Estimation of controlled direct effects of polyunsaturated fatty acids on colorectal adenoma risk using inverse probability weighting

Abstract Omega-3 and omega-6 polyunsaturated fatty acids exert putatively divergent effects on inflammation, partly through differential production of prostaglandin E2 (PGE-2), a pro-inflammatory metabolite implicated in colorectal carcinogenesis. To elucidate polyunsaturated fatty acid effects independent of downstream PGE-2, we estimate the controlled direct effect (CDE) of five polyunsaturated fatty acids on colorectal adenoma risk while holding the major urinary metabolite of PGE-2 (or PGE-M) levels constant. We analyzed data from 819 participants from the Tennessee Colorectal Polyp Study (TCPS), a colonoscopy-based case–control study conducted in Nashville, Tennessee between 2003 and 2010. The analytic sample included participants with measurements of red blood cell polyunsaturated fatty acids, PGE-M concentration, and relevant covariates. Total effects were estimated using multivariable regression and CDEs were estimated using inverse probability weighting (IPW). Our analysis suggested a notable 34% reduction (ORCDE = 0.66, 95% CI = 0.54, 0.82) in colorectal adenoma risk for the controlled direct effect of one standard deviation increase in omega-3 eicosapentaenoic acid (EPA) levels, independent of PGE-M levels. Our findings suggest a putative effect of EPA on reducing colorectal adenomas independent of changes in downstream PGE-M levels in the TCPS population, highlighting its potential role in colorectal adenoma prevention.