Gene-expression data integration to squamous cell lung cancer subtypes reveals drug sensitivity

Abstract

Background: Squamous cell lung cancer (SqCC) is the second most common type of lung cancer in the United States. Previous studies have used gene-expression data to classify SqCC samples into four subtypes, including the primitive, classical, secretory and basal subtypes. These subtypes have different survival outcomes, although it is unknown whether these molecular subtypes predict response to therapy. Methods: Here, we analysed RNAseq data of 178 SqCC tumour samples and characterised the features of the different SqCC subtypes to define signature genes and pathway alterations specific to each subtype. Further, we compared the gene-expression features of each molecular subtype to specific time points in models of airway development. We also classified SqCC-derived cell lines and their reported therapeutic vulnerabilities. Results: We found that the primitive subtype may come from a later stage of differentiation, whereas the basal subtype may be from an early time. Most SqCC cell lines responded to one of five anticancer drugs (Panobinostat, 17-AAG, Irinotecan, Topotecan and Paclitaxel), whereas the basal-type cell line EBC-1 was sensitive to three other drugs (PF2341066, AZD6244 and PD-0325901). Conclusion: Compared with the other three subtypes of cell lines, the secretory-type cell lines were significantly less sensitive to the five most effective drugs, possibly because of their low proliferation activity. We provide a bioinformatics framework to explore drug repurposing for cancer subtypes based on the available genomic profiles of tumour samples, normal cell types, cancer cell lines and data of drug sensitivity in cell lines.