Leveraging High-Throughput Data to estimate Leukocyte Telomere Length in the Million Veteran Program

Telomeres are highly repeated DNA sequences that stop chromosome ends from fraying. They shorten as we age, making them markers of biological aging. In this study, we employed an innovative approach to estimate leukocyte telomere length (LTL) through the utilization of the bioinformatics tool TelSeq, leveraging existing whole-genome sequencing data sourced from the Department of Veteran Affairs Million Veteran Program (MVP) encompassing a cohort of 102,646 diverse individuals. Using the TelSeq-estimated LTL, we conducted parametric and non-parametric statistical analyses with demographic descriptors such as age, sex, and genetically inferred ancestry (GIA). Our results revealed a near-linear yet modest inverse correlation between age and LTL ({rho}= -0.20, p < 1E-300). Specifically, each year increase in age corresponded to an 8bp LTL decrease (p < 1E-300) after adjusting for sex and GIA groups. Females had longer LTL than males, even after controlling for ancestry and age ({beta} = 58 bp, p < 1E-300). Additionally, differences in LTL were observed among ancestry groups (Median LTLAfrican = 2.03kb, Median LTLEuropean = 2.01kb, Median LTLAdmixedAmerican = 2.00kb), however, these differences were not statistically significant between the two largest groups - African and European ancestry. To validate TelSeq-estimated LTL, we utilized methylation profiles of 140 cytosine-phosphate-guanine dinucleotides (CpGs) on 28,669 individuals to derive DNA methylation-based telomere length estimator (DNAmTL). DNAmTL had a modest correlation with TelSeq-estimated LTL ({rho}=0.21, p < 1E-300). Consistent with Telseq-estimated LTL, DNAmTL was inversely associated with age where each year increase corresponded to 16bp DNAmTL decrease (p < 1E-300) after adjusting for sex and GIA groups. After controlling for these parameters, males were also found to exhibit shorter DNAmTL than females ({beta} = -98 bp, p < 1E-300), similar to observations from TelSeq-estimated LTL. Lastly, individuals of African and Admixed-American ancestry had longer DNAmTL than those of European ancestry. These findings align with existing literature elucidating the effects of age, sex, and ancestry on telomere length variations, and also represent the first comparison of TelSeq-derived LTL with methylation-derived DNAmTL. This valuable resource is now accessible to the broader MVP community, facilitating further telomere research.