Abstract 4066: Development and validation of a gene mutation-based signature to predict response to PD-1 inhibitors in nonsquamous NSCLC

Abstract

Abstract Background: Genetic variations in nonsquamous NSCLC displayed significant impact on immune microenvironment and response to programmed cell death protein 1 (PD-1) blockade immunotherapy. We undertook an unbiased analysis to develop a gene mutation-based signature (GMS) and predict the efficacy of anti-PD-1 treatment. Methods: Two independent cohorts (MSK-IMPACT and CheckMate 012) consist of 265 nonsquamous NSCLC patients treated with anti-PD-1 were analyzed for gene mutation via next-generation sequencing. A GMS was built in a randomly selected 123 samples from MSK-IMPACT training cohort, using multivariate cox analysis of high-frequency mutation genes (≥10%) associated with progression-free survival (PFS) after anti-PD-1 treatment. We then validated our findings in the remaining 83 samples (internal validation cohort) and in CheckMate-012 external validation cohort (n=59). Results: A GMS that consisted of 6 genes (KRAS, EGFR, TP53, STK11, PTPTD and KMT2C), was generated to classify patients into high and low GMS groups in the training cohort. Patients with high GMS in the training cohort had longer PFS (p<0.001) compared with low GMS. We noted equivalent findings in the internal validation cohort (p=0.004) and in the external validation cohort (p=0.001). The GMS was an independent predictive factor for anti-PD-1 treatment (Table 1). Furthermore, GMS can successfully predict PFS in anti-PD-1 monotherapy (p<0.001) as well as combined treatment (p<0.001), and in high PD-L1 (p=0.004) as well as low PD-L1 (p=0.006) expression subgroups. When combined the GMS and PD-L1, those with GMS high/PD-L1high had the strongest objective response rates and favorable survival than other subgroups. Conclusion: Our study highlights the potential predictive value of GMS for immunotherapeutic benefit in nonsquamous NSCLC. The combination of GMS and PD-L1 may be a feasible and promising biomarker in guiding treatment decisions for anti-PD-1 therapy. Table 1Univariate and multivariable Cox regression analysis of predictive factorsUnivariate analysisMultivariate analysisVariableHR95%CIP-valueHR95%CIP-valueMSK-IMPACT cohortGMS (High vs. low)0.360.19-0.680.0020.380.20-0.740.004PD-L1 (High vs. low)0.520.31-0.860.0110.570.33-0.960.034TMB (High vs. low)0.590.36-0.990.046Gender (Male vs. female)0.810.49-1.350.423Age (≥ 65 vs. <65 years)1.090.65-1.820.753Smoking (Ever vs. Never)0.500.28-0.910.022CheckMate-012 cohortGMS (High vs. low)0.270.11-0.650.0030.270.11-0.650.003PD-L1 (High vs. low)1.440.74-2.830.284TMB (High vs. low)0.400.21-0.790.008Gender (Male vs. female)1.040.53-2.050.902Age (≥ 65 vs. <65 years)0.870.45-1.680.671Smoking (Ever vs. Never)0.600.28-1.280.183 Citation Format: Zhong-Yi Dong, Xin-Ran Tang, Li Liu, De-Hua Wu. Development and validation of a gene mutation-based signature to predict response to PD-1 inhibitors in nonsquamous NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4066.