Using human genetics to understand the causes and consequences of circulating cardiac troponin I in the general population

Marta R Moksnes, Helge Røsjø, Anne Richmond, Magnus N Lyngbakken, Sarah E Graham, Ailin Falkmo Hansen, Brooke N Wolford, Sarah A Gagliano Taliun, Jonathon LeFaive, Humaira Rasheed, Laurent F Thomas, Wei Zhou, Ida Surakka, Nicholas J Douville, Archie Campbell, David J Porteous, Paul Welsh, Naveed Sattar, George Davey Smith, Lars G Fritsche, Jonas B Nielsen, Bjørn Olav Åsvold, Kristian Hveem, Caroline Hayward, Cristen J Willer, Ben M Brumpton, Torbjørn Omland
doi: https://doi.org/10.1101/2020.09.04.20187401

Abstract

Circulating cardiac troponin proteins are associated with structural heart disease and predict incident cardiovascular disease in the general population. However, the genetic contribution to cardiac troponin I (cTnI) concentrations and its causal effect on cardiovascular phenotypes is unclear. We combine data from the Trøndelag Health Study and the Generation Scotland Scottish Family Health Study and perform a genome-wide association study of highsensitivity cTnI concentrations with 48 115 individuals.

We identified 12 genetic loci (8 novel) associated with cTnI concentrations. Associated protein-altering variants highlighted putative functional genes: CAND2, HABP2, ANO5, APOH, FHOD3, TNFAIP2, KLKB1 and LMAN1. Using two-sample Mendelian randomization we confirmed the non-causal role of cTnI in acute myocardial infarction, but could not rule out a causal role for cTnI in heart failure. Using genetically informed methods for causal inference of cTnI helps inform the role and value of measuring cTnI in the general population.

Cardiac specific troponins I (cTnI) and T (cTnT) are structural proteins involved in cardiac muscle contraction. During acute myocardial ischemia, cardiac troponin proteins are released into the blood stream and are used to diagnose acute myocardial infarction (AMI)1. In AMI, cardiac troponin release is a result of direct ischemic cardiomyocyte injury. In contrast, the pathobiology of cardiac troponin release outside of AMI is not known, although low-level cardiac troponin elevations are associated with future cardiovascular events in the general population, leading some clinicians to suggest a need for wider use of cardiac troponins in risk stratification in general2. Hence, a better understanding of low-level cardiac troponin release is needed, and the study of large population-based cohorts with genetic data could bring new insights.

The genetic determinants of cardiac troponin in the general population have not been studied in detail, and the downstream biology, including possible causal effects on cardiovascular disease, is unclear. The cTnI and cTnT proteins are encoded by the TNNI3 and TNNT2 genes3, respectively, and mutations in these genes are recognized as disease-causing mutations in patients with familial cardiomyopathies4,5. Patients with cardiomyopathies demonstrate increased concentrations of circulating cardiac troponins6, and these conditions are strongly associated with heart failure (HF) development7. We recently demonstrated a close association between cTnI and left ventricular hypertrophy8, which represents an intermediate stage prior to symptomatic HF9. Moreover, as experimental studies have found immune responses targeting cTnI to aggravate acute cardiac damage and to induce myocardial dysfunctions and HF10,11, cTnI may have a direct or indirect causal effect on HF development in humans. Circulating cardiac troponin concentrations are also influenced by cardiac troponin degradation and clearance12,13. A previous genome-wide association study (GWAS) in the general population from the Generation Scotland Scottish Family Health Study (GS:SFHS) reported five genetic loci associated with cTnI concentrations and four other loci associated with cTnT concentrations14, however without replication in an independent cohort. More knowledge about the genetic determinants of circulating cardiac troponins could reveal new pathobiology important for HF development. In contrast, as AMI type 1 is a result of an acute epicardial coronary artery occlusion and cTnI is only expressed in cardiac myocytes, no causal role for cTnI is expected for incident AMI. Due to limited knowledge of the genetic contribution to circulating cardiac troponins, no previous studies have used genetically informed methods of causal inference, namely Mendelian randomization (MR), to assess the causal effect of cardiac troponin concentrations on AMI or HF15.

High-sensitivity cTnI has been cross-sectionally and repeatedly measured in three surveys of the Trøndelag Health Study (HUNT) from 1995 to 2019. Additionally, recent efforts to sequence 53 831 diverse genomes in the NHLBI TOPMed Program16 have enabled the imputation of more than 26 million variants in HUNT and GS:SFHS combined, including less common loss-of-function variants that have not been previously investigated14.

In order to validate previous findings and discover novel genetic variants associated with circulating cTnI in the general population, we conducted a genome-wide meta-analysis of cTnI concentration in 48 115 individuals from HUNT and GS:SFHS. Additionally, we assessed changes in cTnI with time in 5 178 HUNT participants with repeated measures. Finally, we performed MR to investigate the causal effects of circulating cTnI on AMI and HF.

Results

Discovery of genetic loci associated with cTnI

After a discovery genome-wide association meta-analysis of HUNT (n = 29 839) and GS:SFHS (n = 18 276), we identified 12 genetic loci associated (p-value < 5×10−8) with cTnI concentrations (Table 1, Supplemental Tables S1, S2, Supplemental Figures F1, F2). All 12 loci were well imputed in both HUNT and GS:SFHS and had consistent directions and effect sizes across both studies. Eight of the 12 loci were previously unreported, while four were previously reported, but not yet replicated14. Additionally, we discovered 11 novel, rare variants (MAF < 0.001, p-value < 5×10−8) that were only imputed in HUNT and could not therefore be replicated (Supplemental Table S1). Excluding individuals with a known history of comorbidities did not materially change the results (Supplemental Table S4). No genetic loci were genome-wide significant for changes in cTnI concentrations with time, however three loci were suggestive (p-value < 10−6) (Supplemental Tables S3, S5).

View this table:
Table 1:

Index variants in genome-wide significant cTnI loci

We found three protein-altering index variants in cTnI loci that have previously been associated with cardiovascular health or with muscle development and injury: Firstly, a protein-altering variant in the fibrinolysis related gene HABP2 (rs7080536, MAF = 0.04, effect size = −0.10, p-value = 1×10−12) that reduces the activity of the translated protein17, coagulation factor VII activating protease, and has been associated with carotid stenosis18 and venous thromboembolism19. Secondly, a protein-altering variant in APOH (rs1801690, MAF = 0.06, effect size = 0.10, p-value = 1×10−14) that has been associated with the concentrations of creatine kinase20, which is an unspecific biomarker for acute muscle injury, and with activated partial thromboplastin time20, a measure of coagulation. Third, a previously reported protein-altering variant in ANO5 (rs7481951, MAF = 0.41, effect size = 0.05, p-value = 2×10−19) that has been suggested to be causal for adult-onset muscular dystrophy21. Furthermore, we identified a rare protein-altering variant (rs151313792, MAF = 0.004, effect size = 0.26, p-value = 1×10−8) in FHOD3, a gene that regulates actin assembly in cardiac sarcomeres22. This variant has been associated with dilated cardiomyopathy23,24 and left ventricular ejection fraction25.

Most of the remaining index variants are in close proximity to genes previously associated with cardiovascular disease (ALPK3, VCL, CAND2, KLKB1, CYP4V2, HLA-DQB1)4,23,26–29, cardiovascular development, structure and function (ALPK3, FHOD3, VCL, CAND2, TNFAIP2)27,30–33, unspecific muscle injury (CAND2, ANO5)20, or processes such as coagulation/fibrinolysis, blood-pressure regulation and vascular inflammation (F12, KLKB1, HLA-DQB1, LMAN1)29,34,35. The F12/GRK6 locus has also been associated with chronic kidney disease36. Two index variants were in perfect LD (r2 = 1.00) with protein-altering variants in KLKB1 and LMAN1, as described in the section on fine-mapping below.

