Abstract PD5-03: Relationship between tumor-infiltrating lymphocytes (TILs) and outcomes in the KEYNOTE-119 study of pembrolizumab vs chemotherapy for previously treated metastatic triple-negative breast cancer (mTNBC)

Abstract

Abstract Background: Several studies show that the presence of TILs has prognostic significance in TNBC, with greater lymphocytic infiltration associated with clinical outcomes. In the phase 2 KEYNOTE-086 study, TIL levels were a surrogate marker of preexisting antitumor immunity and were independent predictors of response to pembrolizumab (pembro) monotherapy. Here, we analyzed the relationship between the presence of TILs and outcomes in the phase 3 KEYNOTE-119 (NCT02555657) study of pembro monotherapy vs single-agent chemotherapy (chemo) in patients (pts) with previously treated mTNBC. Methods: From October 2015 to April 2017, 622 pts from 31 countries were enrolled. Key eligibility criteria included centrally confirmed TNBC, 1 or 2 prior systemic treatments for metastatic disease, documented progression on most recent therapy, prior treatment with an anthracycline and/or a taxane, and provision of a tumor sample for central determination of triple-negative status and PD-L1 expression. Pts with active brain metastases were excluded. Pts were stratified by PD-L1 status (positive vs negative; PD-L1 positivity was defined as a CPS ≥1) and history of prior neoadjuvant/adjuvant treatment vs de novo metastatic disease at initial diagnosis. Pts were randomly assigned 1:1 to pembro 200 mg Q3W or single-agent chemo per investigator’s choice of capecitabine, eribulin, gemcitabine, or vinorelbine, with a maximum enrollment cap of 60% for each, administered per local product label, until progression, intolerable toxicity, or investigator/patient decision. Response was assessed every 9 wk for 1 y and every 12 wk thereafter. Primary end points were OS in the PD-L1 CPS ≥10, CPS ≥1, and total populations. Secondary end points were PFS, ORR, duration of response, and DCR in the PD-L1 CPS ≥10, CPS ≥1, and total populations, and safety. The presence of TILs in archival tumor samples was assessed by light microscopy of hematoxylin and eosin-stained sections using a predefined method. The relationship between TILs and OS and PFS was analyzed using Cox regression; the relationship between TILs and ORR was analyzed using logistic regression. The interaction between treatment and TILs was assessed. Results: The study was negative for the primary end point. TILs were evaluable in 536/622 (86.2%) treated pts (273 [pembro]; 263 [chemo]). The median TILs distribution was 5% (IQR, 14%). TIL levels were significantly higher in responders vs nonresponders in the pembro arm, but not the chemo arm. TIL levels as a continuous variable were significantly (P< 0.05) associated with all clinical outcomes tested in the pembro but not the chemo arm (Table). For pts with TILs <5%, assessment of OS yielded a HR of 1.50 (95% CI, 1.14-1.97); for pts with TILs ≥5%, the HR was 0.75 (95% CI, 0.59-0.96). Median OS and 18-month OS rate (pembro vs chemo) for pts with TILs <5% was 5.9 vs 8.8 mo and 15% vs 27%, respectively; for pts with TILs ≥5%, this was 12.5 vs 11.3 mo and 35% vs 27%. The correlation between TILs vs CPS was moderate at 0.45; multivariate modeling of TILs and CPS showed independent predictive value. Conclusions: High TILs were significantly associated with better clinical outcomes with pembro but not chemo. Efficacy estimates at the prespecified median TIL cut point (≥5%) suggest that a subset of later-line advanced pts with TNBC can derive prolonged survival benefit from pembro over chemo. Association of TILs With Clinical OutcomesTreatment ArmContinuousBORDCRPFSOSVariableORRP*DCRP*Event RateP*Event RateP*Pembro (n=273),TILs25 (9.2)0.000732 (11.7)0.0058241 (88.3)0.0002239 (87.5)0.0001n (%)Chemo (n=263),TILs29 (11.0)0.180950 (19.0)0.3026191 (72.6)0.234236 (89.7)0.3321n (%)*P values are 1-sided for pembro and 2-sided for chemo.The stratification variable, prior (neo)adjuvant therapy vs de novo metastatic disease at initial diagnosis, was used as a covariate in the model.TILs was square root transformed.BOR, best overall response; chemo, chemotherapy; DCR, disease control rate; OS, overall survival; pembro, pembrolizumab; PFS, progression-free survival; TILs, tumor-infiltrating lymphocytes. Citation Format: Sherene Loi, Eric Winer, Oleg Lipatov, Seock-Ah Im, Anthony Goncalves, Javier Cortes, Keun S Lee, Peter Schmid, Laura Testa, Isabell Witzel, Shoichiro Ohtani, Nicholas Turner, Stefania Zambelli, Nadia Harbeck, Fabrice Andre, Rebecca Dent, LingKang Huang, Jaime Mejia, Vassiliki Karantza, Roberto Salgado. Relationship between tumor-infiltrating lymphocytes (TILs) and outcomes in the KEYNOTE-119 study of pembrolizumab vs chemotherapy for previously treated metastatic triple-negative breast cancer (mTNBC) [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr PD5-03.