167 Subsite Specific Risk of Colorectal Cancer Risk in Patients Treated for Hodgkin's Lymphoma: A Long-Term Follow-up Study in the Netherlands

Abstract

BACKGROUND. Recent genome-wide association studies have identified at least 20 germline single nucleotide polymorphisms (SNPs) significantly associated with colorectal cancer incidence. Common genetic variation is likely also related to survival after colorectal cancer diagnosis; however, the genome-wide analysis approach has not been widely applied to the study of colorectal cancer survival. METHODS. Full genome scans, and imputation to HapMap, were conducted for 2,822 men and women with invasive colorectal cancer who were enrolled in one of six studies: the Health Professionals Follow-up Study (N=173), the Nurses' Health Study (N=298), the Physicians' Health Study (N=325), the Vitamins and Lifestyle Study (N=285), the Postmenopausal Hormones Study (N=280), and the Women's Health Initiative (N=1,461). We used Cox proportional hazards regression to assess SNPspecific associations with disease-specific survival and overall survival in models adjusted for age, sex, and principal components. SNPs were modeled additively to reflect associations per copy of the minor allele. Results were combined across studies via fixed-effects metaanalysis. To account for multiple-testing, a p-value threshold of p,1x10-8 was required for genome-wide significance. RESULTS. No common SNPs (i.e., minor allele frequency ≥5%) were significantly associated with survival outcomes, including previously identified colorectal cancer susceptibility SNPs. Two SNPs of low minor allele frequency were associated with disease-specific survival at a level exceeding or approaching genome-wide significance: the minor allele in one SNP near APC2, with a frequency of 1%, was associated with significantly poorer survival (p=1.2x10-9), and the minor allele in another SNP near PRIM2, present in 3% of cases, attained borderline significance in relation to poorer survival (p= 2.0x10-8). Two other SNPs, one near MAML2 and one near ZNF804B, exhibited adverse associations of borderline significance with both overall survival (p=8.4x10-7 and 6.7x107, respectively) and disease-specific survival (p=1.1x10-7 and 5.0x10-7); both of these SNPs also had low minor allele frequencies (1% and 2%, respectively). CONCLUSIONS. Common genetic variation, particularly with respect to polymorphisms present in ≥5% of the population, does not appear to be significantly associated with colorectal cancer survival. There is still, however, some indication that germline genetic variation is related to colorectal cancer survival outcomes. Our findings for SNPs with lowminor allele frequencies provide suggestive evidence of adverse survival outcomes in those with specific, less common polymorphisms. In light of the low frequency of the minor alleles in these SNPs, false-positive results are possible; thus, further follow-up is needed.