Abstract
One of the primary causes of global cancer-associated mortality is lung cancer (LC). Current improvements in the management of LC rely mainly on the advancement of patient stratification, both molecularly and clinically, to achieve the maximal therapeutic benefit, while most LC screening protocols remain underdeveloped. In this research, we first employed two algorithms (ESTIMATE and xCell) to calculate the immune/stromal infiltration scores. This helped identify the altered immune infiltration landscapes in lung adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC). Afterward, based on their immune-related characteristics, we successfully stratified the LUAD and LUSC into 2 and 3 clusters, respectively. Different from the conventional bioinformatic approaches that start from the investigation of differential expression of single genes, differentially enriched curated gene sets identified through gene set variation analyses (GSVA) were curated, and gene names were reconstructed afterward. Furthermore, weighted gene correlation network analyses (WGCNA) were used to reveal hub genes highly connected with the clustering process. Actual expression levels of critical hub genes among different clusters were compared and so were the functional pathways these genes enriched into. Lastly, a computational method was applied to predict and compare the responses of each cluster to primary therapeutic agents. The heterogeneity presented in our study, along with the drug responses expected for identified clusters, may shed light on future exploration of combination immunochemotherapy that facilitates the optimization of individualized therapy.