Genome wide gene-environment interaction study identifies a CYP24A1-related variant as a modifier of colorectal cancer risk associated with menopausal estrogen plus progesterone therapy. X. Garcia-Albeniz1, A. Rudolph2, C. M. Hutter3, Y. Lin3, E. White3, H. Brenner4, G. Casey5, S. Gallinger6, L. Hsu3, T. J. Hudson7, L. Le Marchand8, J. Potter3, M. Slattery9, B. Zanke10, P. A. Newcomb3, A. T. Chan11, U. Peters3, J. Chang-Claude2 1) Epidemiology, Harvard School of Public Health, Boston, MA; 2) Epidemiology, German Cancer Research Center, Heidelberg, Germany; 3) Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA; 4) Clinical Epidemiology and Aging Research, German Cancer Research Center, Heidelberg, Germany; 5) Preventive Medicine, University of Southern California, Keck School of Medicine, Los Angeles, CA; 6) Surgery, Mount Sinai Hospital, New York, NY; 7) Medical Biophysics, University of Toronto, Toronto, Canada; 8) Epidemiology, University of Hawaii Cancer Center, Honolulu, HI; 9) Internal Medicine, University of Utah Health Sciences Center, Salt Lake City, UT; 10) Clinical Epidemiology. Ottawa Hospital Research Institute, Ottawa, ON; 11) Gastroenterology, Massachusetts General Hospital and Harvard Medical School, Boston, MA.
Postmenopausal hormone therapy use (PMH) has been consistently associated with a decreased risk of colorectal cancer (CRC). The underlying pathways involved in prevention of CRC by PMH are largely unknown. Our aim was to use a genome-wide gene-environment analysis to identify variants modifying the effect of PMH on the risk of CRC. We included 9 studies from the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO) and the Colon Cancer Family Registry. Genome-wide genotype array data from each study was imputed to the HapMap II population (CEU). PMH use was evaluated as use at reference time of any PMH, estrogen only (E-only), and estrogen+progesterone (E+P). To test for multiplicative (GxPMH) interactions, we used a case-control logistic regression analysis as well as a two-stage Cocktail method that includes a screening step to prioritize SNPs followed by a testing step for GxPMH interactions using weighted multiple hypothesis testing. In secondary analyses, we applied another two stage method that jointly tests marginal association and GxE interaction as well as an Empirical Bayes method. Final results for each method were obtained by combining study level results with fixed-effects meta-analyses, comprising up to 10,835 postmenopausal women (type of compound known on 9,674). The genome-wide interaction analysis using case-control logistic regression did not yield any genome-wide significant (p<5.0x10-8) interaction with PMH use. Based on the Cocktail method, we observed a significant interaction of the variant rs964293 with E+P (OR=0.65, p=2.8x10-5; threshold for significance=3.1x10-4). This variant also showed significant interaction with E+P use on CRC risk using the Empirical Bayes method (OR=0.62, 95% CI 0.53-0.73, p=9.1x10-9). We did not detect any genome-wide significant interaction when considering E-only or any PMH use. The variant rs964293 is located in an intergenic region 28 kb upstream of CYP24A1 on chromosome 20q13.2. Among women not using PMH, rs964293 was marginally associated with an increase in risk for CRC (per allele OR=1.08, 95% CI 1.00-1.16, p=0.05) whereas among women taking E+P, rs964293 was associated with a lower risk of CRC (per allele OR=0.72, 95% CI 0.59-0.88, p=0.0011). We identified a CYP24A1-related variant as effect modifier of CRC risk associated with use of E+P, using a genome-wide approach. This finding offers insight in the role of PMH and its downstream pathways in the etiopathogenesis of CRC.
You may contact the first author (during and after the meeting) at