Association of high-sensitivity troponin T with outcomes in asymptomatic non-severe aortic stenosis: a post-hoc substudy of the SEAS trial

Summary

Background

High-sensitivity Troponin T (hsTnT), a biomarker of cardiomyocyte overload and injury, relates to aortic valve replacement (AVR) and mortality in severe aortic stenosis (AS). However, its prognostic value remains unknown in asymptomatic patients with AS. We aimed to investigate if an hsTnT level >14 pg/mL (above upper limit of normal 99th percentile) is associated with echocardiographic AS-severity, subsequent AVR, ischaemic coronary events (ICE), and mortality in asymptomatic patients with non-severe AS.

Methods

In this post-hoc sub-analysis of the multicentre, randomised, double-blind, placebo-controlled SEAS trial (ClinicalTrials.gov, NCT00092677), we included asymptomatic patients with mild to moderate-severe AS. We ascertained baseline and 1-year hsTnT concentrations and examined the association between baseline levels and the risk of the primary composite endpoint, defined as the first event of all-cause mortality, isolated AVR (without coronary artery bypass grafting (CABG)), or ICE. Multivariable regressions and competing risk analyses examined associations of hsTnT level >14 pg/mL with clinical correlates and 5-year risk of the primary endpoint.

Findings

Between January 6, 2003, and March 4, 2004, a total of 1873 patients were enrolled in the SEAS trial, and 1739 patients were included in this post-hoc sub-analysis. Patients had a mean (SD) age of 67.5 (9.7) years, 61.0% (1061) were men, 17.4% (302) had moderate-severe AS, and 26.0% (453) had hsTnT level >14 pg/mL. The median hsTnT difference from baseline to 1-year was 0.8 pg/mL (IQR, −0.4 to 2.3). In adjusted linear regression, log(hsTnT) did not correlate with echocardiographic AS severity (p = 0.36). In multivariable Cox regression, a hsTnT level >14 pg/mL vs. hsTnT ≤14 pg/mL was associated with an increased risk of the primary composite endpoint (HR, 1.41; 95% CI, 1.18–1.70; p = 0.0002). In a competing risk model of first of the individual components of the primary endpoint, a hsTnT level >14 pg/mL was associated with ICE risk (HR 1.71; 95% CI, 1.23–2.38; p = 0.0013), but not with isolated AVR (p = 0.064) or all-cause mortality (p = 0.49) as the first event.

Interpretation

hsTnT level is within the reference range (≤14 pg/mL) in 3 out of 4 non-ischaemic patients with asymptomatic mild-to-moderate AS and remains stable during a 1-year follow-up regardless of AS-severity. An hsTnT level >14 pg/mL was mainly associated with subsequent ICE, which suggest that hsTnT concentration is primarily a risk marker of subclinical coronary atherosclerotic disease.

Funding

Merck & Co., Inc., the Schering-Plough Corporation, the Interreg IVA program, Roche Diagnostics Ltd., and Gangstedfonden. Open access publication fee funding provided by prof. Olav W. Nielsen and Department of Cardiology, Bispebjerg University Hospital, Denmark.

Research in context.

Evidence before this study

We performed forward citation searching from PubMed for articles published from database inception to October 1, 2022, using the following MeSH and text search terms without restrictions on study type or language: “aortic valve stenosis”, “asymptomatic”, “mild to moderate”, “non severe”, “troponin”, “high sensitivity troponin”, “high sensitive troponin”, “serial measurements”, “biological variation”, “sex specific threshold troponin “, “ischemic coronary disease”, “coronary artery disease”, “atherosclerosis”, “atherosclerosis risk factor”, “aortic stenosis risk factors”, “guidelines valvular heart disease “, “guidelines aortic stenosis”, “guidelines ischemic heart disease”, “cardiac amyloidosis and aortic stenosis”. Backward citation searching was conducted manually by inspecting the reference list of systematic reviews, guidelines and manufacturer's method sheet to retrieve specific cited information from the original source.

We found that increased cardiac troponins are widely used diagnostic and prognostic biomarkers of not only acute myocardial ischaemia but also other acute and chronic cardiovascular conditions, and mortality. In aortic valve stenosis (AS), cardiac troponins have been associated to AS progression, the need for aortic valve replacement surgery (AVR) and postoperative outcomes in symptomatic patients with moderate-to-very severe AS. To our knowledge, there were no studies to examine the prevalence of increased cardiac troponins in asymptomatic patients with non-severe AS, and no serial measurements over long-term follow-up (months to years) to determine temporal changes in relation to AS related outcomes. Moreover, no studies examined the predictive value of increased cardiac troponins for isolated AVR without prior or concomitant revascularisation.

