Telomere length, Metabolic Traits, and Myocardial Infarction: A Mediation Mendelian Randomization Analysis

doi:10.21203/rs.3.rs-2503938/v1

Download PDF

Article

Telomere length, Metabolic Traits, and Myocardial Infarction: A Mediation Mendelian Randomization Analysis

https://doi.org/10.21203/rs.3.rs-2503938/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The associations between leukocyte telomere length (LTL) and common metabolic traits for myocardial infarction (MI) are unclear. We aimed to evaluate the causal effect of LTL on MI and the mediating role of metabolic traits in the pathway from LTL to MI. Summary statistics for LTL, twelve metabolic traits and MI were obtained from large consortia of genome-wide association studies. A two-sample two-step MR was used to determine 1) the causal effect of LTL on MI and twelve metabolic traits; 2) causal effects of metabolic traits on MI after adjusting for LTL; and 3) mediation effects of these metabolic traits. We observed genetically predicted longer LTL was strongly associated with lower risk of MI (OR[95% CI]:0.832[0.750,0.924]; P<0.001), lower FI (β[95% CI]:-0.041[-0.062,-0.020]; P<0.001), higher SBP (1.558[0.778,2.338]; P<0.001), and higher DBP (0.785[0.223,1.347]; P<0.001)but not associated with other metabolic traits. SBP, DBP and FI were positively associated with MI after adjusting for LTL. Mediation analysis showed evidence of positively indirect effect of LTL on MI through SBP and DBP, while an inversely indirect effect through FI. The direct effect of LTL on MI (OR[95% CI]:0.778[0.666,0.909]; P=0.002) was observed after adjusting for SBP, DBP and FI. Our study identified the independent causal role of LTL on MI and the mediating effects of SBP, DBP and FI in the causal pathway from LTL to MI.

Health sciences/Cardiology

Health sciences/Endocrinology

Health sciences/Risk factors

Health sciences/Medical research/Genetics research

telomere length

metabolic traits

myocardial infarction

Mendelian randomization

mediation

Myocardial infarction (MI), or heart attack in lay terms, is usually caused by acute blockage of coronary arteries, resulting in heart muscle and cardiac injury ¹. It is a life-threatening disease with a high global prevalence and remains a source of high morbidity and mortality ². The prevalence of MI increases dramatically with age, therefore telomere length, one of the best biomarkers of biological aging, is considered to be a potential predictor of MI occurrence ^3,4. Telomeres are composed of G-rich repetitive DNA sequences that are located at the terminal of linear eukaryotic chromosomes, maintaining chromosomal stability and controlling cell proliferation ⁵. In normal human somatic cells, telomeres shorten gradually as cells divide, but when reaching a critical level, cells no longer divide and undergo apoptosis or senescence ⁶. Subsequently, senescent cells can acquire a proinflammatory senescence-related secretory phenotype, which promotes the risk of aging-related cardiovascular diseases ⁷.

In recent years, high throughout assays to measure peripheral leucocytes telomere length (LTL) have allowed researchers to study the role of biological aging in MI evolution ⁸. Although compelling evidence from previous studies has consistently shown that shorter LTL may increase the risk of CVD, these studies lack definitive evidence regarding the causal relationship between LTL and MI ^9–11. Meanwhile, a prospective cohort study counterintuitively found that a longer telomere has been associated with a high risk of incident MI, but this result was limited by small sample sizes and could not correct the effects of confounding factors ¹². Furthermore, these studies might be biased by other confounders and reverse causation, thus causal inferences cannot be drawn on the basis of existing observational evidence.

Epidemiological studies also highlighted the important role of metabolic factors in telomere attrition and MI. For example, a meta-analysis identified a strong association between LTL and the risk of MI, and underlined that cardiometabolic risk factors (such as body mass index (BMI), hypertension and diabetes) could contribute to telomere attrition and therefore accelerates biological aging ¹³. A cross-sectional study of US adults also found that LTL was strongly associated with adiposity and related biomarkers including BMI, waist circumference, high-density lipoprotein cholesterol, triglycerides, insulin resistance, systolic blood pressure, and diastolic blood pressure and so on ¹⁴. Broadly speaking, multiple metabolic risk factors, such as abnormal of blood pressure, blood glucose, blood lipids, waist circumference, and other factors, have been confirmed to be closely related to increased MI risk ^15,16 as well as telomere shortening ^17–20. Thus, we hypothesize these metabolic traits may act as potential mediators that lie in the pathway from LTL to MI.

Mendelian randomization (MR) is an epidemiological technique that uses genetic variants as instrumental variables to estimate the causal effect of exposures ²¹. Unlike traditional observational studies, MR can overcome some of the limitations of reverse causation bias and confounding. Previous MR studies have proved the causal effect of LTL on MI ²², the causal effect on several metabolic risk factors, such as blood pressure ²³, glycemic progression ²⁴, and BMI²⁵. On the other hand, MR studies also supported that metabolic risk factors may increase the risk of CVD ^26,27. These results indicated that metabolic factors may partly account for the causal effects of LTL on CVD, but none of them have explored their potential mediating roles in the pathway from LTL to MI. Therefore, to evaluate the causal relationship between LTL and MI and further explored the possible mediators of metabolic traits, we conducted a two-step and two-sample MR based on large-scale summarized genome-wide association study (GWAS) data that are publicly available.

