Imaging-based organ-specific aging clock predicts human diseases and mortality

Background: Organ-specific aging clocks hold great potential in reflecting the health status of individual organs and promoting longevity. The existing plasma-derived organ-specific biological ages have been developed based on either empirical knowledge of pre-defined organ markers or organ-enriched molecules using in vitro tissues. Such frameworks could be refined by leveraging more objective organ markers. Specifically, in vivo imaging-based measurements are inherently organ-specific and could delineate structural and functional characteristics more objectively. However, there is a lack of systematic evaluation of imaging-based aging clocks. Methods : This study utilized 1,777 multimodal imaging-derived phenotypes (IDPs) obtained through magnetic resonance imaging, optical coherence tomography, and dual-energy x-ray absorptiometry from 11,000 healthy participants. Based on this unprecedented dataset, a machine-learning-based pipeline was established to assess the organ-specific biological age of seven organs: brain, heart, liver, kidney, pancreas, eye, and body composition. By defining the age gap as the difference between biological and chronological age, we evaluated their contributions to diseases and mortality, shared and distinct metabolic and proteomic bases, modifiable factors and potential drug targets. Findings : Our results found selective interrelationships between the aging of brain, heart and body composition. The organ-specific age gap was primarily associated with incident diseases and mortality related to corresponding organs. For instance, individuals with accelerated brain aging exhibited nearly onefold elevated risk (hazard ratio, HR=1.99, [95% confidence interval, CI,1.67-2.38]) of incident dementia, while those with accelerated heart aging exhibited a 22% increased risk of incident arrhythmias (HR=1.22, [95% CI, 1.13-1.32]). The top-contributing IDPs to biological age could serve as potential biomarkers for incident disease predictions, achieving an area under the curve (AUC) greater than 0.8 for dementia (AUC=0.82) and obesity (AUC=0.85), respectively. Subsequent analysis of proteins revealed 966 shared and 507 organ-specific molecular signatures for the biological aging of different organs. In particular, GDF15 protein was consistently associated with accelerated brain aging. Finally, we identified key modifiable factors and 14 drug targets for organ-specific aging. Interpretation : This study extends imaging-based measurement of aging from brain to other organs. Moreover, we highlight that imaging-based aging clocks demonstrate organ-specificity not only at macro scale but also at micro scales, which could refine the knowledge of organ-specific aging process and promote personalized intervention and treatment of organ aging. Funding: National Natural Science Foundation of China.