Tumor microenvironment evaluation to predict anti–PD-1 response of advanced gastric cancer: Results from a multicenter prospective clinical trial.

Abstract

415 Background: Clinical trials support the efficacy of immune checkpoint blockades (ICBs) plus chemotherapy in a subset of patients with metastatic gastric cancer (mGC). With the aim of identifying determinants of response, we developed a TMEscore assay (PCR assay with 30 genes) to assess tumor microenvironment, which was previously proved to be a robust predictive biomarker for patients treated with ICBs. Methods: Advanced GC patients treated with ICB combined with chemotherapy as first-line regimen were included in this multi-center prospective clinical trial (NCT#04850716). 65 tumor specimens obtained from 8 medical centers before treatment were applied to estimate TMEscore, PD-L1(CPS) and mismatch repair deficiency (MMR). Of 65 patients, 43 patients with treatment response and 34 patients with progression-free survival (PFS) were used to compare the predictiveness of biomarkers. Results: Theoverall response rate (ORR) and median PFS of this cohort were 39.9% and 4.67 months.Enhanced ORR was observed in TMEscore-high mGC patients (ORR = 61%), but only 16.7% in the low group. Regressive tumors (partial response, PR) had markedly higher TMEscore than stable and progressive tumors (stable disease, SD; progressive disease, PD, P = 0.001). By applying ROC curve analysis, the TMEscore was found to be an encouraging predictive biomarker that surpassed MMR and CPS statistically (AUC = 0.863, 0.525, and 0.519, respectively; Delong test, P = 0.003). Importantly, the Kaplan-Meier survival analysis demonstrated that a high TMEscore was significantly related to a more favorable PFS ( P = 0.0026, HR=0.18, 95%CI 0.06-0.55). Compared to TMEscoreA (immune score), TMEscoreB (stromal score) presented a dramatically higher hazard ratio for PFS (HR = 12.25, P = 0.0001), implicating stromal activation as a critical mechanism of intrinsic resistance to ICB plus chemotherapy. Meanwhile, PD-L1 (CPS) was not statistically associated with PFS, whether 1 or 5 were used as a cutoff to divide patients ( P = 1 and 0.22). Conclusions: This prospective clinical study indicated that the TMEscore assay is a robust biomarker for screening mGC patients who may derive survival benefit from ICB plus chemotherapy. Our data also suggest that TMEscore may be a more accurate predictive biomarker than MSI and CPS (PD-L1) for mGC patients, which warrants additional investigations in larger cohorts. Clinical trial information: NCT#04850716 .