High-sensitivity cardiac troponin T for detection of subtle abnormalities of cardiac phenotype in a general population of elderly individuals




To investigate the association between circulating cardiac biomarkers and minor abnormalities in cardiac phenotype [left ventricular (LV) mass and midwall fractional shortening (MFS)] in elderly individuals in a general population sample.

Design and setting

We examined the relationship between plasma concentrations of high-sensitivity cardiac troponin T (hs-cTnT) or N-terminal probrain natriuretic peptide (NT-proBNP) and elevated LV mass (LV mass/body surface area >95 g m−2 for women and 115 g m−2 for men), reduced MFS (<15%) or isolated LV diastolic dysfunction in 1973 elderly subjects (mean age 73 ± 5 years, range 65–84) resident in the Lazio region of Italy and enrolled in the PREDICTOR study.


Overall, 24.8% of subjects had elevated LV mass, and 30.4% had reduced MFS. Median [quartile 1–3] plasma concentrations of hs-cTnT and NT-proBNP were higher in individuals with elevated than those with normal LV mass: 6.6 [3.5–11.6] and 147 [64–296] ng L−1 vs. 4.6 [3.0–8.1] and 79 [41–151] ng L−1 respectively (< 0.001). There was a graded increase in median hs-cTnT concentrations across clinical categories of LV hypertrophy: 4.6 [3.0–8.1], 5.8 [3.1–10.2], 7.6 [3.8–13.7] and 8.4 [3.8–17.6] ng L−1 for subjects with normal LV mass and mild, moderate or severe LV hypertrophy respectively (< 0.0001); hs-cTnT also increased with increasing quartiles of MFS or grades of isolated LV diastolic dysfunction.


Within an extremely low range of concentrations, increased hs-cTnT amongst community-dwelling elderly subjects is associated with subtle alterations in cardiac phenotype, suggesting that minor injury to cardiac myocytes and subsequent release of troponin reflect subclinical pathophysiological LV deterioration in this population.


American College of Cardiology/American Heart Association


high-sensitivity C-reactive protein


high-sensitivity cardiac troponin T


left ventricular


body surface area


midwall fractional shortening


N-terminal probrain natriuretic peptide


receiver-operating characteristic


High-sensitivity cardiac troponins allow an early and reliable diagnosis of acute myocardial infarction [1]; they are also robust predictors of fatal events in patients with chronic coronary disease [2] and heart failure (HF) [3, 4]. In studies in the general population, extremely low levels of release of cardiac troponin into the bloodstream have been detected in some individuals, with relatively higher levels associated with higher risk of mortality [5-8]. The same relationship had previously been observed using less-sensitive reagents [9, 10]. The exact mechanism(s) of release of troponin is not fully understood, but may be the result of normal myocyte turnover or due to myocyte necrosis and apoptosis, cellular release of proteolytic troponin degradation products, increased cell wall permeability and/or formation and release of blebs [11], as well as noncardiac factors [12]. Advanced age, male sex, reduced renal function and other cardiovascular disease risk factors can influence troponin release in the general population [7-11].

Alterations in left ventricular (LV) structure and function are risk factors for cardiovascular events and are associated with increased circulating levels of troponins in the general population. For instance, LV mass was found to be an independent determinant of elevated troponin in 3546 individuals aged 30–65 years enrolled in the Dallas Heart Study [6] and in 1005 elderly subjects in a Swedish population [13]. There is a graded relationship between subtle alterations in cardiac phenotype and discrete increases in circulating cardiac marker concentrations in the general population [6].

Here we report a head-to-head evaluation of high-sensitivity cardiac troponin T (hs-cTnT) and the natriuretic peptide N-terminal probrain natriuretic peptide (NT-proBNP), in addition to biomarkers of inflammation [high-sensitivity C-reactive protein (hsCRP)] and renal function (cystatin C), in community-dwelling elderly individuals who underwent detailed echocardiographic investigation in a multicentre cross-sectional epidemiological study. We also assessed the ability of biomarkers to correctly identify individuals with elevated LV mass, abnormal midwall fractional shortening (MFS) or isolated LV diastolic dysfunction.


