https://orcid.org/0000-0002-0660-4943
Abstract
Cardiac troponin I and T measured by high-sensitivity assays (hs-cTn) are detectable in most ambulatory adults. Hs-cTnI and T may perform differently for prediction of cardiovascular (CV) events. In adults with myopathies and advanced age, hs-cTnT can lose CV specificity. We undertook a meta-analysis of ambulatory studies measuring hs-cTn to determine whether hs-cTnI is a superior predictor of CV outcomes.
Articles evaluating hs-cTnI or T for incident heart failure (HF), myocardial infarction (MI), CV, and all-cause death in ambulatory adults were screened. Adjusted HRs were extracted standardized as hs-cTn tertile 3 vs 1. Pooled effects were calculated, and heterogeneity assessed. Predefined subgroup analyses for hs-cTnI vs T included age and prevalent CV disease.
A total of 5499 studies were screened and 54 met inclusion criteria with up to 277 498 participants followed for 8.1 ± 4.5 years. Pooled estimates for HF, MI, CV, and all-cause death were reported as 2.30 [95% confidence interval (CI) 2.04, 2.60], 1.64 (95%CI 1.47, 1.86), 2.07 (95%CI 1.82, 2.35), and 1.66 (95%CI 1.51, 1.83). No differences in HRs for subgroups based on hs-cTnI vs hs-cTnT (except all-cause death), age, or prevalent CV disease were observed. Heterogeneity between studies was high with an I2 > 60% for all endpoints for both hs-cTn assays. In a sensitivity analysis of studies measuring both hs-cTnI and T there was also no differences for the prediction of any CV endpoint.
Hs-cTnI and T prediction of multiple CV endpoints are not significantly different irrespective of age and pre-existing CV disease in ambulatory adults.
Introduction
Assays for cardiac troponin I (cTnI) and T (cTnT) were introduced in the 1990s as a specific and sensitive methodology for diagnosing myocardial infarction (MI) and have become the cornerstone of the Universal Definition of MI (1). The sensitivity of the assays evolved in the late 2000s to what are now termed high-sensitivity (hs) assays with an increase in low-end analytical precision that results in 10–100-fold increase in sensitivity compared to prior generation “sensitive” or “conventional” assays while maintaining excellent specificity (2, 3). By definition, for an assay to be designated as an hs assay, the percent coefficient of variation at the 99th percentile upper reference limit should be ≤10% and measurable concentrations should be attainable at a concentration at or above the assay’s limit of detection for ≥50% of both healthy men and women (4). Progressively higher levels of either cTnI or T identify ambulatory individuals without known cardiovascular disease (CVD) are at increased risk of incident CVD (5–10). These findings led to the incorporation of hs-cTn as part of guidelines to define subclinical disease (11, 12). However, the interchangeability of hs-cTnI and T results for the prediction of future CVD events in ambulatory populations is controversial (13–15). For example, the PEACE study found both cTnI and T predict death and incident heart failure (HF), but only cTnI predicts incident MI (13, 16). In contrast, the HOPE study found both cTnI and T could predict incident MI (17, 18). To address this equipoise, we planned an up-to-date systematic review and meta-analysis to specifically test the hypothesis that cTnI and T can differentially predict specific CVD outcomes in studies of ambulatory adult cohorts with and without prevalent CVD.
Materials and Methods
The systematic review protocol was registered with the International Prospective Register of Systematic Reviews on February 13, 2023 (Registration ID CRD42023387204). This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols) 2015 checklist (19). The analysis of participant level de-identified hs-cTnI and T data from visit 4 with clinical outcomes from the Atherosclerosis Risk in Communities (ARIC) study was approved by the ARIC paper proposal committee.
Eligibility Criteria and Search Strategy
A search was conducted in the following databases: PubMed, Web of Science, and Embase throughout April 2024. The full search strategy can be found in Supplemental Table 1. Inclusion was limited to articles published in the English language. We included cohort studies and secondary data analysis of randomized controlled trials that met the following criteria: (a) the study included enrollment of adult participants from a community-based setting; (b) the primary exposure of interest was either hs-cTnI, T or both; and (c) outcomes of interest included at least one of the following incident outcomes: MI, HF, CV death, all-cause death, or major cardiovascular events (MACE) that could include some or all prior CVD events and/or stroke. Variability of the definition of MACE in observational studies has been previously noted (20). We excluded studies with COVID-19 patients, chronic kidney disease patients on dialysis, and hospitalized or emergency department-based patients.
