Abstract 222P
Background
Immune checkpoint inhibitors (ICIs) achieved good efficacy in treatment of non-small cell lung cancer (NSCLC). However, many patients can not maintain durable clinical benefit (DCB). Variant allele frequency (VAF), defined as the percentage of alternate alleles within a genomic locus, was considered as a new potential prognostic and predictive biomarker. We aimed to investigate the predictive role of VAF in NSCLC using ICIs.
Methods
Genomic data of 921 NSCLC patients from the cBioPortal database, and 23 NSCLC patients from the local cohort were included in model construction and evaluation. Transcriptomic data of 514 NSCLC patients from TCGA were included in mechanism analysis.
Results
Frequency of mutation <5% was eliminated. Then we identified 15 differential mutational genes between DCB and NDB, including PTPRT (P = 0.001), EGFR (P = 0.002), EPHA3 (P = 0.002), ERBB4 (P = 0.002), ATRX (P = 0.006), PTPRD (P = 0.007), AMER1 (P = 0.013), TERT (P = 0.020), HGF (P = 0.022), POLE (P = 0.022), STK11 (P = 0.023), NTRK3 (P = 0.032), EPHA5 (P = 0.037), ALK (P = 0.040), and PGR (P = 0.041). Using LASSO algorithm to construct the prediction model. ROC-AUC reached 0.703, 0.690, and 0.674 in the training cohort (n=313), test-1 cohort (internal validation, n=133), and test-2 cohort (external validation, n=157), respectively. Then we divided the patients into high- and low- score groups. K-M survival curve showed the low score group president a longer PFS (Training: P < 0.0001, Test-1: P = 0.0025, Test-2: P = 0.0230) and longer OS (Test-3 [n=341]: P = 0.0009). After multivariate Cox regression correction, the risk model was still statistically significant. In addition, the model was independent of PD-L1 expression. WGCNA and enrichment analyses of 514 NSCLC from TCGA showed that the low score group was associated with lymphocyte differentiation, mononuclear cell differentiation, immune response-regulating cell surface receptor signaling pathway, activation of immune response, and leukocyte mediated immunity.
Conclusions
Mutational VAF profile was a promising novel biomarker in the prediction and prognosis of NSCLC. We established a predictive model of ICI response, which was of great significance in tumor precision therapy.
Legal entity responsible for the study
National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.