PREDICTOR (Valutazione della PREvalenza di DIsfunzione Cardiaca asinTOmatica e di scompenso caRdiaco conclamato in un campione di popolazione di età ≥65 anni nel Lazio) was a cross-sectional, population-based study to evaluate the prevalence of asymptomatic LV dysfunction and HF in elderly subjects, as previously described [14]. A random sample of 5940 residents (aged 65–84 years) from five cities (Rome, Civitavecchia, Frosinone, Ostia and Viterbo) in the Lazio region of Italy (about 5.5 million inhabitants) was identified based on data from the regional health registry. Participants were referred to eight cardiology centres in Lazio for a clinical examination, blood tests, electrocardiography, comprehensive Doppler echocardiography and blood sampling to measure circulating biomarkers. A complete list of participating centres and investigators has been published [14]. Approval for this study was obtained from the local ethics committee.

Echocardiography and outcome

Colour Doppler echocardiography was performed in peripheral participating centres using commercially available machines, according to a predefined acquisition protocol [14]. Echocardiograms were recorded using standard DICOM format on digital supports and sent to a central laboratory (European Imaging Laboratory, Rome) for reading. Linear measurements of cardiac chambers were obtained either from the two-dimensional parasternal long axis view or, when available, from the M-mode parasternal short axis recording according to the recommendations of the American Society of Echocardiography (ASE) [15] and from the apical four-chamber view. When linear LV measures were not available due to inadequate views or because of wall motion abnormalities, LV volumes were obtained only from the apical four-chamber view and the ejection fraction (EF) was calculated by the modified Simpson's rule. LV systolic function was calculated at either the endocardial level (EF) or the midwall level (MFS) using a previously reported modified ellipsoidal model [14, 16]. LV mass was calculated according to the recommendations of the ASE and the European Association of Echocardiography (EAE) [17]. LV hypertrophy was defined as LV mass/body surface area (BSA) >95 g m−2 in women and >115 g m−2 in men. Four categories of LV mass were identified as normal (43–95 and 49–115 g m−2 for women and men respectively), and slightly (96–108 and 116–131 g m−2 respectively), moderately (109–121 and 132–148 g m−2 respectively) and severely abnormal (≥122 and ≥149 g m−2 respectively) [17]. LV midwall dysfunction was defined as MFS <15%; this cut-off level is commonly used in studies of high-risk patients [18] and has demonstrated prognostic relevance in hypertensive subjects [19].

Diastolic dysfunction was defined as mild, moderate or severe using a Doppler-derived multiparametric algorithm including Doppler-derived indexes of transmitral flow and pulmonary vein flow, and tissue Doppler imaging of the lateral mitral annulus (E/e′) as an indirect estimate of increased LV filling pressures [20], as previously described [14, 21]. Stages of HF were defined according to the criteria of the American College of Cardiology Foundation/American Heart Association (ACC/AHA) [22].

Circulating biomarkers

Fasting venous blood samples were collected in tubes containing ethylenediamine tetraacetic acid tripotassium salt, with subjects resting in the supine position for at least 15 min. Blood was centrifuged at 4 °C within 10 min and aliquots of plasma were immediately frozen and subsequently shipped on dry ice to a central laboratory. Samples were stored at −70 °C until required for assay. Plasma concentrations of NT-proBNP, hs-cTnT, hsCRP and cystatin C were measured in a central laboratory by electrochemiluminescence immunoassay using commercial reagents (Elecsys 2010, Roche Diagnostics GmbH), by staff blinded to the identity of the subjects. hs-cTnT was measured within the range 3–10 000 ng L−1; the 99th percentile cut-off from a healthy reference population was 14 pg mL−1 [23].

Statistical analyses

Sociodemographic and clinical characteristics were compared between subjects with or without abnormal LV mass, and with or without reduced midwall dysfunction. Differences between subjects were compared using the t-test for normally distributed continuous variables, Mann–Whitney test for non-normal continuous variables and chi-square test for categorical variables. Linear regression models were performed to assess the association between circulating biomarker concentrations and LV mass/BSA, with adjustment for age, gender, body mass index (BMI), hypertension, diabetes, atrial fibrillation and levels of serum cholesterol and creatinine. Circulating biomarkers were considered as log-transformed continuous variables (except cystatin C). The ability of biomarkers to correctly identify individuals with elevated LV mass or abnormal MFS was assessed by receiver-operating characteristic (ROC) curves. SPSS (release 18.0) and STATA (release 10) software were used for statistical analyses.


