Mendelian randomization methods allow for causal interpretations of the effect of a phenotype, here TL, on an outcome, here AD, by the use of genetic variants to generate instrumental variable (IV) estimates. A, The effects of TL-associated variants and genetic risk score (GRS) on AD using data from the International Genomics of Alzheimer’s Project Consortium in the Lambert et al5 study. The OR could be interpreted as changes in odds per TL-decreasing allele for AD. B, IV estimates of TL on AD for each variant and GRS. The OR could be interpreted as changes in odds per SD increase of TL for AD. SNP indicates single-nucleotide polymorphism.
Zhan Y, Song C, Karlsson R, Tillander A, Reynolds CA, Pedersen NL, Hägg S. Telomere Length Shortening and Alzheimer Disease—A Mendelian Randomization Study. JAMA Neurol. 2015;72(10):1202-1203. doi:10.1001/jamaneurol.2015.1513
Telomeres are sequences of repetitive nucleotides at the end of the chromosomes, which protect them from fusion with neighboring chromosomes.1 Observational studies have found associations between shorter telomeres and Alzheimer disease (AD).2 However, these studies could have residual confounding or reverse causation, making it difficult to draw conclusions on whether telomere length (TL) is causally associated with AD. For the past decades, instrumental variable (IV) analysis has been developed for assessing causality using genetic variants in epidemiological research under the name of mendelian randomization (MR).3 In the present study, we investigated the causal effect of TL on AD by applying the MR method to summary genome-wide association study (GWAS) data from Codd et al4 and from the International Genomics of Alzheimer’s Project Consortium.5
We used 7 lead single-nucleotide polymorphisms (SNPs) identified to be of genome-wide significance4 as IVs for TL and extracted the final estimates of these SNPs on AD from the International Genomics of Alzheimer’s Project Consortium.5 For the causal effect of TL on AD using single SNPs as IVs, we calculated the estimates by the division of each SNP’s effect on AD and the corresponding SNP’s effect on TL. To estimate the causal effect of TL on AD using a genetic risk score as the IV, we applied the MR method using summary data.6 Statistical analyses were performed in R 3.0 (R Project for Statistical Computing).
The effects of each TL SNP, or genetic risk score, on AD and the overall IV estimates are described in the Table and Figure. Two SNPs (TERT and OBFC1) were associated with AD (P = .002 and .02, respectively). The genetic risk score for TL was significantly associated with shorter TL (β, −0.07 SD decrease of TL per allele; 95% CI, −0.06 to −0.08; P = 4.9 × 10−90) and higher risk for AD (odds ratio, 1.02 for per TL-decreasing allele; 95% CI, 1.01 to 1.04; P = .004). The MR analysis showed that shorter TL was causally associated with a higher risk for AD (odds ratio, 1.36 per SD decrease of TL; 95% CI, 1.12 to 1.67; P = .002).
In the present study, for the first time to our knowledge, we used IV techniques and provided support for a causal effect of TL on the risk for AD using summary GWAS data. The MR analysis has the advantage of being independent of any measured or unmeasured confounders by using genetic variants as IVs. However, an MR study also requires a large sample size to gain sufficient statistical power. To achieve this, we took advantage of the summary GWAS data from the International Genomics of Alzheimer’s Project Consortium.5 Based on the original TL article,4 we roughly estimate 1 SD decrease to be equal to an attrition rate of 1226 TL base pairs in 40 to 60 years, corresponding to a 36% higher risk for AD.
Leukocyte TL, as measured in the GWAS, is correlated with TL in neurons. Thus, it should be considered a proxy of neuronal TL. We tested the MR assumption regarding the pleiotropic effects of the 7 SNPs and did not find any evidence for violations of these assumptions. Additionally, we performed sensitivity analysis by applying likelihood-based methods6 for the final MR estimate and obtained similar results.
In summary, our study provided evidence for a causal relationship between TL and AD. Further elucidation of this association could provide insights into the physiological roles of telomeres in AD pathogenesis.
Corresponding Author: Sara Hägg, PhD, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, 171 77 Stockholm, Sweden (firstname.lastname@example.org).
Author Contributions: Dr Hägg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Pedersen, Hägg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Zhan, Hägg.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Zhan, Song.
Obtained funding: Pedersen, Hägg.
Administrative, technical, or material support: Pedersen, Hägg.
Study supervision: Pedersen, Hägg.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded by a Karolinska Institutet delfinansiering grant for doctoral student (Dr Zhan), the Loo & Hans Osterman Foundation, the Foundation for Geriatric Diseases, the Swedish Council for Working Life and Social Research (2013-2292), and the Swedish Research Council (521-2013-8689).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the European Union Framework 7 ENGAGE Project (HEALTH-F4-2007-201413) for providing telomere summary data, and the International Genomics of Alzheimer’s Project Consortium for providing Alzheimer disease summary data for these analyses.