Biomarkers Associated With Response And Immune-related Adverse Events For Immune Checkpoint Inhibitors

In this study,we aim to identify biomarkers associated with response and immune-related adverse events risk during immune checkpoint inhibitors therapy.The work is composed of two sections below:Section 1 Association of MUC16 mutation with response to immune checkpoint inhibitors in solid tumorsImmune checkpoint inhibitor(ICI)-based immunotherapy,which mainly targets cytotoxic T-lymphocyte associated protein 4(CTLA4),programmed cell death 1(PD-1),and its ligand(PD-L1),has created a paradigm shift and shown impressive clinical benefits to patients with advanced-stage cancers.Yet,therapeutic response to ICI varies among patients.Although multiple biomarkers of response have been reported,these markers have limitations.As the third most frequently mutated gene in cancers,the association between MUC16 mutation and response to ICI remains unclear.This study used multidimensional genomic data,including whole-exome and mRNA sequencing data,of 10195 patients from The Cancer Genome Atlas across 30 solid tumor types,56 patients from a non-small cell lung cancer(NSCLC)cohort,145 patients from a melanoma cohort,and 15 patients from a pan-solid tumors cohort,aiming to investigate the association of MUC16 mutation with response to ICI.In the pan-cancer data set,the mutational frequency of MUC16 was 19.68%;patients with MUC16 mutation had higher tumor mutational burden(median,230 mutations vs 48 mutations;P<0.001,Mann-Whitney U test)and neoantigen load(median,179 neoantigens vs 48 neoantigens;P<0.001,Mann-Whitney U test)than those without mutations;the tumor immune microenvironment with dual-positive CD8A and PD-L1 was overrepresented in MUC16-mutated tumors compared with wild-type ones(43.8%vs 32.4%;P<0.001),with greater extent of immune cells infiltration observed in the microenvironment;additionally,of the 40 immune-related genes,which were classified into three categories:immune checkpoints,T cell receptor signalling,and T-effector and interferon-y gene signature,36(90%)exhibited significantly upregulated expression in the microenvironment of mutated vs wild-type tumors,such as CTLA4,PD-L1,PD-1,LAG3,TIGIT,OX40,and ICOS.Survival analyses were further performed to demonstrate that MUC16 mutation was associated with improved overall survival in both the NSCLC(hazard ratio,0.34;95%CI,0.12-0.99;P=0.04)and melanoma(hazard ratio,0.57;95%CI,0.36-0.90;P=0.02)cohorts.The improvement persisted after adjusting for confounders including age,sex,and dominant mutational signatures in the melanoma cohort(hazard ratio,0.57;95%CI,0.33-0.96;P=0.04),using the Cox proportional hazards model.Gene set enrichment analysis revealed that eight gene sets regarding cell proliferation,immune response,and mTORC1 signaling,were enriched in MUC16-mutated tumors,providing biological insights concerning the link between MUC16 mutation and ICI response.Taken together,these findings support the hypothesis that high immunogenicity and a responsive tumor immune microenvironment are a hallmark of MUC16-mutated tumors.MUC16 mutation may serve as a prognostic stratification biomarker to help optimize the application of personalized ICI immunotherapy.Section 2 Multi-omics predictive biomarkers of immune-related adverse events riskDespite significantly prolonged survival provoked by blockade of immune checkpoints in cancer patients,this inhibition induces autoimmune toxicities,termed immune-related adverse events(irAEs),due to consequent immune activation in patients.The most common irAEs are observed in skin and gastrointestinal tract,but any organ can be affected and some infrequent irAEs may be serious and fatal;thus,these toxicities require early detection and positive prevention.However,predictive biomarkers for irAEs occurrence risk remain elusive.To identify biomarkers that contribute to irAEs risk prediction,we comprehensively analyzed pharmacovigilance data of 4865522 reports from the US Food and Drug Administration Adverse Event Reporting System and multi-omics data from The Cancer Genome Atlas Pan-Cancer Atlas cohort with 9104 samples across 21 cancer types,using the general linear regression model.Risk of irAEs occurrence was evaluated as reporting odds ratio(ROR).The performance of predicting irAEs risk was measured as Pearson coefficient correlation(R)for models,which was obtained using leave-one-out cross-validation.Log-likelihood ratio test was applied for comparisons of goodness of fit between different models,and multicollinearity among variables in the model was evaluated by variance inflation factor.Multiple testing correction was performed controlling the false discovery rate(FDR)by the Benjamini-Hochberg method.We found 23 biomarkers predictive of irAEs ROR(FDR<0.05),including dendritic cells(DC)abundance(R,0.69),tumor mutational burden(TMB)(R,0.63),CD4+naive T-cells abundance(R,0.55),expression abundances of IRF4(R,0.847)and TCL1A(R,0.82),SHC phosphorylation level on Tyr317(SHC-pY317)(R,-0.75),and miR-155-3p(R,0.73).Of those,the transcription factor IRF4 showed the highest correlation,possibly relating to its essential roles in many aspects of B-cells,T-cells,and DC differentiation and function.Additionally,IRF4 deletion may induce transplant acceptance and resistance to several autoimmune diseases,and its expression can be inhibited by trametinib.These evidences suggest a potential therapeutic strategy for irAEs.Compared with single biomarker,combination of these candidates offered superior predictive performance,with no sign of multicollinearity detected.Specifically,the combined DC,TMB,and CD4+naive T-cells linear model accounted for 66%of ROR variance across cancer types(R,0.81;FDR=1.1×10-4)with pronounced model promotion in comparison with bivariate models(log-likelihood ratio test,P=8.7×10-4 relative to TMB-DC model;P=2.8×10-4 relative to TMB-CD4+naive T-cells model).Further,a trivariate model combining IRF4,TCL1 A,and SHC-pY317 factors explained 76%of ROR variance(R,0.87;FDR=2.5×10-6),with significant improvement of predictive accuracy as compared to the IRF4-TCL1A bivariate model(P=0.03).Collectively,these findings provide a valuable resource of potential predictive biomarkers for irAEs and may deepen our understanding of its pathogenesis,holding promise for aiding in early detection of high-risk patients in the clinical practice.