T-cell inflamed gene expression to predict prognosis and response to immune checkpoint inhibitors.

Abstract

e15264 Background: To prove that T-cell-inflamed gene-expression profile (GEP) can predict prognosis and response to immune checkpoint inhibitors (ICIs) in the real-world study. Methods: 36 patients admitted from January 2018 to December 2018 were enrolled, including 15 patients with advanced NSCLC, 12 with advanced bladder cancer and 9 with advanced liver cancer. All patients were treated with nivolumab or pembrolizumab and provided informed written consent for analysis. Tumor total RNA was isolated from pretreatment FFPE samples and RNA-seq was performed on the Illumina NovaSeq 6000 system. GEP was calculated as a weighted sum of normalized expression values for 18 genes, as described in the article written by Ayer M, et al. in 2017. The data cutoff date was July 30, 2019. Primary end points were objective response rate (ORR) and disease control rate (DCR). Secondary end point was progression-free survival (PFS) assessed per RECIST v1.1. 95% CIs and P values for ORR and DCR were evaluated by the binomial exact method. Correlation between GEP and ORR, DCR and PFS was assessed by Wilcoxon rank sum test and Cox regression model, respectively. Survival curves were estimated by Kaplan-Meier methods. Cut-off value of GEP was confirmed by the Youden Index. The comparisons of DCR and PFS were performed by Fisher’s exact test and log-rank test, respectively. Results: ORR was 30.6% (95% CI, 16.3%-48.1%) and DCR was 50.0% (95% CI, 32.9%-67.1%). Median PFS was 3.9 months (95% CI, 0.8-NR). We found that GEP was higher in patients who achieved DCR and had longer PFS. And then GEP was significantly associated with DCR (p = 0.019) and PFS (p = 0.005), but not significantly with ORR (p = 0.42). Based on Youden Index, cut-off value of GEP was confirmed as -0.368, with 94.4% of sensitivity and 66.7% of specificity. According to cut-off value, 23 patients were evaluated as GEP-H. GEP-H group had higher DCR and longer PFS (DCR: 73.9% vs 7.69%, p = 0.0003; mPFS: 7.0m vs 2.1m, p = 0.00049). Conclusions: GEP can predict prognosis and response to ICIs in multiple cancers in our real-world study. Interpretation of these data may be limited by small sample sizes. Further studies are being conducted.