Genome-wide significant risk associations for mucinous ovarian carcinoma

Abstract

Genome-wide association studies have identified several risk associations for ovarian carcinomas but not for mucinous ovarian carcinomas (MOCs). Our analysis of 1,644 MOC cases and 21,693 controls with imputation identified 3 new risk associations: rs752590 at 2q13 (P = 3.3 × 10(-8)), rs711830 at 2q31.1 (P = 7.5 × 10(-12)) and rs688187 at 19q13.2 (P = 6.8 × 10(-13)). We identified significant expression quantitative trait locus (eQTL) associations for HOXD9 at 2q31.1 in ovarian (P = 4.95 × 10(-4), false discovery rate (FDR) = 0.003) and colorectal (P = 0.01, FDR = 0.09) tumors and for PAX8 at 2q13 in colorectal tumors (P = 0.03, FDR = 0.09). Chromosome conformation capture analysis identified interactions between the HOXD9 promoter and risk-associated SNPs at 2q31.1. Overexpressing HOXD9 in MOC cells augmented the neoplastic phenotype. These findings provide the first evidence for MOC susceptibility variants and insights into the underlying biology of the disease.

Figure 1

Manhattan plots showing association between risk of MOC and the genotypes of SNPs in a 1Mb region of re-imputation surrounding the most significantly associated SNP at (a) 2q13 (top SNP: rs752590), (b) 2q31.1 (top SNP: rs711830) and (c) 19q13.2 (top SNP: rs688187). Sample size is 1,644 cases and 21,693 controls. Red dots indicate a genotyped SNP in COGS, progressively darker grey dots indicate SNPs with pre-phased imputation r² values between 0.30 and 0.60, 0.60 and 0.80 and 0.80 to 0.95, respectively, and black dots indicate SNPs with pre-phased imputation r² values between 0.95 and 1.0.

Figure 2

The epigenetic landscape of MOC risk regions. All candidate causal SNPs are shown, the most significantly associated risk SNP is indicated in red. Data for normal immortalized ovarian epithelial (IOE) cells, serous ovarian cancer cells (CaOV3, UWB1.289, used in the absence of analogous data for mucinous ovarian cancer cells) colon cancer cells (HCT-116) and normal tissue from Hniez et al (colonic crypts, sigmoid colon and ovary). At both (a) 2q31.1 and (b) 2q13 there is extensive overlap between regulatory biofeatures and risk SNPs. We also included collated ENCODE ChIPseq data for CTCF at 2q13, since PAX8 is rarely expressed in invasive mucinous ovarian cancer, SNPs that coincide with repressor marks could be the most relevant for this disease subtype.

Figure 3

Expression quantitative trait locus (eQTL) at MOC risk regions. Boxplots show the median (horizontal line), 1^st to 3^rd quartiles of expression (box) and 1.5 × the interquartile range (whiskers) in arbitrary units (a.u.). Tumor data are from The Cancer Genome Atlas for 339 high grade serous ovarian cancers (HGSOC) and 121 colorectal cancers (CRC). Significant associations were found between HOXD9 gene expression at the 2q31.1 region and genotypes of the risk SNP rs711830 in (a) CRC (P = 0.01) and (b) HGSOC (P = 4.95 × 10⁻⁴). (c) A significant association was found between PAX8 gene expression at the 2q13 region and genotypes of the risk SNP rs6542127 in CRC (P = 0.03). P-values are corrected for false-discovery-rate and considered to be a significant below 0.1.

Figure 4

Chromosome conformation capture (3C) at the 2q31.1 region containing HOXD9 performed in the EFO27 mucinous ovarian cancer cell line. Csp6i restriction enzyme fragments containing the MOC risk SNPs rs4972504 (T2 fragment), rs2857532 (T10 fragment) and rs2072590 (T16 fragment) show evidence of interaction with the HOXD9 promoter region, defined as 1.5 kb upstream of the transcription start site. (a) Schematic showing the locus, all SNPs with a 1:100 chance of being the causal variant, the locations of the interacting fragments (horizontal red bars), the risk SNPs located within each interacting region (vertical red bars) and the HOXD9 promoter bait region (blue bar). (b) PCR confirmation of bait-target interactions (+ lanes) of the predicted size (L=ladder) and absence in controls (– lanes) without the addition of ligase. DNA sequencing chromatograms show interactions between the bait region (blue sequence) ligated to the target region (red sequence) and the intervening Csp6i restriction enzyme site (black box).

Figure 5

Modeling the effects of HOXD9 overexpression in two in vitro models of MOC. Overexpression of HOXD9 was confirmed by (a) qPCR in arbitrary units (a.u.), error bars are standard deviation (SD) and represent two independent experiments and (b) immunofluorescence microscopy (200 × magnification). Overexpression of HOXD9 is associated with increased anchorage independent growth. (c) Data shown are mean relative colony numbers from three independent experiments ± SD. Two-tailed paired t-test. (d) Phase-contrast images of colony growth. Examples of colonies are indicated with arrowheads.