Systematic analysis of noncoding somatic mutations and gene expression alterations across 14 tumor types

Erik Larsson and colleagues present an analysis pipeline for identifying likely transcription-altering noncoding somatic mutations in cancer using publicly available data from 505 tumor genomes across 14 cancer types. They find that TERT promoter mutations show strong associations to altered transcriptional levels and identify recurrent promoter mutations in DPH3 and PLEKHS1 . Somatic mutations in noncoding sequences are poorly explored in cancer, a rare exception being the recent identification of activating mutations in TERT regulatory DNA. Although this finding is suggestive of a general mechanism for oncogene activation, this hypothesis remains untested. Here we map somatic mutations in 505 tumor genomes across 14 cancer types and systematically screen for associations between mutations in regulatory regions and RNA-level changes. We identify recurrent promoter mutations in several genes but find that TERT mutations are exceptional in showing a strong and genome-wide significant association with increased expression. Detailed analysis of TERT across cancers shows that the strength of this association is highly variable and is strongest in copy number–stable cancers such as thyroid carcinoma. We additionally propose that TERT promoter mutations control expression of the nearby gene CLPTM1L . Our analysis provides a detailed pan-cancer view of TERT transcriptional activation but finds no clear evidence for frequent oncogenic promoter mutations beyond TERT .