Charting the transcriptomic landscape of primary and metastatic cancers in relation to their origin and target normal tissues

Most cancer deaths are caused by metastasis but understanding how metastatic cancers adapt from their origin tissues to their target tissues remains a fundamental challenge. Here, we ask if the gene expression of primary and metastatic tumors is more similar to their tissue of origin or to their target tissue. We analyzed the gene expression profiles of both paired and unpaired primary and metastatic tumor patient cohorts across multiple cancer types, studying single-cell and bulk RNA-seq data. To quantify the similarities between tumor samples and their normal, noncancerous origin and targeted tissues, we computed the Euclidean distance between their expression profiles, termed their transcriptomic distances, at the whole-genome and pathway levels. While the expression of most primary tumors is more similar to origin than to their target tissues, the expression patterns of metastases shift towards their target tissue. In specific cases, we chart the alterations occurring at the pathway level, pointing to key adaptive transcriptomic changes. Notably, our analysis indicates that primary cancers across several cancer types are primed for metastasis before the process begins. We further reproduce our findings using a new, unpublished RNA-seq data from colorectal cancer patients with liver metastases This first of its kind systematic analysis of the expression landscape of primary tumors and metastases with respect to their normal origin and target tissues provides a transcriptomic-wide view of cancer tumors adaptations to their metastatic niches