Determinants of bile acids.

Abstract

632 Background: Several epidemiological studies have demonstrated an association between bile acids (BA) and colorectal cancer. Despite this evidence, there are limited studies that examine the determinants of bile acids in humans. We conducted a secondary, cross-sectional analysis on 735 adenoma formers from a phase III adenoma prevention trial of ursodeoxycholic acid (URSO) who provided a fecal sample at their baseline visit. Methods: Fecal measures of primary bile acids cholic acid (CA) and chenodoxycholic acid (CDCA), as well as secondary bile acids, deoxycholic acid (DCA), lithocholic (LCA) were the outcome variables of interest. Dietary, non-dietary, medication use, genetic variation in the CYP7A1 gene (as single nucleotide polymorphisms (SNPs) and haplotypes), and physical activity, and characteristics of baseline adenoma were evaluated as correlates of fecal BAs. We used maximum likelihood methods to incorporate left-censoring in the BAs that had concentrations below the detection limit (BDL). These log-normal parametric survival models were used to fit potential predictors of fecal BA levels, stratifying by sex. The Akaike information criterion (AIC) was used to choose the best predictive model from each BA outcome. Model accuracy was assessed using leave-one-out cross validation and a likelihood-based measure of R2 and the Brier score. Results: LDL and cholesterol were predictors for primary bile acid CDCA in men and women, as was the use of pain medication (use of medication lowered BA levels). Genetic variation in CYP7A1 (as haplotypes and individual SNP's) was a significant determinant of CA level in both men and women. BMI was an important predictor for both men and women for DCA and CA. Presentation with tubular-villous histology was associated with LCA levels whereas proximal adenoma was an important predictor for CA levels, in men and women. Conclusions: This secondary analysis of a uniquely large dataset validates previous studies that suggest lifestyle and genetic factors influence colonic exposure to carcinogenic BAs. These findings also importantly highlight gender-specific differences in BA determinants that may explain colorectal cancer incidence differences between men and women.