Abstract
Epigenetic aging clocks represent contemporary aging biomarkers that predict age using methylomic data. These models can be categorized as first-generation clocks that estimate chronological age or next-generation clocks that are designed to associate with health, lifestyle, and/or outcomes. Recently, we created a next-generation buccal clock called CheekAge that associates with all-cause mortality risk in older adults. To better understand our model, we collated 25 Infinium MethylationEPIC datasets in the Gene Expression Omnibus database and analyzed the ability of CheekAge and five other well-known clocks to associate with distinct health and disease signals. CheekAge outcompeted every other clock tested by significantly associating with a total of 33 different disease and health variables, including human immunodeficiency virus, major depressive disorder, psychological trauma, prediabetes, body mass index, non-alcoholic fatty liver disease, pulmonary fibrosis, exposure to the chemical endocrine disruptor PBB-153, and various cancers and tumors. Of the six clocks tested, the next-generation clocks outperformed the first-generation clocks. To better understand the underlying biology of CheekAge, we iteratively removed CpG inputs to identify DNA methylation sites that promoted or antagonized each association. Finally, we performed detailed enrichment analyses on these sites to unveil overrepresented biological processes and transcription factor targets.