Study Design

We applied a two-step, two-sample MR method using publicly available datasets to explore the potential mediated roles of metabolic traits (blood pressure, lipid traits, glycemic traits, and measures of adiposity) between LTL and the MI in the population of European ancestry (Fig. 1). Specifically, we first estimated the causal effect of LTL on MI, and then the effects of potential mediation using two-step MR. In step one, we tested causal effects of LTL on potential mediators using univariable MR (UVMR) and in step two we investigate the causal effects of potential mediators on MI adjusted for LTL using multivariable MR (MVMR).

Data Sources

The exposure was LTL and the study outcome was MI in the present study. A total of 12 metabolic traits were considered as potential mediators, including systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), fasting insulin (FI), fasting glucose (FG), glycated hemoglobin (HbA1c), waist circumference (WC), waist-to-hip ratio (WHR), and BMI. For each trait, summary-level data (effect estimates, standard errors, and P-values) were obtained from recent large European GWASs (Table 1). The studies included in cited GWASs were all approved by a relevant review board.

Table 1

Data sources were used in the MR analyses for the current study.
Phenotype	Participants included in analysis	Ancestry	Measurement	Adjustments	Source
Exposure
Leukocyte telomere length	472,174 individuals	European	multiplex qPCR methodology (T/S ratio)	Age, sex, genotyping array, and 10 PCs	UK Biobank (https://figshare.com/s/caa99dc0f76d62990195)²⁸
Mediators
SBP	757,601 individuals	European	Automated or manual measurement	Sex, age, genotyping array, and BMI	UK Biobank, ICBP consortium (https://gwas.mrcieu.ac.uk/datasets/ieu-b-38/)³⁰
DBP	757,601 individuals	European	Automated or manual measurement	Sex, age, genotyping array, and BMI	UK Biobank, ICBP consortium (https://gwas.mrcieu.ac.uk/datasets/ieu-b-39/)³⁰
TC	187,365 individuals	European	Directly measure after > 8h of fasting	Age, sex, study-specific covariates, and PCs	GLGC consortium (http://csg.sph.umich.edu/willer/public/lipids2013/)³¹
LDL-C	173,082 individuals	European	Directly measure after > 8h of fasting or use Friedewald formula	Age, sex, and PCs	GLGC consortium (http://csg.sph.umich.edu/willer/public/lipids2013/)³¹
HDL-C	187,167 individuals	European	Directly measure after > 8h of fasting	Age, sex, and PCs	GLGC consortium (http://csg.sph.umich.edu/willer/public/lipids2013/)³¹
TG	177,861 individuals	European	Directly measure after > 8h of fasting	Age, sex, and PCs	GLGC consortium (http://csg.sph.umich.edu/willer/public/lipids2013/)³¹
FI	151,013 individuals	European	Measure in pmol/L	BMI, study-specific covariates, and PCs	MAGIC consortium (https://magicinvestigators.org/downloads/)³²
FG	200,622 individuals	European	Measure in mmol/L	BMI, study-specific covariates, and PCs	MAGIC consortium (https://magicinvestigators.org/downloads/)³²
HbA1c	146,806 individuals	European	Measure in %	Study-specific covariates, and PCs	MAGIC consortium (https://magicinvestigators.org/downloads/)³²
WC	231,353 individuals	European	Waist circumference measure at baseline	Age, study-specific covariates, and BMI	GIANT consortium (https://portals.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium_data_files)³³
WHR	210,082 individuals	European	Waist circumference and hip circumference at baseline	Age, study-specific covariates, and BMI	GIANT consortium (https://portals.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium_data_files)³³
BMI	681,275 individuals	European	Height and weight measured at baseline	Age, sex, recruitment center, genotyping batches and 10 PCs	UK Biobank, GIANT consortium (https://portals.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium_data_files)³⁴
Outcome
MI for primary analysis	61,505 cases and 577,212 controls	European	Measure in ICD-10 code	Age, sex, the first 20 principal components, and genotyping array	UK Biobank, CARDIoGRAMplusC4D Consortium (https://www.ebi.ac.uk/gwas/docs/about)³⁵
MI for sensitivity analysis	43,676 cases and 128,199 controls	European	Measure in ICD-10 code	Age, sex, and the first 20 principal components.	CARDIoGRAMplusC4D Consortium (https://gwas.mrcieu.ac.uk/datasets/ieu-a-798/)³⁹
Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; FG, fasting glucose; FI, fasting insulin; GIANT, genetic investigation of anthropometric traits; GLGC, the global lipids genetics consortium; ICBP, international consortium for blood pressure; ICD-10, International Classification of Diseases version-10; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MAGIC, the meta-analyses of glucose and insulin-related traits consortium; MI, myocardial infarction; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; WC, waist circumference; WHR, waist-to-hip ratio.

Leukocyte telomere length Summary-level data for LTL was extracted from the meta-analysis of GWAS from the UK Biobank, including 472,174 individuals of European ancestry ²⁸. Using the multiplex qPCR methodology, LTL is measured as the ratio of telomere repeat copy number (T) relative to that of a single copy gene ²⁹. The LTL values have been log-transformed and Z-standardized for each individual. The genetic associations between SNPs and LTL were adjusted for age, sex, genotyping array and the 10 genetic principal components (PCs).

