CASP8 at 2q33 is an established common, low risk susceptibility locus for breast cancer but functional variants are yet to be elucidated. Previously we used haplotype mining of 45 tSNPs in 3778 breast cancer cases and controls to identify a risk haplotype maximizing association evidence. A protective haplotype was also suggested. Here, we perform a focused sequencing study to identify the critical sequence variant/s.

To not exclude potential regulatory variants, we designed a custom Agilent SureSelect capture array to target all non-repetitive sequence in the 1 Mb region surrounding CASP8. The resulting 41 libraries for germline DNA (24 breast cancer cases homozygous for the risk haplotype, 17 cancer-free women carrying the protective haplotype) were sequenced on the Life Technologies SOLID system. A total of 1464 sequence variants were prioritized by their propensity to lie on the risk or protective haplotypes, resulting in the selection of 42 variants to follow-up with expression-QTL analysis (MAFs 0.1-0.4). Our RNA sequencing panel comprised 156 tissue samples from 88 women (68 breast tumor, 88 normal breast). Illumina TruSeq stranded mRNA sequencing was performed on an Illumina HiSeq 2000. Genotyping for the 42 selected variants was performed using Illumina BeadXpress. RNAseq eQTL analysis focused on 4 genes: CFLAR, CASP10, CASP8, ALS2CR12 using counts standardized to fragments per kilobase per million bases (FPKM). Differences in FPKM between carriers and non-carriers of the 42 DNA sequence variants were assessed in both normal and tumor tissues. Replication analyses were carried out on available TCGA data and published blood eQTL microarray data.

In normal tissue, an intronic variant in ALS2CR12 on the risk haplotype was best associated with CASP8 expression (19% decreased FPKM, p=9x10-5). Eight other variants (all in ALS2CR12 or CASP8) showed similar CASP8 decreases (p<10-3). Increased ALS2CR12 expression was also observed for the same variants, and decreased CASP8 expression was seen in tumor. Despite the difference in technologies, lower coverage and sparser genotyping, the decreased CASP8 and increased ALS2CR12 levels were replicated in both in the TCGA normal tissue (p=3x10-4) and blood eQTL data (p=4x10-14).

In tumor, our strongest eQTL results were for decreased expression of CFLAR (p=1x10-4) and CASP10 (p=4x10-5), identified for variants on the protective haplotype (including rs1045485). Examination of TCGA breast tumors did not replicate these findings; however read depths were substantially lower for these genes in the TCGA data.

We find evidence for two separate paths to 2q33 breast cancer risk. Reduction of CASP8 expression, which may lead to reduced apoptosis and an environment conducive to tumor development. Reduction of CFLAR-CASP10 expression in tumor may implicate autophagy leading to a challenging environment for tumors, but requires more replication.

Citation Format: Nicola J. Camp, Wei-Yu Lin, Alex Bigelow, Marina A. Parry, Tim Mosbruger, George Burghel, Venkatesh Rajamanickam, Sushilaben H. Rigas, Rachel Cosby, Dan Connley, Guoying Wang, Tresa George, Rosalie Waller, Lisa A. Cannon-Albright, Brandt Jones, Rob Sargent, Malcolm W.R. Reed, Angela Cox. Targeted DNA and RNA sequencing identifies breast cancer risk variants associated with differential expression of CASP8 and CFLAR/CASP10. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 3267. doi:10.1158/1538-7445.AM2014-3267