Advancing precision medicine in colorectal cancer through a transcriptome-wide gene expression outlier analysis

Novel therapies, tailored on the heterogeneous molecular features of human tumours, are urgently needed to increase survival of patients with metastatic colorectal cancer (mCRC). Mutations and genomic alterations have always been the main source of information in clinical decision making, but recent studies have suggested that characterization of aberrant cancer gene expression events could facilitate the detection of new vulnerabilities in cancer patients. In this work, we developed a gene expression outlier detection pipeline capable to identify transcriptome-wide overexpression (positive outliers) and loss of expression (negative outliers) events, and we applied it to a large collection of CRC cell lines. The reliability of the proposed approach was demonstrated for both positive and negative outliers by comparing our results with experimental data reported by previous studies, which also supported the clinical relevance of validated expression outliers. By coupling our outlier pipeline with publicly available gene and drug annotations, we observed that the catalogue of genes coding for protein endowed with enzymatic activity and scoring as positive gene expression outliers that might be druggable extended beyond the list of genes currently established as drug targets in CRC. Furthermore, we propose that negative gene expression outliers may unveil cancer cell vulnerabilities that are still therapeutically unexplored in CRC, an example of such a case is loss of expression of the MTAP gene. Finally, we reported that many gene expression outliers identified in our CRC cell lines and validated in an independent pan-cancer cell line dataset are significantly associated with different sensitivity to anticancer drugs, including cases without a matching between the gene involved in the gene expression alteration and the drug target.