Analysis of event-free survival (EFS) of stromal tumor infiltrating lymphocytes (sTILs), PD-L1 expression, and their early dynamics in the NeoTRIP trial.

Abstract

597 Background: We demonstrated that pre-treatment sTILs and PD-L1, and on treatment sTILs but not PD-L1 were associated with pathological Complete Response (pCR) to neoadjuvant therapy in patients (pts) with triple-negative breast cancer (TNBC) enrolled in NeoTRIP (NCT02620280) trial (Bianchini G ESMO 2020). Here we assess the association between the same biomarkers and EFS. Methods: NeoTRIP randomized 280 pts to nab-paclitaxel/carbo for 8 cycles (CT) or with atezolizumab (CT/A). As Per-Protocol Population, 258 pts were evaluable for EFS, the primary endpoint of the study.We collected samples at baseline (n=258/258; 100%) and on Day 1 Cycle 2 (D1C2) (n=230/258; 89.2%]. We centrally assessed sTILs (cut-off ≥30% to be considered high) and PD-L1 expression (SP142) on immune cells (IC) (IC≥1% considered positive). Association with EFS was investigated. Imaging mass cytometry (IMC) (Wang Nature 2023) was used to investigate the biology linked to PD-L1 dynamic over-expression. Results: Median follow-up was 54 months. Pre-treatment high sTILs and PD-L1+ were associated with lower risk of recurrence in CT arm (HR 0.35 [0.12-0.99], p=0.049 and HR 0.31 [0.15-0.64], p=0.001, respectively). In CT/A arm neither pre-treatment marker was significantly associate with EFS. At D1C2 in CT arm, high sTILs but not PD-L1+ was significantly associated with lower risk of recurrence (HR 0.34 [0.15-0.81], p=0.015 and HR 0.65 [0.28-1.48], p=0.30, respectively). In CT/A arm both high sTILs and PD-L1+ were strongly associated with lower risk of recurrence (HR 0.23 [0.07-0.78], p=0.019 and HR 0.25 [0.11-0.57], p=0.001, respectively). Only in CT/A arm, on-treatment PD-L1 provided prognostic information independent of baseline biomarkers, on-treatment sTILs and pCR (adjHR = 0.25 [0.11-0.58], p=0.001). In patients with baseline PD-L1- tumors and paired D1C2 samples (n=87), conversion from PD-L1- to on-treatment PD-L1 positivity occurred in 64.3% and 17.8% in CT/A and CT arms, respectively (p= 7.39x10 -6 ). The tumors converted to PD-L1+ status were significantly associated with better outcome in CT/A arm (HR 0.23 [0.08-0.68], p=0.008) but not in CT arm (HR 0.82 [0.24-2.85], p=0.76). Notably, in CT/A arm, patients with baseline PD-L1- tumors had similarly low pCR rate regardless of PD-L1 status on D1C2 (20% and 25.9% in PD-L1- and PD-L1+ D1C2 groups, respectively). The IMC results for the biological characterization of tumors with induction of PD-L1+ will be presented. Conclusions: In NeoTRIP on-treatment high sTILs (both CT and CT/A arms) and PD-L1 positivity (only in CT/A arm) were significantly associated with lower risk of recurrence independently of pCR and baseline biomarkers. The findings suggest up-regulation of PD-L1 as new candidate pharmacodynamic biomarker of benefit from atezolizumab. Whether this observation hold also for pembrolizumab remain to be defined.