Potential mediators Associations of SNPs with 12 metabolic traits were derived from publicly available GWAS consortiums. Genetic associations with SBP and DBP were extracted from the meta-analysis of GWAS of UK Biobank (N = 458,577) and the international consortium of blood pressure (N = 299,024) ³⁰. SBP and DBP values were added 15 and 10 mmHg for users of lowering blood pressure medication, respectively. Effect sizes of SNPs associated with TC, LDL-C, HDL-C, and TG were extracted from the meta-analysis of GWAS of global lipid genetic consortium (GLGC) without users of lipid-lowering medication (N = 188,577) ³¹. Blood lipid levels were typically measured after > 8 hours of fasting. Associations of SNPs with FI, FG, and HbA1c were derived from the Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC) of European ancestry individuals without diabetes and users of diabetes-relevant medication ³². Measures for FG and FI taken from whole blood were corrected to plasma level using the correction factor 1.13. Data for WC and WHR were derived from a meta-analysis of GWAS of up to 230,000 individuals from the genetic investigation of anthropometric traits (GIANT) consortium ³³. For BMI, the genetic associations were extracted from a meta-analysis of GWAS of up to 700,000 individuals from the UK Biobank and GIANT consortium ³⁴. More detailed information was described in Table 1.

Study outcome Data on MI was obtained from the hitherto largest published meta-analysis of GWAS among individuals of European ancestry, which was composed of 638,717 individuals (61,505 individuals with MI and 577,212 controls) from UK Biobank and CARDIoGRAMplusC4D consortium, with adjustment for age, sex, 20 principal components, and genotyping array ³⁵. The outcome of MI was defined by International Classification of Diseases version-10 (ICD-10) codes (I21, I22, I23, I25.2), doctor-diagnosed, and self-reported ³⁵.

Selection of instrumental variable

We used single nucleotide polymorphisms (SNPs) as genetic instrumental variables (IVs) in MR analyses and selected IVs according to the following criteria. First, we selected all SNPs which reached genome-wide significance (P < 5×10^− 8) as IVs for exposures (LTL, potential mediators, and MI). Second, we excluded SNPs with minor allele frequency < 0.01, linkage disequilibrium (LD) coefficient r² > 0.001 within 10 Mb. Third, we calculated F-statistics for each SNP and selected SNPs with F-statistics > 10 to ensure the strength of the IVs. Finally, for those SNPs associated with exposures or mediators, but were unavailable in the public GWAS of the outcomes, we used proxies with r² > 0.80 in the genotype data of European from Phase 3 of the 1000 Genomes Project. The IVs used to conduct multivariable MR (MVMR) were combinations of SNPs associated with the corresponding exposures. The criteria for MVMR were similar to the UVMR criteria.

Statistical Analyses

Two-step MR ³⁶, a novel approach for causal mediation analysis, was employed to investigate the mediation effects of metabolic traits between LTL and MI. First, we estimate the causal effects of LTL on MI and potential mediators by performing two-sample UVMR analyses. The inverse-variance weighted (IVW) method (when there is heterogeneity, we used the IVW with random effects, otherwise we used fixed effects) was performed as main method while the heterogeneity was assessed by the Cochran Q test and I² index. To assess the robustness of the findings, we also performed sensitivity analyses using methods with different assumptions about horizontal pleiotropy, including MR-Egger regression, weighted median, MR-robust adjusted profile score (MR-RAPS), and MR-pleiotropy residual sum and outlier (MR-PRESSO). And I²_GX was calculated to quantify the strength of ‘no measurement error’ violation for MR-Egger and I²_GX greater than 0.9 represents the MR-Egger estimate is unlikely to be biased by violation of the ‘no measurement error’ ³⁷. Detailed advantages of the methods were described in Supplementary Text S1. In order to pass our sensitivity analyses, at least three of these four methods must agree with the primary IVW results.

Second, we estimate the causal effects of the potential mediators on the outcome of MI. Only metabolic traits that were strongly significant in the first step were considered as potential mediators. For each mediator, we estimated the direct effects on MI adjusted for LTL using two-sample MVMR. Similarly, we carried out random-effect or fixed-effect MVMR-IVW analyses as main analysis to assess the direct effects according to the existence of heterogeneity. The MVMR-Egger and MVMR-robust were used as sensitivity analyses. Detailed information was described in Supplementary Text S1.

Finally, we used the product of coefficients method to estimate the indirect effect. This involved first estimating the effect of LTL on each metabolic trait, then multiplying this with the effect of that metabolic trait on the outcome after adjusting for LTL. Standard errors for mediation effects were derived by using delta method ³⁸. In sum, we considered all mediators and obtained a causal effect of LTL on MI after adjusting for all possible mediators. The mediated effect by all the mediators was calculated by the total effect of LTL on MI minus the direct effect of LTL on MI with adjustment for all mediators.

For selection of potential mediators, the Bonferroni corrected P-values 0.0042 (0.05/12, 12 metabolic traits were considered as potential mediators) was used and P-values below 0.0042 were considered as evidence of associations (Fig. 1). For other analyses, results with P-values < 0.05 indicated statistically significant causal association. All analyses were two-sided and performed using R (version 3.6.3). Two-step MR was performed using the “TwoSampleMR” (version 0.5.6) and “MendelianRandomization” (version 0.3) packages.

