Research On Predictive Biomarkers For Clinical Benefit From Checkpoint Inhibitors In Lung Cancer Patients With Advanced-stage

Background: Lung cancer patients with advanced-stage disease usually harbor multiple genetic variation.Previous studies presented that EGFR mutation and ALK rearrangement had an unfavorable impact on clinical benefit of immune checkpoint inhibitors(ICIs)therapy.The underlying mechanism is still uncertain.PD-L1 expression and tumor mutation burden(TMB)are approved biomarkers for ICIs therapy.However,some patients with positive PD-L1 expression and high TMB still have no response to checkpoint inhibitors.It is urgent to develop novel predictive biomarker for patient selection.Combined biomarkers and predictive model by machine learning have become two main research directions,aiming at improving prediction efficiency and explaining potential biological mechanism.Part I: Clinical relevance of PD-L1 expression and CD8+ T cells infiltration in patients with EGFR-mutated and ALK-rearranged lung cancerObjective: PD-L1 expression and CD8+ T cell infiltration are two important factors of tumor microenvironment(TME).We tried to use this combined biomarker with PD-L1 expression and CD8+ T cell infiltration to explore the underlying mechanism of impaired response to ICIs therapy in EGFR-mutant and ALK-rearranged patients with advanced-stage lung cancer,and confirmed the predictive and prognostic role of TME subtypes.Main result: 1.Patients with EGFR mutation and ALK rearrangement had different TME subtypes distribution compared with wildtype patients,especially having lower proportion of PD-L1+/CD8+ TME subtype.2.Patients with EGFR 19 deletion had lower proportion of PDL1+/CD8+ TME subtype than that with EGFR 21L858 R mutation,and patients with T790 M positive mutation had lower proportion of PD-L1+/CD8+ TME subtype than that with T790 M wild-type.3.PD-L1+/CD8+ TME predicted worst overall survival in mutant patients,and maybe had an association with clinical response to ICIs therapy.4.Single biomarker had less prognostic power than combined biomarker.Conclusion: Low proportion of PD-L1 and CD8 co-expression in TME could be the underlying mechanism of impaired benefit to ICIs therapy in patients with EGFR mutations or ALK rearrangements and could explain the heterogeneous response to immunotherapy in different EGFR subtypes.Part II: Building a predictive model using RNA sequencing data and machine learning for checkpoint blockade therapy in lung cancer patientsObjective: With the consideration of three important factors of antitumor immunity,such as tumorigenesis,immune cell infiltration and tumor-immune interaction,we used RNA sequencing data of multiple cancer and machine learning to build a predictive model,aiming at achieving robust predictive power than traditional single markers.Main result: 1.Compared with PD-L1 expression and TMB,the model LITES has better predictive efficacy of clinical response in lung cancer patients who were treated with checkpoint inhibitors,and is independent of PD-L1 expression and TMB.2.Six features of LITES have synergistic effect on prediction efficiency.3.The predictive value of LITES for immunotherapy has tumor-type specificity and pan-cancer applicability,and its prognostic value has immunotherapy specificity.4.The oncogenes NRAS and PDPK1 may be related to primary resistance of ICIs therapy that needs to be further explored.Conclusion: LITES has better predictive performance in lung cancer patients who were treated with checkpoint inhibitors,when comparing with traditional biomarkers,and could effectively identify patients who had no response to ICIs therapy...