Fine-mapping for causal variants in the cTnI loci

We identified 14 credible sets (Supplemental Table S9) of genetic variants in the cTnI loci using SuSiE37. For each credible set there was a 95% cumulative posterior probability that it contained at least one causal variant. The sizes of the credible sets ranged from 1 variant to 64 variants (in the HLA region and one region on chromosome 10), providing varying evidence for the importance of specific variants. Nine of the cTnI index variants (rs7650482, rs12331618, rs7080536, rs7481951, rs4283165, rs8024538, rs1801690, rs151313792 and rs12604424) were found in a credible set, and one of these credible sets only contained the index variant itself (rs7080536), suggesting that this was the most likely causal variant in the locus. The two loci with the lowest p-values (chromosomes 10 and 11) gave rise to two credible sets each, indicating multiple independent signals. However, the credible sets were in close proximity or overlapping and therefore the seemingly independent signals might have resulted from uncaptured LD structures.

Annotation of the variants in each credible set (Supplemental Table S10) revealed four additional protein-altering variants in moderate to high LD with non-exonic index variants (Table 1): rs11718898 in CAND2, correlation r2 in HUNT (r2HUNT) = 0.91 with index variant rs7650482 (Figure 1a); rs3733402 in KLKB1, r2HUNT = 1.00 with index variant rs12331618 (Figure 1b); rs1132339 in TNFAIP2, r2HUNT = 0.65 with index variant rs4283165 (Figure 1c); rs2298711 in LMAN1, r2HUNT = 1.00 with index variant rs12604424 (Figure 1d). Furthermore, we found a rare exonic frameshift substitution (rs137854521 in ANO5, r2HUNT = 0.003, D’ = 1.00 with the more common index variant rs7481951) in the second credible set in the same region as rs7481951, and a variant in another credible set was located in the F12 5’ untranslated region (UTR) and has been associated with coagulation factor XII deficiency (rs1801020, r2HUNT = 0.99 with rs2731672). The rs2298711 protein-altering variant was reported as index variant in the previous GWAS study in GS:SFHS14.

Figure 1:
Figure 1:

LocusZoom plots of cTnI loci with protein-altering variants in moderate or high LD with the index variant. Chromosome positions (GRCh38) are given on the x-axes, and the meta-analysis –log10(p-value) for the variants are given on the y-axis. The genome-wide significance p-value threshold (5 × 10−8) is indicated by the dotted line. The correlation r2 for each variant is indicated by colors relative to the index variants (LD Ref Var) in each panel: Panel a: The CAND2 locus on chromosome 3 with index variant rs7650482. A protein-altering variant in CAND2, rs11718898 (pVal77Ala), is in high LD with the index variant. Panel b: The KLKB1 locus on chromosome 4 with index variant rs12331618. A protein-altering variant in KLKB1, rs3733402 (p.Ser143Asn, p.Ser105Asn), is in perfect LD with the index variant. Panel c: The TNFAIP2 locus on chromosome 14 with index variant rs4283165. A protein-altering variant in TNFAIP2, rs1132339 (p.Gln282Glu), is in moderate LD with the index variant. Panel d: The LMAN1 locus on chromosome 18 with index variant rs12604424. A protein-altering variant in LMAN1, rs2298711 (p.Met410Leu), is in perfect LD with the index variant.

Prioritization of genes, pathways and tissues

To further explore the underlying biology of the cTnI associated loci, we used DEPICT38 to identify candidate causal genes at each locus. For suggestive (p-value < 1×10−5) cTnI loci (excluding the HLA region), we identified gene sets enriched for genes in cTnI loci, prioritized genes in each locus that shared predicted function with the genes in the remaining cTnI loci more often than expected by chance, and identified tissues and cell types in which cTnI locus genes were highly expressed (Supplemental Tables S6-S8). After removing results with false discovery rate above 5%, we found an enriched expression of suggestive cTnI genes in heart and heart ventricle tissues, and in tooth development. Among 14 461 reconstituted gene sets, one gene set (“abnormal placenta morphology”) was enriched with genes in significant cTnI loci. DEPICT prioritized the genes KLHL21, THAP3, CAND2, AOPEP, CTNNA3, SYNPO2L, VCL, ANXA11, MYOF, NRAP, CASP7, ANO5, ALPK3 and ZNF592 based on shared predicted functions with genes from the other associated loci. Most of these (CAND2, VCL, ANO5, ALPK3, NRAP, CASP7, SYNPO2L, ZNF592) were in significant (p-value < 1×10−8) cTnI loci, but not always near the index variant.

Heritability of cTnI concentration

We estimated the proportion of phenotypic variance explained by all genotyped autosomal variants using GCTA and found a narrow-sense SNP heritability of cTnI concentrations in HUNT (variance explained, Vg/Vp ± 1SE) to be 0.15 ± 0.01, p-value 2.63×10−32. The HUNT cohort has a high degree of relatedness and is therefore more genetically homogeneous than cohorts with fewer relatives. Including more rare variants from the TOPMed panel would further increase the homogeneity of the dataset. This might, in part, explain the lower heritability observed in HUNT than in previous studies39,40.

Variants associated with tissue-specific cis-eQTLs

Five of the cTnI index variants were significantly associated with the expression of one or more genes in at least one tissue type (Supplemental Table S11). Two of these variant-gene associations were found in the cardiovascular system: rs7481951 was associated with the expression of ANO5 in aortic and tibial arteries and rs12331618 was associated with the expression of ADK in tibial arteries.

To discover shared causal variants for cTnI concentration and expression levels of nearby genes in different tissues, we used the cis-eQTL data from GTEx v8 and performed a Bayesian colocalization analysis of the gene expression signals and cTnI concentration signals. Four cTnI loci were significantly colocalized (posterior probability of common causal variant > 75%) with the cis-eQTL signals for a gene in at least one tissue (Supplemental Table S12): The expression of coagulation factor F11 in tibial arteries (rs12331618), the expression of coagulation factor F12 in liver (rs2731672), the expression of BMS1P4 (pseudogene) in whole blood (rs2270552), the expression of ADK in the sigmoid colon (rs2270552), as well as the expression of lncRNA gene Lnc-WDR73-10 in the left heart ventricle, mammary breast tissue and brain frontal cortex (rs8024538). The association signal for ADK was also near the posterior probability cut-off of 75% in aortic tissue (rs2270552) (Supplemental Table S12).

PheWAS of cTnI loci and cTnI polygenic risk score

Four of the index variants associated with increased cTnI concentrations were significantly associated with diseases or continuous traits in UK Biobank (Figures 2 and 3, Supplemental Tables S13, S14) after correction for multiple testing: The variant rs7650482 was associated with cardiac arrhythmias, in particular atrial fibrillation and flutter, as well as waist hip ratio, standing height and basal metabolic rate. In addition, the variant rs12331618 was associated with pulmonary heart disease and phlebitis/thrombophlebitis, the variant rs2731672 was associated with standing height and basal metabolic rate, and the variant rs9391746 was mainly associated with autoimmune inflammatory conditions.

The polygenic risk score (PRS) based on genome-wide significant index variants (n = 12, variance explained = 1.2%) was associated with seven phecodes and continuous traits in UK Biobank after correction for multiple testing (Figures 2 and 3, Supplemental Tables S15, S16): cardiac arrhythmias in general, and the sub-phenotype atrial fibrillation and flutter in particular, as well as standing height, aspartate aminotransferase 1, whole body fat free mass, whole body water mass and basal metabolic rate.