Effect of LTL on the risk of MI

According to the selection of IV, we selected 153 independent SNPs at a genome-wide significant level to genetically predicted LTL. After harmonizing, a total of 137 SNPs were selected as IVs for the UVMR analysis of causal inference between LTL and MI (Supplementary Table S1). The F-statistics of Genetic IVs for the LTL in the current MR studies ranged from 28.65 to 1043.70, suggesting the weak instruments bias is minimal. Because indication of heterogeneity (P for Cochran Q test < 0.001, I² = 68.12%; Supplementary Table S2) in the IVW analysis, we selected IVW with random effects to alleviate the problem. In the primary IVW method, genetically predicted LTL was inversely associated with the risk of MI (OR [95% CI]: 0.832 [0.750,0.924]; P < 0.001; Fig. 2). It means an increase of one standard derivation (SD) in genetically predicted LTL led to a 16.8% [95% CI: 7.6%, 25.0%] lower risk of MI. Consistent causal associations were also observed by using the methods of weighted median (0.857 [0.776,0.946]; P < 0.001), MR-RAPS (0.848 [0.778,0.925], P < 0.001), and MR-PRESSO (0.832 [0.750,0.924]; P < 0.001; Fig. 2). MR-Egger intercept test and MR-PRESSO distortion test both indicated no evidence of horizontal pleiotropy (all P > 0.05). Scatter plots were presented in Supplementary Fig. 1A. In the leave-one-out analysis where we iteratively removed one SNP at a time and performed IVW using the remaining SNPs, we did not observe apparent outlying SNPs and the odds ratios were in accordance with our primary findings, aggregating closely around the expected value of estimation (Supplementary Fig. 1B).

Effects of LTL on metabolic traits

Detailed information on the genetic IVs for UVMR of LTL on metabolic traits after LD clumping and harmonization are represented in Supplementary Table S3. Genetically predicted increased LTL appeared to be strongly associated with lower levels of FI (β [95% CI]: -0.041[-0.062, -0.020]; P < 0.001), higher SBP (1.558 [0.778, 2.338]; P < 0.001), and higher DBP (0.785 [0.223, 1.347]; P < 0.001) (Fig. 3). Weighted median, MR-RAPS, and MR-PRESSO methods provided consistent results of LTL on FI, SBP, and DBP (all P < 0.05; Supplementary Table S4). Only suggestive evidence for the causal associations of LTL on TG (0.065 [0.007,0.123]; P = 0.028) and WC (0.066 [0.015, 0.117]; P = 0.011; Fig. 3), while MR-Egger and weighted median don’t provide significant results (all P > 0.05; Supplementary Table S4). No evidence was observed for the causal relationships of LTL on LDL-C, HDL-C, TG, TC, FG, HbA1c, BMI, and WHR (all P > 0.05; Fig. 3) and other UVMR methods also showed no evidence for the effect of LTL on these traits (Supplementary Table S4). I²_GX were greater than 0.9 except for TC, indicating the expected relative bias of the MR-Egger causal estimate is minimal (Supplementary Table S4). No horizontal pleiotropy was detected by MR-Egger intercept test and MR-PRESSO distortion test (Supplementary Table S4).

Effects of metabolic traits on the risk of MI

Because causal relationships of LTL with FI, SBP, and DBP were observed, we considered FI, SBP, and DBP as potential mediators. Detailed information on Genetic IVs for the MVMR model with exposures of FI and LTL, exposures of SBP and LTL, and exposures of DBP and LTL were presented in Supplementary Table S5, Table S6, and Table S7, respectively. After adjusting for LTL, MVMR analyses demonstrated that direct positive effect of genetically predicted FI (OR [95% CI]: 2.091 [1.568,2.787]; P = 0.026), SBP (1.034 [1.028, 1.039]; P < 0.001), and DBP (1.059 [1.050, 1.069]; P < 0.001) on the risk of MI (Fig. 4). MVMR-Egger and MVMR-robust methods provided robust and consistent results (Supplementary Table S8). MVMR-Egger intercept test indicated no evidence of horizontal pleiotropy (all P > 0.05; Supplementary Table S8).

Mediation effects of metabolic traits on MI

Since FI, SBP, and DBP were causally influenced by LTL and also presented causal effects on MI, we conducted mediation analyses to investigate whether the effect of LTL on MI was mediated by FI, SBP, and DBP. The causal association between LTL and MI was mediated by FI, with an inverse mediation effect of -0.030 ([95%CI: -0.049, -0.011]), whilst the positive mediation effects of SBP and DBP were observed (0.052 [0.024, 0.079] and 0.045 [0.012, 0.078] for SBP and DBP, respectively; Fig. 5). The total mediation effect of the combination of SBP, DBP, and FI on MI was 0.068 ([0.017, 0.118]). Notably, the direct effect of LTL on MI was observed (OR [95% CI]: 0.778 [0.666, 0.909]; P = 0.002; Supplementary Table S8) with adjustment for SBP, DBP, and FI.

Sensitivity analysis

To orient the direction of causality between LTL and MI, UVMR method was also performed to investigate the causal relationship of MI on LTL. Genetic IVs for MI were presented in Supplementary Table S1. There was no causality between genetically determined MI and LTL (Supplementary Table S2). Thus, the direction of causal relationship between LTL and MI is from LTL to MI.

Given the population overlap of MI with LTL, SBP, and DBP in the UK Biobank, we extracted the summary-level data of MI from the CARDIoGRAMplusC4D GWAS and repeated the analyses involving the traits ³⁹ (Table 1). Detailed information was described in Supplementary Text S1. Results were similar to the primary analysis. UVMR results supported the causal effect of LTL on MI, while no evidence of the causal effect of MI on LTL (Supplementary Table S9). MVMR results showed positive direct effects of FI, SBP, and DBP on MI adjusting for genetic association with LTL (Supplementary Table S10). Mediation analysis showed an inversely indirect effect of FI on MI and positive indirect effects of SBP and DBP on MI on the pathway from LTL to MI (Supplementary Figure S2). The total mediation effect of the combination of SBP, DBP, and FI on MI was positive (Supplementary Figure S2) and the direct effect of LTL on MI was observed with adjustment for DBP, SBP, and FI (Supplementary Table S10).

