Abstract B8: Blood lipid concentrations and colorectal adenomatous polyps: A systematic review, meta-analysis, and meta-regression of endoscopy studies, 1986–2010

Abstract Objective: To summarize the association between blood lipid concentrations and the prevalence of colorectal adenoma in order to inform ongoing research on cholesterol-lowering interventions and colorectal disease. Data sources and analyses: We systematically searched MEDLINE for studies that performed colonoscopy or sigmoidoscopy for any indication and measured lipid concentrations from a fasting blood draw at the time of endoscopy. In these studies, individuals found to have adenoma of any size were considered cases, and those without any colorectal lesion were considered controls. Included studies reported results of serum or plasma concentrations of at least one of the following lipid measurements: total cholesterol (TC), low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), or triglycerides (TG). Studies were excluded if there were less than 30 adenoma cases or 30 controls, if it was unclear that all controls were screened negative of any colorectal lesion, or if reporting errors were suspected. Differences in the mean blood concentration in mg/dL for each lipid component were pooled using DerSimonian-Laird random effects models with inverse variance weighting. Random effects meta-regression was used to estimate pooled effect estimates that adjusted for study-specific covariates (calendar year of endoscopy, proportion of study participants who were men, difference in mean age between cases and controls, and whether the study excluded individuals that had undergone previous polypectomy). Results: Eighteen of 121 endoscopy studies identified in our MEDLINE search that were conducted between 1986 and 2010 met the inclusion criteria. Combined, these studies included 6,645 adenoma cases and 21,335 controls. In total, 15 of 18 studies measured TC and TG, 12 of 18 measured HDL, and 6 of 18 measured LDL. Adenoma cases had significantly higher TC than controls (adjusted mean difference (MD) = 5.13 mg/dL; 95% confidence interval (CI): 3.43, 6.84), and higher TG than controls (MD = 18.42 mg/dL; 95% CI: 11.71, 25.12). HDL levels were 2.10 mg/dL (95% CI: −1.37, −2.83) lower in individuals with adenoma compared to controls, but LDL was not significantly different (MD = 1.37; 95% CI: −1.67, 4.41). Collectively, the four covariates accounted for a statistically significant portion of heterogeneity in meta-regression analyses for TC, HDL, and LDL, but not for TG (P = 0.55, 0.28, 0.46, and < 0.001, respectively, from a partitioned Q-statistic for the residual heterogeneity of the fully-adjusted models). Confounding by risk factors such as smoking status and body mass index could not be ruled out, but sub-group analyses based on these factors provided similar results. Conclusions: In a meta-analysis of endoscopy patients, individuals with colorectal adenoma had higher TG and lower HDL than patients found to be free from adenoma (overall effect P < 0.001 for both). These results suggest that LDL, the lipoprotein most affected by common cholesterol-lowering medications such as statins, was not associated with the prevalence of adenoma. Citation Information: Cancer Prev Res 2011;4(10 Suppl):B8.