic, 6% apocrine and 5% others. Median sTILs was 20% (0-90%). 55% pts received AdjCT [21% anthracycline (A), 19% A + taxane, and 15% other]. Median follow-up for IDFS and OS were 7.4 and 10.6 yrs, respectively. Multivariate analyses demonstrated that higher N stage (p < 0.01), lower sTILs (p = 0.01) and no AdjCT (p < 0.01) were independently associated with worse IDFS and OS. Histology (medullary subtype) was associated with better IDFS in univariate (HR 0.56, 95% CI, 0.35-0.89) but not in multivariate analyses, once sTILs were accounted for. Among systemically untreated pts (n = 182), higher N (p < 0.01) and lower sTILs (p = 0.04) were associated with worse IDFS. For systemically untreated T1N0 pts (n = 111), the 5-yr IDFS was 70% (95% CI, 61-81) [T1a: 83% (95% CI, 63-100), T1b: 68% (95% CI, 52-88) and T1c: 67% (95% CI, 55-83)], compared to 78% (95% CI, 68-84) for T1N0 pts treated with AdjCT. Conclusions: In TNBC pts, N stage, sTILs and receipt of AdjCT were independently prognostic for IDFS and OS. sTILs remained prognostic for IDFS in systemically untreated TNBC. In N0 TNBC, the risk of recurrence or death was substantial in the absence of chemotherapy, even for those with T1 tumors.

TPS17 Background: Emerging studies suggest that breast cancer, particularly triple negative breast cancer (TNBC), may be sensitive to immunotherapy. However, the response rate of single agent immune checkpoint blockade agent in TNBC is rather low. Previous genomic study in residual tumor after neoadjuvant chemotherapy showed inverse correlations between MEK activation signature and the amount of tumor infiltrating lymphocytes (TILs) in residual disease samples as well as poor outcome. Preclinical study also showed that the combination of MEK inhibitor and anti-PD-L1 antibody in mouse model can eradicate TNBC tumors. Methods: This is a single arm, Phase I/II trial of Pembrolizumab (P) in combination with Binimetinib (B) in patients with unresectable locally advanced or metastatic TNBC. This trial is currently opened for accrual at Mayo Clinic in Florida and Minnesota. Patients with TNBC defined as ER ≤ 10% and PR ≤ 10% who received ≤ 3 prior lines with measurable disease will be enrolled. The primary objective of the Phase I part is to determine the maximum tolerated dose of B in combination with P and for the Phase II part is objective response rate (ORR) by RECIST criteria. The secondary endpoints include ORR by irRECIST, progression free survival, and overall survival. The total sample size is 15-38 patients with 6-12 patients in Phase I with 2 dose levels and 9-26 patients in Phase II. Simon’s Two-Stage Optimal Design is used to test the null hypothesis that this two-drug combination has an ORR of at most 15% vs. the alternative hypothesis that it has an ORR of at least 35%. Patients will receive single agent B for 2 weeks prior to starting P. A mandatory biopsy will be performed before starting B and an optional biopsy will be performed after 2 weeks of B. Tumor tissue will be evaluated for the amount and phenotypes of TILs, PD-L1 expression, and gene expression analysis using PanCancer Immune Profiling Panel, and PDJ amplification. Peripheral blood will be evaluated for circulating immunoregulatory cells, cytokine profiling, circulating tumor cells (CTCs), as well as p-ERK and PD-L1 expression on CTCs. Clinical trial information: NCT03106415.

5 Background: Triple-negative breast cancer (TNBC) is an immunogenic breast cancer subtype, with a greater percentage of tumors containing tumor-infiltrating lymphocytes (TILs) compared to non-TNBC. TILs are prognostic for recurrence and predictive for chemotherapy benefit in TNBC, but not in estrogen receptor (ER)-positive tumors. A subset of TNBC expresses the androgen receptor (AR), and resembles ER-positive breast cancer in terms of demographics (older age) and clinical course (later recurrences that tend to involve bone). However, little is known regarding the association between TILs and AR expression in TNBC. Methods: From a cohort of 9982 women with surgically-treated non-metastatic breast cancer, patients who met criteria for TNBC (ER/PR < 1% and HER2-negative) by centralized-pathology review were included as previously published [Leon-Ferre et al, Breast Cancer Res Treat 2017]. Stromal TILs in these tumors were quantified on full-face hematoxylin and eosin sections from the surgical specimen, following the 2014 TIL Working Group recommendations [Salgado et al, Ann Oncol 2014]. The expression of AR [EPR1535(2), Abcam] was scored as a continuous variable (1% increments), and categorized as absent (0%), low (1-25% ), low-moderate (26-50%) moderate (51-75%) and high (> 75%) and correlated with TIL density. Results: 605 patients met criteria for TNBC. The median age was 56 years. Most tumors were T1-2 (95%), N0-1 (86%), and high grade (88%). The median stromal TIL content was 20% (0-90%). Tumor tissue was available for AR staining in 509/605 patients. 165 (32%) tumors expressed AR as follows: low: 12%, low-moderate: 7%, moderate: 5%, and high: 8%. Apocrine tumors (n = 30) expressed AR most frequently (86%) and at higher levels (AR high: 50%). For all 509 tumors, the median stromal TIL content according to AR expression was AR absent: 20%, AR low: 20%, AR low-moderate: 15%, AR high: 10%. AR and stromal TILs (both as continuous variables) demonstrated a weak inverse correlation (correlation coefficient -0.1064, p = 0.017). Conclusions: AR expression in TNBC is associated with lower levels of stromal TILs. Additional data correlating the AR, stromal TILs and outcomes in this cohort will be presented at the meeting.

Background The clinical utility of a new biomarker should ideally be established in a prospective randomized clinical trial. However, such trials are not always practical. As such, it is common for investigators to identify promising biomarkers using archived specimens and clinical data collected from previously completed therapeutic trials. Simon et al. defined such biomarker studies as prospective–retrospective studies and proposed specific conditions to satisfy for such evaluations to be more than hypothesis generating. One condition they proposed is that archived tissues must be available on a sufficiently large number of patients from the pivotal trials to ensure adequately powered analyses. Purpose The purpose of this article is to provide a new perspective on how to estimate power for assessing the prognostic and predictive values of a single binary biomarker in prospective–retrospective biomarker studies. Computer programs are provided to facilitate the use of these methods in practice. Methods The proposed methods utilize additional information that becomes available during the course of the treatment trial including sample size, accrual time, additional follow-up time, and the observed number of events at time of biomarker analysis. These methods involve solving for the exponential hazard rates that give rise to the event numbers that are consistent with those observed while satisfying other design parameter constraints. Conclusion Simon et al. proposed a new paradigm for biomarker design, conduct, analysis, and evaluation in prospective–retrospective studies that offers a more efficient alternative than fully prospective biomarker studies. As a general rule, they suggest that samples from at least two-thirds of the patients be available for analysis. In this article, I expand on this issue and provide a methodological tool useful for estimating study power in prospective–retrospective biomarker studies. It is my hope that these incremental efforts to improve the quality and statistical rigor in biomarker studies will hasten the introduction of useful tumor biomarkers into clinical practice.