This study provided genetic support for a causal relationship between LTL and MI, with a 17% reduction in MI risk for each 1-SD increase in LTL. When considering the roles of metabolic factors in the pathway from LTL to MI within a mediation MR framework, we identified three factors (SBP, DBP, and FI) that significantly mediated the effect of LTL on MI. Furthermore, the causality between LTL and MI remained robust and consistent in all sensitivity analyses and MVMR analyses, suggesting the relationship was independent of preformed mediators.

Consistent with our findings, previous studies convinced that LTL has a protective causal effect on MI ^9,40. A meta-analysis including 32 studies, with a total of 144 610 participants, showed that the shortest LTL was associated with a 39% higher risk of incident MI after adjusting by traditional covariates of chronical age, and sex ^13,41. MR studies have also supported that longer telomeres reduce the risk for CVD, whereby per 1-SD increase in change in genetically increased LTL was responsible for a 12% lower risk of coronary heart disease ⁴². Moreover, a network MR analysis identified a mediating role of insulin on the causal pathway of short telomeres and coronary heart disease pathogenesis ²⁶. However, the association between LTL and MI has not been studied exhaustively until our two-sample bidirectional MR analysis revealed the causal relationships and oriented the causal direction. Several mechanisms have been proposed to account for the independent role of LTL in MI. First, telomere attrition and dysfunction may contribute to increased chromosomal instability, causing senescence and destruction of some cells ⁴³. Accumulation of senescent cells leads to vascular changes and dysregulation of blood flow, triggering atherosclerosis and forthcoming cardiovascular disease ⁴⁴. Second, experimental evidence suggested that telomere attrition provokes pro-inflammatory cytokines secretion ⁴⁵, while inflammation has long been regarded as a key contributor to MI. Finally, short LTL has been associated with other aging-associated disorders in humans, which could be a general reflection of vascular aging ⁴⁶.

Furthermore, we also explored the causal mechanism underlying the pathway from LTL to MI with an emphasis on a wide range of metabolic factors. In the MR mediation analysis, we observed that genetically predicted increased LTL was associated with higher SBP, higher DBP, and lower FI, respectively. Then, as expected, the genetically predicted SBP, DBP and FI all showed a positive association with MI. It suggests that LTL disparities may produce additional burden of MI partly by increasing SBP, DBP or decreasing FI. The positive association of LTL with SBP and DBP seemed paradoxical, since several prior studies showed that measured LTL was inversely associated with BP traits, and that the number of LTL shortening alleles was inversely associated with SBP ^47,48. Conversely, two studies reported an increasing trend in SBP with genetically increased telomere length, despite the observational associations of TL with SBP and DBP being minimal ^14,49. Besides, a study has also shown inconsistent results that mean telomere length was related to SBP and DBP in a non-linear manner ⁵⁰. Thus, no definite conclusions can be drawn from these results. By contrast, a recent MR study provided causal evidence that longer LTL increased lung function and blood pressure traits among middle-aged U.K. Biobank participants, which was in accordance with our mediation analysis ²³. Unexceptionally, another MR study including 78 592 Europeans suggested that genetically instrumented longer LTL was associated with raised SBP and DBP, while no unbalanced horizontal pleiotropy was detected ⁵¹. A possible explanation for this stems from an epidemiological study in the Framingham Offspring Cohort that longer LTL was associated with increased aldosterone-to-renin ratio and eventually causes hypertension ⁵². With respect to FI, our results were consistent with previous observational and MR studies, which found associations between shorter telomeres and higher fasting insulin levels ^53,54. A possible explanation for the mediating role of blood pressure stems from an epidemiological study in the Framingham Offspring Cohort, showing that longer LTL was associated with an increased aldosterone-to-renin ratio, which directly causes increased blood pressure and thus increasing the risk of MI ⁵². Another general perception for LTL and FI is that oxidative stress and inflammation are unifying factors for the associations of a relatively short LTL with atherosclerosis and with insulin resistance ^55,56. And then pancreatic islets respond to the increased insulin demand due to insulin resistance, maintaining homeostasis by increasing insulin output and expanding beta cell mass ⁵⁷. Ultimately, both the insulin-resistant state and the resulting metabolic disturbances increase the risk of developing MI ⁵⁸.

