Germline copy number variants and endometrial cancer risk

Abstract

Known risk loci for endometrial cancer explain approximately one third of familial endometrial cancer. However, the association of germline copy number variants (CNVs) with endometrial cancer risk remains relatively unknown. We conducted a genome-wide analysis of rare CNVs overlapping gene regions in 4115 endometrial cancer cases and 17,818 controls to identify functionally relevant variants associated with disease. We identified a 1.22-fold greater number of CNVs in DNA samples from cases compared to DNA samples from controls (p = 4.4 × 10^-63). Under three models of putative CNV impact (deletion, duplication, and loss of function), genome-wide association studies identified 141 candidate gene loci associated (p < 0.01) with endometrial cancer risk. Pathway analysis of the candidate loci revealed an enrichment of genes involved in the 16p11.2 proximal deletion syndrome, driven by a large recurrent deletion (chr16:29,595,483-30,159,693) identified in 0.15% of endometrial cancer cases and 0.02% of control participants. Together, these data provide evidence that rare copy number variants have a role in endometrial cancer susceptibility and that the proximal 16p11.2 BP4-BP5 region contains 25 candidate risk gene(s) that warrant further analysis to better understand their role in human disease.

Fig. 1

Manhattan plots for CNV-GWAS of 4,115 endometrial cancer cases and 17,818 controls. Genome-wide p-values for deletion-only (top), duplication-only (middle) and loss of function CNVs (bottom). Dashed line indicates Bonferroni derived genome-wide significance thresholds at 8.31 × 10^–6 for deletion-only, 5.97 × 10^–6 for duplication-only and 5.81 × 10^–6 for loss of function

Fig. 2

Overlap of putative endometrial cancer risk copy number variants with previously identified endometrial cancer risk and type II diabetes risk variants. Copy number deletions (red) and duplications (blue) in the region of HNF1B

Fig. 3

Significantly over-represented pathways for candidate genes derived from duplication-only (DUP) and loss of function (LOF) CNV_GWAS. Significantly enriched pathways are ordered by adjusted p-value (most-to-least significant) of 58 and 116 candidate genes derived from duplication-only and loss of function GWAS. Reactome (REAC) (Fabregat et al., 2018), KEGG (Kanehisa et al., 2019), WikiPathways (WP) (Slenter et al., 2018), Gene Ontology (GO) (Ashburner et al., 2011), Human Phenotype Ontology (HP) (Köhler et al., 2019) were selected as annotation databases. Heatmap on left depicts which CNV-GWAS candidate genes were overrepresented. Gene sets on right side of figure encompass multiple genes: 16p11.2A = SPN, QRPT, C16orf54, ZG16, MAZ, MVP, CDIPT, SEZ6L2, ASPHD1, KCTD13, TMEM219, HIRIP3, DOC2A, C16orf92, ALDOA, TBX6, GDPD. Stress = CYP1B1, FGF12, PPARA, BCLAF1, POLQ, FANCM, ERCC2, GML. Membrane = SLC6A3, SLCO1B3, DLG2, TMEM231, SLC19A1, SLC4A7. Genes denoted with * denote additional gene loci identified via recurrent 16p deletion identified in LOF CNV-GWAS but were also represented in other enriched pathways

Fig. 4

Gene set enrichment analysis for candidate risk genes derived from duplication-only (DUP) and loss of function (LOF) CNV-GWAS. FUMA gene set enrichment analysis results for candidate genes derived from DUP and LOF CNV-GWAS (n = 58 and n = 116, respectively). Adjusted p-values presented. Gene sets on right side encompass two sets of genes, all of which are at 16p11.2 and driven by recurrent deletion identified. 16p11.2.A = SEZ6L2, ASPHD1, KCTD13, TMEM219, TAOK2, HIRIP3, INO80E, DOC2A, ALDOA, PPP4C, TBX6, YPEL3, GDPD3. 16p11.2.B = TMEM219, TAOK2, HIRIP3, INO80E, DOC2A, ALDOA, PPP4C