Telomere length (TL) is a marker of biological aging associated with several health outcomes. High throughput reproducible TL measurements are needed for large epidemiological studies. We compared the novel DNA methylation-based estimator (DNAmTL) with the high-throughput quantitative PCR (qPCR) and the highly accurate flow cytometry with fluorescent in situ hybridization (flow FISH) methods using blood samples from healthy adults. We used Pearson's correlation coefficient, Bland Altman plots and linear regression models for statistical analysis. Shorter DNAmTL was associated with older age, male sex, white race, and cytomegalovirus seropositivity (p<0.01 for all). DNAmTL was moderately correlated with qPCR TL (N=635, r=0.41, p < 0.0001) and flow FISH total lymphocyte TL (N=144, r=0.56, p < 0.0001). The agreements between flow FISH TL and DNAmTL or qPCR were acceptable but with wide limits of agreement. DNAmTL correctly classified >70% of TL categorized above or below the median, but the accuracy dropped with increasing TL categories. The ability of DNAmTL to detect associations with age and other TL-related factors in the absence of strong correlation with measured TL may indicate its capture of aspects of telomere maintenance mechanisms and not necessarily TL. The inaccuracy of DNAmTL prediction should be considered during data interpretation and across-study comparisons.
Telomere length (TL) comparisons from different methods are challenging due to differences in laboratory techniques and data configuration. This study aimed to assess the validity of converting the quantitative polymerase chain reaction (qPCR) telomere/single copy gene (T/S) ratio to TL in kilobases (kb). We developed a linear regression equation to predict TL from qPCR T/S using flow cytometry with fluorescence in situ hybridization (flow FISH) TL data from 181 healthy donors (age range = 19–53) from the National Marrow Donor Program (NMDP) biorepository. TL measurements by qPCR and flow FISH were modestly correlated (R2 = 0.56, p < 0.0001). In Bland-Altman analyses, individuals with the shortest (≤10th percentile) or longest (≥90th) flow FISH TL had an over- or under-estimated qPCR TL (bias = 0.89 and −0.77 kb, respectively). Comparisons of calculated TL from the NMDP samples and 1810 age- and sex-matched individuals from the National Health and Nutrition Examination Survey showed significant differences (median = 7.1 versus 5.8 kb, respectively, p < 0.0001). Differences in annual TL attrition were also noted (31 versus 13 bp/year, respectively, p = 0.02). Our results demonstrate that TL calculated in kb from qPCR T/S may yield biased estimates for individuals with the shortest or longest TL, those often of high clinical interest. We also showed that calculated TL in kb from qPCR data are not comparable across populations and therefore are not necessarily useful.