Circulating biomarkers and LV hypertrophy

As expected, median [quartile 1–3] circulating biomarker concentrations in this general population sample were very low: hs-cTnT, 5.5 [3.0–9.6] ng L−1; NT-proBNP, 92 [48–185] ng L−1. The hs-cTnT concentration was higher than 14 ng L−1 (the 99th percentile in apparently healthy individuals) in 12.7% of subjects, but particularly in men (18.1% vs. 7.1% in women). The hs-cTnT concentration was below the detection limit (3 ng L−1) in 30.1% of subjects (41.7% of women and 19.1% of men). Furthermore, hsCRP concentration was above 10 ng L−1 in 8.0% of subjects (9.4% of women and 7.1% of men), suggesting the presence of an active inflammatory processes. Clinical characteristics according to hs-cTnT concentration are shown in Table S1.

Table 1 shows the clinical characteristics of the 1604 participants with normal or elevated LV mass. Overall, 24.8% of subjects had LV hypertrophy, and 28.6% had previous cardiovascular disease (6.1% myocardial infarction and 7.6% atrial fibrillation). Subjects with elevated LV mass were more likely to be women, with higher blood pressure and BMI than those with normal LV mass (Table 1). Hypertension, previous myocardial infarction, atrial fibrillation, HF and LV systolic or diastolic dysfunction were more common in those with LV hypertrophy. Lipid profiles and levels of blood glucose and serum creatinine were similar in the two groups. Plasma concentrations of the two cardiac markers hs-cTnT and NT-proBNP, but not of hsCRP or cystatin C, were significantly higher in individuals with elevated LV mass (Table 1). There were progressive rises in the median plasma concentrations of hs-cTnT and NT-proBNP with increasing severity of LV hypertrophy in the entire population (Fig. 1, P < 0.0001 for trend for both markers). In univariate analysis, hs-cTnT was linearly correlated with LV mass/BSA in all subjects, irrespective of ACC/AHA stage of HF (Table S2), but only in those with stages B and C after adjustment for covariates.

Table 1. Clinical characteristics and presence of LV hypertrophy and reduced MFS
 LV massMFS
NormalElevated P NormalReduced P
  1. SBP, systolic blood pressure; DBP, diastolic blood pressure; LBBB, left bundle branch block; RBBB, right bundle branch block; LVFS, left ventricular fractional shortening.

  2. Elevated LV mass was defined as LV mass/BSA > 95 g m−2 for women and 115 g m−2 for men and reduced MFS was defined as <15%. Systolic dysfunction was defined as LVEF < 50%. Diastolic function was defined as normal if at least three of the following conditions were satisfied: E/A ratio > 0.75 and <1.5; deceleration time of E wave > 140 ms and <280 ms; peak velocity of pulmonary venous (PV) systolic (S) and diastolic (D) flow such as PV peak S > PV peak D, the peak velocity of PV backward flow at atrial contraction (PVa) and the PVa wave duration (PVa dur) such as PVa dur-Adur difference <0, and the ratio of early mitral inflow to early mitral annular velocity E/e′ < 8 (Reference [14]). Categories were compared using the t-test for normally distributed continuous variables, Mann–Whitney test for non normal continuous variables and chi-square test for categorical variables.