Figure 2:
Figure 2:

PRS-PheWAS: Associations of PRS (increasing cTnI concentrations) with 1 688 phecodes in UK Biobank.

Figure 3:
Figure 3:

PRS-PheWAS: Associations of PRS (increasing cTnI concentrations) with 83 continuous traits and biomarkers in UK Biobank. ECG: Electrocardiogram, MRI: Magnetic resonance imaging.

Mendelian Randomization

To investigate the causal effect of circulating cTnI on both AMI and HF, we performed twosample MR using 11 of 12 index variants from our meta-analysis as an instrument for cTnI (F-statistic = 44, r2 = 0.012 in the independent sample [N = 3 424]), and summary-level data on AMI (5 948 AMI cases, 355 246 controls from UK Biobank) and HF (47 309 cases and 930 014 controls from the Heart Failure Molecular Epidemiology for Therapeutic Targets consortium [HERMES]).

There was no evidence of a causal effect of circulating cTnI on AMI (Inverse Variance Weighted (IVW) odds ratio (OR) = 1.00, 95% Confidence Interval (CI) [0.99, 1.00]). Bidirectional MR however suggested evidence for a potential causal effect of cTnI on HF (IVW OR = 1.20, 95% CI: [1.00, 1.45]) in addition to the well-known effect of HF on cTnI concentration (IVW effect size = 0.14, 95% CI: [0.00, 0.28]) (Supplemental Table S17). Sensitivity analyses showed similar results between all methods (weighted mode, weighted median, penalized weighted median, MR-Egger), although wider confidence intervals due to low power, and Steiger filtering did not change the results (Supplemental Table S17, Figures 4–5). The MR-Egger intercept was close to zero for both outcomes, giving no strong evidence of direct effects of the index variants on AMI or HF that are not mediated by cTnI (horizontal pleiotropy). However, there was some evidence for heterogeneity in the analysis of HF (heterogeneity statistic Q = 27, p-value = 0.003), but not for the analysis of AMI (Q = 9, pvalue = 0.50). The heterogeneity in the HF analysis could possibly reflect horizontal pleiotropy, which would violate the instrumental variable analysis assumptions, or indicate multiple underlying mechanisms.

Figure 4:
Figure 4:

Mendelian randomization analysis of the causal effect of cTnI on AMI based on summary statistics from the meta-analysis of HUNT and GS:SFHS (n = 48 115) for cTnI and from a GWAS of AMI in UK Biobank from the Neale laboratory (cases/controls = 5 9948/355 246). Effect allele-cTnI associations given on x-axis and effect allele-AMI associations given on y-axis. All associations are given with 1 standard deviation error bars.

Figure 5:
Figure 5:

Mendelian randomization analysis of the causal effect of cTnI on HF based on summary statistics from the meta-analysis of HUNT and GS:SFHS (n = 48 115) for cTnI and from a meta-analysis of HF in the HERMES consortium (cases/controls = 47 309/930 014). Effect allele-cTnI associations given on x-axis and effect allele-HF associations given on y-axis. All associations are given with 1 standard deviation error bars. The outlier variant with a large effect size for both traits is rs9391746 from the HLA region.

Discussion

Our GWAS meta-analysis of 48 115 individuals in HUNT and GS:SFHS reveals a genetic contribution to circulating protein concentrations of cTnI with associations to several biological processes. We identified 12 loci consistently associated with cTnI concentrations across both studies, thereby confirming four of five previously unreplicated loci14. Furthermore, we suggest several novel loci for change in cTnI with time, confirm cTnI as a non-causal biomarker for AMI and observe indications that cTnI may have a causal effect on HF development.

Cardiac troponin I and T are important biomarkers for cardiovascular disease, and especially used to diagnose AMI in patients with symptoms of acute myocardial ischemia1,41. Moreover, with the development of high-sensitivity assays, cTnI and cTnT have been found to provide strong and independent information regarding future risk for cardiac morbidity and mortality among subjects of the general population39. Of note, while cardiac troponins are excellent markers of AMI in the acute setting, high-sensitivity troponins seem more related to incident heart failure events than to future AMI in the general population42–45. Hence, data from several independent cohorts support a model of cardiac troponins as either directly or indirectly associated with heart failure development. This model is supported by highsensitivity troponin concentrations being correlated with indices of structural heart disease, and most prominently left ventricular mass43,45–47.

Using genetic data from two large cohorts from the general population (HUNT and GS:SFHS), we now provide new and additional information related to a role for cardiac troponins in cardiovascular disease. We identify several index variants in or near genes previously related to the cardiovascular system, either through disease associations or links to cardiovascular development and function. ALPK3, FHOD3, VCL and CAND2 has been implicated heart development and/or function27,30–32, and TNFAIP2 is central in blood vessel formation33. Both ALPK3 and VCL are associated with cardiomyopathy4,23,27, and CAND2 is associated with atrial fibrillation28. CAND2 and ANO5 are also associated with creatine kinase concentrations20, which is a biomarker for unspecific muscle injury. Furthermore, the KLKB1/CYP4V2 locus has been associated with atherosclerosis26, and the KLKB1 and F12 genes are both involved in several proteolytic reaction cascades in the cardiovascular system34. These cascades are associated with coagulation/fibrinolysis, blood pressure regulation and inflammation34. LMAN1 has also been associated with coagulation35, and both TNFAIP2 and HLA-DQB1 are associated with vascular inflammation29,33. Both coagulation and atherosclerosis are pathobiological components crucial to the progress of acute exacerbations of coronary artery disease, and blood pressure regulation will impact left ventricular mass and is therefore comparable with previous data on cardiac troponins and left ventricular structure and function. Several of the protein-altering and regulatory region variants that we identified had previously also been linked to some of the above-mentioned processes17–21,26,48,49. Excluding individuals with impaired kidney function or a history of diabetes, atrial fibrillation and flutter, cardiomyopathy or HF, did not attenuate the effect sizes of any of the 12 index variants, supporting that the genetic associations with cTnI concentrations were not driven by these clinical endpoints.

Fine-mapping of the cTnI associated loci pointed to protein-altering and regulatory region variants, and further indicated two separate causal signals in each of the two most significant cTnI loci. A rare ANO5 frameshift substitution (rs137854521) was possibly driving the second independent signal in the most significant locus as indicated by the fine-mapping analysis. The protein-altering variant in HABP2 was alone in its credible set, providing additional support for its contribution to circulating cTnI concentrations in the general population. On the other hand, a large credible set from the barely significant HLA region contained only intergenic variants, none of which were the index variant, weakening the evidence for a role of this specific variant in troponin biology. This is not surprising, given that variants in this region are known to be difficult to both impute and interpret50.

With a small number of significant cTnI loci, the DEPICT analysis was limited by the variant database. We filtered on variants in the DEPICT database so that all cTnI loci were represented, but with the cost that the results were not necessarily based on the most significant variants within the loci. Despite this limitation, we found that the expression of genes in cTnI loci with p-value < 10−5 was most significantly enriched in the heart (particularly in the ventricles), which is as expected given that cTnI is only expressed in cardiac myocytes, which is also the basis for the excellent performance of cardiac troponins as cardiovascular biomarkers51. Additionally, DEPICT prioritized several genes in the cTnI loci that were associated with heart development, structure and function, and with cardiovascular diseases: NRAP is involved in sarcomere assembly during cardiomyocyte development52 and is associated with familial cardiomyopathies53, while CASP7 is involved in activating apoptosis54 and is associated with cardiac electrophysiology31. Both are located in the HABP2 locus. Likewise, SYNPO2L, which is involved in sarcomeric signal transduction55 and cardiac development56, is located in the second most significant locus on chromosome 10, but further away from the index variant than VCL. The associations with sarcomere structure and function are plausible based on the biology as cardiac troponins are essential components of the sarcomere. Accordingly, our findings could indicate that our meta-analysis had only moderate statistical power to detect a wider range of the genetic associations with circulating cTnI concentrations.

