Genome-wide identification of differentially methylated promoters and enhancers associated with response to anti-PD-1 therapy in non-small cell lung cancer

Abstract

Although approved programmed cell death protein (PD)-1 inhibitors show durable responses, clinical benefits to these agents are only seen in one-third of patients in most cancer types. Therefore, strategies for improving the response to PD-1 inhibitor for treating various cancers including non-small cell lung cancer (NSCLC) are urgently needed. Compared with genome and transcriptome, tumor DNA methylome in anti-PD-1 response was relatively unexplored. We compared the pre-treatment methylation status of cis-regulatory elements between responders and non-responders to treatment with nivolumab or pembrolizumab using the Infinium Methylation EPIC Array, which can profile ~850,000 CpG sites, including ~350,000 CpG sites located in enhancer regions. Then, we analyzed differentially methylated regions overlapping promoters (pDMRs) or enhancers (eDMRs) between responders and non-responders to PD-1 inhibitors. We identified 1007 pDMRs and 607 eDMRs associated with the anti-PD-1 response. We also identified 1109 and 1173 target genes putatively regulated by these pDMRs and eDMRs, respectively. We found that eDMRs contribute to the epigenetic regulation of the anti-PD-1 response more than pDMRs. Hypomethylated pDMRs of Cytohesin 1 Interacting Protein (CYTIP) and TNF superfamily member 8 (TNFSF8) were more predictive than programmed cell death protein ligand 1 (PD-L1) expression for anti-PD-1 response and progression-free survival (PFS) and overall survival (OS) in a validation cohort, suggesting their potential as predictive biomarkers for anti-PD-1 immunotherapy. The catalog of promoters and enhancers differentially methylated between responders and non-responders to PD-1 inhibitors presented herein will guide the development of biomarkers and therapeutic strategies for improving anti-PD-1 immunotherapy in NSCLC.

Fig. 1. Methylomic features of the anti-PD-1 response in NSCLC patients.

a Overview of the study design and summary of the results. b Proportion of differentially methylated promoter or enhancer regions normalized by the total number of promoters or enhancers for each chromosome.

Fig. 2. KEGG pathways enriched for DMR target genes.

a Proportion of all promoters and enhancers that overlap with DMRs (i.e., pDMRs and eDMRs, respectively). b Proportion of all genes targeted by pDMRs and eDMRs. c–e Significantly enriched (q value < 0.01) KEGG pathways for all DMR target genes c for eDMR target genes d and for pDMR target genes e. Immune-related, oncogenic, and metabolic-regulation KEGG pathways; cancer-associated and other pathways are marked by different color codes.

Fig. 3. Differential methylation of HLA gene enhancers for the anti-PD-1 response.

a Integrative genomic view showing HLA gene clusters of human chromosome 6. Arches represent enhancer–promoter interactions. b Megascopic views of enhancer-containing regions (I–V) to highlight sequence region overlap among HLA genes, DMRs, enhancers, and super-enhancers. c Comparison of mean beta values of eDMRs targeting MHC-II between responders (R) and non-responders (NR). d, e Comparison of xCell scores between responders (R) and non-responders (NR) for CD4⁺ effector memory T cells d and CD8⁺ T cells. e The significance of difference between two groups was tested by a Wilcoxon one-sided rank sum test.

Fig. 4. Evaluation of pDMRs for CYTIP and TNFSF8 to predict the anti-PD-1 response.

a, b Comparison of methylation levels of pDMR for CYTIP between responders (R) and non-responders a and those for TNFSF8 b in the validation cohort. c Comparison of the PD-L1 expression level between responders (R) and non-responders (NR) in the validation cohort. d, e Proportion of patients with objective response (blue) or no objective response (red) to anti-PD-1 therapy for the methylation level threshold of pDMR for CYTIP (40%) or that for TNFSF8 (50%), respectively. “Pos” indicates patients predicted to respond, and “Neg” indicates those predicted to not respond to anti-PD-1 therapy. f Analysis as for d, e with 1% of PD-L1 IHC expression level. g Analysis as for d, e with combined use of pDMRs for CYTIP and TNFSF8.

Fig. 5. Evaluation of pDMRs for CYTIP and TNFSF8 to predict survival after anti-PD-1 therapy.

a–d Comparison of progression-free survival (PFS) between the patients’ group based on the methylation level of pDMR for CYTIP (methylation of CYTIP promoter ≤ 40% vs. > 40%) a, methylation level of pDMR for TNFSF8 (methylation of TNFSF8 promoter ≤ 50% vs. > 50%) b, PD-L1 expression (PD-L1 ≥ 1% vs. < 1%) c, or combined use of both pDMRs (methylation of CYTIP ≤ 40% and TNFSF8 ≤ 50% promoter vs. the others) d. e–h Comparison of overall survival (OS) between the patients’ group based on the methylation level of the pDMR for CYTIP (methylation of CYTIP promoter ≤ 40% vs. > 40%) e, the methylation level of the pDMR for TNFSF8 (methylation of TNFSF8 promoter ≤ 50% vs. > 50%) f, PD-L1 expression (PD-L1 ≥ 1% vs. < 1%) g, or combined use of both pDMRs (methylation of CYTIP ≤ 40% and TNFSF8 ≤ 50% promoter vs. the others) h.