n (%)1207 (75.2)397 (24.8) 1113 (69.6)485 (30.4) 
Age (year, mean±SD)72.8 ± 4.972.4 ± 5.0<0.000172.2 ± 4.873.7 ± 5.2<0.0001
Female (%)46.663.2<0.000151.847.20.10
SBP (mmHg, mean ± SD)136.6 ± 17.2142.7 ± 17.0<0.0001136.6 ± 16.9161.7 ± 17.9<0.0001
DBP (mmHg, mean±SD)79.9 ± 9.482.1 ± 9.1<0.000180.0 ± 9.481.6 ± 9.30.0001
Heart rate (beats/min, mean ± SD)70.7 ± 11.370.5 ± 12.50.7969.9 ± 11.272.2 ± 12.4<0.0001
BMI (kg m−2, mean±SD)25.9 ± 3.926.9 ± 4.1<0.000126.0 ± 4.026.4 ± 3.80.09
BMI ≥ 30 kg m−2 (%)13.721.4<0.000115.416.10.72
Comorbidities (%)  
Chronic pulmonary disease7.28.60.387.28.90.25
Cardiovascular disease25.538.5<0.000125.936.1<0.0001
Previous MI4.78.10.0113.311.3<0.0001
Atrial fibrillation6.412.6<0.00016.212.0<0.0001
ACC/AHA stage (%)  
A23.60 25.23.3 
B55.387.4 55.879.8 
C4.112.6 2.215.7 
ECG findings  
LV hypertrophy (%)4.316.2<0.00016.010.3<0.0001
LBBB (%)1.76.8<0.00011.27.2<0.0001
RBBB (%)
LV mass/BSA (g m−2)82.6 ± 14.4120.9 ± 21.2<0.000187.1 ± 19.8104.1 ± 26.1<0.0001
LV diastolic diameter/BSA (cm m−2)2.52 ± 0.252.75 ± 0.29<0.00012.60 ± 0.252.51 ± 0.30<0.0001
Relative wall thickness0.40 ± 0.070.44 ± 0.08<0.00010.39 ± 0.060.47 ± 0.08<0.0001
Left atrial area/BSA (mm2 m−2)18.2 ± 4.421.1 ± 5.5<0.0118.7 ± 4.519.6 ± 5.60.006
LVEF (%)67.0 ± 6.363.6 ± 9.6<0.000168.6 ± 4.860.4 ± 9.1<0.0001
HF (%)4.112.6<0.00012.215.7<0.0001
Systolic dysfunction (%)1.77.6<0.0001010.7<0.0001
Diastolic dysfunction (%)36.558.7<0.000130.479.6<0.0001
Laboratory values (mean ± SD) 
Total cholesterol (mg dL−1)205.1 ± 39.0202.2 ± 40.90.21205.6 ± 38.2202.8 ± 42.50.03
Triglycerides (mg dL−1)118.4 ± 58.7121.8 ± 61.30.33114.4 ± 52.3128.8 ± 71.3<0.0001
Serum creatinine (mg dL−1)0.9 ± 0.20.9 ± 0.30.9450.9 ± 0.31.0 ± 0.30.003
Blood glucose (mg dL−1)103.5 ± 29.5105.5 ± 26.90.245101.6 ± 24.9110.0 ± 36.4<0.0001
Biomarkers (median [Q1–Q3])  
hs-cTnT (ng L−1)4.6 [3.0–8.1]6.6 [3.5–11.6]<0.00014.5 [3.0–8.1]6.5 [3.3–11.1]<0.0001
NT-proBNP (ng L−1)79 [41–151]147 [64–296]<0.000185 [44–162]107 [53–256]<0.0001
hsCRP (mg L−1)1.7 [0.9–3.7]1.6 [0.8–3.2]0.0271.7 [0.9–3.6]1.8 [0.8–3.7]0.84
Cystatin C (mg L−1)1.1 [1.0–1.3]1.1 [1.0–1.3]0.0841.1 [1.0–1.3]1.1 [1.0–1.3]0.13
Figure 1.

Relation between circulating cardiac biomarker concentrations and LV mass. Median circulating concentrations of hs-cTnT (upper panel) or NT-proBNP (lower panel) in relation to categories of LV mass (LV mass/BSA, g m−2) in the whole sample, or separately in men and women. The number of subjects in each category is shown. P < 0.0001 across categories for all, except NT-proBNP in women (P = 0.003), by Kruskal-Wallis nonparametric test.