While we identified protein-altering variants in more than half of the loci, several statistically significant variants were located outside of the coding regions. Non-coding variants might however have regulatory functions, and we therefore investigated their associations with variation in gene expression across 49 tissues. Four cTnI loci with non-coding index variants were highly colocalized with the expression signals for a nearby gene in at least one tissue, including the expression of coagulation factors in arteries and liver, and a protein degradation activator in blood. The colocalization with expression signals for adenosine kinase was also quite strong in the aorta. This protein has been implicated in cardiomyocyte microtubule dynamics and cardiac adaption to high blood pressure and is found to attenuate cardiac hypertrophy57. These findings corroborate the role of these loci in cTnI biology, possibly in relation to the response to cardiac injury and myocardial stress, and to the breakdown of circulating cTnI.

The single-variant phenome-wide association tests in the UK Biobank indicated some level of pleiotropy of cTnI variants: Seven of 12 index variants were associated with other traits than cTnI. These traits included cardiovascular diseases, waist-hip ratio, standing height, aspartate aminotransferase 1 concentrations and basal metabolic rate. Waist-hip ratio has been linked to future risk of cardiovascular outcomes58, standing height correlates with blood pressure, cardiac output and vascular resistance59, and the basal metabolic rate could potentially affect cTnI degradation. Aspartate aminotransferase 1 is an enzyme released from the heart in patients with AMI60, but is also related to amino acid metabolism in the liver and is currently mainly used as an unspecific biomarker for liver cell damage61. This enzyme could thus both be related to cardiovascular disease and possibly to degradation and clearance of circulating cTnI in the liver. However, there were no associations between cTnI variants and the more liver-specific enzyme alanine aminotransferase 1, which could indicate that the association is related to the cardiovascular system, rather than to liver damage. Hence, the mechanism whereby aspartate aminotransferase 1 influences circulating cTnI concentrations will require additional studies. The identified HLA region variant in our study was mainly associated with autoimmune and inflammatory diseases, and this is relevant for cardiovascular diseases62, but more studies are needed also here, especially since interpretation of specific HLA region variants is difficult.

The PRS combining the 12 cTnI index variants explained 1.2% of variation in cTnI concentrations in an independent sample from HUNT. In UK Biobank it was associated with some of the same traits as the single variants, which is not surprising: standing height, basal metabolic rate, aspartate aminotransferase 1 concentrations and cardiac arrhythmias, including atrial fibrillation. Furthermore, the PRS was associated with whole body fat-free mass and whole-body water mass which was not seen for any single variant. By performing two-sample MR, we confirmed a non-causal role of cTnI in AMI. In contrast, we found indications of a potential causal effect of cTnI in HF development, although the estimate was imprecise. This is surprising given the current opinion of cTnI as a passive biomarker of cardiac injury63, but in line with some animal studies suggesting that circulating cTnI can trigger immune responses causing HF10,11. We could expect a small degree of residual bias producing apparent bi-directional effects in the MR setting when there is a strong effect in one direction. However, the reverse relationship of HF causing cTnI was not particularly strong, and Steiger filtering did not attenuate the effect estimate, as we would expect if an effect of HF on cTnI was the sole causal direction. The MR-Egger intercept did not provide any strong evidence of horizontal pleiotropy and the effect estimate was consistent, but the test had low statistical power. We did however observe evidence for heterogeneity, which could indicate horizontal pleiotropy. Finally, the observation could result from biological antecedents of cTnI release, but this was not possible to evaluate with our data.

In conclusion, our study highlights diverse genetic determinants of cTnI concentration in the general population and confirms cTnI as a non-causal biomarker for AMI. Furthermore, it suggests several novel loci for change in cTnI over time and suggests a potential causal effect of cTnI on HF development, although future studies are required to validate these findings and to explore possible mechanisms.

Methods

Description of cohorts

HUNT

The HUNT Study is a longitudinal population-based health study conducted in the county of Trøndelag, Norway. Data and samples have been collected through four surveys (HUNT1 [1984-1986], HUNT2 [1995-1997], HUNT3 [2006-2008] and HUNT4 [2017-2019])64. Approximately 123 000 individuals (aged ≥ 20 years) have participated in one or more HUNT surveys. More than 70 000 of these participants have been genotyped using one of three Illumina HumanCoreExome arrays: 12 v.1.0, 12 v.1.1 and 24 with custom content (UM HUNT Biobank v1.0). Sample and variant quality control (QC) was performed using standard practices and has been reported in detail elsewhere65. All variants were imputed from the TOPMed reference panel (freeze 5)16 using Minimac4 v1.0 (https://genome.sph.umich.edu/wiki/Minimac4). The reference panel is composed of 53 831 multi-ethnic samples and 308 107 085 single nucleotide polymorphisms (SNPs) and indel variants at high depth (mean read depth 38.2X).

GS:SFHS

The recruitment and design of GS:SFHS has been previously been reported in detail66. In summary, potential participants (aged 35–65 years) were identified and invited at random from collaborating general medical practices in Scotland. Participants were asked to identify ≥ 1 first-degree relative aged ≥ 18 years who would also be able to participate. Participants were genotyped using either Illumina HumanOmniExpressExome-8v1_A or v1-2, and 20 032 passed QC criteria. Sample and variant QC has been previously been reported elsewhere67. All variants were imputed from the TOPMed reference panel (freeze 5b)16 using Minimac4 v1.0 (https://genome.sph.umich.edu/wiki/Minimac4).

Cardiac Troponin I (cTnI)

HUNT

Non-fasting serum samples were drawn in 1995-1997 (HUNT2), 2006-2008 (HUNT3) and 2017-2019 (HUNT4). The HUNT2 and HUNT3 samples were centrifuged at room temperature and stored at −80°C after collection. The HUNT2 samples were thawed and refrozen once in 2008 and subsequently stored at −20°C. High-sensitivity cTnI concentrations were analyzed in 2014 from HUNT2, in 2015 from HUNT3 and was analyzed consecutively in 2017-2019 from fresh serum samples in HUNT4. The analysis (Architect STAT, Abbott Laboratories, Abbott Diagnostics) was performed using the manufacturers’ protocol and quality assurance scheme.

We combined cTnI measurements from all studies into one variable for cTnI concentration: If an individual had measurements from multiple HUNT surveys, we prioritized the earliest measurement, assuming participants to be less exposed to potential confounding factors at a younger age. After prioritization, 29 839 individuals of European ancestry provided measurements of cTnI in HUNT, of which 23 990 were above or equal to the lower limit of detection (1.2 ng/L) reported for the assay68. Measurements below 1.2 ng/L are imprecise on an individual level but were included since they collectively could add information to the GWAS analysis.

We defined change in cTnI concentrations per time as the difference in cTnI concentrations (ng/L) from HUNT2 to HUNT3 divided by the follow-up time. In total, 5 178 participants of European ancestry had cTnI measurements in both studies and were included in the analysis. The median follow-up time was 10 years and 7 months.

GS:SFHS

Fasting blood samples were collected between 2006 and 2010. Serum samples were centrifuged, and serum aliquots stored at –80°C until future biochemical analyses. Highsensitivity cTnI concentration (Architect STAT, Abbott Laboratories, Abbott Diagnostics) was measured using the manufacturers’ protocol and quality assurance scheme.