Screening and Data Extraction
Eligible studies were imported into the Covidence platform, a web-based tool to manage and screen all abstracts and full-text articles (21). Duplicates were automatically removed by Covidence based on the title, year, volume, and author. All titles and abstracts were screened by a content expert (C.d.F.) and one additional independent author. Any discrepancies were resolved by a third author. Studies that met the initial screening criteria were then moved to full-text review and reviewed by a third author with discrepancies resolved through a consensus discussion among authors (excluding B.C. and C.B.). Studies investigating overlapping subsets of the same cohorts with the same assay were excluded. Data extraction into an Excel spreadsheet was independently conducted by 3 authors. The third author reviewed the independent extraction results of the other 2 authors. Discrepancies were resolved by consensus among all authors.
The following was extracted from each study: year of publication, sample size, gender distribution, mean (or median) age, cohort baseline characteristics, year of hs-cTn collection, cardiac troponin isoforms (I vs T), troponin assay manufacturer, reported outcomes and number of events, the corresponding hazard ratios (HR) and 95% CI, median follow-up time in years, the level of statistical model adjustment, and the variables that were included in the reported regression models. We extracted the HRs from each study from the most adjusted regression model which typically included demographics and traditional CV risk factors, as well as reported estimated glomerular filtration rate when available (n = 18) and, when measured, circulating biomarker concentrations including C-reactive protein (n = 20) and natriuretic peptides (n = 20) as covariates.
Quality Assessment
To assess the quality of the included cohort studies, 2 authors independently evaluated each study using the Newcastle-Ottawa Scale (22). Outcomes were evaluated concerning the adequacy of cohort follow-up, the assessment of the outcome, and the adequacy of the follow-up period for the outcome of interest. Scores could range from 0 to 9 with a score of 7–9 considered high quality.
Data Synthesis and Statistical Analysis
To facilitate interpretation, we standardized hazard ratios (HRs) across all studies, given that HRs were reported in varying formats either on a continuous log scale—either per standard deviation (SD) change or per unit on the log scale—or on a categorical scale, often in 3 or 4 groups of equal size, although some used 2 or 5 groups of unequal sizes. Following the method proposed by Hemingway et al., we categorized the biomarker distribution into tertiles and reported the HR of the highest tertile relative to the lowest tertile (23). Hemingway et al. suggest using tertiles for biomarker distribution to avoid a low number of events. Studies included in the review but not reporting details of the distribution required to calculate scaling factors were not included in the meta-analysis. For details regarding converting heterogeneous HR to tertiles see the Supplemental Material section “Data Synthesis, Harmonizing Hazard Ratios.”
A meta-analysis of the converted HRs was performed using random effects. To infer the statistical heterogeneity among the eligible studies, the I2 statistic and visual checks of forest plots were used. Publication bias across the selected studies was evaluated using both a funnel plot and the Egger test. For each endpoint of the analysis, the estimated HR are divided by cTnI and T with a final aggregated result. Studies with participants with both cTnI and cTnT measured contributed to the overall HR for each assay. For the ARIC study, a major contributor to the general population literature with respect to hs-cTn risk-prediction, the format of the published articles, did not allow for extraction and conversion to tertiles for both hs-cTnI and T. We requested access to the primary data with measures of both cTn assays at visit 4 (1996–1998) and directly conducted the tertile analysis for all outcomes with both hs-cTn assays.
Predefined subgroup analyses were performed for all endpoints including the presence or absence of prevalent CVD, and age. Age was evaluated as a categorical variable, divided into 2 categories of ≥70 years and <70 years, and with meta-regression as a continuous variable. In addition, we performed 2 sensitivity analyses not originally proposed in our analysis plan. First, we analyzed each outcome inclusive of only studies that measured both hs-cTnI and T on the same participants at the same time point (n = 3 to 7 studies based on outcome). Second, we conducted a meta-regression based on the mean year of the blood collection potentially reflecting differences in standard-of-care for primary and secondary prevention across studies. The level of statistical significance was a two-sided P < 0.05. All statistical analyses were performed using Stata v.17.0 BE (StataCorp). The Do files for the STATA code used for this meta-analysis are provided in the Supplemental Material.