Added value of this study

Using data from the largest existing randomised trial of asymptomatic patients with non-severe AS and no history of coronary artery disease, our study provides novel and clinically relevant insights into high sensitivity Troponin T (hsTnT) trajectory over a 1-year follow-up and its relation to AS severity. Furthermore, we are the first to explore associations between increased vs. normal hsTnT level and risk of isolated AVR, ischaemic coronary events (ICE), and mortality in asymptomatic patients with AS.

Implications of all the available evidence

Our data indicate that hsTnT levels are more influenced by latent coronary artery disease and left ventricular mass than by the echocardiographic grading of AS severity. Our study shows that clinicians should expect normal hsTnT levels in 3 out of 4 patients. Furthermore, in these non-ischaemic patients with asymptomatic non-severe AS, the mean annual change is below 1 pg/mL, regardless of the AS severity. Still, an increased vs. normal hsTnT level is strongly and consistently associated with a high 5-year risk of ICE-related outcomes independently of traditional cardiovascular risk factors and lipid-lowering treatment. Therefore, increased hsTnT is associated with the risk of AVR, and the mechanism may be through latent coronary artery disease. There is a need for prospective studies to examine whether an increased hsTnT may guide revascularisation before AVR in patients with non-severe AS.

Introduction

Degenerative calcific aortic valve stenosis (AS) is a growing^1,2 global health problem affecting up to 12.4%³ of the projected 1.5 billion elderly aged above 65 years in 2050.⁴ Due to a shared link with age and atherosclerotic risk factors, coronary artery disease (CAD) coexists in up to 75% of older adults with AS.^5,6 The association between prevalent AS and CAD^2,5,7, 8, 9, 10, 11, 12 is unmistakable as 1 of 3 symptomatic patients undergo concomitant aortic valve replacement (AVR) and coronary revascularisation.^5,13,14 Consequently, it may be challenging to determine whether new-onset symptoms of angina pectoris or heart failure are due to progression of AS or CAD in patients with initially asymptomatic AS.

High-sensitivity cardiac troponin is an established biomarker of ischaemia and cardiomyocyte injury due to cardiac pressure overload.15, 16, 17, 18, 19 Still, there are no guideline recommendations for high-sensitivity troponin T (hsTnT) guided monitoring during the watchful waiting of patients with AS. The current literature mainly consists of observational studies in heterogeneous populations investigating whether single measurements of cardiac troponin levels can be used as a marker of left ventricular remodelling on the pathway to end-stage heart failure and a need for AVR in severe AS.20, 21, 22, 23, 24 In the clinical setting, it is undetermined which hsTnT levels can be expected in an outpatient population of stable patients with asymptomatic non-severe AS. Similarly, it is unclear if hsTnT primarily associates with CAD or AS severity in patients without hemodynamically significant AS and if serial hsTnT measurements may aid further in this regard.

We, therefore, conducted an observational post-hoc analysis of the large, randomised SEAS trial²⁵ (Simvastatin and Ezetimibe in Aortic Stenosis), which investigated the effect of lipid-lowering therapy in 1873 patients with asymptomatic mild to moderate-severe AS, preserved systolic function and no history or symptoms of CAD. Our main objectives were: 1) to explore what patient characteristics are associated with hsTnT concentration; 2) hsTnT trajectory during one year of follow-up; and 3) to assess if hsTnT concentration at baseline is associated with adverse clinical outcomes, particularly isolated AVR, ischaemic coronary events (ICE), and all-cause mortality. We hypothesised that an hsTnT level >14 pg/mL at baseline would be associated with more severe AS and herald a higher risk of isolated AVR, ICE, and mortality.

Methods

Study design and participants

The Simvastatin and Ezetimibe in Aortic Stenosis (SEAS) trial was a multicenter (173 centers in Northern Europe), double-blind, placebo-controlled randomised clinical trial which examined the effect of combined Simvastatin and Ezetimibe (oral 40 + 10 mg) in patients with asymptomatic AS. A total of 1873 eligible patients were randomised between January 6, 2003, and March 4, 2004. The main inclusion criteria were age 45–85 years, normal left ventricular (LV) ejection fraction, and an investigator-assessed transaortic maximal velocity (Vmax) 2.5–4.0 m/s determined by transthoracic Doppler echocardiography. The main exclusion criteria were pre-existing CAD, diabetes, and/or other indications for lipid-lowering therapy.^25,26 The present post-hoc analysis examined the association of hsTnT concentrations at baseline with the severity of AS and cardiovascular outcomes over a 5-year follow-up from baseline and whether hsTnT concentration remained stable during the first year of follow-up. We included 1739 patients with a baseline hsTnT measurement and at least one in-study Vmax measurement.