A total of 18 276 individuals of European ancestry provided measurements of cTnI in GS:SFHS. Measurements below the reported limit of detection (1.2 ng/L)68 were reported as half of the limit of detection (0.6 ng/L) since the original measurements were unavailable.

Association analyses

We used a linear mixed model regression under an additive genetic model for each variant as implemented in BOLT-LMM v.2.3.469 to control for relatedness between the samples in all analyses.

We included age, sex, genotyping batch and the first 10 genetic principal components of ancestry (PCs) as covariates in both HUNT and GS:SFHS. Additionally, we included as a covariate in HUNT, the survey from which the serum sample was collected for cTnI measurement (HUNT2, HUNT3 or HUNT4). We applied rank-based inverse normal transformation for the cTnI concentration after adjusting for age and sex using linear regression (Additionally adjusting for survey in HUNT). Variants with a minor allele count (MAC) < 10 or imputation r2 < 0.3 were excluded, leaving 21 233 433 variants in the HUNT analysis and 21 677 569 variants in the GS:SFHS analysis. We performed genomic control correction after each analysis.

Meta-analysis

We performed a fixed-effect inverse-variance weighted meta-analysis using METAL70. A total of 27 194 041 variants imputed in up to 48 115 individuals were included in the analysis. We considered genetic loci reaching a p-value < 5×10−8 in the meta-analysis for follow-up analysis.

Sensitivity analyses

To investigate if the identified genetic loci could be explained by clinical endpoints associated with chronically elevated cardiac troponin levels, we performed a sensitivity analysis of cTnI concentration in HUNT. We excluded 5 253 individuals with a known history of atrial fibrillation (ICD9 427.3, ICD10 I48), diabetes mellitus (self-reported), HF (ICD9 428, ICD10 I50) or cardiomyopathy (ICD9 425, ICD10 I42), or with impaired kidney function (estimated glomerular Filtration rate [eGFR] < 60 ml/min/1.73m2) in the HUNT survey where the serum sample was collected for cTnI measurement.

Heritability estimation

We estimated the narrow-sense (additive) single nucleotide polymorphism (SNP) heritability of cTnI in HUNT using genome-wide complex trait analysis (GCTA) for family data71–73.

We created a genetic relationship matrix (GRM) based on all genotyped autosomal variants in the 29 839 HUNT participants with troponin measurements, followed by a second GRM that was identical to the first except that the off-diagonal elements that were < 0.05 were set to zero. We used both GRMs with GCTA-GREML (genomic-relatedness-based restricted maximum-likelihood) to estimate the variance in cTnI concentration explained by the variants. Age, sex, genotyping batch and HUNT survey were used as covariates in the analysis, and the cardiac troponin measurement was transformed to normality with rankbased inverse normal transformation after regression on age, sex and HUNT survey prior to analysis.

Variant annotation and fine-mapping

We annotated genetic variants using ANNOVAR74. For fine-mapping of the cTnI loci we applied SuSiE37, which uses iterative Bayesian stepwise selection to estimate the number of causal variants in each locus and create credible sets with a 95% cumulative posterior probability for containing a causal variant. The linkage disequilibrium (LD) matrix used with SuSiE was calculated from the imputed variants of 5000 unrelated individuals in HUNT, and variants only imputed in GS:SFHS were therefore excluded from the fine-mapping. We visualized the fine-mapped regions with LocusZoom75.

Tissue, gene and gene set enrichment tests

We performed gene set enrichment, gene prioritization and tissue/cell type enrichment tests on the cTnI loci using DEPICT38. To ensure that DEPICT could identify variants in all loci we filtered the meta-analysis results to the 3 906 791 variants overlapping with the DEPICT reference files. We used plink v1.976 to identify two sets of independent variants by clumping the filtered results into 500 kb regions of correlated variants (r2 < 0.05) meeting two different p-value cut-offs (p-value < 1×10−5 and p-value < 5×10−8). The two sets of independent variants were used as input for DEPICT and analyzed separately. All cTnI loci were included in the analyses except variants in the HLA region.

Expression quantitative trait locus tests

To identify cTnI index variants associated with the expression level of a gene in specific tissue types, we downloaded single tissue cis-eQTL data from the GTEx portal, data set v8 (https://www.gtexportal.org). The data contained variant-gene expression associations (excluding associations with false discovery rate > 0.05 within each gene) based on 714 samples of European ancestry across 49 tissues. We identified all significant associations between cTnI index variants and gene expression levels in any tissue types, using a significance threshold of p-value < 1×10−9 (5×10−8/49 tissues).

To further assess if any cis-eQTL signals and GWAS meta-analysis signals were consistent with a shared causal variant for cTnI concentration and gene expression, we used Bayesian colocalization analysis as implemented in the R package coloc77. For each tissue type, we analyzed all genes whose expression was associated (p-value < 5×10−8) with at least one cTnI associated variant (p-value < 5×10−8), using effect sizes and standard errors for each varianttrait association as input. The coloc software estimated the variance in each trait (cTnI concentration or gene expression level) from the sample sizes and minor allele frequencies. We set the prior probability of a genetic variant being associated with only cTnI concentrations, only gene expression or both traits to be 10−4, 10−4 and 10−6 respectively. We considered posterior probabilities (PP4) above 75% to give strong support for a common causal variant for the two traits in a given tissue, posterior probability between 50% and 75% to be inconclusive or to give weak support for a common causal variant, and posterior probabilities < 50% to indicate no support for a common causal variant between the two traits.

Polygenic risk score for cTnI

We used R v3.6.278 to construct a polygenic risk score (PRS) for cTnI, by summing the product of the effect size and the estimated allele count (dosage) for the 12 index variants in genome-wide significant loci (p-values < 5×10−8).

To test the association between the PRS and cTnI in an independent sample, we performed the GWAS on cTnI concentrations in 26 415 HUNT participants, excluding 3 424 individuals who were unrelated (no 3rd degree relatives or closer) both to each other and to the individuals in the GWAS. We used KING v2.1.579 to estimate the relatedness. The smaller GWAS in HUNT (N = 26 415) was meta-analyzed with the full GWAS from GS:SFHS to obtain effect estimates independent of the excluded individuals. We constructed the PRS from the effect size estimates from this meta-analysis and used a linear regression model to test the association between the PRS and serum concentrations of cTnI (inverse rank transformed to normality) in the independent sample of 3 424 unrelated HUNT participants.

Phenome-wide association tests (PheWAS)

We tested the association of the PRS (PRS-PheWAS), as well as the individual index variants (single variant PheWAS), with 1 688 phecodes and 83 continuous biomarkers and traits in participants of white British ancestry in UK Biobank80. A PRS was calculated for each participant as described in the above paragraph, using TOPMed imputed estimated allele counts and effect size from the meta-analysis. We used a logistic or linear regression model respectively to assess the association of the single variant estimated allele counts or inverse normalized PRS and each phecode or continuous biomarker. For the PRS-pheWAS we included as covariates sex and birth year for binary traits, and sex and age at measurement for continuous traits. For the single variant pheWAS we used GWAS summary statistics generated with SAIGE v.29.4.281, with sex and the first four PCs as covariates in addition to age at initial assessment for quantitative traits and birth year for binary traits. To correct for multiple testing of phecodes and continuous variables we used a Bonferroni corrected significance cut-off of 2.8×10−5.