Results
After screening, 80 studies were included in the systematic review, out of which 54 studies were included in the meta-analysis. The exclusion of 26 studies was predominantly due to inadequate results to pool either continuous or categorical biomarker data with the outcomes of interest (Fig. 1). There were 206 489, 262 124, 130 329, 219 602, and 277 498 participants included in the analysis for the outcomes of incident HF, MI, CV death, all-cause death, and MACE, respectively. The mean age of the pooled studies is 61.8 ± 8.2 years. The mean follow-up was 8.1 ± 4.5 years. Of the 54 studies, 23 measured only hs-cTnT (all Roche Diagnostics’ assay), 23 measured only hs-cTnI [17(74%) Abbott Diagnostics’ assay], and 8 measured both hs-cTnI and T (all Abbott and Roche, respectively). Baseline characteristics of the 54 studies included in the analysis are summarized in Supplemental Table 2. The quality of the published studies using the Newcastle-Ottawa Scale was high with 6 studies scored as 9, 42 studies scored as 8, and 6 studies scored as 7.
Consort diagram of studies included in the systematic review and meta-analysis. Color figure available at https://academic.oup.com/clinchem.
Association of Cardiac Troponin Concentrations With Heart Failure
The overall pooled estimate of adjusted HR for incident HF was 2.30 [95% confidence interval (CI) 2.04, 2.60] when comparing the participants in the top tertile with the bottom tertile of hs-cTn concentration (Fig. 2A). Overall, significant heterogeneity was observed between studies with an I2 of 81.2%. Subgroup analysis of studies analyzing hs-cTnI vs hs-cTnT did not show a difference in predicting incident HF with an adjusted HR of 2.28 (95%CI 1.85, 2.81) and 2.31 (95%CI 2.00, 2.68) for hs-cTnI and hs-cTnT, respectively. Publication bias was observed on inspection of the funnel plot where multiple studies showed a HR greater than the pooled estimate and Egger’s test showed significant publication bias (P = 0.001) (Supplemental Fig. 1A and B).
(A), Pooled estimates for baseline troponin I and T for outcomes HF; (B), Pooled estimates for baseline troponin I and T for outcome MI; (C), Pooled estimates for baseline troponin I and T for the outcome of cardiovascular death; (D), Pooled estimates for baseline troponin I and T for the outcome all-cause death; (E), Pooled estimates for baseline troponin I and T for the outcome MACE. Population characteristics (0, No CVD; 2, prevalent stable coronary artery disease; 4, mixed prevalent and no known coronary artery disease); LOA, level of adjustment (0, no adjustment, 2, demographics + risk factors + estimated glomerular filtration rate; 3, 2 + biomarkers (natriuretic peptides and/or C-reactive protein). YOP, year of publication. See references in Supplemental Data.
Association of Cardiac Troponin Concentrations With Myocardial Infarction
The overall pooled estimate of the HR for MI was 1.64 (95%CI 1.47, 1.86) comparing patients in the highest tertile to the lowest tertile of hs-cTn concentration (Fig. 2B). There was notable heterogeneity overall, with an I² of 90.7%. Subgroup analysis revealed no difference between hs-cTnI and hs-cTnT measured concentrations for predicting incident MI, with HRs of 1.62 (95%CI 1.41, 1.86) for hs-cTnI and 1.66 (95% CI 1.37, 2.01) for hs-cTnT. Publication bias was evident upon inspection of the funnel plot, where several studies reported HRs higher than the pooled estimate, and Egger’s test indicated significant publication bias (P < 0.0001) (Supplemental Fig. 2A and B).
Association of Cardiac Troponin Concentrations With Cardiovascular Death
The pooled estimate of the adjusted HR for CV death in the highest tertile of hs-cTn concentrations compared to the lowest tertile was 2.11 (95%CI 1.86, 2.40) (Fig. 2C). The overall heterogeneity was moderate with an I2 of 63.3%. Subgroup analysis showed no significant heterogeneity in the prediction of CV death between hs-cTnI and hs-cTnT with a HR of 2.03 (95%CI 1.70, 2.41) for hs-cTnI vs 2.19 (95%CI 1.82, 2.65) for hs-cTnT. There was no significant evidence of publication bias for the endpoint of CV death. Visual inspection of the funnel plot did not show an asymmetric distribution, which was further confirmed by a nonsignificant Egger’s test with a P value of 0.517 (Supplemental Fig. 3A and B).
