Genetically Inferred Telomere Length and Testicular Germ Cell Tumor Risk

Telomeres are AGGGTT nucleotide repeats that protect chromosomes from degradation (1). Excessively long telomere length, with upregulated telomerase activity, may result in immortalized cells with unlimited potential for growth and proliferation, promoting tumorigenesis (2). Several recent studies have identified peripheral leukocyte telomere length (LTL) as a biomarker correlated with tissue-specific telomere length and associated with solid tumor risk (1, 3–5).

Recent genome-wide association studies (GWAS) have identified genetic variants associated with peripheral LTL that have utility as a surrogate genetic measure in deciphering associations with cancer risk (1, 4–6). As the association between LTL and testicular germ cell tumor (TGCT) risk is poorly understood (1, 6), we used a polygenic risk score (PRS) of telomere length–associated variants and Mendelian randomization approaches to examine whether genetically inferred LTL is associated with TGCT risk.

The Testicular Cancer Consortium (TECAC) was utilized to investigate the relationship between inferred LTL and TGCT risk (7). In total, TECAC includes imputed individual-level data from the NCI, UPENN, and United Kingdom cohorts (2,049 cases and 6,921 controls), as well as additional GWAS meta-analysis summary statistics from all 17,529 consortium participants (3,558 cases and 13,971 controls). Data used in this analysis are from individuals with European ancestry and are available through direct application to TECAC (www.tecac.org) and in dbGAP phs001349.v1.p1.

LTL was genetically inferred for participants with available genotype data (NCI, UPENN, and United Kingdom cohorts), using a PRS (1, 5) containing 22 germline variants associated with LTL (Supplementary Table S1; ref. 8):

where |PR{S_i}$| is the PRS for individual |i$|⁠, |{w_j}$| is the estimated LTL-associated variant weight in bp per LTL increasing allele, and |{x_{ij}}$| is the number of LTL increasing alleles for the ith individual and the jth LTL variant. The PRS was standardized to have mean 0 and SD of 1 and association tests were conducted separately for the NCI, UPENN, and United Kingdom cohorts using logistic regression. Results were combined using fixed effects meta-analysis.

Using available GWAS meta-analysis summary statistics from TECAC (7), we extracted associations between LTL-associated variants and TGCT risk. Summary statistics–based Mendelian randomization analyses were conducted merging LTL-associated variants into a genetic instrument across all available TECAC participants (1, 5, 6). As SEs were not available for the original LTL-associated variants (8), other published variants with detailed summary statistics were used (6). Mendelian randomization-Egger regression was utilized to evaluate heterogeneity and potential pleiotropy of included LTL variants (1, 6).

All statistical analyses were performed in R version 3.6.3 with two-sided significance levels (P < 0.05).

Of 17 available LTL-associated variants, four were found to be nominally associated with TGCT risk (P < 0.05), with only one (rs28616016) demonstrating a positive association (Table 1; Supplementary Table S2; Supplementary Fig. S1). There was no overall association between LTL-associated variants and TGCT risk (P = 0.72; Fig. 1) and no association between the LTL PRS and TGCT risk [OR, 1.02; 95% confidence interval (CI), 0.97–1.07; P = 0.45; P_het = 0.34; Supplementary Fig. S2]. Stratified analyses by TGCT histologic group (seminoma, 824 cases and nonseminoma germ cell, 1,046 cases) also suggested no association between the LTL PRS and TGCT risk (seminoma: OR, 1.04; 95% CI, 0.96–1.12; P = 0.33 and nonseminoma germ cell: OR, 1.01; 95% CI, 0.95–1.08; P = 0.71; Supplementary Fig. S3). Likewise, Mendelian randomization analyses indicated no directionality between the LTL-associated variants and TGCT risk (maximum likelihood method OR, 0.63; 95% CI, 0.22–1.80; P = 0.39; Supplementary Tables S3 and S4; Supplementary Fig. S4). The Mendelian randomization-Egger regression intercept was not statistically significant (P = 0.56), indicating no pleiotropic effects (Supplementary Table S4).

As two included LTL-associated variants (rs7705526 TERT and rs28616016 RFWD3) are in linkage disequilibrium with previously identified TGCT GWAS variants (rs2736100 TERT and rs4888262 RFWD3; R²_EUR = 0.51 and 0.69, respectively; ref. 7), we conducted sensitivity analyses removing these variants to investigate any potential influence of these variants on the overall LTL PRS association with TGCT risk (Supplementary Figs. S5 and S6). These sensitivity analyses with TGCT published variants removed indicated suggestive evidence for an association between the remaining LTL-associated variants and TGCT risk (TERT removed, P = 0.12 and TERT and RFWD3 removed, P = 0.03; Supplementary Figs. S5A and S6A) and between the variant removed LTL PRS and TGCT risk (TERT removed, P = 0.005; P_het = 0.42 and TERT and RFWD3 removed, P = 0.049; P_het = 0.29; Supplementary Figs. S5B and S6B).

Our study of 3,558 TGCT cases and 13,971 controls provides little overall evidence for an association between LTL-associated variants and TGCT risk. This lack of association was consistently observed among individual telomere length–associated variants, as well as within PRS and Mendelian randomization analyses. The TERT and RFWD3 variants, in linkage disequilibrium with previously identified TGCT GWAS variants, were the only variants to demonstrate evidence suggesting an association with LTL (indirect and direct association, respectively). Sensitivity analyses removing the TERT and RFWD3 variants suggested a marginal association between LTL and TGCT risk, indicating the relationship between LTL and TGCT may be complex.

