An epigenetic biomarker of aging for lifespan and healthspan

Abstract

Identifying reliable biomarkers of aging is a major goal in geroscience. While the first generation of epigenetic biomarkers of aging were developed using chronological age as a surrogate for biological age, we hypothesized that incorporation of composite clinical measures of phenotypic age that capture differences in lifespan and healthspan may identify novel CpGs and facilitate the development of a more powerful epigenetic biomarker of aging. Using an innovative two-step process, we develop a new epigenetic biomarker of aging, DNAm PhenoAge, that strongly outperforms previous measures in regards to predictions for a variety of aging outcomes, including all-cause mortality, cancers, healthspan, physical functioning, and Alzheimer's disease. While this biomarker was developed using data from whole blood, it correlates strongly with age in every tissue and cell tested. Based on an in-depth transcriptional analysis in sorted cells, we find that increased epigenetic, relative to chronological age, is associated with increased activation of pro-inflammatory and interferon pathways, and decreased activation of transcriptional/translational machinery, DNA damage response, and mitochondrial signatures. Overall, this single epigenetic biomarker of aging is able to capture risks for an array of diverse outcomes across multiple tissues and cells, and provide insight into important pathways in aging.

Keywords: epigenetic clock; DNA methylation; biomarker; healthspan.

Figure 1

Roadmap for developing DNAm PhenoAge. The roadmap depicts our analytical procedures. In step 1, we developed an estimate of ‘Phenotypic Age’ based on clinical measure. Phenotypic age was developed using the NHANES III as training data, in which we employed a proportional hazard penalized regression model to narrow 42 biomarkers to 9 biomarkers and chronological age. This measure was then validated in NHANES IV and shown to be a strong predictor of both morbidity and mortality risk. In step 2, we developed an epigenetic biomarker of phenotypic age, which we call DNAm PhenoAge, by regressing phenotypic age (from step 1) on blood DNA methylation data, using the InCHIANTI data. This produced an estimate of DNAm PhenoAge based on 513 CpGs. We then validated our new epigenetic biomarker of aging, DNAm PhenoAge, using multiple cohorts, aging-related outcomes, and tissues/cells. In step 3, we examined the underlying biology of the 513 CpGs and the composite DNAm PhenoAge measure, using a variety of complementary data (gene expression, blood cell counts) and various genome annotation tools including chromatin state analysis and gene ontology enrichment.

Figure 2

Mortality Prediction by DNAm PhenoAge. (A) Using five samples from large epidemiological cohorts—two samples from the Women’s health Initiative, the Framingham Heart Study, the Normative Aging Study, and the Jackson Heart Study—we tested whether DNAm PhenoAge was predictive of all-cause mortality. The Fig. displays a forest plot for fixed-effect meta-analysis, based on Cox proportional hazard models, and adjusting for chronological age. Results suggest that DNAm PhenoAge is predictive of mortality in all samples, and that overall, a one-year increase in DNAm PhenoAge is associated with a 4.5% increase in the risk of death (p=9.9E-47). This is contrasted against the first generation of epigenetic biomarkers of aging by Hannum and Horvath, which exhibit less significant associations with lifespan (p=1.7E-21 and p=4.5E-5, respectively). (B and C) Using the WHI sample 1, we plotted Kaplan-Meier survival estimates using actual data from the fastest versus the slowest agers (panel B). We also applied the equation from the proportional hazard model to predict remaining life expectancy and plotted predicted survival assuming a chronological age of 50 and a DNAm PhenoAge of either 40 (slow ager), 50 (average ager), or 60 (fast ager) (panel C). Median life expectancy at age 50 was predicted to be approximately 81 years for the fastest agers, 83.5 years for average agers, and 86 years for the slowest agers.

Figure 3

Chronological age versus DNAm PhenoAge in a variety of tissues and cells. Although DNAm PhenoAge was developed using methylation data from whole blood, it also tracks chronological age in a wide variety of tissues and cells. (A) The correlation across all tissues/cells we examined is r=0.71. (B-ZJ) report results in different sources of DNA as indicated in panel headings. The numbers correspond to the data sets from (Horvath 2013). Overall, correlations range from r=0.35 (breast, panel O) to r=0.92 (temporal cortex in brain, panel L).

Figure 4

DNAm PhenoAge measured in dorsolateral prefrontal cortex relates to Alzheimer’s disease and related neuropathologies. Using postmortem data from the Religious Order Study (ROS) and the Memory and Aging Project (MAP), we find a moderate/high correlation between chronological age and DNAm PhenoAge (panel A). We also estimate the Pproportion of neurons via the CETS algorithm and show that it correlates with DNAm PhenoAge (B). Further furtherthe correlation between chronological agen and DNAm PhenoAge is increased after adjusting for the estimated proportion on neurons in each sample (panel C). We also find that DNAm PhenoAge is significantly higher (p=0.00046) among those with Alzheimer’s disease versus controls (panel D), and that it positively correlates with amyloid load (p=0.012, panel E), neuritic plaques (p=0.0032, panel F), diffuse plaques (p=0.036, panel G), and neurofibrillary tangles (p=0.0073, panel H).

Figure 5

Lifestyle factors versus DNAm PhenoAge acceleration in blood in the WHI. In this cross- sectional analysis, the correlation test analysis (bicor, biweight midcorrelation) between select variables and DNAm PhenoAgeAccel reveals that education, income, exercise, proxies of fruit/vegetable consumption, and HDL cholesterol are negatively associated (blue) with DNAm PhenoAge, i.e. younger epigenetic age. Conversely, CRP, insulin, glucose, triglycerides, BMI, waist-to-hip ratio, systolic blood pressure, and smoking have a positive association (red) with DNAm PhenoAge. All results have been adjusted for ethnicity and batch/data set. Similar results based on multivariate regression models can be found in Supplement 1: Fig. S6B.