Epigenetic prediction of response to anti-PD-1 treatment in non-small-cell lung cancer: a multicentre, retrospective analysis

Abstract

Anti-programmed death-1 (PD-1) treatment for advanced non-small-cell lung cancer (NSCLC) has improved the survival of patients. However, a substantial percentage of patients do not respond to this treatment. We examined the use of DNA methylation profiles to determine the efficacy of anti-PD-1 treatment in patients recruited with current stage IV NSCLC. In this multicentre study, we recruited adult patients from 15 hospitals in France, Spain, and Italy who had histologically proven stage IV NSCLC and had been exposed to PD-1 blockade during the course of the disease. The study structure comprised a discovery cohort to assess the correlation between epigenetic features and clinical benefit with PD-1 blockade and two validation cohorts to assess the validity of our assumptions. We first established an epigenomic profile based on a microarray DNA methylation signature (EPIMMUNE) in a discovery set of tumour samples from patients treated with nivolumab or pembrolizumab. The EPIMMUNE signature was validated in an independent set of patients. A derived DNA methylation marker was validated by a single-methylation assay in a validation cohort of patients. The main study outcomes were progression-free survival and overall survival. We used the Kaplan-Meier method to estimate progression-free and overall survival, and calculated the differences between the groups with the log-rank test. We constructed a multivariate Cox model to identify the variables independently associated with progression-free and overall survival. Between June 23, 2014, and May 18, 2017, we obtained samples from 142 patients: 34 in the discovery cohort, 47 in the EPIMMUNE validation cohort, and 61 in the derived methylation marker cohort (the T-cell differentiation factor forkhead box P1 [FOXP1]). The EPIMMUNE signature in patients with stage IV NSCLC treated with anti-PD-1 agents was associated with improved progression-free survival (hazard ratio [HR] 0·010, 95% CI 3·29 × 10-4-0·0282; p=0·0067) and overall survival (0·080, 0·017-0·373; p=0·0012). The EPIMMUNE-positive signature was not associated with PD-L1 expression, the presence of CD8+ cells, or mutational load. EPIMMUNE-negative tumours were enriched in tumour-associated macrophages and neutrophils, cancer-associated fibroblasts, and senescent endothelial cells. The EPIMMUNE-positive signature was associated with improved progression-free survival in the EPIMMUNE validation cohort (0·330, 0·149-0·727; p=0·0064). The unmethylated status of FOXP1 was associated with improved progression-free survival (0·415, 0·209-0·802; p=0·0063) and overall survival (0·409, 0·220-0·780; p=0·0094) in the FOXP1 validation cohort. The EPIMMUNE signature and unmethylated FOXP1 were not associated with clinical benefit in lung tumours that did not receive immunotherapy. Our study shows that the epigenetic milieu of NSCLC tumours indicates which patients are most likely to benefit from nivolumab or pembrolizumab treatments. The methylation status of FOXP1 could be associated with validated predictive biomarkers such as PD-L1 staining and mutational load to better select patients who will experience clinical benefit with PD-1 blockade, and its predictive value should be evaluated in prospective studies. "Obra Social" La Caixa, Cellex Foundation, and the Health and Science Departments of the Generalitat de Catalunya.