See Point by Park et al., p. 1898
We agree with Park and colleagues that methodologic limitations such as confounding, reverse causation, and collider stratification bias could influence the association between body mass index (BMI) and mortality following a cancer diagnosis. We addressed some of the limitations as mentioned in our review published in this issue and in our prior work using causal diagrams to guide analysis of the associations of pre-, peri-, and post-diagnosis BMI with mortality after colorectal cancer (1). As shown in Table 1 of our review, when we exclude from the analysis ever-smokers (to address confounding) or patients who decreased a BMI category prior to diagnosis (to address reverse causation), the inverse association of overweight with mortality after colorectal cancer was attenuated and became nonsignificant; however, in no instance did an adverse association with overweight emerge.
Park and colleagues also emphasized the need for precise measures of adipose tissue distribution within racial/ethnic subgroups as body fatness varies across race at a given BMI and consequently BMI may associate differently with cancer survival across race. We agree. In fact, we previously examined mortality associations of both BMI and visceral adiposity measured by computed tomography (CT) by race (2) in our breast cancer cohort (Fig. 1). We found that BMI at diagnosis had no association with mortality in either black or white breast cancer patients, independent of age, smoking, alcohol, and comorbidity status, and tumor and treatment characteristics. In contrast, visceral adiposity was associated with overall mortality in black but not white breast cancer patients. Clearly, as we demonstrate and Park and colleagues assert, “BMI is an inadequate measure of body fatness” and one size does not fit all. For patients with chronic conditions such as cancer, more precise measures of visceral and subcutaneous adipose tissue compartments as well as muscularity are needed both to improve individual risk stratification and to further our understanding of the role of obesity in cancer prognosis. Fortunately, these measures are often available because cancer patients routinely undergo CT scans for diagnosis and surveillance purposes that produce highly accurate estimates of muscle and visceral and subcutaneous adipose tissue, such as those in Fig. 1. Furthermore, automated methods of body composition assessment from CT are increasingly available and will help not only to accelerate research in this area but to aid in translation of this knowledge into clinical practice.
While we agree with Park and colleagues that more research is needed to examine precise measures of body fatness on survival by race, sex, age, cancer site, stage, and subtype, there is increasingly strong evidence that adequate muscularity is consistently associated with better survival across many of these risk strata and may in part explain the observation that higher BMI is often associated with better survival (3).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This work was supported by a grant from the National Cancer Institute of the National Institutes of Health (R01 CA184953).