A Connexin-Based Biomarker Model Applicable for Prognosis and Immune Landscape Assessment in Lung Adenocarcinoma

Gap junction protein (Connexin) family is the basic unit of cellular connection, whose multiple members were recently demonstrated to be associated with tumor progression. However, the expression pattern and prognostic value of connexin in lung adenocarcinoma (LUAD) have not yet been elucidated.

Consensus cluster algorithm was first applied to determine a novel molecular subtype in LUAD based on connexin genes. The differentially expressed genes (DEGs) between two clusters were obtained to include in Cox regression analyses for the model construction. To examine the predictive capacity of the signature, survival curves and ROC plots were conducted. We implemented GSEA method to uncover the function effects enriched in the risk model. Moreover, the tumor immune microenvironment in LUAD was depicted by CIBERSORT and ssGSEA methods.

The integrated LUAD cohort (TCGA-LUAD and GSE68465) were clustered into two subtypes (C1 = 217 and C2 = 296) based on 21 connexins and the clinical outcomes of LUAD cases in the two clusters showed remarkable discrepancy. Next, we collected 222 DEGs among two subclusters to build a prognostic model using stepwise Cox analyses. Our proposed model consisted of six genes that accurately forecast patient outcomes and differentiate patient risk. GSEA indicated that high-risk group was involved in tumor relevant pathways were activated in high-risk group, such as PI3K/AKT signaling, TGF-β pathway, and p53 pathway. Furthermore, LUAD cases with high-risk presented higher infiltration level of M2 macrophage and neutrophil, suggesting high-risk group were more likely to generate an immunosuppressive status.

Our data identified a novel connexin-based subcluster in LUAD and further created a risk signature which plays a central part in prognosis assessment and clinical potency.