Investigating the association between anthropometry and colorectal cancer survival: a two-sample Mendelian randomization analysis

Abstract

Background: Observational epidemiologic studies on the association of anthropometric traits and colorectal cancer (CRC) survival provide inconsistent results, and potential limitations prohibit the investigation of causality. We examined the associations between seven genetically predicted anthropometric traits [height, body mass index (BMI), waist circumference (WC), hip circumference (HC), waist-hip circumference ratio, birth weight and body fat percentage] and CRC-specific mortality among CRC cases using two-sample Mendelian randomization (MR).

Methods: Analyses were performed using 16 964 CRC cases, out of which 4010 died due to their disease, from the Genetics and Epidemiology of Colorectal Cancer Consortium and Colon Cancer Family Registry. We further conducted stratified analyses by anatomical site and stage. We applied the inverse variance weighted approach, and sensitivity analyses were conducted to assess the impact of potential violations of MR assumptions and adjust for collider bias.

Results: One standard deviation (SD 13.4 cm) higher genetically predicted levels of WC were associated with worse CRC survival [hazard ratio (HR); 1.22, 95% confidence interval (CI); 1.02-1.47]. Positive associations were further observed for a SD higher genetically predicted BMI (SD; 4.8 kg/m2, HR; 1.5, 95% CI; 1.15-1.95) and HC (SD; 9.2 cm, HR; 1.32, 95% CI; 1.02-1.73) and CRC-specific mortality in cases of stages II/III. The latter associations were generally robust to sensitivity analyses. Positive but imprecisely estimated associations were found for most other anthropometric traits.

Conclusions: Despite the limitations of cancer survival research, our findings support that CRC cases should avoid obesity. Further research should inform the development of recommendations targeting overweight/obesity management during cancer surveillance.

Keywords: Mendelian randomization; birth weight; body fat percentage; body mass index; colorectal cancer; height; hip circumference; survival; waist circumference; waist–hip circumference ratio.

Figure 1.

Directed acyclic graph demonstrating a typical example of collider bias in the anthropometry-colorectal cancer survival association in the two-sample Mendelian randomization setting. The genetic instrument is strongly associated with anthropometry, anthropometry is associated with risk of colorectal cancer incidence, and a measured or unmeasured confounder works as a common cause of both colorectal cancer incidence and survival (black solid lines). Conditioning on colorectal cancer incidence induces the association between the previously independent genetic instrument and the confounder (red dashed line). This would cause the violation of the second Mendelian randomization assumption, and hence threaten the validity of the survival estimate.

Figure 2.

Mendelian randomization estimates for the association between genetically predicted anthropometric traits and colorectal cancer-specific mortality, overall and stratified by anatomical site and tumor stage. Hazard ratios and corresponding 95% confidence intervals are presented for each anthropometric trait using the main inverse variance weighted method and three approaches accounting for collider bias: Corrected Weighted Least Squares, Slope-hunter, and Simulation Extrapolation. Each dot represents the point estimate, and horizontal lines denote 95% confidence intervals.