The hs-cTnT levels were higher in men than in women, and in subjects with normal or increased LV mass, at equivalent grades of LV hypertrophy. For both genders, hs-cTnT and NT-proBNP rose significantly with the grade of abnormal LV mass, although more markedly in men (Fig. 1). Circulating concentrations of hsCRP or cystatin C were not related to the grade of LV hypertrophy in the whole sample, or after stratification by gender (data not shown).

Similar investigations were carried out for sensitivity analysis in a subgroup of 686 normotensive subjects with a median hs-cTnT concentration of 4.4 [3.0–7.5] ng L−1. hs-cTnT increased progressively in subjects with normal LV mass (= 571, 83.2%; 3.0 [3.0–5.7] ng L−1), and slightly (= 68, 9.9%; 3.7 [3.0–6.3] ng L−1), moderately (= 29, 4.2%; 3.7 [3.0–5.4]) or severely abnormal LV mass (= 18, 2.6%; 5.4 [3.3–12.6] ng L−1, < 0.0001 across categories). The hs-cTnT was also associated with LV mass in normotensive subjects without a known history of diabetes, chronic obstructive pulmonary disease, myocardial infarction or atrial fibrillation (data not shown).

Circulating biomarkers, HF and LV hypertrophy

Participants were further stratified according to ACC/AHA stages of HF and presence of LV hypertrophy. There was a graded increase in plasma concentrations of hs-cTnT in subjects with normal LV mass across different stages of HF (Fig. 2, Table S3). Differences between subjects with normal LV mass and those in stages A or B HF were more marked for hs-cTnT than for NT-proBNP, particularly amongst men.

Figure 2.

Relations between circulating cardiac biomarker concentrations and both ACC/AHA stage and normalized LV mass. Subjects were classified according to ACC/AHA stage and normal (LV mass/BSA: women 43–95, men 49–115 g m−2) or abnormal normalized LV mass. Median biomarker concentrations are shown. P-values refer to nonparametric comparison of concentrations in subjects with normal (coloured bars) or elevated (black bars) normalized LV mass in each ACC/AHA stage of HF. *P < 0.05; **P < 0.001.

Subjects in ACC/AHA stages B or C HF with LV hypertrophy had significantly higher circulating concentrations of hs-cTnT and NT-proBNP than those with normal LV mass (Fig. 2, Table S3). The difference was more pronounced for NT-proBNP amongst male subjects in stages B (+114%) or C (+144%) HF.

Circulating biomarkers and MFS

To further investigate the ability of cardiac markers to detect early alterations in LV phenotype in this elderly population, subjects were divided into two groups based on normal (≥15%) or reduced (<15%) MFS; data from 1598 participants were available for this analysis. Table 1 shows the clinical characteristics of these two subgroups of subjects, 485 (30.4%) of whom had below normal MFS. These subjects were older, with higher blood pressure and a higher prevalence of diabetes, hypertension, previous myocardial infarction, atrial fibrillation, HF and LV systolic or diastolic dysfunction than those with normal MFS. In addition, circulating levels of triglycerides, glucose, hs-cTnT and NT-proBNP were higher.

Next subjects were divided according to quartiles of MFS (Table S4). hs-cTnT concentrations were significantly higher in men and women with reduced MFS (< 0.0001 for both genders, Fig. 3) whereas NT-proBNP was higher in men (< 0.0001), but not women (= 0.11). hs-cTnT and NT-proBNP concentrations were significantly elevated in normotensive and hypertensive individuals with below normal MFS (data not shown). Concentrations of hsCRP or cystatin C were not different between the two groups.

Figure 3.

Relation between circulating cardiac biomarker concentrations and MFS. Subjects were divided into two groups with abnormal (<15%, grey bars) or normal (≥15%, white bars) MFS. Median concentrations of hs-cTnT and NT-proBNP are shown. P-values refer to nonparametric comparison of concentrations. The number of subjects is indicated in each bar.

The proportion of subjects with reduced MFS (<15%) rose progressively with the level of hs-cTnT, to 29.9% of men and 32.6% of women in the highest tertile (P for trend = 0.001 for men, <0.0001 for women, Table 2). Higher NT-proBNP was associated with subnormal MFS in men (< 0.0001), but not in women (= 0.22).

