See Counterpoint by Caan, et al., p. 1906

Dr. Caan and colleagues state that the overweight paradox does in fact exist and that overweight cancer patients have better survival than normal weight cancer patients because of muscle mass and a potential role of subcutaneous fat as nutritional reserves. This raises the question “Is the body composition of overweight cancer patients the same as that of noncancer overweight individuals?” Figure 1 in Dr. Caan's study shows a significant heterogeneity in body composition of colorectal cancer patients in the normal (body mass index, BMI 18.5–<25) and overweight (BMI 25–<30) categories. The proportion of patients with normal body composition is noticeably higher in the overweight category (approximately 60%) than in the normal weight category (20%–50%). Furthermore, the majority of normal weight cancer patients have low muscle mass. Should we categorize cancer patients with BMI 18.5–<25 as normal weight or underweight? Are the cancer patients with BMI 25–<30 overweight or normal weight?

As the scientific consensus was made to define overweight and obesity using the BMI in mid-1990s, BMI has been the most useful and practical tool to identify individuals and groups at risk for chronic diseases and mortality. The BMI cutoffs for overweight and obesity were chosen based on their associations with increased health risk in generally healthy populations in Western Europe or the United States. The use of BMI standardized our approach to defining overweight and obesity and significantly increased the comparability of findings across studies. However, it is well acknowledged that BMI is not a perfect measure of adiposity because it does not directly measure fat mass or percentage of body fat. Also, for a given BMI, the body composition could differ by age, sex, and race/ethnicity and possibly by diseases that lead to weight change. Considering weight loss, including muscle loss, often precedes clinical recognition of cancer, the BMI cutoffs commonly used in noncancer populations would not properly characterize body adiposity in cancer patients. Furthermore, one-time assessment of BMI at the time of or close to cancer diagnosis would not appropriately evaluate a patient's lifetime adiposity that is known to affect cancer prognosis and survival.

The result in Table 1 is based on the analysis using the standard definition of normal weight, BMI 18.5–<25, as a reference group. However, normal weight patients looked more like underweight than normal weight (as seen in Figure 1). If a higher BMI cutoff for the reference group, for example, 22.5–<25, is chosen to be more representative of normal body composition, more appropriate comparisons would have been made. In addition, the relative risk of 0.87 with 95% confidence interval from 0.65 to 1.17 in overweight never smokers compared with normal weight suggests the study may not have sufficient power to detect an association, either positive or negative, in this analysis. A significant misclassification of exposure and insufficient power to detect a modest association will produce spurious associations. As studies of overweight and all-cause mortality needed very large observational studies to detect a modest but significant association, we will need larger and carefully designed studies with better assessment of body adiposity in cancer patients. Meanwhile, when we weigh the evidence for body adiposity and cancer survival, we should be aware of the significant limitations in current literature as discussed in our article.

No potential conflicts of interest were disclosed.

This work was funded, in part, by The Foundation for Barnes-Jewish Hospital, and by grants to G.A. Colditz by the NCI (U54 CA155496) and the Siteman Cancer Center at Washington University School of Medicine support grant (P30 CA091842).