Organ Aging Clocks Make Alcohol Look Protective, but Confounding May Be Fooling Us

Alcohol is clearly harmful at heavy levels, yet observational studies often show apparent benefits for light or moderate drinkers. This 'healthy drinking' paradox usually reflects confounding: abstainers include former drinkers and people who stopped for health reasons, and drinking patterns track socioeconomic status. Organ-specific proteomic aging clocks, which use blood proteins to estimate the biological age of individual organs, offer a new way to study alcohol's effects, but early results seem to echo the paradox.

This Perspective compares three recent studies, all using UK Biobank Pharma Proteomics Project data (about 50,000 adults) and the Olink platform. Goeminne et al. (2025) trained clocks to predict mortality and used alcohol intake in grams per day. Alcohol was associated with faster aging in brain, kidney and intestine clocks but slower aging in lung and artery clocks. Wang et al. (2026) trained clocks on chronological age and used drinking frequency, finding negative associations with kidney, lung, artery, muscle and intestine clocks. Wang et al. (2025), adjusting only for age and sex with Bonferroni correction, found faster aging in most organs, except a non-significant intestine association. In that study, drinking with meals was linked to slower aging, which may reflect moderate, social drinking or socioeconomic confounding. Goeminne et al. also found wine 'protective' and beer linked to faster aging even after adjusting for the Townsend deprivation index.

The author attributes the inconsistencies to several factors: different protein sets in each clock, different training outcomes (mortality vs chronological age), different alcohol definitions, and different adjustment and multiple-testing strategies. Because the two Wang studies used the same training target and exposure definition, their divergence likely stems from protein selection and from the stronger socioeconomic adjustment in the 2026 study. For brain clocks, both Goeminne and Wang found synaptic signatures but only Goeminne found an alcohol association, which suggests a methodological rather than biological cause. Clocks may also capture different biology, and protein weights chosen independently of alcohol may reflect the alcohol sensitivity of individual proteins rather than true organ aging.

The author recommends comparing clocks' functional biology directly, using several clocks together, and pairing proteomic clocks with organ-specific epigenetic clocks, which may better capture long-term alcohol imprints. Longitudinal designs would help, ideally in biobanks like FinnGen with medication data to limit sick-quitter bias. Twin and family studies could separate genetic from environmental influences, and alcohol should be studied against both chronological-age and mortality-related aging.

The key caveat is that these are cross-sectional observational associations, and a Perspective rather than new data. Apparent organ-level benefits should not be read as evidence that alcohol slows aging, as that interpretation could understate its known harms.