Understanding which cancers are genuinely driven by excess body fat — and which associations are artifacts of confounding — has long been contested in oncology. A sweeping new synthesis now provides the clearest quantitative picture to date, with implications for how obesity is framed as a modifiable cancer risk factor across populations and sexes.
This systematic review and meta-analysis, published in Nature Metabolism, synthesized 226 prospective studies encompassing 1.5 million incident cancer cases drawn from PubMed, EMBASE, and Scopus through April 2025. Elevated BMI was positively associated with 19 of 25 cancer types examined, and inversely associated with three. Crucially, four cancer types — leukemia, non-Hodgkin lymphoma, bladder cancer, and glioma — emerged as positively associated with adiposity in a way not previously recognized by major cancer consensus bodies such as the World Cancer Research Fund or IARC. The analysis also found that waist circumference performed comparably to BMI as a predictor, suggesting central adiposity carries similar risk information. Geographic and sex-based heterogeneity was notable: postmenopausal breast and ovarian cancer associations were stronger in East Asian populations, colorectal cancer showed stronger associations in men, and gallbladder cancer associations were attenuated in East Asia and among men. Mendelian randomization data — which help reduce confounding by using genetic proxies for adiposity — broadly supported the observational findings.
This is a methodologically rigorous and substantively important contribution. The inclusion of Mendelian randomization evidence strengthens causal inference beyond what observational data alone can support. The identification of four newly flagged cancer types warrants follow-up, particularly because the biological mechanisms linking adiposity to lymphoid malignancies and glioma are less established than those for hormone-sensitive cancers. Key limitations include the BMI instrument's inability to distinguish fat mass from lean mass, limited imaging-based data, and residual population-level confounding. Nevertheless, at this scale and with genetic corroboration, the findings are more than incremental — they suggest current clinical guidelines on obesity and cancer risk may need updating.