Association of Cardiac Troponin Concentrations With All-cause Death
When comparing all-cause death in the highest tertile of hs-cTn concentrations to the lowest tertile, the overall pooled estimate of the adjusted HR was 1.63 (1.50, 1.78) (Fig. 2D). The analysis exhibited high heterogeneity with an I² of 84.3%. Subgroup analyses based on hs-cTn assay indicated a difference in the prediction for all-cause death between hs-cTnI and hs-cTnT, with a HR of 1.46 (95%CI 1.36, 1.57) for hs-cTnI vs 1.93 (95%CI 1.59, 2.34) for hs-cTnT. Visual inspection of the funnel plot suggested the presence of publication bias, which was further substantiated by Egger’s test with P = 0.010 (Supplemental Fig. 4A and B).
Association of Cardiac Troponin Concentrations With Major Adverse Cardiovascular Events
The overall pooled adjusted HR for MACE in the highest vs the lowest tertile of hs-cTn concentrations was 1.69 (95%CI 1.57, 1.82) (Fig. 2E). There was significant overall heterogeneity with an I2 of 84.4%. There was no difference observed in the composite MACE prediction for subgroups based on the hs-cTn assay measured. The pooled HR for hs-cTnI was 1.60 (1.49, 1.73) compared to the pooled HR of 1.79 (1.57, 2.05) for hs-cTnT. Publication bias was evident from visual inspection of the funnel plot and further confirmed by Egger’s test with P < 0.0001 (Supplemental Fig. 5A and B).
Subgroup Analysis and Meta-regression
We first examined the potential influence of prevalent CVD. For all 5 outcomes of interest, there was no significant heterogeneity seen with the adjusted HR for the highest vs lowest tertile of hs-cTn concentrations based on the presence or absence of prevalent CVD (Supplemental Figs. 6A, 7A, 8A, 9A, 10A). Furthermore, when we evaluated age as a dichotomous variable (mean study participant age ≥ 70 years vs <70 years) there was no heterogeneity with respect to the adjusted HR of the highest vs lowest tertile for any of the 5 outcomes (Supplemental Figs. 6B, 7B, 8B, 9B, 10B).
Using meta-regression to examine the effect of mean age as a continuous variable with each hs-cTn assay on the pooled estimates for each endpoint we observed that the hs-cTn assay affected the magnitude of association between hs-cTn concentration and all-cause death with trends in the opposite direction for I and T such that advancing age was associated with a lower HR for hs-cTnT and death with a test of interaction between age and hs-cTn assay of P = 0.050. (Fig. 3). In contrast, using meta-regression we found no significant interaction of the mean age of the cohorts with hs-cTnI or T concentrations on the adjusted HR for the highest to lowest tertile for incident HF, MI, CV death, and MACE.
(A), Meta-regression bubble plot of studies analyzing the endpoint of all-cause death using cardiac troponin I for random-effects meta-regression using mean age as covariate for all-cause death; (B), Meta-regression bubble plots of studies analyzing the endpoint of all-cause death using cardiac troponin T for random-effects meta-regression using mean age as covariate for all-cause death. The statistical interaction for cardiac troponin T and all-cause death is P = 0.050.
Sensitivity Analyses of Cohorts that Measured Both hs-cTnI and T
A sensitivity analysis was performed including only the studies measuring concentrations of both hs-cTnT and hs-cTnI at the same timepoint. The number of studies for analysis varied by endpoint with 3 studies for incident HF (n = 31 141 total participants), 5 studies for incident MI (n = 34 779 total participants), 6 studies for CV death (n = 35 284 total participants), 6 studies for all-cause death (n = 43 245 total participants), and 4 studies for MACE (n = 17 310 total participants). There was no significant difference observed in the HR of subgroup hs-cTnT vs hs-cTnI for incident HF, MI, CV death, all-cause death, and MACE (Supplemental Fig. 11A–E).
Meta-Regression Sensitivity Analysis of Outcomes based on Year of Blood Collection
A meta-regression was performed to assess the impact of mean year of blood collection as a continuous variable and hs-cTn assay on the pooled estimates for each endpoint. There was no interaction of mean year of blood draw and hs-cTnI vs T on the adjusted HR for highest to lowest tertiles for all endpoints.
Discussion
Our meta-analysis comparing adjusted HR for 4 clinical CV outcomes and all-cause death in ambulatory adults and an analysis of primary data within the ARIC study evaluated over a quarter of a million unique participants from 54 studies identified several important findings. First, contrary to our hypothesis, we did not find a difference between hs-cTnI and T concentration HRs for the 4 incident CV outcomes. This finding was supported in our sensitivity analysis inclusive of studies that had both hs-cTnI and T measured at the same timepoint. Second, we did find a difference in the HR for all-cause death with hs-cTnT being a stronger predictor than hs-cTnI. Third, heterogeneity between studies for most outcomes was high (typically defined as an I2 > 75%) (24). To identify sources of heterogeneity we performed subgroup analyses based on age, prevalent CVD, and year of blood collection as these factors are potentially associated with differences in circulating cTnI and T concentrations. These clinical features could not explain the heterogeneity between studies for CV endpoints. Fourth, publication bias was found for multiple endpoints.
