Association between body mass index and cancer survival in a pooled analysis of 22 clinical trials

Abstract

Background: Data are inconsistent on the association between body mass index (BMI) at time of cancer diagnosis and prognosis. We used data from 22 clinical treatment trials to examine the association between BMI and survival across multiple cancer types and stages. Methods: Trials with ≥5 years of follow-up were selected. Patients with BMI <18.5 kg/m2 were excluded. Within a disease, analyses were limited to patients on similar treatment regimens. Variable cutpoint analysis identified a BMI cutpoint that maximized differences in survival. Multivariable Cox regression analyses compared survival between patients with BMI above versus below the cutpoint, adjusting for age, race, sex, and important disease-specific clinical prognostic factors. Results: A total of 11,724 patients from 22 trials were identified. Fourteen analyses were performed by disease site and treatment regimen. A cutpoint of BMI =25 kg/m2 maximized survival differences. No statistically significant trend across all 14 analyses was observed (mean HR =0.96; P = 0.06). In no cancer/treatment combination was elevated BMI associated with an increased risk of death; for some cancers there was a survival advantage for higher BMI. In sex-stratified analyses, BMI ≥ 25 kg/m2 was associated with better overall survival among men (HR =0.82; P =0.003), but not women (HR =1.04; P =0.86). The association persisted when sexspecific cancers were excluded, when treatment regimens were restricted to dose based on body surface area, and when earlystage cancers were excluded. Conclusion: The association between BMI and survival is not consistent across cancer types and stages. Impact: Our findings suggest that disease, stage, and genderspecific body size recommendations for cancer survivors may be warranted.