Table 2. Elevated LV mass or reduced MFS according to levels of hs-cTnT or NT-proBNP
  Elevated LV mass/BSAReduced MFS
  1. Percentage of subjects with abnormal LV mass/BSA (>95 g m−2 for women and 115 g/ m−2 for men) or MFS (<15%) in relation to concentration of hs-cTnT and NT-proBNP. Participants were divided into four categories by quartiles of NT-proBNP concentration. Alternatively, participants were divided into four categories for hs-cTnT: first those with a concentration below the limit of the assay blank (LOB, <3 ng L−1), and then the remainder into tertiles of measured concentration. P-values indicate trend across categories.

hs-cTnT (ng L−1)<3 (LOB)1933986.
 3.0–5.7 (T1)2242348.929.120.523.5
 5.8–9.8 (T2)27818715.529.927.027.3
 9.9–142.4 (T3)33414421.340.329.932.6
 P for trend  <0.0001<0.00010.001<0.0001
NT-proBNP (ng L−1)5–47 (Q1)3091937.119.220.421.2
 48–92 (Q2)2572498.924.120.624.9
 93–185 (Q3)21327712.723.520.220.2
 186–6702 (Q4)25324529.235.938.328.2
 P for trend  <0.0001<0.0001<0.00010.22

Circulating biomarkers and isolated LV diastolic dysfunction

The hs-cTnT concentration increased stepwise in subjects with mild or moderate-to-severe isolated LV diastolic dysfunction compared to those with normal diastolic function (< 0.0001, Table 3), in the whole population and after stratification by gender. NT-proBNP concentration was significantly increased in subjects with isolated LV diastolic dysfunction.

Table 3. Circulating biomarkers according to severity of isolated LV diastolic dysfunction
 NormalDiastolic dysfunction
 MildModerate–severe P
  1. Plasma concentrations of circulating biomarkers (median [Q1–Q3]) in subjects with mild or moderate-to-severe isolated LV diastolic dysfunction. P for comparison by nonparametric Mann–Whitney test.

All subjects
hs-cTnT (ng L−1)4.6 [3–7.9]5.8 [3–9.8]6.4 [3.1–12.4]<0.0001
NT-proBNP (ng L−1)78 [41–154]89 [49–162]194 [97–532]<0.0001
hsCRP (mg L−1)1.7 [1.8–3.6]1.5 [0.8–3.4]2.2 [1.6–4.7]0.01
Cystatin C (mg L−1)1.1 [1.0–1.3]1.1 [1.0–1.3]1.1 [1.0–1.3]0.11
hs-cTnT (ng L−1)5.9 [3.1–9.6]7.2 [4.6–11.3]7.7 [4.2–15.8]<0.0001
NT-proBNP (ng L−1)64 [33–127]80 [43–162]211 [85–612]<0.0001
hsCRP (mg L−1)1.7 [0.8–3.5]1.6 [0.8–3.3]1.8 [1.0–4.9]0.19
Cystatin C (mg L−1)1.1 [1.0–1.3]1.1 [1.0–1.3]1.1 [1.0–1.3]0.51
hs-cTnT (ng L−1)3.3 [3.0–6.3]4.1 [3.0–7.9]5.0 [3.0–9.2]0.001
NT-proBNP (ngL−1)98 [51–179]94 [55–162]172 [104–358]<0.0001
hsCRP (mg L−1)1.7 [0.8–3.7]1.5 [0.8–3.6]2.5 [1.1–4.7]0.06
Cystatin C (mg L−1)1.1 [1.0–1.3]1.1 [1.0–1.3]1.1 [1.0–1.3]0.16

Diagnostic accuracy of circulating biomarkers to detect cardiac phenotype

The ability of hs-cTnT and NT-proBNP to correctly identify individuals with elevated LV mass and/or abnormal MFS was compared in the whole cohort and stratified by gender (Table 4). Overall, we found that the areas under the ROC curves (AUCs) were between 0.54 and 0.71, with no significant differences between the two markers (except for a slight superiority of hs-cTnT compared with NT-proBNP for detecting abnormal MFS in women only, = 0.02). The diagnostic accuracy of cardiac biomarkers to detect isolated LV diastolic dysfunction was poor.

