Across 204,494 UK Biobank participants followed a median 10.5 years, each 1-SD increase in the share of energy from ultra-processed food (UPF) carried a 12% higher hazard of hospital-recorded myeloid neoplasms (HR 1.12; 95% CI 1.05-1.21) among 848 cases. Myeloproliferative neoplasms showed the clearest subtype signal (HR 1.15). The association held using UPF weight percentage and among people without detectable clonal hematopoiesis (CHIP) at baseline, with lipid, inflammatory and myeloid molecular correlates.
Myeloid cancers such as MPN, MDS and AML are rarely studied in nutritional epidemiology, so this is a genuinely new outcome for the UPF literature, which has mostly centered on cardiometabolic disease, colorectal cancer and mortality. The CHIP analysis is the most interesting design choice: it suggests UPF may matter beyond simply accelerating expansion of pre-existing mutant clones, though the confidence interval nearly touches 1.0 and cannot settle the question. Caveats are substantial. Diet came from the Oxford WebQ, a 24-hour recall that misclassifies habitual intake, and UK Biobank volunteers are healthier and more affluent than average. Residual confounding by obesity, smoking and overall diet quality is hard to exclude, and outcomes came from hospital records, so missed or mislabeled cases are likely. An effect of this size is modest and could be compatible with bias. The molecular analyses are exploratory and the authors do not claim causation. Editorially, this is incremental: it adds a plausible hematologic endpoint to an already broad UPF risk pattern. It supports existing advice to limit UPF but does not justify specific dietary targets for blood cancer prevention.