Differences between cTnI and T measured by hs assays appear to have little clinical implications for the accuracy of the diagnosis of MI with a high correlation between concentrations, and guideline recommendations without a preference for the use of either cTnI measurable by several hs assays or hs-cTnT (25–27). However, hs-cTn concentrations, typically lower than those seen with acute MI presentations, are incorporated into HF guidelines to define stage B chronic HF, at increased risk for incident symptomatic HF and CV death (11). Recent guidelines for people with diabetes also recommend that hs-cTn can be used for annual screening and elevated concentrations can trigger further evaluation including echocardiography (12). Therefore, differences between the 2 assays, particularly in subsets of patients such as older adults or those with prevalent CVD, could guide the choice of hs-cTn assay to use in this setting. A case for a difference between cTnI and T in the ambulatory setting is supported by the finding of preferentially elevated cTnT concentrations in patients with a variety of different types of myositis that is supported by their demonstration of cTnT mRNA expression in the skeletal muscle (28). The identification of cTnT mRNA expression in the skeletal muscle of older adults without overt myositis or sarcopenia raised the possibility that cTnT could have a lower specificity for incident CVD events compared to cTnI as people age (29). Recent findings also identify cTnI but not T can predict asymptomatic left ventricular remodeling (30). Findings from a large cohort study of ambulatory participants without prevalent CVD and a clinical trial of ambulatory individuals with prevalent CVD suggested there were differences in ischemic cardiovascular outcomes (13, 15). Our findings from this meta-analysis including a sensitivity analysis of only studies measuring both assays, and the primary analysis of both cTnI and T in a single large cohort do not support a contention that there are meaningful differences in cTnI or T concentrations even in older adults that would alter the prognostic association with any of the 4 clinical endpoints we evaluated.
Prior meta-analyses have also evaluated prognostic differences between hs-cTnI and T for all-cause mortality and incident CV outcomes in participants from ambulatory cohorts. We have summarized these findings reported in the individual meta-analyses in Supplemental Table 3. Interestingly, the most recent meta-analysis from Neuman et al. using a participant level analysis consistently found that progressively larger cTnT and not I concentrations have a higher HR for incident CVD outcomes. Participant level analysis has an advantage of greater precision and flexibility in uniformly adjusting for confounders. However, study selection for Neuman et al. was restricted to access to cohorts with availability of participant level data and as such includes 19 studies with 125 334 participants (5). Potentially the inclusion of >150 000 additional participants from 34 additional studies in the present analysis could have resulted in increasing generalizability, moving the difference between the HR estimates for cTnI and T toward the null. This could be a true finding but may also be the result of less precision when pooling study level results vs patient level data particularly with high heterogeneity noted between studies. Consistent with an earlier meta-analysis by Van der Linden as well as the analysis by Neuman et al. we showed that hs-cTnT was a stronger prognosticator for all-cause mortality than hs-cTnI (5, 10). Understanding mechanistically that cTnT is expressed in the skeletal muscle of older adults and patients with overt myopathies we might expect cTnT to represent some combination of progressive frailty as well as CVD. This may explain our finding that hs-cTnT concentrations are more strongly associated with all-cause mortality than hs-cTnI.
Limitations
We found significant between study heterogeneity similar to prior study level meta-analyses (6–10). Factors such as the population studied, ascertainment and definitions of endpoints, adjustment covariates, duration of follow-up and year of sample collection for cTn measurement (reflecting both participant medical care and sample stability), assay vendor (for hs-cTnI), as well as pre-analytical sample handling could contribute. Interestingly, we did not find differences based on presence or absence of prevalent CVD. One factor to consider is that participants with prevalent CVD were often participants in ambulatory randomized controlled trials who tend to be healthier overall than participants in observational cohorts with similar inclusion. We saw no statistical difference in studies that included measurements with both assays in the same participants. However, for incident MI the difference in hs-cTnI and T among studies that measured both approached significance such that with the addition of future studies testing both assays, this conclusion may change. Last, as recently shown in NHANES, different hs-cTnI assays perform differently for prognostication, particularly at the low-end of the assays’ measurement range that could result in different tertile 1 to 3 HR in the same population (29). Though most ambulatory adult hs-cTnI studies use the same assay from Abbott, the inclusion of other hs-cTnI could have introduced heterogeneity between studies.