Table 4. Diagnostic accuracy of detection of cardiac phenotype alterations in the total population and by gender
 n (%)hs-cTnTNT-proBNPP (hs-cTnT vs. NT-proBNP)
AUC [95% CI]Optimal cut-offNPV (%)PPV (%)AUC [95% CI]Optimal cut-offNPV (%)PPV (%)
  1. The accuracy of hs-cTnT or NT-proBNP in detecting elevated LV mass (LV mass/BSA > 95 g m−2 for women and 115 g m−2 for men), abnormal LV MFS (<15%) or isolated diastolic dysfunction was evaluated by receiver operating characteristic (ROC) curves. The area under the ROC curve (AUC) was compared for the two markers using the method of Hanley and McNeil. NPV, negative predictive value; PPV, positive predictive value.

Elevated LV mass
All396 (24.8)0.62 [0.59–0.65]5.981.232.50.66 [0.62–0.69]12982.237.20.06
Men146 (18.5)0.69 [0.64–0.73]7.389.628.70.71 [0.66–0.76]12989.434.40.32
Women250 (30.9)0.64 [0.60–0.68]3.678.940.80.60 [0.56–0.65]12974.539.40.13
Abnormal MFS
All484 (30.4)0.61 [0.58–0.64]5.677.038.80.58 [0.55–0.61]14574.139.60.12
Men256 (32.3)0.61 [0.57–0.65]6.775.140.00.62 [0.58–0.66]9274.440.60.68
Women228 (28.4)0.60 [0.55–0.64] [0.49–0.58]14574.132.90.02
Elevated LV mass or abnormal MFS
All670 (44.6)0.61 [0.58–0.63]5.664.153.40.61 [0.58–0.63]13061.455.20.76
Men312 (39.7)0.63 [0.59–0.67]5.972.849.40.64 [0.60–0.68]8755.155.40.76
Women358 (44.6)0.61 [0.58–0.65]3.565.654.50.56 [0.52–0.60]14459.752.50.01
Isolated diastolic dysfunction
All780 (43.6)0.56 [0.53–0.59]4.959.150.70.54 [0.51–0.58] 9057.449.50.22
Men390 (42.7)0.57 [0.53–0.61]6.462.747.60.56 [0.52–0.59]7364.148.70.64
Women390 (44.5)0.56 [0.52–0.60]3.060.848.30.51 [0.48–0.55]9156.945.70.04


We conducted a cross-sectional epidemiological study including almost 2000 community-dwelling elderly persons (65–84 years) living in the region of Rome, Italy. We found, first, that circulating levels of hs-cTnT were generally very low (median 5.5 ng L−1), and troponin levels were below the detection limit in a substantial proportion of the population (18% of men and 41% of women). Secondly, despite this very low concentration, hs-cTnT increased gradually in relation to small alterations in LV mass, MFS or isolated LV diastolic dysfunction. Thirdly, the differences in concentration were more marked for hs-cTnT than for NT-proBNP in women. Fourthly, the diagnostic performance of the two cardiac biomarkers for identifying subjects with elevated LV mass, abnormal MFS or isolated diastolic dysfunction was comparable but poor. Finally, circulating levels of markers of systemic inflammation (hsCRP) or renal function (cystatin C) were not related to alterations of cardiac phenotype.

Recent studies in the general population have reported very low levels of circulating cardiac troponins, measured with high-sensitivity methods. The proportion of subjects with detectable troponin levels varies widely according to the demographic characteristics of the population (e.g. mean age or ratio of males/females) and assay performance: 19% in the DETECT study [8], 25% in the Dallas Heart Study [6], 66% in the Atherosclerosis Risk in Communities (ARIC) study [7] and 81% in a study of middle-aged Japanese men [5]. In the present sample of community-dwelling elderly individuals, the high proportion with detectable hs-cTnT levels (70%) is probably explained by their advanced age; the concentration of hs-cTnT was above 14 ng L−1, which is currently considered the upper limit of normal for the assay (99th percentile) [23, 24], in about 13% of subjects.