Mendelian Randomization of cTnI on AMI and HF

To evaluate causal associations of circulating cTnI on AMI or HF, we performed inversevariance weighted (IVW) two-sample MR analyses using R v3.6.278 with the TwoSampleMR82 and MRInstruments83 libraries. We used the effect sizes of 11 index variants from the meta-analysis as genetic instrument for cTnI, while we found the respective associations and effect sizes of the instrument variants for AMI (ICD10 I21) and HF (ICD10 I50) from independent genome-wide summary level data (5 948 AMI cases vs 355 246 controls from UK Biobank80 available from the Neale laboratory http://www.nealelab.is/ukbiobank), and 47 309 HF cases vs 930 014 controls from the HERMES consortium56. The rare cTnI index variant rs151313792 was excluded from both analyses because it was not present in either outcome data sets, and we were unable to identify a proxy variant. In the analysis of HF, the variant rs8039472 was used as a proxy for rs8024538 (r2HUNT = 0.98) because rs8024538 was not present in the HF data. We performed sensitivity analyses using Steiger filtered IVW, weighted mode, weighted median, penalized weighted median and MREgger. To evaluate the reverse causal association of HF on cTnI concentration, we also performed an IVW two-sample MR analysis using the effect sizes of 11 index variants from the same HERMES summary statistics56 as a genetic instrument for HF, and the respective associations and effect sizes of the same variants from our GWAS meta-analysis of cTnI.

Ethics

Participation in HUNT is based on informed consent, and the study has been approved by the Norwegian Data Protection Authority and the Regional Committee for Medical and Health Research Ethics in Central Norway (REK Reference number: 2015/2294).

The GS:SFHS obtained written informed consent from all participants and received ethical approval from the National Health Service Tayside Committee on Medical Research Ethics (REC Reference number: 05/S1401/89).

Data Availability

Data generated or analyzed during this study are available from the corresponding authors upon reasonable request.

Data Availability

Data generated or analyzed during this study are available from the corresponding authors upon reasonable request.

Author contributions

M.R.M analyzed the data and wrote the first draft of the manuscript. H.Røsjø contributed to the writing of the manuscript with knowledge on cardiac biology and cardiac troponin measurements. A.R, M.N.L, S.E.G, A.F.H, B.N.W, S.A.G.T, J.LF, H.Rasheed, L.F.T, W.Z, L.G.F, J.B.N and C.H contributed to analyses. B.M.B. and T.O. conceived and designed the study. All authors interpreted results and revised the paper.

Competing Interests

None declared.

Acknowledgments

The Trøndelag Health Study (The HUNT Study) is a collaboration between HUNT Research Center (Faculty of Medicine and Health Sciences, NTNU, Norwegian University of Science and Technology), Trøndelag County Council, Central Norway Regional Health Authority, and the Norwegian Institute of Public Health. The genotyping in HUNT was financed by the National Institutes of Health; University of Michigan; the Research Council of Norway; the Liaison Committee for Education, Research and Innovation in Central Norway; and the Joint Research Committee between St Olav’s hospital and the Faculty of Medicine and Health Sciences, NTNU.