To the best of our knowledge, our study is the first to use two-sample MR analysis to investigate the causal role of metabolic traits in explaining the effect of telomere length on the risk of myocardial infarction. Our study has several strengths. First, we assess the causal role of mediators by using a two-step two-sample MR method with large-scale GWAS summary-level data, which suffered less bias from reverse causation, confounders, and measurement error than observational mediation analysis ³⁶. Second, we perform MVMR to estimate the direct effect of LTL on MI by adjusting for all significant mediators together and emphasize the effect of LTL on MI is independent of blood pressure and fasting insulin. These findings established that telomeres length might serve as a promising biomarker in predicting the risk of MI. Third, multiple sensitivity analyses were performed for enhancing the robustness of the results, testing the validity of the assumptions, and accounting for sample overlap, thus minimizing the possibility of biased results. Nevertheless, several potential limitations should also be considered. First, unknown causal mediated pathways in the association between LTL and MI may affect the results, despite the lack of evidence for pleiotropy in the analyses and twelve metabolic traits were considered in the study. Second, there was sample overlap in the GWAS of LTL, SBP, DBP and outcome of MI, which might increase false positive error and lead to bias ⁵⁹. Whereas, the analysis of MI GWAS summary-level data from only CARDIoGRAMplusC4D consortium generated similar results and large F-statistics (> 10) ensure the bias was minimal. Third, MR is not equivalent to a randomized trial and MR simulates life-course exposure due to genetic variants are randomly assigned to the offspring and remain constant after conception. Therefore, causal mediated effects derived from MR analyses may differ in magnitude of random trial, and should be explained as life-course effects. Fourth, measurement error and interaction between telomere length and potential mediators may bias the results. However, the size of the biases in mediation MR study was smaller than mediation observational analyses ³⁶. Fifth, weak instrument bias could attenuate the causal estimate towards the null. However, we only used genetic variants with F-statistic > 10 to minimize the weak instruments bias ⁶⁰. Finally, we cannot exclude the possibility of metabolic traits playing sex-, age- and ethnicity-specific roles in the pathway from telomere length to MI due to current summary-level genetic data cannot capture these effects. These limitations reiterate the need to interpret our MR results as evidence of the causal relationship between telomeres and MI and exploration of mediating metabolic risk factors, rather than precise estimates of telomere effects. In the future, we hope that data from specific strata by age, sex, and ethnicity in individual-level data will be available to address these unresolved issues.

In conclusion, we identified causal roles of LTL on MI is independent of blood pressure and fasting insulin. The LTL may influence the risk of MI via blood pressure and fasting insulin, which implies new mechanisms of the pathway from telomere length to MI. The findings provide novel insights into the mechanisms of LTL in reducing the risk of MI and evidence to support future clinical trials of LTL and MI.

Acknowledgements

We appreciate all investigators for sharing these published GWAS summary-level data.

Author contributions

Z.L.Z., Y.X.Z., and Z.Y.C. contributed to the research questions and study design. Z.L.Z., Y.X.Z., and Z.Y.C. contributed to the data curation. Z.L.Z., Y.X.Z., and Z.Y.C. contributed to methodology development. Z.L.Z., Y.X.Z., and Z.Y.C. conducted statistical analyses. Z.L.Z., Y.X.Z., J.B.L., and Z.Y.C. helped validate and perform sensitivity analyses. Z.L.Z., Y.X.Z., J.B.L., and Z.Y.C. interpreted the results and wrote the original draft of the manuscript. Z.Y.C., Z.L.Z., Y.X.Z., and J.B.L. helped review and edit the final draft of the manuscript. The author read and approved the final manuscript.

Funding

This research received no external funding.

Ethics declarations

Competing interests

The authors declared no competing interests.

Data availability statement

All the data used in this study are available at GWAS datasets (the data source was listed in Table 1).

