Studies of associations between waist circumference (WC) and waist-hip ratio (WHR) with cancer risk are limited. We estimated the risk of overall and site-specific cancers with selected anthropometric measures and compared the discriminative value of each measure for the risk of cancer.

This study included 60,107 adults from ANZDCC, a pooled cohort with objectively measured BMI, WHR and WC, followed for cancer from 1983-2010. Cancer outcomes were ascertained by linkage to the Australian Cancer Database. Cox proportional models analysed associations between quintiles of each measure and overall, breast, colorectal, and prostate cancers, adjusting for covariates. Areas under the receiver operative curve (AUROC) were produced to determine which anthropometric measure best discriminated cancer risk in linear models.  

 During 892,066 person-years of follow-up, 4,876 men and 4,052 women developed cancer. Among men, overall cancer risk was elevated for Q5 vs. Q1 for WC [HR: 1.24(95%CI:1.01 –1.52)] and WHR [1.98(1.09–3.58)], but not BMI; for colorectal cancer, Q5 vs. Q1 of BMI was the only predictor [1.75(1.37–2.43)]. There was no association with prostate cancer. Among women, overall cancer risk was elevated for Q5 vs. Q1 of BMI [1.21(1.11–1.33)], WC [1.41(1.26–1.57)] and WHR [1.26(1.15-1.38)]; for colorectal cancer, Q5 vs. Q1 of WC [1.43(1.08–1.88)] and WHR [1.78(1.19-2.66), but not BMI were associated with an increased risk; for breast cancer, Q5 vs. Q1 of BMI [1.18(1.00-1.40)] and WC [1.35(1.10-1.68)], but not WHR were associated with an increased risk. BMI, WC and WHR adequately predicted the risk of overall and colorectal cancer, AUROC ≥0.70, but no measure was better than another.

BMI, WC, and WHR similarly discriminate risk of cancer, but those with marked central adiposity have a greater magnitude of risk associated with overall, colorectal (women), and breast cancer. Associations between anthropometric measures and cancer are complex, and further work is needed to understand them better.