Local ethic committees had approved the SEAS trial, which adhered to the Declaration of Helsinki and all participants provided written informed consent (ClinicalTrials.gov unique identifier, NCT00092677). The reporting of the study adhere to the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) statement.²⁷

Troponin measurements

Blood samples were centrifuged and stored at −80 °C until analysed in a central laboratory between November 18 and December 2, 2016. There were 1739 hsTnT measurements at baseline, but 125 participants had missing hsTnT measurements at year 1 (see Supplementary eTable S1 for baseline characteristics). The hsTnT was measured by electrochemiluminescence immunoassay, Elecsys Troponin T hs 200, on a COBAS e 601 (Roche Diagnostics), with a manufacturer-reported measuring range of 3–10 000 pg/mL, an intermediate precision between 1.3–8.6% and the 99th percentile cut-off for a healthy reference population of 14 pg/mL. In sensitivity analyses, we also examined sex-specific 99th percentile cut-offs above 8.9 pg/mL for women and 15.5 pg/mL for men.^28,29

We calculated the relative annual hsTnT change as

$$\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{c}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}=\frac{\left[\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{a}\mathrm{t}\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}1\right]}{\left[\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{a}\mathrm{t}\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}\right]}$$

and the absolute annual hsTnT change as

$$\mathrm{a}\mathrm{b}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{c}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}\left(\Delta \mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\right)=\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{a}\mathrm{t}\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{1}—\mathrm{h}\mathrm{s}\mathrm{T}\mathrm{n}\mathrm{T}\mathrm{a}\mathrm{t}\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}$$

The primary exposure variable was baseline hsTnT level ≤14 pg/mL vs. hsTnT level >14 pg/mL, where a cut-off point of 14 pg/mL was previously proposed to define prognostically relevant myocardial injury.^18,28 Secondary exposure variables were hsTnT level persistently within reference range (≤14 pg/mL) vs. hsTnT level persistently above upper limit of normal (>14 pg/mL) at both baseline and year 1, examined at two thresholds: hsTnT >14 pg/mL and hsTnT >50 pg/mL.

Echocardiography

The SEAS echocardiographic core laboratory analysed the echocardiograms after clinical data blinding,²⁶ and according to published guidelines.^25,30 In this analysis, we defined moderate-severe AS as 2 or more of the following baseline characteristics: aortic valve area (AVA) <1 cm², Vmax ≥4.0 m/s, mean aortic valve pressure gradient (Pmean) ≥40 mmHg, and velocity ratio <0.25, and otherwise as non-severe AS.³¹

Outcomes

The primary endpoint in this substudy was a composite of the first event of all-cause mortality, isolated AVR without coronary artery bypass grafting (CABG), and ICE. Secondary endpoints were all-cause mortality, AVR with or without concomitant revascularisation, and ICE. The decision to perform isolated AVR or AVR with concomitant CABG was at the discretion of the local Heart Team in keeping with contemporary guidelines and without knowledge of the hsTnT concentrations. In the present analysis, we defined isolated AVR without concomitant CABG or percutaneous coronary intervention (PCI) as a proxy for pure AS progression. Similarly, we defined ICE as a composite of the first event of myocardial infarction (MI), PCI, or CABG with concomitant AVR.

Statistical analysis

The sample size of our study was predetermined by the sample size of patients included in the main SEAS trial and power calculations on biomarkers studies to estimate the prognostic impact of elevated vs. normal levels of biomarkers on adverse outcomes during up to 5 years follow-up. Returning estimates were based on 90% power and a significance level of 5% and the largest sample size needed of 2∗600 patients to detect isolated events. With 664 events from the primary composite endpoint and 494 of any AVR, our study is well powered to examine the association between hsTnT concentrations and outcomes, simultaneously adjusted with up to 20 covariates if needed.

We evaluated normality for continuous variables by visual inspections of histograms. Continuous variables were presented as means with standard deviation (SD) for normally distributed values or medians with interquartile range (IQR) for skewed variables. Categorical variables were presented as numbers and percentages. As appropriate, descriptive statistics were used to analyse baseline characteristics, including ANOVA, Chi-square, Kruskal–Wallis, and Wilcoxon tests. Missing at random values, aside from hsTnT, at baseline or year 1 were imputed by multivariate imputation using the method of the chained equation.³²

To examine the correlation between baseline hsTnT concentrations and clinical variables, we used a multivariable linear regression model with log-transformed hsTnT and eGFR. The standardised clinically relevant independent variables were age, sex, Framingham 10-year cardiovascular risk score, body mass index (BMI), systolic blood pressure, ejection fraction, AS severity (moderate-severe vs. non-severe AS), Pmean, left ventricular mass index (LVMi), interventricular septum thickness at end-diastole, and log(eGFR)³³ level (see Supplementary eMethods).