Shen, J.-F. et al. Sacubitril Valsartan Enhances Cardiac Function and Alleviates Myocardial Infarction in Rats through a SUV39H1/SPP1 Axis. Oxid Med Cell Longev 2022, 5009289, doi:10.1155/2022/5009289 (2022).
Yang, K. et al. Effect of neuron-derived neurotrophic factor on rejuvenation of human adipose-derived stem cells for cardiac repair after myocardial infarction. J Cell Mol Med 23, 5981-5993, doi:10.1111/jcmm.14456 (2019).
Yazdanyar, A. & Newman, A. B. The burden of cardiovascular disease in the elderly: morbidity, mortality, and costs. Clin Geriatr Med 25, doi:10.1016/j.cger.2009.07.007 (2009).
Brouilette, S., Singh, R. K., Thompson, J. R., Goodall, A. H. & Samani, N. J. White cell telomere length and risk of premature myocardial infarction. Arteriosclerosis, Thrombosis, and Vascular Biology 23, 842-846 (2003).
Liu, N. et al. Telomere dysfunction impairs epidermal stem cell specification and differentiation by disrupting BMP/pSmad/P63 signaling. PLoS Genet 15, e1008368, doi:10.1371/journal.pgen.1008368 (2019).
Huang, Z. et al. Dynamics of leukocyte telomere length in adults aged 50 and older: a longitudinal population-based cohort study. Geroscience 43, 645-654, doi:10.1007/s11357-020-00320-y (2021).
Schuliga, M. et al. Mitochondrial dysfunction contributes to the senescent phenotype of IPF lung fibroblasts. J Cell Mol Med 22, 5847-5861, doi:10.1111/jcmm.13855 (2018).
Allende, M. et al. Short Leukocyte Telomere Length Is Associated With Cardioembolic Stroke Risk in Patients With Atrial Fibrillation. Stroke 47, 863-865, doi:10.1161/STROKEAHA.115.011837 (2016).
Weischer, M. et al. Short telomere length, myocardial infarction, ischemic heart disease, and early death. Arterioscler Thromb Vasc Biol 32, 822-829, doi:10.1161/ATVBAHA.111.237271 (2012).
Fitzpatrick, A. L. et al. Leukocyte telomere length and cardiovascular disease in the cardiovascular health study. Am J Epidemiol 165, 14-21 (2007).
Haycock, P. C. et al. Leucocyte telomere length and risk of cardiovascular disease: systematic review and meta-analysis. BMJ 349, g4227, doi:10.1136/bmj.g4227 (2014).
Østhus, I. B. Ø., Lydersen, S., Dalen, H., Nauman, J. & Wisløff, U. Association of Telomere Length With Myocardial Infarction: A Prospective Cohort From the Population Based HUNT 2 Study. Prog Cardiovasc Dis 59, 649-655, doi:10.1016/j.pcad.2017.04.001 (2017).
Emami, M. et al. Accelerated Biological Aging Secondary to Cardiometabolic Risk Factors Is a Predictor of Cardiovascular Mortality: A Systematic Review and Meta-analysis. Can J Cardiol 38, 365-375, doi:10.1016/j.cjca.2021.10.012 (2022).
Rehkopf, D. H. et al. Leukocyte Telomere Length in Relation to 17 Biomarkers of Cardiovascular Disease Risk: A Cross-Sectional Study of US Adults. PLoS Med 13, e1002188, doi:10.1371/journal.pmed.1002188 (2016).
Akhabue, E. et al. New and emerging risk factors for coronary heart disease. Am J Med Sci 347, 151-158, doi:10.1097/MAJ.0b013e31828aab45 (2014).
Mozaffarian, D., Wilson, P. W. & Kannel, W. B. Beyond established and novel risk factors: lifestyle risk factors for cardiovascular disease. Circulation 117, 3031-3038, doi:10.1161/CIRCULATIONAHA.107.738732 (2008).
Chen S, Y. F., Lin J, Matsuguchi T, Blackburn E, Lee ET, Howard BV, Zhao J. Short leukocyte telomere length is associated with obesity in American Indians the Strong Heart Family study. Aging (Albany NY) 6(5):380-9, doi:10.18632/aging.100664 (2014 May).
Chen, W. et al. Leukocyte telomere length is associated with HDL cholesterol levels: The Bogalusa heart study. Atherosclerosis 205, 620-625, doi:10.1016/j.atherosclerosis.2009.01.021 (2009).
Elks CE, S. R. The long and short of telomere length and diabetes. Diabetes 63(1):65-7, doi:10.2337/db13-1469 (2014).
Gardner, J. P. et al. Rise in insulin resistance is associated with escalated telomere attrition. Circulation 111, 2171-2177, doi:10.1161/01.CIR.0000163550.70487.0B (2005).
Hemani, G., Tilling, K. & Davey Smith, G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet 13, e1007081, doi:10.1371/journal.pgen.1007081 (2017).
Deng, Y. et al. Telomere length and the risk of cardiovascular diseases: A Mendelian randomization study. Front Cardiovasc Med 9, 1012615, doi:10.3389/fcvm.2022.1012615 (2022).
Demanelis, K., Tong, L. & Pierce, B. L. Genetically Increased Telomere Length and Aging-Related Traits in the U.K. Biobank. J Gerontol A Biol Sci Med Sci 76, 15-22, doi:10.1093/gerona/glz240 (2021).
Cheng, F. et al. Shortened Leukocyte Telomere Length Is Associated With Glycemic Progression in Type 2 Diabetes: A Prospective and Mendelian Randomization Analysis. Diabetes Care 45, 701-709, doi:10.2337/dc21-1609 (2022).
Rode, L., Nordestgaard, B. G., Weischer, M. & Bojesen, S. E. Increased body mass index, elevated C-reactive protein, and short telomere length. J Clin Endocrinol Metab 99, E1671-E1675, doi:10.1210/jc.2014-1161 (2014).
Zhan, Y. et al. Exploring the Causal Pathway From Telomere Length to Coronary Heart Disease: A Network Mendelian Randomization Study. Circ Res 121, 214-219, doi:10.1161/CIRCRESAHA.116.310517 (2017).
Huang, Y. et al. Associations of Visceral Adipose Tissue, Circulating Protein Biomarkers, and Risk of Cardiovascular Diseases: A Mendelian Randomization Analysis. Front Cell Dev Biol 10, 840866, doi:10.3389/fcell.2022.840866 (2022).
Codd, V. et al. Polygenic basis and biomedical consequences of telomere length variation. Nature Genetics 53, 1425-1433, doi:10.1038/s41588-021-00944-6 (2021).
Cawthon, R. M. Telomere length measurement by a novel monochrome multiplex quantitative PCR method. Nucleic Acids Res 37, e21, doi:10.1093/nar/gkn1027 (2009).
Evangelou, E. et al. Genetic analysis of over 1 million people identifies 535 new loci associated with blood pressure traits. Nat Genet 50, 1412-1425, doi:10.1038/s41588-018-0205-x (2018).
Willer, C. J. et al. Discovery and refinement of loci associated with lipid levels. Nat Genet 45, 1274-1283, doi:10.1038/ng.2797 (2013).
