Background: Markers of somatic variation including shorter telomere length (TL), presence of clonal hematopoiesis of indeterminate potential (CHIP), and lower mitochondrial DNA copy number (mtDNAcn) can be assessed from whole genome sequencing (WGS) and reflect biological aging. To assess association of these dynamic genome metrics with frailty, we used WGS generated through the NHLBI Trans-Omics for Precision Medicine (TOPMed) Program.
Methods: Four TOPMed cohorts (ARIC, CHS, FHS, WHI) were included in this study. Frailty was assessed with the Fried criteria (scored 0-5: 0 classified as robust, 1-2 as pre-frail, and 3-5 as frail). Association analyses used a three-level ordinal model. Measures of the dynamic genome were centrally called at the TOPMed Informatics Research Center. Cross-sectional multivariable ordinal regression models were run in each cohort with frailty as the dependent variable and TL, mtDNAcn, or CHIP as the independent variable of interest, adjusting for the difference between age at blood draw and age at frailty assessment, age at frailty assessment, sex, self-reported race, education, BMI, smoking, and cohort-specific covariates (clinic, pedigree). Results were meta-analyzed using regression coefficients and standard errors.
Results: Participants (n=8788: n=727 frail, n=4350 prefrail and n=3711 robust) were 65.2 ± 10 yrs at time of blood draw, 73.6 ± 7.6 yrs at time of frailty assessment, with mean age difference of 8.4 ± 9.4 yrs between the two timepoints; 61.7% female; 85% European ancestry, 14.3% African ancestry and 0.7% Hispanic. Adjusting for all covariates, lower mtDNAcn was significantly associated with risk of worse frailty status in FHS (; 95% CI = -0.466; -0.7927, -0.1393) and the meta-analysis across the four cohorts (Z = -2.5, p = 0.01, three of four studies with consistent negative effect). Shorter TL showed significant association with risk of worse frailty in unadjusted models for CHS (-0.15; -0.25, -0.01) and FHS (-0.22; -0.38, -0.07); with the adjustment for all covariates effects were no longer significant. Neither TL nor CHIP showed significant association with frailty in meta-analysis.
Conclusions: Only mtDNAcn revealed association with frailty after covariate adjustment, perhaps due to survival bias or interrelatedness of metrics. Future work will extend investigation to mosaic chromosomal alterations, epigenetic aging, and collapsed accelerated aging across multiple measures. Identification of specific aging related biological changes that may influence frailty may help to identify novel pathways for prevention and intervention development.