Time-to-event analyses started at the baseline and ended at year 5. The association of the primary composite endpoint and baseline hsTnT levels was examined with event-free Kaplan–Meier survival estimate curves and log-rank test for (a) quartiles of hsTnT and (b) baseline hsTnT levels ≤14 pg/mL vs. hsTnT levels >14 pg/mL. The association between outcomes as a function of baseline hsTnT levels ≤14 pg/mL vs. >14 pg/mL was further examined by (i) the observed cumulative incidence rates (per 100 patient-years), presented with 95% confidence intervals (95% CI), (ii) the cause-specific cumulative incidence functions for competing events, presented in %, and (iii) uni- and multivariable Cox proportional hazard regression models, presented as hazard ratios (HR) with 95% CI. The multivariable model, stratified by the center of inclusion and lipid-lowering treatment, was adjusted for sex and standardised baseline values of age, creatinine, and mean aortic gradient. Sensitivity analysis tested for an independent association of hsTnT with outcomes by additional model adjustment for the Framingham 10-year cardiovascular risk score and treatment (lipid-lowering vs. placebo). In addition, we explored the interaction effects of hsTnT level >14 pg/mL and sex, and hsTnT level >14 pg/mL and AS severity on all outcomes.

All tests were two-tailed, and p-values <0.05 were considered statistically significant. Statistical analyses were performed in Stata software (Stata/IC, version 13.1, StataCorp, College Station, Texas, USA; packages estout, table1, stpm2, stpm2cif, mi impute chained, std, ckdepi, framingham).

Role of the funding source

The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. EH and OWN had access to dataset and had final responsibility for the decision to submit manuscript for publication.

Results

Baseline characteristics

We included 1739 of the 1873 SEAS patients (92.8%) with available baseline hsTnT measurements. Of those, 678 were women (39.0%) and 1061 men (61.0%), with a mean (SD) age of 67.5 (9.7) years and LVEF of 65% (8%). As shown in Table 1, 302 patients (17.4%) had moderate-severe AS with a mean (SD) aortic valve area of 0.8 (0.2) cm², Vmax 3.7 (0.5) m/s, Pmean 33.2 (8.8) mm Hg, and velocity ratio 0.22 (0.03).

Table 1.

Demography at baseline.

	All patients	hsTnT at baseline		p value
		≤14 pg/mL	>14 pg/mL
	No.	No. (%)	No. (%)
Variable	1739	1286 (74.0)	453 (26.0)
Year 0
Age, mean (SD), y	67.5 (9.7)	65.6 (9.4)	72.8 (8.2)	<0.0001
Sex at birth, No. (%)
Men	1061 (61.0)	724 (56.3)	337 (74.4)	<0.0001
Women	678 (39.0)	562 (43.7)	116 (25.6)
Blood pressure, mean (SD), mm Hg
Systolic	144.7 (20.0)	143.2 (19.3)	148.8 (21.3)	<0.0001
Diastolic	82.0 (10.3)	82.0 (10.3)	82.3 (10.3)	0.57
Heart rate, mean (SD), bpm	68.0 (10.4)	68.0 (10.3)	67.7 (10.6)	0.57
BMI, mean (SD), kg/m2	26.5 (4.4)	26.3 (4.4)	26.9 (4.4)	<0.0088
Antihypertensive treatment, No. (%)	1318 (75.8)	936 (72.8)	382 (84.3)	<0.0001
Active smoking, No. (%)	330 (19.0)	264 (20.5)	66 (14.6)	0.0054
Framingham 10-year risk score,a median (IQR), %	25.8 (16.1, 38.3)	22.2 (14.6, 34.1)	34.6 (25.2, 46.2)	<0.0001
Medical history, No. (%)
Hypertension	891 (51.2)	614 (47.7)	277 (61.1)	<0.0001
Prior atrial fibrillation	163 (9.4)	93 (7.2)	70 (15.5)	<0.0001
Baseline measurements
Atrial fibrillation on baseline ECG, No. (%)	54 (3.1)	25 (1.9)	29 (6.4)	<0.0001
Echocardiography
Aortic valve area, mean (SD), cm2	1.3 (0.5)	1.3 (0.5)	1.2 (0.4)	0.053
Aortic valve area index (BSA), mean (SD), cm2/m2	0.7 (0.2)	0.7 (0.2)	0.6 (0.2)	0.0014
Ejection fraction, mean (SD), %	65.8 (8.0)	66.2 (7.9)	64.8 (8.4)	0.0014
Left ventricular mass index, mean (SD), g/m2	101.5 (30.6)	97.5 (28.0)	112.8 (34.5)	<0.0001
Transaortic maximal velocity (Vmax), mean (SD), m/s	3.1 (0.5)	3.1 (0.5)	3.2 (0.6)	0.0002
Mean aortic gradient (Pmean), mean (SD), mm Hg	22.9 (8.8)	22.4 (8.6)	24.3 (9.1)	<0.0001
Stroke volume index <35 mL/m2, No. (%)	445 (25.6)	326 (25.3)	119 (26.3)	0.43
Non-severe aortic stenosis,b No. (%)	1437 (82.6)	1076 (83.7)	361 (79.7)	0.055
Moderate-severe aortic stenosis,b No. (%)	302 (17.4)	210 (16.3)	92 (20.3)
Laboratory values
hsTnT, median (IQR), pg/mL	9.7 (6.7, 14.4)	8.1 (6.1, 10.5)	18.6 (15.8, 24.6)	<0.0001
eGFR, mean (SD), mL/min/1.73 m2	66.9 (13.0)	68.8 (12.5)	61.5 (13.0)	<0.0001
Creatinine, mean (SD), mg/dL	1.1 (0.2)	1.0 (0.2)	1.1 (0.2)	<0.0001
High-sensitive C-reactive protein (hsCRP), mean (SD), mg/dL	0.4 (0.8)	0.4 (0.7)	0.5 (0.9)	0.016
Total cholesterol, mean (SD), mg/dL	221.6 (39.3)	222.6 (39.1)	218.7 (39.7)	0.073
Low-density lipoprotein (LDL), mean (SD), mg/dL	138.4 (35.2)	139.0 (35.1)	136.8 (35.4)	0.26
High-density lipoprotein (HDL), mean (SD), mg/dL	58.1 (16.6)	58.5 (16.9)	57.3 (15.4)	0.18
Year 1
Laboratory values (n = 1614)
hsTnT, median (IQR), pg/mL	10.7 (7.4, 15.6)	8.8 (6.5, 11.8)	19.4 (16.0, 25.7)	<0.0001
hsTnT, No. (%)
≤14 pg/mL	1125 (64.7)	1064 (82.7)	61 (13.5)	..
>14 pg/mL	489 (28.1)	130 (10.1)	359 (79.2)
hsTnT change from baseline to year 1 (ΔhsTnT),c median (IQR), pg/mL	0.8 (−0.4, 2.3)	0.8 (−0.2, 2.0)	0.6 (−2.1, 3.7)	0.079
hsTnT change from baseline to year 1 (hsTnT change),d median (IQR)	1.1 (1.0, 1.2)	1.1 (1.0, 1.3)	1.0 (0.9, 1.2)	<0.0001