Chen, J. et al. The trans-ancestral genomic architecture of glycemic traits. Nat Genet 53, 840-860, doi:10.1038/s41588-021-00852-9 (2021).
Shungin, D. et al. New genetic loci link adipose and insulin biology to body fat distribution. Nature 518, 187-196, doi:10.1038/nature14132 (2015).
Yengo, L. et al. Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum Mol Genet 27, 3641-3649, doi:10.1093/hmg/ddy271 (2018).
Hartiala, J. A. et al. Genome-wide analysis identifies novel susceptibility loci for myocardial infarction. Eur Heart J 42, 919-933, doi:10.1093/eurheartj/ehaa1040 (2021).
Carter, A. R. et al. Mendelian randomisation for mediation analysis: current methods and challenges for implementation. Eur J Epidemiol 36, 465-478, doi:10.1007/s10654-021-00757-1 (2021).
Bowden, J. et al. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int J Epidemiol 45, 1961-1974, doi:10.1093/ije/dyw220 (2016).
Carter, A. R. et al. Understanding the consequences of education inequality on cardiovascular disease: mendelian randomisation study. BMJ 365, l1855, doi:10.1136/bmj.l1855 (2019).
Nikpay, M. et al. A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nat Genet 47, 1121-1130, doi:10.1038/ng.3396 (2015).
De Meyer, T. et al. Telomere Length as Cardiovascular Aging Biomarker: JACC Review Topic of the Week. J Am Coll Cardiol 72, 805-813, doi:10.1016/j.jacc.2018.06.014 (2018).
Brouilette, S. W. et al. Telomere length, risk of coronary heart disease, and statin treatment in the West of Scotland Primary Prevention Study: a nested case-control study. Lancet 369, 107-114 (2007).
Haycock, P. C. et al. Association Between Telomere Length and Risk of Cancer and Non-Neoplastic Diseases: A Mendelian Randomization Study. JAMA Oncol 3, 636-651, doi:10.1001/jamaoncol.2016.5945 (2017).
Freitas-Simoes, T.-M. et al. Walnut Consumption for Two Years and Leukocyte Telomere Attrition in Mediterranean Elders: Results of a Randomized Controlled Trial. Nutrients 10, doi:10.3390/nu10121907 (2018).
Chen, M. S., Lee, R. T. & Garbern, J. C. Senescence mechanisms and targets in the heart. Cardiovasc Res 118, 1173-1187, doi:10.1093/cvr/cvab161 (2022).
Coppé, J.-P. et al. Senescence-associated secretory phenotypes reveal cell-nonautonomous functions of oncogenic RAS and the p53 tumor suppressor. PLoS Biol 6, 2853-2868, doi:10.1371/journal.pbio.0060301 (2008).
Yetim, E. et al. The association between telomere length and ischemic stroke risk and phenotype. Sci Rep 11, 10967, doi:10.1038/s41598-021-90435-9 (2021).
Rode, L., Nordestgaard, B. G. & Bojesen, S. E. Peripheral blood leukocyte telomere length and mortality among 64,637 individuals from the general population. J Natl Cancer Inst 107, djv074, doi:10.1093/jnci/djv074 (2015).
Scheller Madrid, A., Rode, L., Nordestgaard, B. G. & Bojesen, S. E. Short Telomere Length and Ischemic Heart Disease: Observational and Genetic Studies in 290 022 Individuals. Clin Chem 62, 1140-1149, doi:10.1373/clinchem.2016.258566 (2016).
Brown, L. L., Zhang, Y. S., Mitchell, C. & Ailshire, J. Does Telomere Length Indicate Biological, Physical, and Cognitive Health Among Older Adults? Evidence from the Health and Retirement Study. J Gerontol A Biol Sci Med Sci 73, 1626-1632, doi:10.1093/gerona/gly001 (2018).
Huang, Y.-Q. et al. The relationship between mean telomere length and blood pressure: results from the National Health and Nutrition Examination Surveys. Ann Transl Med 8, 535, doi:10.21037/atm.2020.03.205 (2020).
Loh, N. Y., Noordam, R. & Christodoulides, C. Telomere length and metabolic syndrome traits: A Mendelian randomisation study. Aging Cell 20, e13445, doi:10.1111/acel.13445 (2021).
Vasan, R. S. et al. Association of leukocyte telomere length with echocardiographic left ventricular mass: the Framingham heart study. Circulation 120, 1195-1202, doi:10.1161/CIRCULATIONAHA.109.853895 (2009).
Verhulst, S. et al. A short leucocyte telomere length is associated with development of insulin resistance. Diabetologia 59, 1258-1265, doi:10.1007/s00125-016-3915-6 (2016).
Gardner, J. P. et al. Rise in insulin resistance is associated with escalated telomere attrition. Circulation 111, 2171-2177 (2005).
Avogaro, A., de Kreutzenberg, S. V. & Fadini, G. P. Insulin signaling and life span. Pflugers Arch 459, 301-314, doi:10.1007/s00424-009-0721-8 (2010).
Aviv, A. Genetics of leukocyte telomere length and its role in atherosclerosis. Mutat Res 730, 68-74, doi:10.1016/j.mrfmmm.2011.05.001 (2012).
Dai, C. et al. Pancreatic islet vasculature adapts to insulin resistance through dilation and not angiogenesis. Diabetes 62, 4144-4153, doi:10.2337/db12-1657 (2013).
Dunder, K., Lind, L., Zethelius, B., Berglund, L. & Lithell, H. Increase in blood glucose concentration during antihypertensive treatment as a predictor of myocardial infarction: population based cohort study. BMJ 326, 681 (2003).
Burgess, S., Davies, N. M. & Thompson, S. G. Bias due to participant overlap in two-sample Mendelian randomization. Genet Epidemiol 40, 597-608, doi:10.1002/gepi.21998 (2016).
Burgess, S. & Thompson, S. G. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40, 755-764, doi:10.1093/ije/dyr036 (2011).

No competing interests reported.

SupplementaryTables.xlsx
Supplementary Tables S1-S10.
SupplementaryTextsFigures.docx
Supplementary Text S1 and Figures S1-S2.

Download PDF

Version 1

posted

You are reading this latest preprint version

Telomere length, Metabolic Traits, and Myocardial Infarction: A Mediation Mendelian Randomization Analysis

Status:

Version 1

Abstract

Figures

Introduction

Materials And Methods

Study Design

Data Sources

Selection of instrumental variable

Statistical Analyses

Results

Effect of LTL on the risk of MI

Effects of LTL on metabolic traits

Effects of metabolic traits on the risk of MI

Mediation effects of metabolic traits on MI

Sensitivity analysis

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1