Studies with comparable sample sizes with our investigation detected associations between inferred LTL and cancer risk (1, 4). Our genetic approach does not contain the biases typically associated with studies of measured LTL (e.g., differences in LTL by DNA extraction approach; ref. 5) and only utilized weights from a single large LTL GWAS (8), ensuring improved PRS weighting and more accurate downstream analyses. Some LTL variants were not included into our analysis due to unavailable data (rs547680822 and rs4027719) or low-quality imputation (rs188891454, rs144510686, and rs28372734). It is not likely these exclusions significantly affected our overall findings as these variants cumulatively explain approximately 0.2% of the variation in measured LTL (8). Current LTL variants identified by GWAS (N = 75,000 individuals) explain a small percentage of the variance in measured telomere length (6, 8), suggesting GWAS in larger samples may discover additional LTL variants that could better explain telomere length in testicular tissue or capture relevant aspects of telomere length more important for TGCT risk.

Our study finds no compelling evidence for an overall relationship between current LTL variants and TGCT risk, suggesting LTL is likely not a strong biomarker of TGCT risk. However, some components of LTL, specifically when removing TERT and RFWD3 variants, do demonstrate evidence for a relationship with TGCT risk, suggesting specific genetic elements of LTL may be relevant for TCGT risk.

Kristian Almstrup, Copenhagen University Hospital – Rigshospitalet, Copenhagen, Denmark.

Javier Benitez, Spanish National Cancer Centre, Madrid, Spain.

D. Timothy Bishop, University of Leeds, Leeds, England, United Kingdom.

Victoria K. Cortessis, University of Southern California, Los Angeles, CA.

Alberto Y. Ferlin, University of Brescia, Brescia, Italy.

Jourik A. Gietema, University of Groningen, Groningen, the Netherlands.

Mark H. Greene, NCI, Rockville, MD.

Tom Grotmol, Cancer Registry of Norway, Oslo, Norway.

Robert Hamilton, University of Toronto and The Princess Margaret Cancer Centre, Toronto, Ontario, Canada.

Michelle A. T. Hildebrandt, University of Texas MD Anderson Cancer Center, Houston, TX.

Peter A. Kanetsky, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL.

Lambertus A. Kiemeney, Radboud University Medical Center, GA Nijmegen, the Netherlands.

Davor Lessel, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.

Katherine A. McGlynn, NCI, Rockville, MD.

Katherine L. Nathanson, University of Pennsylvania, Philadelphia, PA.

Thorunn Rafnar, deCODE Genetics, Reykjavík, Iceland.

Lorenzo Richiardi, University of Turin and CPO-Piemonte, Turin, Italy.

Stephen M. Schwartz, Fred Hutchinson Cancer Research Center, Seattle Washington.

Rolf I. Skotheim, University of Oslo, Oslo, Norway.

Clare Turnbull, Queen Mary University, London, England, United Kingdom.

Fredrik Wiklund, Karolinska Institute, Stockholm, Sweden.

Tongzhang Zheng, Brown University, Providence, RI.

P.A. Kanetsky reports grants from NCI during the conduct of the study. No disclosures were reported by the other authors.

The opinions expressed by the authors are their own and this material should not be interpreted as representing the official viewpoint of the U.S. Department of Health and Human Services, the NIH, or the NCI.

D.W. Brown: Conceptualization, formal analysis, visualization, methodology, writing–original draft, writing–review and editing. Q. Lan: Conceptualization, writing–review and editing. N. Rothman: Conceptualization, writing–original draft. J. Pluta: Resources, writing–review and editing. K. Almstrup: Resources, writing–review and editing. M.D. Dalgaard: Resources, writing–review and editing. M.H. Greene: Conceptualization, writing–review and editing. T. Grotmol: Resources, writing–review and editing. C. Loveday: Resources, writing–review and editing. S.M. Schwartz: Resources, writing–review and editing. C. Turnbull: Resources, writing–review and editing. F. Wiklund: Resources, writing–review and editing. P.A. Kanetsky: Resources, writing–review and editing. K.L. Nathanson: Resources, writing–review and editing. K.A. McGlynn: Conceptualization, resources, writing–review and editing. M.J. Machiela: Conceptualization, supervision, methodology, writing–original draft, writing–review and editing.

This work was supported by the Intramural Research Program of the NCI and used the computational resources of the NIH's High-Performance Computing Biowulf cluster. The Testicular Cancer Consortium was supported by NIH grant U01 CA164947 (to K.L. Nathanson, P.A. Kanetsky, and S.M. Schwartz). The Penn GWAS was supported by the Abramson Cancer Center at the University of Pennsylvania (P30 CA016520) and NIH grant CA114478 (to K.L. Nathanson and P.A. Kanetsky). The Danish GWAS was supported by Villum Kann Rasmussen Foundation, a NABIIT grant from the Danish Strategic Research Council, the Novo Nordisk Foundation, the Danish Cancer Society, and the Danish Childhood Cancer Foundation. This work was supported by the Norwegian Cancer Society (418975), the Nordic Cancer Union (S-12/07), and the Swedish Cancer Society (CAN2011/484 and CAN2012/823). The Norwegian/Swedish study was supported by the Norwegian Cancer Society (grant nos. 418975 – 71081 – PR-2006–0387 and PK01–2007–0375), the Nordic Cancer Union (grant no. S-12/07), and the Swedish Cancer Society (grant nos. 2008/708, 2010/808, 2011/484, and CAN2012/823). S.M. Schwartz was supported by NCI grant R01CA085914 and contracts CN-67009 and PC-35142, and Fred Hutchinson Cancer Research Center institutional funds. C. Turnbull was supported by the Movember Foundation. The authors received no specific funding for this work.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.