PhenoAge (DNAm PhenoAge Epigenetic Clock)

PhenoAge is a second-generation epigenetic clock trained on nine clinical markers of mortality risk rather than on chronological age alone. Morgan Levine and colleagues first built a composite phenotypic age from albumin, creatinine, fasting glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red blood cell distribution width, alkaline phosphatase, and chronological age, then found 513 CpG methylation sites that best predicted that composite [1]. The claim: because it is trained on clinical mortality markers instead of chronological age, PhenoAge should track functional biological decline more closely than first-generation clocks.

The 2018 validation paper reported strong associations between DNAm PhenoAge and all-cause mortality, cardiovascular disease and coronary heart disease mortality, cancer incidence and mortality, and Alzheimer's disease risk, using data from 456 participants aged 21 to 100 plus large-scale replication cohorts [1]. A later comparative review found 41 of PhenoAge's 513 CpGs overlap with the Horvath clock and only 6 overlap with the Hannum clock, indicating substantial biological independence between clock generations [4]. In the uncontrolled TRIIM pilot trial (n=9), PhenoAge fell by an average that contributed to the four-clock combined 1.5-year reversal after 12 months of growth hormone, DHEA, and metformin, though this was one small open-label study with no placebo arm [2].

How to measure it: a standard blood draw is processed on a methylation array, either 450K or EPIC, then the Levine et al. algorithm converts array data into the DNAm PhenoAge score, typically reported in years alongside age acceleration [3]. Direct-to-consumer panels bundle this with GrimAge, Horvath, and DunedinPACE [5].

How to intervene on it: no drug is approved to specifically lower PhenoAge. Since it is trained on inflammatory and metabolic clinical markers (CRP, glucose, albumin), interventions that improve those underlying markers, weight loss, glycemic control, anti-inflammatory diet, plausibly move the score, though this is inferred from the training markers rather than demonstrated in a dedicated RCT targeting PhenoAge itself. The TRIIM pilot is the only published human intervention trial reporting a PhenoAge change, and it was small and uncontrolled [2].

Critics point out that because PhenoAge is built from routine clinical labs, some of its predictive power may reduce to those labs' well-established associations with mortality, elevated CRP and glucose, low albumin, rather than a distinctly epigenetic signal [6]. Its performance also varies by ethnicity and cohort composition, since the original training population was not fully representative of global populations.

The plain takeaway: PhenoAge is a well-validated cohort predictor of mortality and disease risk that improves on first-generation clocks, but it has only one small, uncontrolled human trial showing it can be moved by an intervention, and no proven drug or supplement changes it in a large randomized trial.