References

  1. 1.↵
    Thygesen, K. et al. Fourth Universal Definition of Myocardial Infarction (2018). Circulation vol. 138 (2018).
  2. 2.↵
    Farmakis, D., Mueller, C. & Apple, F. S. High-sensitivity cardiac troponin assays for cardiovascular risk stratification in the general population. Eur. Heart J. 1–8 (2020) doi:10.1093/eurheartj/ehaa083.
    OpenUrlCrossRef
  3. 3.↵
    Sheng, J. J. & Jin, J. P. TNNI1, TNNI2 and TNNI3: Evolution, regulation, and protein structure-function relationships. Gene 576, 385–394 (2016).
    OpenUrlCrossRefPubMed
  4. 4.↵
    Ingles, J. et al. Evaluating the Clinical Validity of Hypertrophic Cardiomyopathy Genes. Circ. Genomic Precis. Med. 12, 57–64 (2019).
    OpenUrl
  5. 5.↵
    Richard, P. et al. Hypertrophic cardiomyopathy: Distribution of disease genes, spectrum of mutations, and implications for a molecular diagnosis strategy. Circulation 107, 2227–2232 (2003).
    OpenUrlAbstract/FREE Full Text
  6. 6.↵
    McCarthy, C. P. et al. Myocardial Injury in the Era of High-Sensitivity Cardiac Troponin Assays. JAMA Cardiol. 4, 1034–1042 (2019).
    OpenUrl
  7. 7.↵
    Seferović, P. M. et al. Heart failure in cardiomyopathies: a position paper from the Heart Failure Association of the European Society of Cardiology. Eur. J. Heart Fail. 21, 553–576 (2019).
    OpenUrl
  8. 8.↵
    Lyngbakken, M. N. et al. Gender, High-Sensitivity Troponin I, and the Risk of Cardiovascular Events (from the Nord-Trøndelag Health Study). Am. J. Cardiol. 118, 816–821 (2016).
    OpenUrl
  9. 9.↵
    Lazzeroni, D., Rimoldi, O. & Camici, P. G. From left ventricular hypertrophy to dysfunction and failure. Circ. J. 80, 555–564 (2016).
    OpenUrlCrossRefPubMed
  10. 10.↵
    Göser, S. et al. Cardiac troponin I but not cardiac troponin T induces severe autoimmune inflammation in the myocardium. Circulation 114, 1693–1702 (2006).
    OpenUrlAbstract/FREE Full Text
  11. 11.↵
    Okazaki, T. et al. Autoantibodies against cardiac troponin I are responsible for dilated cardiomyopathy in PD-1-deficient mice. Nat. Med. 9, 1477–1483 (2003).
    OpenUrlCrossRefPubMedWeb of Science
  12. 12.↵
    Hammarsten, O., Mair, J., Möckel, M., Lindahl, B. & Jaffe, A. S. Possible mechanisms behind cardiac troponin elevations. Biomarkers 23, 725–734 (2018).
    OpenUrl
  13. 13.↵
    Shi, Q. et al. Degradation of cardiac troponin I in serum complicates comparisons of cardiac troponin I assays. Clin. Chem. 45, 1018–1025 (1999).
    OpenUrlAbstract/FREE Full Text
  14. 14.↵
    Welsh, P. et al. Cardiac Troponin T and Troponin I in the General Population. Circulation 139, 2754–2764 (2019).
    OpenUrl
  15. 15.↵
    Smith, G. D. & Ebrahim, S. ‘Mendelian randomization’: Can genetic epidemiology contribute to understanding environmental determinants of disease? Int. J. Epidemiol. 32, 1–22 (2003).
    OpenUrlCrossRefPubMedWeb of Science
  16. 16.↵
    Taliun, D. et al. Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. Biorxiv 1–46 (2019).
  17. 17.↵
    Olsson, M. et al. Genome-wide analysis of genetic determinants of circulating factor VII-activating protease (FSAP) activity. J. Thromb. Haemost. 16, 2024–2034 (2018).
    OpenUrl
  18. 18.↵
    Willeit, J. et al. Marburg I polymorphism of factor VII-activating protease: A prominent risk predictor of carotid stenosis. Circulation 107, 667–670 (2003).
    OpenUrlAbstract/FREE Full Text
  19. 19.↵
    Ahmad-Nejad, P. et al. The G534E-polymorphism of the gene encoding the Factor VIIactivating protease is a risk factor for venous thrombosis and recurrent events. Thromb. Res. 130, 441–444 (2012).
    OpenUrlCrossRefPubMed
  20. 20.↵
    Kanai, M. et al. Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases. Nat. Genet. 50, 390–400 (2018).
    OpenUrlCrossRefPubMed
  21. 21.↵
    Hicks, D. et al. Europe PMC Funders Group Europe PMC Funders Author Manuscripts A founder mutation in Anoctamin 5 is a major cause of limb girdle muscular dystrophy. 134, 171–182 (2014).
    OpenUrl
  22. 22.↵
    Matsuyama, S. et al. Interaction between cardiac myosin-binding protein C and formin Fhod3. Proc. Natl. Acad. Sci. U. S. A. 115, E4386–E4395 (2018).
    OpenUrlAbstract/FREE Full Text
  23. 23.↵
    Esslinger, U. et al. Exome-wide association study reveals novel susceptibility genes to sporadic dilated cardiomyopathy. PLoS One 12, (2017).
  24. 24.↵
    Wooten, E. C. et al. Formin homology 2 domain containing 3 variants associated with hypertrophic cardiomyopathy. Circ. Cardiovasc. Genet. 6, 10–18 (2013).
    OpenUrlAbstract/FREE Full Text
  25. 25.↵
    Pirruccello, J. P. et al. Analysis of cardiac magnetic resonance imaging in 36,000 individuals yields genetic insights into dilated cardiomyopathy. Nat. Commun. 11, 2254 (2020).
    OpenUrl
  26. 26.↵
    Inouye, M. et al. Novel Loci for Metabolic Networks and Multi-Tissue Expression Studies Reveal Genes for Atherosclerosis. PLoS Genet. 8, e1002907 (2012).
    OpenUrlCrossRefPubMed
  27. 27.↵
    Phelan, D. G. et al. ALPK3-deficient cardiomyocytes generated from patient-derived induced pluripotent stem cells and mutant human embryonic stem cells display abnormal calcium handling and establish that ALPK3 deficiency underlies familial cardiomyopathy. Eur. Heart J. 37, 2586–2590 (2016).
    OpenUrlCrossRefPubMed
  28. 28.↵
    Nielsen, J. B. et al. Biobank-driven genomic discovery yields new insight into atrial fibrillation biology. Nat. Genet. 50, 1234–1239 (2018).
    OpenUrlCrossRefPubMed
  29. 29.↵
    Xiong, Y. et al. Association between HLA-DQB1 alleles and susceptibility to coronary artery disease in Southern Han Chinese. Hum. Immunol. 78, 540–546 (2017).
    OpenUrl
  30. 30.↵
    Gaudet, P., Livstone, M. S., Lewis, S. E. & Thomas, P. D. Phylogenetic-based propagation of functional annotations within the Gene Ontology consortium. Brief. Bioinform. 12, 449–462 (2011).
    OpenUrlCrossRefPubMed
  31. 31.↵
    Norland, K. et al. Sequence variants with large effects on cardiac electrophysiology and disease. Nat. Commun. 10, 1–10 (2019).
    OpenUrlCrossRefPubMed
  32. 32.↵
    Shiraishi, S. et al. TBP-interacting Protein 120B (TIP120B)/Cullin-associated and Neddylation-dissociated 2 (CAND2) inhibits SCF-dependent ubiquitination of myogenin and accelerates myogenic differentiation. J. Biol. Chem. 282, 9017–9028 (2007).
    OpenUrlAbstract/FREE Full Text
  33. 33.↵
    Jia, L. et al. The roles of TNFAIP2 in cancers and infectious diseases. J. Cell. Mol. Med. 22, 5188–5195 (2018).
    OpenUrlCrossRef
  34. 34.↵
    Björkqvist, J., Jämsä, A. & Renné, T. Plasma kallikrein: The bradykinin-producing enzyme. Thromb. Haemost. 110, 399–407 (2013).
    OpenUrlCrossRefPubMed
  35. 35.↵
    Nichols, W. M. et al. Mutations in the ER-Golgi intermediate compartment protein ERGIC-53 cause combined deficiency of coagulation factors V and VIII. Cell 93, 61–70 (1998).
    OpenUrlCrossRefPubMedWeb of Science
  36. 36.↵
    Köttgen, A. et al. Multiple New Loci Associated with Kidney Function and ChronicKidney Disease: The CKDGen consortium. Nat. Genet. 42, 376–384 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  37. 37.↵
    Wang, G., Sarkar, A. K., Carbonetto, P. & Stephens, M. A simple new approach to variable selection in regression, with application to genetic fine-mapping. bioRxiv (2018) doi:10.1101/501114.
    OpenUrlAbstract/FREE Full Text
  38. 38.↵
    Pers, T. H. et al. Biological interpretation of genome-wide association studies using predicted gene functions. Nat. Commun. 6, 5890 (2015).
    OpenUrlCrossRefPubMed
  39. 39.↵
    Willeit, P. et al. High-Sensitivity Cardiac Troponin Concentration and Risk of First-Ever Cardiovascular Outcomes in 154,052 Participants. J. Am. Coll. Cardiol. 70, 558–568 (2017).
    OpenUrlFREE Full Text
  40. 40.↵
    Bossard, M. et al. Factors independently associated with cardiac troponin I levels in young and healthy adults from the general population. Clin. Res. Cardiol. 106, 96–104 (2017).
    OpenUrl
  41. 41.↵
    Lyngbakken, M. N., Myhre, P. L., Røsjø, H. & Omland, T. Novel biomarkers of cardiovascular disease: Applications in clinical practice. Crit. Rev. Clin. Lab. Sci. 56, 33–60 (2019).
    OpenUrl
  42. 42.↵
    Omland, T. et al. A sensitive cardiac troponin T assay in stable coronary artery disease. N. Engl. J. Med. 361, 2538–2547 (2009).
    OpenUrlCrossRefPubMedWeb of Science
  43. 43.↵
    Omland, T. et al. Prognostic value of cardiac troponin I measured with a highly sensitive assay in patients with stable coronary artery disease. J. Am. Coll. Cardiol. 61, 1240–1249 (2013).
