Heterogeneity of tumor immune microenvironment and real-world analysis of immunotherapy efficacy in lung adenosquamous carcinoma

Abstract

Lung adenosquamous carcinoma (ASC) is an uncommon histological subtype. We aimed to characterize the tumor immune microenvironment (TIME) in lung ASC and estimate patient response to immune checkpoint inhibitors (ICIs), which have never been systematically investigated. In cohort I, we collected 30 ASCs from a single center for analysis of TIME characteristics, including immuno-phenotyping, tumor mutation burden (TMB), T-cell receptor (TCR) repertoires, tumor-infiltrating lymphocytes (TILs), and immune checkpoint expression. Twenty-two (73.3%) patients were EGFR-positive. The TIME was defined by immune-excluded (60%) and immune-desert phenotype (40%). Strikingly, programmed cell death-ligand 1 (PD-L1) and programmed cell death-1 (PD-1) were predominantly expressed in squamous cell carcinoma components (SCCCs) versus adenocarcinoma components (ACCs), where enhanced CD4⁺ FOXP3⁺ regulatory T cell and attenuated CD57⁺ natural killer cell infiltration were present, consistent with a landscape of fewer innate immune cells, more immunosuppressive cells. SCCCs had higher TMB, higher TCR clonality, and lower TCR diversity than ACC. In cohort III, the efficacy of ICI-based therapy was estimated using a real-world data of 46 ASCs from 11 centers. Majority of 46 patients were driver genes negative and unknown mutation status, 18 (39%) and 18 (39%), respectively. The overall objective response rate of 28%, median progression-free survival of 6.0 months (95% confidence interval [CI] 4.3-7.7), and median overall survival of 24.7 months (95% CI 7.2-42.2) were observed in the ICI-based treatment. This work ascertains suppressive TIME in lung ASC and genetic and immuno-heterogeneity between ACCs and SCCCs. Lung ASC patients have a moderate response to ICI-based immunotherapy.

Keywords: PD-L1 expression; heterogeneity; immune checkpoint inhibitor; lung adenosquamous carcinoma; tumor immune microenvironment.