To the Editor:
Leukocyte telomere length is a widely studied, but inconsistent, marker of disease risk. As telomere length varies by blood cell subtype,1 measurements represent a weighted average across constituent blood cells. Proportions of leukocyte subtypes can differ greatly between individuals and these differences may introduce extraneous inter-individual variation in telomere length estimates. The degree of variation attributed to differences in leukocyte subtype composition is largely unknown, in part because cell sorting methods require fresh blood samples rarely available in epidemiologic studies.
Studies of blood DNA methylation face similar challenges. Like telomeres, blood DNA methylation differs by leukocyte subtype, and there are well-documented examples where failure to adjust for leukocyte proportions demonstrably leads to biased effect estimates.2 A method widely employed in epidemiologic studies to disentangle leukocyte subtype proportions using patterns of DNA methylation3 may be useful in research on leukocyte telomere length. Here, we use methylation and telomere length measurements from the same blood DNA samples to examine the effect of leukocyte subtype composition on telomere length measurements.
We used existing data on relative leukocyte telomere length (rLTL)4 and genome-wide DNA methylation5 measured in the same blood samples from a subsample of 445 non-Hispanic white women enrolled in the Sister Study (median age, 57; interquartile range, 36–64). rLTL was assessed using multiplex quantitative polymerase chain reaction and standardized as z-scores.4 The study was approved by the institutional review boards of the National Institute of Environmental Health Sciences and the Copernicus Group. We assessed leukocyte composition by applying a validated deconvolution approach to HumanMethylation450 BeadChip data to estimate proportions of six distinct subtypes (CD8+ and CD4+ T-cells, B-cells, natural killers, monocytes, and granulocytes).3 We first assessed Pearson correlations between rLTL and each estimated leukocyte proportion. As age is strongly, inversely associated with telomere length,6 we used the age-telomere length relationship to study the inter-individual variability attributable to leukocyte composition. We used nested linear regression models of age and telomere length with leukocyte proportions included (“full model”) and excluded (“reduced model”) and calculated the difference in the model coefficients of determination, adjusted for the number of predictors (“adjusted R2”). Finally, we used the likelihood ratio test (LRT) to assess whether inclusion of leukocyte components improved model fit.
rLTL was positively correlated with the blood sample proportion of CD8+ T-cell (ρ = 0.19), B-cell (ρ = 0.15), and monocyte components (ρ = 0.11) and negatively correlated with granulocyte components (ρ = –0.14) (Figure). In the reduced model, age was inversely associated with rLTL (per 10-year increase in age: β = –0.20; 95% confidence interval [CI] = –0.30, –0.10) and explained a small absolute proportion of telomere length variation (adjusted R2 = 0.03). In the full model, the age association was slightly attenuated (β = –0.18; 95% CI = –0.28, –0.08) and a higher proportion of rLTL was explained (adjusted R2 = 0.09). Comparing the nested models, inclusion of the leukocyte components explained an additional 6% of the inter-individual variability in telomere length—a tripling compared with age alone—and considerably improved model fit (LRT: X2df = 5 = 34.2; P = 2.2 × 10–6).
We find that leukocyte subtype composition contributes to inter-individual variation in telomere length measurements. Leukocyte telomeres appear longer in blood samples with higher proportions of CD8+ T-cells or B-cells. These lymphocyte subtypes are reported to have the longest telomeres.1 Our analysis supports prior findings that age explains a tiny fraction of inter-individual telomere length variability6 and suggests that leukocyte subtype proportions explain a much higher amount of measurement variation. Factors that impact leukocyte composition, including air pollution,7 may contribute noise or measurement error-based confounding when telomere length measurements are used as biomarkers, reducing comparability between studies. Although they explain only a small amount of the variation, genetic polymorphisms can be used as predictors of telomere length.8 Such polymorphism-based predictions do not reflect acquired changes in length from age or exposure while offering the distinct advantage of avoiding confounding by cell-type composition.
We would like to thank Katie M. O’Brien for providing the internal review of the article.
Jacob K. Kresovich
Christine G. Parks
Dale P. Sandler
Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC
Clarice R. Weinberg
Computational Biology and Biostatistics Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC
Jack A. Taylor
Epidemiology Branch; and, Epigenetic and Stem Cell Biology Laboratory, National Institute of Environmental Health Sciences, Research Triangle Park, NC, email@example.com
1. Lin J, Cheon J, Brown R, et al. Systematic and cell type-specific telomere length
changes in subsets of lymphocytes. J Immunol Res. 2016;2016:5371050.
2. Koestler DC, Marsit CJ, Christensen BC, et al. Peripheral blood immune cell methylation profiles are associated with nonhematopoietic cancers. Cancer Epidemiol Biomarkers Prev. 2012;21:1293–1302.
3. Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86.
4. Kim S, Sandler DP, Carswell G, et al. Telomere length
in peripheral blood and breast cancer risk in a prospective case-cohort analysis: results from the Sister Study. Cancer Causes Control. 2011;22:1061–1066.
5. Kresovich JK, Xu Z, O’Brien KM, Weinberg CR, Sandler DP, Taylor JA. Methylation-based biological age and breast cancer risk. J Natl Cancer Inst. 2019;111:1051–1058.
6. Wang Y, Savage SA, Alsaggaf R, et al. Telomere length
calibration from qPCR measurement: limitations of current method. Cells. 2018;7:183.
7. Gao X, Colicino E, Shen J, et al. Impacts of air pollution, temperature, and relative humidity on leukocyte distribution: an epigenetic perspective. Environ Int. 2019;126:395–405.
8. Codd V, Nelson CP, Albrecht E, et al.; CARDIoGRAM consortium. Identification of seven loci affecting mean telomere length
and their association with disease. Nat Genet. 2013;45:422–427, 427e1–e2.