Fig. 3: Pan-cancer DNA copy number alteration correlates of patient survival.

a Across the TCGA pan-cancer cohort, the global extent of DNA copy number alteration (CNA) is associated with worse OS. The CNA index is the standard deviation of the gene-level CNA values across genes in the tumor profile. Stratified log-rank p value corrects for cancer type. b The numbers of genes significantly associated with patient survival are indicated (by Cox), with or without correction for cancer type, and with cancer type correction and with RNA patient survival association (FDR < 10%) in the same direction. c A top set of 1253 genes (out of 24,776) had both RNA and CNA significantly associated with patient survival in the same direction (FDR < 10% for each, Cox correcting for cancer type) in pan-cancer analyses across 10,843 cancers. For each individual cancer type, the respective gene-level CNA associations with patient survival are indicated for the 1253 genes. d Enriched cytoband regions involving the gene sets from part (c) (worse versus better survival gene sets, with both RNA and CNA association with survival in the same direction). These top regions are enriched with FDR < 10%⁴⁷ by one-sided Fisher’s exact test and with more than one-third of the genes in the cytoband region being included in the pan-cancer survival-associated genes. e The entire set of genes in the 7p15.2 region was applied as a gene signature to eight independent gene expression datasets outside of TCGA (from Fig. 1f). For four datasets, the 7p15.2 signature was associated with worse survival. f Similar to part (e) but applying a signature of all genes in the 10q25.2 region, where four out of eight datasets showed an association with better survival.