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); BSA, body surface area; eGFR, estimated glomerular filtration rate; hsTnT, high-sensitivity Troponin T; IQR, interquartile range; SD, standard deviation.

^aEstimated Framingham 10-year cardiovascular risk score based on age, sex at birth, systolic blood pressure, antihypertensive treatment, smoking, diabetes, total cholesterol, and high-density lipoprotein (HDL).

^bModerate-severe AS was defined as a minimum 2 of the following baseline values: aortic valve area (AVA) <1 cm², transaortic maximal velocity (Vmax) ≥4.0 m/s, mean aortic gradient (Pmean) ≥40 mmHg, and velocity ratio <0.25, otherwise non-severe AS.

^cΔhsTnT = hsTnT at year 1 - hsTnT at year 0.

^dhsTnT change = (hsTnT at year 1)/(hsTnT at year 0).

Factors associated with hsTnT concentration at baseline and after a 1-year follow-up

Baseline hsTnT levels >14 pg/mL were found in 26.0% of patients (453 of 1739) and associated with more cardiovascular risk factors and abnormal echocardiographic parameters (Table 1). The total population's median hsTnT level was within reference range (≤14 pg/mL) at baseline and year 1 (Fig. 1, Supplementary eFig. S1). In multivariable linear regression analysis, log(hsTnT at baseline) was positively correlated with male sex, age, LV mass index, and mean aortic gradient and negatively correlated with log(eGFR) (adjusted R² = 0.36 for the model) (Table 2, Supplementary eFig. S2), but did not correlate with AS severity (moderate-severe vs. non-severe AS, p = 0.36) (not shown).

Fig. 1.

High-sensitivity Troponin T concentration stratified by echocardiographic severity of aortic stenosis. The boxplots show hsTnT concentrations at baseline (grey) and year 1 (bluish) stratified by the severity of aortic stenosis (AS) in non-severe (left) and moderate-severe AS (right). The red dashed lines demarcate hsTnT level >14 pg/mL (above the upper limit of normal 99th percentile) and very high hsTnT level >50 pg/mL. p values reflect a two-sided paired sign test of medians for baseline vs. year 1 and a two-sample unpaired t-test for comparisons of non-severe AS vs. moderate-severe AS, respectively. Abbreviations: AS, aortic stenosis; hsTnT, high-sensitivity Troponin T.

Table 2.

Variables independently associated with high-sensitivity Troponin T concentration in patients with asymptomatic aortic stenosis.