    OpenUrlFREE Full Text
  44. 44.
    McEvoy, J. W. et al. Six-year change in high-sensitivity cardiac troponin T and risk of subsequent coronary heart disease, heart failure, and death. JAMA Cardiol. 1, 519–529 (2016).
    OpenUrl
  45. 45.↵
    de Lemos, J. A. et al. Association of Troponin T Detected With a Highly Sensitive Assay and Cardiac Structure and Mortality Risk in the General Population. JAMA 304, (2010).
  46. 46.
    Chin, C. W. L. et al. High-sensitivity troponin I concentrations are a marker of an advanced hypertrophic response and adverse outcomes in patients with aortic stenosis. Eur. Heart J. 35, 2312–2321 (2014).
    OpenUrlCrossRefPubMedWeb of Science
  47. 47.↵
    Lyngbakken, M. N. et al. Cardiac Troponin I and T Are Associated with Left Ventricular Function and Structure: Data from the Akershus Cardiac Examination 1950 Study. Clin. Chem. 66, 567–578 (2020).
    OpenUrl
  48. 48.↵
    Houlihan, L. M. et al. Common Variants of Large Effect in F12, KNG1, and HRG Are Associated with Activated Partial Thromboplastin Time. Am. J. Hum. Genet. 86, 626–631 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  49. 49.↵
    Verweij, N. et al. Genome-wide association study on plasma levels of midregional-proadrenomedullin and C-terminal-pro-endothelin-1. Hypertension 61, 602–608 (2013).
    OpenUrlCrossRef
  50. 50.↵
    Williams, T. M. Human leukocyte antigen gene polymorphism and the histocompatibility laboratory. J. Mol. Diagnostics 3, 98–104 (2001).
    OpenUrl
  51. 51.↵
    Omland, T. New features of troponin testing in different clinical settings. J. Intern. Med. 268, 207–217 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  52. 52.↵
    Manisastry, S. M., Zaal, K. J. M. & Horowits, R. Myofibril assembly visualized by imaging N-RAP, alpha-actinin, and actin in living cardiomyocytes. Exp. Cell Res. 315, 2126–2139 (2009).
    OpenUrlCrossRefPubMed
  53. 53.↵
    Gehmlich, K., Geier, C., Osterziel, K. J., Van Der Ven, P. F. M. & Fürst, D. O. Decreased interactions of mutant muscle LIM protein (MLP) with N-RAP and α-actinin and their implication for hypertrophic cardiomyopathy. Cell Tissue Res. 317, 129–136 (2004).
    OpenUrlCrossRefPubMedWeb of Science
  54. 54.↵
    Brentnall, M., Rodriguez-Menocal, L., De Guevara, R. L., Cepero, E. & Boise, L. H. Caspase-9, caspase-3 and caspase-7 have distinct roles during intrinsic apoptosis. BMC Cell Biol. 14, (2013).
  55. 55.↵
    van Eldik, W. & Passier, R. Signalling in sarcomeres in development and disease. Netherlands Hear. J. 21, 367–371 (2013).
    OpenUrl
  56. 56.↵
    Shah, S. et al. Genome-wide association and Mendelian randomisation analysis provide insights into the pathogenesis of heart failure. Nat. Commun. 11, 1–12 (2020).
    OpenUrlCrossRefPubMed
  57. 57.↵
    Fassett, J. et al. Adenosine kinase attenuates cardiomyocyte microtubule stabilization and protects against pressure overload-induced hypertrophy and LV dysfunction. J. Mol. Cell. Cardiol. 130, 49–58 (2019).
    OpenUrl
  58. 58.↵
    Huxley, R., Mendis, S., Zheleznyakov, E., Reddy, S. & Chan, J. Body mass index, waist circumference and waist:hip ratio as predictors of cardiovascular riska review of the literature. Eur. J. Clin. Nutr. 64, 16–22 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  59. 59.↵
    Evans, J. M. et al. Body size predicts cardiac and vascular resistance effects on men’s and women’s blood pressure. Front. Physiol. 8, 1–12 (2017).
    OpenUrlCrossRefPubMed
  60. 60.↵
    Panteghini, M. Aspartate aminotransferase isoenzymes. Clin. Biochem. 23, 311–319 (1990).
    OpenUrlCrossRefPubMedWeb of Science
  61. 61.↵
    Pratt, D. S. & Kaplan, M. M. Evaluation of Abnormal Liver-Enzyme Results In Asumptomatic Patients. N. Engl. J. Med. 342, 1266–1271 (2000).
    OpenUrlCrossRefPubMedWeb of Science
  62. 62.↵
    Black, P. H. & Garbutt, L. D. Stress, inflammation and cardiovascular disease. J. Psychosom. Res. 52, 1–23 (2002).
    OpenUrlCrossRefPubMedWeb of Science
  63. 63.↵
    Hoff, J., Wehner, W. & Nambi, V. Troponin in Cardiovascular Disease Prevention: Updates and Future Direction. Curr. Atheroscler. Rep. 18, 12 (2016).
    OpenUrl
  64. 64.↵
    Krokstad, S. et al. Cohort profile: The HUNT study, Norway. Int. J. Epidemiol. 42, 968–977 (2013).
    OpenUrlCrossRefPubMedWeb of Science
  65. 65.↵
    Ferreira, M. A. et al. Shared genetic origin of asthma, hay fever and eczema elucidates allergic disease biology. Nat. Genet. 49, 1752–1757 (2017).
    OpenUrlCrossRefPubMed
  66. 66.↵
    Smith, B. H. et al. Cohort profile: Generation scotland: Scottish family health study (GS: SFHS). The study, its participants and their potential for genetic research on health and illness. Int. J. Epidemiol. 42, 689–700 (2013).
    OpenUrlCrossRefPubMedWeb of Science
  67. 67.↵
    Nagy, R. et al. Exploration of haplotype research consortium imputation for genomewide association studies in 20,032 Generation Scotland participants. Genome Med. 9, 1 (2017).
    OpenUrlCrossRef
  68. 68.↵
    Apple, F. S. & Collinson, P. O. Analytical characteristics of high-sensitivity cardiac troponin assays. Clin. Chem. 58, 54–61 (2012).
    OpenUrlAbstract/FREE Full Text
  69. 69.↵
    Loh, P. R. et al. Efficient Bayesian mixed-model analysis increases association power in large cohorts. Nat. Genet. 47, 284–290 (2015).
    OpenUrlCrossRefPubMed
  70. 70.↵
    Willer, C. J., Li, Y. & Abecasis, G. R. METAL: Fast and efficient meta-analysis of genomewide association scans. Bioinformatics 26, 2190–2191 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  71. 71.↵
    Yang, J., Lee, S. H., Goddard, M. E. & Visscher, P. M. GCTA: A tool for genome-wide complex trait analysis. Am. J. Hum. Genet. 88, 76–82 (2011).
    OpenUrlCrossRefPubMed
  72. 72.
    Yang, J. et al. Common SNPs explain a large proportion of heritability for human height. Nat. Genet. 42, 565–569 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  73. 73.↵
    Zaitlen, N. et al. Using Extended Genealogy to Estimate Components of Heritability for 23 Quantitative and Dichotomous Traits. PLoS Genet. 9, (2013).
  74. 74.↵
    Wang, K., Li, M. & Hakonarson, H. ANNOVAR: Functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38, 1–7 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  75. 75.↵
    Pruim, R. J. et al. LocusZoom: Regional visualization of genome-wide association scan results. Bioinformatics 27, 2336–2337 (2011).
    OpenUrlCrossRef
  76. 76.↵
    Chang, C. C. et al. Second-generation PLINK: Rising to the challenge of larger and richer datasets. Gigascience 4, 1–16 (2015).
    OpenUrlCrossRefPubMed
  77. 77.↵
    Giambartolomei, C. et al. Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics. PLoS Genet. 10, (2014).
  78. 78.↵
    R Core Team. R: A Language and Environment for Statistical Computing. (2019).
  79. 79.↵
    Manichaikul, A. et al. Robust relationship inference in genome-wide association studies. Bioinformatics 26, 2867–2873 (2010).
    OpenUrlCrossRefPubMedWeb of Science
  80. 80.↵
    Sudlow, C. et al. UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age. PLoS Med. 12, 1–10 (2015).
    OpenUrlCrossRefPubMed
  81. 81.↵
    Zhou, W. et al. Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat. Genet. 50, 1335–1341 (2018).
    OpenUrlCrossRefPubMed
  82. 82.↵
    Hemani, G., Haycock, P., Zheng, J., Gaunt, T. & Elsworth, B. TwoSampleMR: Two Sample MR functions and interface to MR Base database. (2020).
  83. 83.↵
    Hemani, G. MRInstruments: Data sources for genetic instruments to be used in MR. (2020).
Back to top
Posted September 05, 2020.
Download PDF

Supplementary Material

Data/Code
Email
Using human genetics to understand the causes and consequences of circulating cardiac troponin I in the general population
Marta R Moksnes, Helge Røsjø, Anne Richmond, Magnus N Lyngbakken, Sarah E Graham, Ailin Falkmo Hansen, Brooke N Wolford, Sarah A Gagliano Taliun, Jonathon LeFaive, Humaira Rasheed, Laurent F Thomas, Wei Zhou, Ida Surakka, Nicholas J Douville, Archie Campbell, David J Porteous, Paul Welsh, Naveed Sattar, George Davey Smith, Lars G Fritsche, Jonas B Nielsen, Bjørn Olav Åsvold, Kristian Hveem, Caroline Hayward, Cristen J Willer, Ben M Brumpton, Torbjørn Omland
medRxiv 2020.09.04.20187401; doi: https://doi.org/10.1101/2020.09.04.20187401
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area