Escape from X-Inactivation Tumor Suppressor (EXITS) Genes Are Associated with Excess Incidence of Cancer in Men

There is a striking male predominance in hematologic malignancies, as well as many solid tumors, but this sex bias is not understood. US males carry an age-adjusted increased risk of 20.4% for developing any cancer, and ≥2:1 male predominance for some cancer types. This risk leads to >150,000 excess new cases of cancer in US men annually. We sequenced DNA from blastic plasmacytoid dendritic cell neoplasm (BPDCN), an aggressive leukemia with significant male predominance (>3:1 M:F). The splicing factor ZRSR2, located on chrX, had acquired loss-of-function mutations in ~30% of BPDCNs, strikingly all from males (P<0.05 vs females). Somatic mutations in ZRSR2 also occur in 3-5% of myelodysplastic syndrome (MDS), which is a male-biased disease (1.6:1 M:F); 100% of previously reported loss-of-function mutations in MDS are in men (Yoshida et al, Nature 2011). These observations highlighted a conundrum. Although females have two X chromosomes (chrX), one copy is inactivated in all somatic cells. Therefore, mutations in chrX tumor suppressors should be equally likely in male and female cancers, because both have only one active copy of each gene vulnerable to a deleterious mutation. However, a small number of genes, including ZRSR2, escape X-inactivation in female cells and are expressed from both maternal and paternal alleles. We hypothesized that mutations in genes that escape X-inactivation could underlie a significant fraction of excess male cancers, as males would require only a single mutation while females would require two. For example, in our analysis, >50% of the excess male risk of BPDCN and 20% of the excess male risk of MDS are associated with ZRSR2 mutation. We termed these genes ‘EXITS’, for Escape from X-Inactivation Tumor Suppressors. A previous study in T-ALL identified KDM6A (UTX) as a possible EXITS gene (Van der Meulen, Blood 2015). To query for EXITS genes in an unbiased manner, we designed computational algorithms to test for male predominance of chrX loss-of-function alterations (SNV, InDel, or copy number loss) in >4100 cancers of 21 types from The Cancer Genome Atlas (TCGA). We discovered putative EXITS genes that were mutated more often in males than females (P<0.001, FDR<0.1), including ATRX, CNKSR2, DDX3X, KDM5C, KDM6A, and MAGEC3. The same approach identified no silent mutations on chrX with male bias, no mutations on chr7 (similar in gene content to chrX) with a male bias, and no mutations on chrX with female bias. Overall, male-predominant loss-of-function mutations were enriched in genes that have been shown to escape X-inactivation (P=0.011). In the case of ATRX, nearly all of the male predominance was seen in glioma, whereas the M:F mutation ratio in all other cancers was essentially equal. Strikingly, using RNA-seq data from the GTEx project we found that ATRX undergoes escape from X-inactivation in female brain but not peripheral blood. This suggests that ATRX escaping X-inactivation in female brain could provide protection against gliomagenesis, and that tissue-specific patterns of escape could influence sex biases between cancer types. A corollary to the EXITS hypothesis is that chrY loss will be more frequent in males who also have mutations in EXITS genes. ChrY harbors evolutionarily ancestral homologs of chrX genes, and chrX genes with Y homologs are more likely to escape X-inactivation. ChrY loss is commonly observed in a range of cancers. Age-related loss of Y is also frequent in non-malignant blood cells, is increased among tobacco users, and is associated with a higher risk for non-hematological cancer. ChrX genes with Y homologs had a male-biased mutation pattern compared to other chrX genes (P=0.018). We classified chrY copy number based on exome sequencing in 949 male tumors and found that cancers with loss of chrY were more likely to have loss-of-function mutations in chrX escape genes than those without loss of Y (30.2% versus 15.9%, P=0.016). Thus, tobacco use may disproportionately increase the risk of some cancers in males, as mutation of chrX genes that have Y homologs but escape X-inactivation would preferentially lead to complete gene inactivation in male cells that have lost chrY by virtue of smoking. In summary, genes escaping X-inactivation may provide protection from cancer in females, and a significant portion of the male predominance of certain cancers, including several hematologic malignancies, may be due to the overrepresentation of EXITS gene mutations in men.