Response variable: log(hsTnT at baseline)	Univariable model					Multivariable model
						Total (N = 1739)
	N	t value	Coef.	(95% Confidence Interval)	p value	t value	Coef.	(95% Confidence Interval)	p value
Male sex	1739	11.83	0.330	0.276, 0.385	<0.0001	15.72	0.393	0.344, 0.442	<0.0001
Agea	1739	19.19	0.247	0.222, 0.272	<0.0001	15.44	0.209	0.182, 0.235	<0.0001
Left ventricular mass index (LVMi)a	1739	12.55	0.170	0.144, 0.197	<0.0001	9.76	0.116	0.093, 0.140	<0.0001
Mean aortic gradient (Pmean)a	1739	4.88	0.069	0.041, 0.096	<0.0001	3.26	0.037	0.015, 0.060	0.0011
log(eGFR)	1739	−13.17	−0.879	−1.010, −0.748	<0.0001	−8.59	−0.589	−0.723, −0.455	<0.0001

Abbreviation: NS, non-significant.

R-squared = 0.3617; adjusted R-squared = 0.3599.

The following non-significant variables were omitted from the multivariable regression model: Systolic blood pressure, body mass index, left ventricular ejection fraction, interventricular septum thickness at end-diastole, severity of aortic stenosis, and Framingham 10-year risk score.

^aStandardised continuous variable = (variable value – mean of variable value)/SD of variable value.

In the whole population, median absolute annual hsTnT change was 0.8 pg/mL [IQR, −0.4 to 2.3] (Table 1, Fig. 1). Inspection of a Sankey diagram revealed that hsTnT levels generally regressed towards the mean over a 1-year follow-up (Supplementary eFig. S1).

At year 1 of follow-up, increased hsTnT levels (>14 pg/mL) were found in 30.3% of patients (489 of 1614) (Table 1). Persistently increased hsTnT levels (>14 pg/mL), both at baseline and 1-year, occurred in 20.6% of patients (359 of 1739). Just 0.6% of patients (10 of 1739) had persistent hsTnT levels >50 pg/mL (Supplementary eTable S2). In 15.5% of patients (250 of 1614), hsTnT level increased from ≤14 pg/mL at baseline to >14 pg/mL at year 1 (Supplementary eFig. S1).

Both patients with persistent hsTnT levels >14 pg/mL, and those with increasing hsTnT levels from baseline, were older, had higher Framingham risk score, more frequent atrial fibrillation, and markedly higher LV mass index at baseline (Supplementary eTable S2).

Associations between baseline hsTnT levels and outcomes

Over a median follow-up of 4.35 years, 652 patients (37.5%) experienced one of the events from the primary composite endpoint (Fig. 2). The primary event consisted of 125 (7.2%) deaths as the first event, 334 patients (19.2%) who underwent isolated AVR without CABG, and 193 patients (11.1%) who experienced ICE. ICE events were due to 43 (2.5%) myocardial infarctions, 15 (0.9%) PCIs, and 135 (7.8%) CABGs with a concomitant AVR. The primary composite endpoint event rate increased with higher quartiles of baseline hsTnT levels (Supplementary eFig. S3). The fourth hsTnT quartile, above 14.4 pg/mL, had the worst outcome (p < 0.0001). An hsTnT level >14 pg/mL was associated with a 41% increased 5-year risk of primary composite endpoint events (multivariable HR, 1.41; 95% CI, 1.18–1.70; p = 0.0002) (Fig. 2) and any AVR (HR 1.50; 95% CI, 1.21–1.87; p = 0.0003) but not death (multivariable p = 0.51) (Supplementary eFig. S4).

Fig. 2.

Multivariable Cox regression models. The figure shows Forest plot with event rates [95% CI] and Hazard Ratios [95% CI] based on Cox regression models examining the association between baseline high-sensitivity Troponin T concentrations ≤14 pg/mL and >14 pg/mL and outcomes: the primary composite end point and its individual components. Models were adjusted for standardised baseline values of age, sex, creatinine, and mean aortic gradient at baseline, stratified by center and lipid-lowering treatment. Follow-up from baseline to year 5. ^aThe first event of All-cause mortality, AVR without CABG, or ICE. ^bThe first event of CABG+AVR, myocardial infarction, or PCI. Abbreviations: AVR, aortic valve replacement; CABG, coronary artery bypass grafting; CI, confidence interval; hsTnT, high-sensitivity Troponin T; ICE, ischaemic coronary events; PCI, percutaneous coronary intervention.

