Across 177,763 UK Biobank adults followed a median 12.6 years, high-quality carbohydrate intake was associated with lower odds of accumulating three or more cardiometabolic diseases (HR 0.92, 95% CI 0.86-0.98), while low-quality carbohydrates raised it (HR 1.11, 1.05-1.19). The signal was clearest for progression from hypertension to type 2 diabetes. Exploratory proteomic and metabolomic mediation pointed to PPAR, PI3K-Akt and MAPK signalling, and genotype-specific interactions were reported at candidate loci.
The central finding fits a long line of evidence that what carbohydrates you eat matters more than how many, and the multi-state modelling is the genuinely useful step: it shows the association concentrates at later transitions, where someone with one condition develops the next, rather than only at first diagnosis. That is practically relevant for people already carrying hypertension. The effect sizes are modest, though. A hazard ratio of 0.98 per transition is small, and food-frequency data collected in a healthier-than-average volunteer cohort leave room for residual confounding and reverse causation, so these are associations, not proof that swapping foods changes outcomes. The mediation percentages of 36-40% are labelled exploratory by the authors and should be read cautiously, since statistical mediation with blood proteins does not establish mechanism. The gene-interaction results, such as the opposite CHD associations by rs17589271 genotype, come from a discovery scan without replication, so they are hypothesis-generating at best. Editorially this is confirmatory: favour whole grains, legumes and fruit over refined grains and sugary foods; personalised genotype-based advice is not supported yet.