Metabolic syndrome and risk of advanced colorectal adenomas in a screening population: Frequentist and Bayesian...

Screening colonoscopy prevents colorectal cancer (CRC) by detecting and removing premalignant lesions. Metabolic syndrome (MetS) has been proposed as an additional risk marker to refine risk stratification beyond age-based screening, but its independent association with advanced colorectal neoplasia remains uncertain.

We conducted a cross-sectional analysis of asymptomatic adults undergoing screening colonoscopy in the Salzburg Colon Cancer Prevention Initiative. Analyses were limited to participants with complete-case data for the Adult Treatment Panel III (ATP III) MetS definition (N = 4891). The primary endpoint was advanced colorectal lesions, defined as advanced adenoma and/or CRC. We estimated incidence rate ratios (IRR) using Poisson regression with robust variance in unadjusted, age/sex-adjusted and fully adjusted models (demographic and lifestyle covariates). Sensitivity analyses applied the International Diabetes Federation (IDF) MetS definition and insulin resistance (HOMA-IR; binary and per doubling). In an exploratory subsample, leptin (per doubling) was evaluated, including joint models with MetS. Bayesian models (non-informative, pessimistic, sceptical priors) quantified posterior effect distributions and equivalence probabilities.

Advanced lesions occurred in 389/4891 participants (7.95%) and were more frequent in ATP III MetS (9.66% vs. 6.97%; p = 0.001). In frequentist analyses, ATP III MetS was associated with advanced lesions in unadjusted models (IRR: 1.39; 95% CI, 1.15-1.68) but not after age/sex adjustment (IRR: 1.09; 95% CI, 0.90-1.32) or full adjustment (IRR: 1.06; 95% CI, 0.83-1.36). Age was the dominant predictor (fully adjusted IRR: 1.73 per decade; 95% CI, 1.53-1.95), while female sex was protective (IRR: 0.57; 95% CI, 0.44-0.73). Findings were concordant using the IDF definition (unadjusted IRR: 1.39; age/sex-adjusted 1.09; fully adjusted 1.07) and for HOMA-IR (binary: no association; per doubling: unadjusted IRR: 1.10, p = 0.052; adjusted null). In the leptin subsample (N = 602), leptin was not independently associated with advanced lesions (per doubling IRR: 0.85 unadjusted; 0.96 age/sex-adjusted; 0.95 fully adjusted) and did not materially alter MetS estimates. Bayesian analyses mirrored attenuation with adjustment; in fully adjusted models, sceptical priors yielded a posterior median IRR of 1.01 (95% CrI: 0.91-1.12) with a 92.9% probability that the MetS effect lay within ±10% of no effect.

The apparent excess risk of advanced colorectal lesions in MetS is explained by age and sex. Across MetS definitions, HOMA-IR, leptin and Bayesian sensitivity analyses, MetS does not provide independent incremental information for CRC screening risk stratification beyond established demographic factors.