Obesity is recognised as a risk factor for advanced colorectal neoplasia (ACN). Body mass index (BMI) is used for obesity, while waist circumference (WC) for abdominal obesity. Several criteria exist for diagnosing obesity in subjects of different races and ethnicities, yet it is unclear which measure of obesity is a better predictor of ACN in colorectal cancer (CRC) screening.Methods
In 2016 and 2017, we performed a prospective, multiple-centre, colonoscopy study that consecutively recruited asymptomatic subjects in seven endoscopy units across North Shanghai, China. Self-reported questionnaires for risk factors of CRC were collected in the study sites. All participants’ body weight, height, and WC were measured by research assistants using standardised methods. All the colonoscopies were completed by senior endoscopists. Taking ACN as the outcome, we conducted univariate binary regression to identify the additional risk of ACN conferred by BMI/MC. We then used their different cut-off levels, comparing separate multivariate regression results to select the one with a highest odds ratio (OR) as the best predictor. The risk scores of all predictors and C-statistics was reported for the regression model with obesity index having the highest OR.Results
A total of 2813 subjects were enrolled (mean age 58.7±8.4 years, 52.0% male), of which 142 (5.0%) was diagnosed with ACN. In univariate regression, several BMI/WC criteria were significant, with age, gender, family history, smoking status and diabetes (table 1). The Asian criteria for overweight (BMI ≥23) attained the highest adjusted OR (1.60, 95% confidence interval (CI) 1.08, 2.36), when compared with international criteria for overweight (BMI ≥25) (OR 1.46, 95% CI 1.01, 2.10) and Asian criteria for abdominal obesity (male ≥85 cm, female ≥80 cm) (OR 1.52, 95% CI 1.03, 2.24) (table 2). The C-statistics of the new model with BMI ≥23 as the predictor for ACN was 0.648 (95% CI 0.600, 0.696).Conclusions
The Asian criteria for obesity had the most discriminatory performance to predict ACN as compared to other obesity measures in this asymptomatic Chinese cohort. These findings imply BMI≥23 may be most suitable ‘obesity-predictor’ of ACN in risk algorithms.