Conclusions and Future Directions
In our meta-analysis of over one quarter of a million ambulatory adult participants from 54 studies we demonstrate that the 2 cardiac troponin isoforms I and T concentrations measured by hs assays do not show different long-term prognostication across unique and CV outcomes. In contrast, cTnT concentration is a more robust prognosticator of all-cause mortality. Future efforts may focus on developing age and assay specific values for clinical implementation in ambulatory adults that may be most relevant for prognostication and eventually stratification of cardiovascular therapies.
Supplemental Material
Supplemental material is available at Clinical Chemistry online.
Nonstandard Abbreviations
hs, high sensitivity; CV, cardiovascular; HF, heart failure; MI, myocardial infarction; cTnI, cardiac troponin I; cTnT, cardiac troponin T; CVD, cardiovascular disease; ARIC, Atherosclerosis Risk in Communities; MACE, major cardiovascular events.
Author Contributions
The corresponding author takes full responsibility that all authors on this publication have met the following required criteria of eligibility for authorship: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Nobody who qualifies for authorship has been omitted from the list.
Yashika Parashar (Conceptualization-Equal, Data curation-Equal, Formal analysis-Equal, Investigation-Equal, Methodology-Equal, Writing—original draft-Equal), Aya Awwad (Conceptualization-Equal, Formal analysis-Equal, Investigation-Equal, Methodology-Equal, Writing—original draft-Equal), Soahum Bagchi (Conceptualization-Equal, Formal analysis-Supporting, Investigation-Equal, Methodology-Supporting, Writing—review & editing-Equal), Brian Claggett (Formal analysis-Supporting, Methodology-Supporting, Writing—review & editing-Supporting), Saman Asad Siddiqui (Data curation-Supporting, Formal analysis-Supporting, Methodology-Supporting, Writing—review & editing-Supporting), Ajari Winifred Ogheneochuko (Formal analysis-Supporting, Methodology-Supporting, Writing—review & editing-Supporting), Christie Ballantyne (Methodology-Supporting, Writing—review & editing-Supporting), and Chris deFilippi (Conceptualization-Lead, Data curation-Equal, Investigation-Lead, Methodology-Equal, Project administration-Equal, Supervision-Lead, Writing—original draft-Lead, Writing—review & editing-Lead).
Authors’ Disclosures or Potential Conflicts of Interest
Upon manuscript submission, all authors completed the author disclosure form.
Research Funding
None declared.
Disclosures
S. Bagchi has received grants or research support from Eli Lilly & Co; consulting fees from Icarus Therapeutics and Max Pharma; and travel/meeting support from Nucleate, Inc. C.M. Ballantyne has received grants or research support from Abbott Diagnostic, Akcea, Amgen, Arrowhead, Ionis, Merck, New Amsterdam, Novartis, Novo Nordisk, Roche Diagnostic, and NIH, and consulting fees from 89Bio, Abbott Diagnostics, Amarin, Amgen, Arrowhead, Astra Zeneca, Denka Seiken, Esperion, Genentech, Illumina, Ionis, Eli Lilly, Merck, New Amsterdam, Novartis, Novo Nordisk, and Roche Diagnostic. B. Claggett has received consulting fees from Alnylam, Cardurion, Cardior, Corvia, Cytokinetics, CVRx, Intellia, and Eli Lilly and participated on a Data Safety Monitoring Board with Novo Nordisk. C. deFilippi has received grants or research support from NIH, Abbott Diagnostics, FujiRebio, Quidel:Ortho, Randox, Roche Diagnostics, and Siemens Healthineers; consulting fees from Abbott Diagnostics, FujiRebio, Quidel:Ortho, Pathfast, Roche Diagnostics, and Siemens Healthineers; honoraria from Pathfast; royalties from UpToDate; Patent issued, 10509044 “Methods for assessing differential risk for developing heart failure,” participated on a Data Safety Monitoring Board with Tosoh, and has an unpaid leadership or fiduciary role with the American College of Cardiology.
Role of Sponsor
No sponsor was declared.
Acknowledgments
We would like to acknowledge and thank Stefania Papatheodorou, Qiaoli Wang, and Paul Bain for their guidance and useful suggestions.
References
Author notes
Previous presentation: Presented in a more limited form as a Poster at the American Heart Association meeting, November 2023, Philadelphia, PA.