When examining competing risk components of the primary endpoint, patients with baseline hsTnT levels >14 pg/mL had a 2-fold higher probability of experiencing an ICE and all-cause mortality as the first event than patients with hsTnT levels within reference range (≤14 pg/mL) (Fig. 3). However, hsTnT levels >14 pg/mL suggested the same probability of isolated AVR as hsTnT levels ≤14 pg/mL (Fig. 3). In a competing risk model, an hsTnT level >14 pg/mL was associated with a 71% higher 5-year ICE risk (HR 1.71; 95% CI, 1.23–2.38; p = 0.0013) (Fig. 2), but not with isolated AVR (univariable p = 0.19, Supplementary eTable S3 and eFig. S5; multivariable p = 0.064, Fig. 2) nor death (p = 0.49) (Fig. 2). In sensitivity analyses, we further adjusted for the cardiovascular risk factors included in the Framingham risk score and lipid-lowering treatment. Still, an hsTnT level >14 pg/mL remained the strongest independent risk factor of ICE (HR, 1.77; 95% CI, 1.28–2.46; p = 0.0006) (Supplementary eFig. S6). There was no significant interaction between hsTnT level >14 pg/mL and active vs. placebo lipid-lowering treatment for either primary composite endpoint (p for interaction = 0.52) or ICE (p for interaction = 0.58). Moreover, we found no significant interaction effects of hsTnT level >14 pg/mL with sex (Supplementary eFig. S7) or AS severity (Supplementary eFig. S8) for any of the outcomes (all p for interaction >0.40). In sensitivity analyses based on sex-specific cut-offs, 31.3% (545 of 1739 patients) had hsTnT levels above 99th percentile at baseline, and 28.5% (460 of 1614) had persistent hsTnT levels >99th percentile (not shown). Apart from generating slightly more patients with hsTnT levels >99th percentile, these sex-specific cut-off values did not associate better to the outcomes as compared with a single cut-off >14 pg/mL (not shown). The annual hsTnT change was within normal biological variation (Fig. 1) and there was no added prognostic value by examining the annual change of hsTnT.

Fig. 3.

Stacked cumulative incidence function plots. The figure shows cumulative incidence of all-cause mortality (ACM) as the first and only event (grey), aortic valve replacement (AVR) without revascularisation (violet), and ischaemic coronary events (ICE) (pink) during follow-up from baseline to year 5 in patients with asymptomatic aortic stenosis, stratified by (a) baseline high-sensitivity Troponin T (hsTnT) levels ≤14 pg/mL and (b) hsTnT levels >14 pg/mL. Abbreviations: ACM, all-cause mortality; AVR, aortic valve replacement; hsTnT, high-sensitivity Troponin T; ICE, ischaemic coronary events.

Discussion

This study presents four novel observations regarding hsTnT in apparently non-ischaemic patients with asymptomatic non-severe AS. First, 3 out of 4 patients with asymptomatic non-severe AS have baseline hsTnT levels within reference range. Second, the hsTnT level increased less than 1 pg/mL during one year of follow-up regardless of the AS severity. Third, patients with baseline hsTnT levels >14 pg/mL had a 41% higher risk of a composite of AVR, ICE and all-cause mortality than those with hsTnT levels ≤14 pg/mL. Fourth, in competing risk analyses, a baseline hsTnT level >14 pg/mL was more closely associated with increased ICE-related risk than isolated AVR or death. The present study, therefore, extends current clinical knowledge on the clinical utility of hsTnT in patients with asymptomatic non-severe AS.

Our study population with a mean age of 67.5 years had a 26% prevalence of hsTnT level above upper limit of normal (>14 pg/mL), which is compatible with the 30% found in a random sample of the general population aged ≥70 years.³⁴ Thus, hsTnT levels seem only minimally affected by aortic valve degeneration in asymptomatic patients whose stenosis is not very severe. This finding contrasts with previous studies that examined more symptomatic patients with a more severe AS.^20,22,24 Instead, our study suggests that hsTnT in asymptomatic patients with non-severe AS is more closely associated with well-known cardiovascular risk factors^20,24,35,36 such as age, male sex, renal function, and LV mass index.

Previous studies described an association between elevated troponin and higher rates of AVR, morbidity, and mortality.^19,20,23,24,37, 38, 39 We confirmed that hsTnT level >14 pg/mL was associated with increased risk of a primary composite endpoint and, apparently, of overall AVR but the competing risk analysis indicates that these associations were primarily related to the ICE component, i.e. CABG as a driver for concomitant AVR. The association with ICE remained significant after adjustment for 10-year Framingham risk score and lipid-lowering treatment, suggesting that hsTnT may be a simple objective risk marker for latent coronary artery disease in these patients. However, lipid-lowering treatment was not modified by hsTnT levels as a predictor of ICE as there were no significant interactions regarding efficacy.

We have previously shown that sequential N-Terminal pro-brain natriuretic peptide measurements provide incremental prognostic value for the risk of aortic valve events in patients with asymptomatic non-severe AS.⁴⁰ Still, the usefulness of sequential hsTnT measurements in this population has not previously been adequately studied. Just one previous study⁴¹ of 25 patients with moderate AS reported a mean annual concentration of 9.2 ng/L and an annual within-individual variation of 11.2% (95% CI, 9.6–13.5) in 16 patients with stable AS. That study was underpowered to report any change of hsTnT over one year. We now substantiate these observations in a large population of stable patients with asymptomatic mild to moderate-severe AS, where hsTnT levels were quite stable during one year of follow-up, with a median annual hsTnT change of 0.8 pg/mL, regardless of AS severity, and therefore too small to reach sufficient statistical power to assess the association of ΔhsTnT with outcomes.

