Metastasis: Leveraging transcriptomics to identify potential therapeutics

By April 18, 2018April 23rd, 2021Publications

Five-year survival rates for patients whose cancer has metastasized are dramatically lower than for those with localized disease. However, few therapeutic development efforts specifically focus on inhibiting the process of metastasis.


Here we perform a systematic pan-cancer analysis of RNA-seq data from The Cancer Genome Atlas to identify targetable gene expression changes in primary tumors that confer metastatic potential. In 4,844 patients comprising 13 different cancer types, we construct a gene expression signature that captures consistent pan-cancer alterations that differ between primary tumors associated with node-negative disease with no recorded distant metastasis (N0 not M1) and primary tumors from patients with node-positive disease (N1, N2, or N3). In order to identify potential therapeutic targets, we query this metastasis signature against drug-induced transcriptomic signatures from CMAP and LINCS. Among the top candidates is Chembl410456, which modulates the TGFβ and Wnt/β-catenin signaling pathways, among others. Genes upregulated in metastatic primary tumors and downregulated by Chembl410456 include SERPINE1 and TWIST1. We observe that low SERPINE1 and TWIST1 expression also associate with improved prognosis in multiple cancers, including colon adenocarcinoma (p=0.007 and p=0.014, respectively). Other top-ranked drugs also target the TGFβ signaling pathway. To summarize, we leverage public RNA-seq data to identify a pan-cancer metastatic signature that can be interrogated to systematically identify potential therapeutics that target metastasis. Building on this, we pinpoint antimetastasis drug candidates including several with mechanisms of action involving modulation of TGFβ and/or Wnt/β-catenin pathways.