A Pan-Cancer Panopticon

With cancer a constant target in research and testing, reporting often focuses on specific cancer types and how tailored treatment can exploit the quirks within individual cancers. But what about studies that cut across cancer categories? Pan-cancer omics makes us step back and consider the wider oncological picture, allowing us to identify mutations, genomic events, and aberrations across the spectrum of cancers.

A key goal in oncology is to curtail mutations that improve cancer’s ability to spread. So far, because modeling such variable mutation rates is no mean feat, most attempts to identify these types of mutations have focused on protein-coding sequences and specific noncoding elements. However, researchers have devised a novel method that uses deep neural networks to search for mutations across the cancer genome. Succinctly known as “Dig,” the technique maps estimates of cancer mutations which are then refined through comparison with predicted mutation counts (1). So far, Dig has mapped mutation rates in 37 cancer types – and the data are even available for web-based exploration.

RNA studies are an equally fruitful avenue for pan-cancer research; previous research has suggested that mRNA content is linked to tumor phenotypes, but technical complexity has prevented further exploration. Recently, a team from Texas developed a technique to measure the amount of total mRNA expression (TmS) from a cancerous tumor – factoring in transcript proportion, purity, and ploidy – and compare it with the amount produced by regular cells to help predict disease progression and tumor phenotypes (2). The method was tested on 6,590 tumors across 15 cancer types, revealing a link between high TmS and risk of cancer progression and death. TmS was also seen to have a relationship with “cancer-specific patterns of gene alteration and intra-tumor genetic heterogeneity (2),” alongside cross-cancer metabolic dysregulation.

And what of chromosomal instability (CIN)? This DNA-affecting process has a long-established association with cancer, yet there is no systematic method to measure CIN types and their effects on cross-cancer phenotypes. New research has led to a comprehensive anthology of CIN origin and diversity – representing over 7,800 tumor specimens from 33 types of cancer. The researchers codified 17 copy number signatures – each exemplifying a different CIN type – that help to forecast drug response and inform potential new treatment options. The finalized compendium highlights the structure underlying genomic complexity in human cancers and provides a useful resource for future studies (3).