Persistent hsTnT levels >14 pg/mL, measured twice during a 1-year follow-up, could indicate a more profound ongoing cardiac injury. In our study, 1 out of 5 patients had hsTnT levels persistently >14 pg/mL, which was associated with higher LV mass as compared to patients with hsTnT levels persistently within reference range ≤14 pg/mL.

In AS, LV mass increases secondary to increased LV afterload and preload due to the progressive aortic valve narrowing,⁴² although previous studies failed to establish a dose–response relationship between the grade of AS severity and amount of LV remodeling.⁴³ Nevertheless, high LV mass might be based on multifactorial conditions commonly coexisting with AS, i.e., cardiac amyloidosis, which affects about 16% of patients referred to AVR for severe AS.⁴⁴ Cardiac amyloidosis is frequently associated with elevated hsTnT levels, and persistent hsTnT levels above 50–54 ng/L are signs of poor outcomes in AS patients referred to AVR.^45,46 Thus, a hsTnT levels persistently >50 pg/mL in our study could be a proxy for the potential presence of cardiac amyloidosis. We were not able to assess whether amyloidosis was a contributing factor in our patients, but only 0.6% of our population had persistent hsTnT levels >50 pg/mL, perhaps due to a low mean age of 67.5 years as opposed to the age of 83.7 years in previous studies.⁴⁴

Our data do not support routine hsTnT screening in AS patients to predict the risk of early AVR. However, if hsTnT levels >14 pg/mL are found in an asymptomatic patient with non-severe AS, it may be a warning sign of unrecognised CAD. Currently, there is a growing interest in performing earlier interventions in less symptomatic patients and in those with moderate AS. Such patients, with less severe AS, may receive aortic valve replacement with or without revascularisation. Our results inspire the hypotheses that hsTnT levels >14 pg/mL could identify patients who might favour isolated PCI before TAVR, but further studies are warranted to investigate the utility of hsTnT in a combined multimarker risk score for risk stratification and surveillance of asymptomatic patients with non-severe AS.

This study has some additional limitations besides those already mentioned. The SEAS trial's inclusion criteria targeted the asymptomatic AS patients in relatively early phases of AS development and excluded patients with a history of CAD, diabetes, heart failure, and kidney failure. Because unselected outpatients with AS often hold these comorbidities, one should be cautious in extrapolating our results to such patients with symptomatic or severe AS, who likely also have a higher prevalence of hsTnT levels >14 pg/mL.

To minimise the bias by indication, we distinguished between an isolated AVR and the double procedure based on the presence of ischaemic events or coronary revascularisations concomitant with AVR.

The biomarker substudy was delayed up to 8 years from closing the SEAS trial, and we could not test for possible temporal degradation of blood samples regarding hsTnT, but our recent study on NT-proBNP based on the same blood samples showed no significant degradation.⁴⁰ A strength of our study is that hsTnT concentration was not routinely used for clinical decision making in patients with AS and the present data did not influence the Heart Teams' clinical decisions.

In this large cohort of 1739 asymptomatic AS patients without overt CAD, we found that hsTnT is within reference range ≤14 pg/mL in 3 out of 4 patients and remains stable during a 1-year follow-up regardless of AS severity. hsTnT levels >14 pg/mL were consistently and most strongly associated with presence of CAD-related events independently of traditional cardiovascular risk factors and lipid-lowering treatment but not with AS severity or death. These findings support the notion of an hsTnT level >14 pg/mL primarily as an independent risk marker of subclinical CAD in asymptomatic patients with non-severe AS.

Contributors

EH, AMG, and OWN conceived and designed the study. MHO, YAK, CAN, SGR, ABR, KW, and OWN contributed to data collection. EH, AMG, AS, MHO, and OWN conducted the data analysis and interpretation. EH, AMG, and OWN drafted the initial manuscript. All authors critically revised the manuscript for important intellectual content and approved the submitted version.

Data sharing statement

The dataset analysed for this study is not publicly available due to privacy or ethical restrictions.

Declaration of interests

EH received research scholar grant from Gangstedfonden, Denmark. MHO received honoraria (lecture fees) from Astra Zeneca and Novo Nordic A/S (outside the submitted work). YAK received study materials and support for article processing. OWN received assays from Roche Diagnostics A/S Denmark for troponin analysis and Interreg IVA grant for collection of database (related to the submitted work), and honoraria (lecture fees) from Orion Pharma, Novartis, and Roche (outside the submitted work). AMG, AS, CAN, SGR, ABR, and KW declare no competing interests.