Cuproptosis-related molecular subtypes direct T cell exhaustion phenotypes and therapeutic strategies for patients with lung adenocarcinoma.

Abstract

Background: T cell exhaustion (TEX) heterogeneity leads to unfavorable immunotherapeutic responses in patients with cancer. Classification of TEX molecular phenotypes is pivotal to overcoming TEX and improving immunotherapies in the clinical setting. Cuproptosis is a novel form of programmed cell death associated with tumor progression. However, the relation between cuproptosis-related genes (CuRGs) and the different TEX phenotypes has not been investigated in lung adenocarcinoma (LUAD). Methods: Unsupervised hierarchical clustering and principal component analysis (PCA) algorithm were performed to determine CuRGs-related molecular subtypes and scores for patients with LUAD. The tumor immune microenvironment (TIME) landscape in these molecular subtypes and scores was estimated using ESTIMATE and ssGSEA algorithms. Furthermore, TEX characteristics and phenotypes were evaluated in distinct molecular subtypes and scores through GSVA and Spearman correlation analysis. Finally, TIDE scores, immunophenoscore, pRRophetic, GSE78220, and IMvigor210 datasets were employed to appraise the distinguishing capacity of CuRGscore in immunotherapy and pharmacotherapy effectiveness. Results: We identified three CuRGclusters, three geneClusters, and CuRGscore based on 1012 LUAD transcriptional profiles from five datasets. Compared with other molecular subtypes, CuRGcluster B, geneCluster C, and low-CuRGscore group with good prognosis presented fewer TEX characteristics, including immunosuppressive cells infiltration and TEX-associated gene signatures, signal pathways, checkpoint genes, transcription and inflammatory factors. These molecular subtypes were also responsive in distinguishing TEX phenotype in the terminal, GZMK+, and OXPHOS- TEX subtypes, but not the TCF7+ TEX subtype. Notably, copper importer and exporter, SLC31A1 and ATP7B, were remarkably associated with four TEX phenotypes and nine checkpoint genes such as PDCD1, CTLA4, HAVCR2, TIGIT, LAG3, IDO1, SIGLEC7, CD274, PDCD1LG2, indicating that cuproptosis was involved in the development of TEX and immunosuppressive environment in patients with LUAD. Moreover, CuRGscore was significantly related to the TIDE score, immunophenoscore, and terminal TEX score (Spearman R = 0.62, p < 0.001) to effectively predict immunotherapy and drug sensitivity in both training and external validation cohorts. Conclusion: Our study demonstrated the extensive effect of cuproptosis on TEX. CuRGs-related molecular subtypes and scores could illuminate the heterogeneity of TEX phenotype as reliable tools in predicting prognosis and directing more effective immunotherapeutic and chemotherapeutic strategies for patients with LUAD.