There seems something wrong with this. I would like to see the original data on biomarkers and audit the process to the conclusions. However, it strikes me as a paper I should highlight as I am not certain it is rubbish. It just appears to have conclusions that don’t seem coherent. Probably underpowered and statistical randomness or something.
https://onlinelibrary.wiley.com/doi/epdf/10.1111/acel.70507
chatGPT:
Summary
This paper tests whether a 4-week controlled dietary intervention can shift Klemera-Doubal Method biological age estimates in older adults. The authors re-analysed data from the Nutrition for Healthy Living study: 104 adults aged 65–75 were randomised to one of four diets: omnivorous high-fat, omnivorous high-carbohydrate, semi-vegetarian high-fat, or semi-vegetarian high-carbohydrate. KDM-derived δAge was calculated as KDM age minus chronological age before and after the intervention.
The reference group was the omnivorous high-fat diet, because it most resembled participants’ usual baseline diet. This group showed no meaningful change in δAge. By contrast, the omnivorous high-carbohydrate diet produced a statistically significant reduction in δAge compared with the omnivorous high-fat diet. The semi-vegetarian groups also tended to reduce δAge, but not all comparisons reached conventional statistical significance.
The authors used two modified KDM algorithms because the NHL dataset did not contain the exact same biomarkers as the original Levine KDM model. One was a substituted version, KDM-S, using seven available or substitute markers; the other, KDM-M, used a broader 15-marker set. They validated these against NHANES data and found that the modified KDM measures remained correlated with chronological age and, particularly in men, predicted mortality risk.
The main interpretation is cautious: δAge is diet-responsive within four weeks, but this does not necessarily mean that biological ageing was reversed. The change may reflect rapid shifts in metabolic, inflammatory, cardiovascular, and body-composition biomarkers rather than durable slowing or reversal of ageing. The authors explicitly state that longer-term trials are needed to know whether these shifts predict reduced age-related disease risk.
What is novel?
The key novelty is that this is presented as the first study of a fully controlled dietary intervention on KDM age in an older population. Prior work has looked at diet and biological-age measures observationally, or used epigenetic clocks in intervention contexts, but this study applies a KDM biomarker-age framework to a tightly controlled diet trial in older adults.
A second novelty is the use of a 2×2 factorial diet design separating two dimensions: protein source, omnivorous versus semi-vegetarian, and macronutrient balance, high-fat versus high-carbohydrate. This lets the authors ask whether shifts in KDM age are more consistent with carbohydrate/fat balance, plant-forward eating, or both.
A third useful contribution is the paper’s framing of KDM-derived δAge as a responsive physiological-state marker, not simply as a biological-age-reversal marker. That is important because many “biological age” intervention papers overinterpret short-term clock changes. Here, the authors are relatively careful: the biomarker score moved, but the biological meaning remains uncertain.
Critique
The strongest feature is the controlled-feeding design. Participants were provided with all meals for four weeks, making this more internally valid than a free-living diet-advice study. The diets also achieved the intended macronutrient and protein-source differences, which strengthens attribution of the biomarker changes to the intervention rather than vague “healthy eating” advice.
The main weakness is that the endpoint is not ageing itself, but a composite biomarker estimate. KDM includes markers that can move quickly with diet, salt intake, inflammation, glucose handling, lipids, hydration, and body weight. Therefore, a 3–4 year apparent reduction in δAge over four weeks should not be read literally as “participants became biologically four years younger.” The authors acknowledge this, but it remains the central interpretive limitation.
The second limitation is duration. Four weeks is enough to change glucose, lipids, blood pressure, inflammatory markers, insulin, and perhaps waist/body composition measures, but it is not enough to show altered ageing trajectory, reduced morbidity, or lower mortality. The study shows biomarker plasticity, not durable geroprotection.
The third limitation is sample size and power. The total sample was 104, and the KDM-M analysis lost one participant. Several effects were directionally favourable but non-significant. That makes it risky to over-rank the four diets. The clearest finding is that the omnivorous high-carbohydrate diet beat the omnivorous high-fat reference; the vegetarian arms look suggestive but less definitive.
There is also a baseline-comparator issue. The OHF diet was chosen as reference because it most resembled participants’ habitual diets. That is reasonable, but it means the observed differences may partly reflect “moving away from usual Western/Australian dietary pattern” rather than a clean high-carb versus high-fat or plant versus animal protein effect. The high-carbohydrate diets used complex carbohydrates and whole/minimally processed foods, so the result should not be generalized to refined-carbohydrate diets.
The modified KDM models are another caveat. KDM-S and KDM-M were validated against NHANES, but they are not identical to the original Levine KDM. Their mortality prediction seemed weaker or sex-dependent, especially with less clear predictive validity in women. Since the NHL cohort was mostly female, this matters for interpreting the age-score changes.
Bottom line
This is a useful and fairly cautious paper. It supports the idea that diet can rapidly shift a KDM-style physiological-age score, especially when moving toward whole-food, higher-complex-carbohydrate or plant-forward patterns. But the study does not prove biological age reversal. The best interpretation is that KDM δAge is a sensitive short-term readout of systemic physiological state; whether sustained changes in this score translate into slower ageing or lower disease risk remains unproven.