The relationship between body mass index and colorectal neoplasia: a systematic review and meta-analysis
Obesity control has been widely accepted as an effective primary preventive measure for colorectal cancer. However, the precise magnitude of the association between physician-measured body mass index (BMI) and colonoscopy-diagnosed colorectal adenoma (CRA) remains unknown. It is also uncertain whether there exist differences in this association among subjects with different characteristics. We aimed to evaluate the association between BMI and CRA, and examined if this association is different according to study design, gender and ethnicity.
We searched the Ovid Medline, EMBASE and the PsychINFO from their inception to September 2016. We included high quality observational studies that evaluated the relationship between physician-measured BMI and colonoscopy-diagnosed CRA. Two reviewers extracted the data independently, and any disagreement was resolved by a third reviewer. We examined the pooled odds ratio between BMI and CRA by a random effects model. Between-study heterogeneity was assessed by the Cochran’s Q statistic and I2 analyses. We also performed subgroup analysis according to study design, gender and ethnicity. The PRISMA statement was adhered to.
We included 17 studies (55,083 subjects) for analysis (Figure 1). Compared to subjects with BMI<25, the prevalence of CRA was significantly higher in those with BMI 25-30 (summary odds ratio [SOR]=1.44, 95% C.I. 1.30-1.61) and BMI>30 (SOR=1.42, 095% C.I. 1.24, 1.63; both p<0.001) (Figure 2). The pooled risks were found to have low to moderate levels of heterogeneity among studies (I2=43.0% and 18.5%, pheterogeneity =0.10 and 0.19, respectively). The Egger’s regression test (p=0.584) reported no publication bias. There was no difference in the pooled risk according to study design in subgroup analysis. The magnitude of association was significantly higher in female (SOR=1.43, 95% C.I. 1.30, 1.58; I2=9.4%, pheterogeneity=0.225) when compared to male subjects (SOR=1.16, 95% C.I. 1.07, 1.24; I2=0%, pheterogeneity=0.374). Different ethnic groups reported different associations (SOR=1.72, 95% C.I. 1.44, 0.88 in whites, Asians and Africans, respectively), which was absent in African subjects (SOR=0.88, 95% C.I. 0.61, 1.29, p=0.516; I2=0%, pheterogeneity=0.680). We evaluated all the potential covariates that might affect heterogeneity of the odds ratios, including age, gender, year of publication, smoking, alcohol drinking, use of aspirin or non-steroidal anti-inflammatory agents, family history of CRC, and presence of diabetes or hypertension for all studies. No covariates were reported to be significant by univariate and multivariate meta-regression analysis.
These findings are based on high quality studies, and showed that the risk for CRA conferred by BMI was significantly higher than that found in existing literature. Their association is higher in female and white individuals, and this bears implications in risk estimation of CRA in primary care settings.