In this randomly sampled general population, LV mass index was elevated, according to the ASE/EAE criteria [17], in 24.8% whilst MFS was abnormal in 30.4%. It has previously been reported that the prevalence of LV systolic dysfunction (EF < 50%) was 5.1% in men and 1.4% in women in this study cohort [14]. Corresponding figures for mild LV diastolic dysfunction were 37.0% and 38.7%. Whereas the prevalence of elevated echocardiographic LV mass and its association with cardiovascular events has been well described in the general population [25-27], data regarding MFS have mainly originated from studies of a few hundred hypertensive patients [16, 28, 29]. We report here for the first time the prevalence of abnormal MFS in a cohort of elderly individuals randomly selected from the general population of a European country; therefore it is difficult to compare these data with findings from large epidemiological studies including thousands of individuals. However, the prevalence of abnormal MFS in this study is close to that observed in a recent study [30] in which LV systolic dysfunction was detectable at the midwall level in one third of diabetic patients without clinically detectable cardiac disease.

In the PREDICTOR study, even minimal alterations in cardiac phenotype or in the ACC/AHA stage of HF were associated with graded increases in circulating troponin levels. The increase in hs-cTnT was also observed in normotensive patients with or without comorbidities. Previous studies have demonstrated a relationship between troponin levels (measured with highly sensitive reagents) and cardiac structure in the general population. In the Dallas Heart Study (3546 individuals aged 30–65 years), increasing hs-cTnT was significantly associated with greater LV hypertrophy (increased LV mass and wall thickness) and worse systolic function [6]. Similarly, the proportion of subjects with LV hypertrophy in the ARIC study (9698 participants) rose progressively with increasing troponin levels [7].

Our findings confirm and extend the results of these previous studies in several ways. First, we demonstrated the association between troponin levels and cardiac phenotypes across clinically relevant categories of LV mass, based on ASE and AEA recommendations, or the ACC/AHA stages of HF development. Second, we stratified the population by gender and verified the relationship between troponin levels and cardiac phenotype in men and women, although absolute levels of circulating troponins and LV mass were lower in women [31]. In a previous study in the general population, a relation between plasma cardiac troponin levels and normalized LV mass was demonstrated in men only, with a nonsignificant trend in women [9]. However, this could be due to the fact that troponin was measured with a traditional assay (sensitivity 10 ng L−1), and circulating troponin levels are lower in women. These observations are consistent with the findings of a population-based study of 1883 individuals in whom LV dysfunction was associated with increases in LV mass and natriuretic peptides in men, but only nonsignificant trends were observed in women [32].

A third feature of our study is that we compared hs-cTnT and NT-proBNP in terms of their associations with cardiac phenotype. We found no major differences between the two cardiac markers in men but, in women, stepwise changes in the cardiac markers with minimal deteriorations in LV structure or function were more pronounced for hs-cTnT than for NT-proBNP. The poor discriminative performance of natriuretic peptides (brain natriuretic peptide or NT-proBNP) within a range of slightly abnormal LV mass has already been reported in elderly subjects [33]. In univariate analyses, hs-cTnT was significantly correlated with LV mass in normal subjects or in those in ACC/AHA stage A HF, whereas NT-proBNP was not. This difference was observed despite the wider dynamic range of concentrations of NT-proBNP (5 to ~15 000 ng L−1) than that of hs-cTnT (3–142 ng L−1) and the higher absolute concentrations of the natriuretic peptide in women compared to men: 103 [55–188] vs. 80 [41–181] ng L−1 (< 0.0001).

A strength of this study is the investigation of the association between two cardiac biomarkers and careful echocardiographic evaluation of MFS, a preclinical stage during the progression to LV hypertrophy and HF of ischaemic and nonischaemic aetiology [18, 34]. In this elderly population, we found no significant relations between markers of systemic inflammation (hsCRP) or abnormal renal function (cystatin C) and cardiac phenotype (LV mass, MFS and isolated diastolic dysfunction) or HF severity.

In our population of elderly individuals, hs-cTnT and NT-proBNP detected elevated LV mass, abnormal MFS or isolated diastolic dysfunction with poor diagnostic accuracy (AUC < 0.7). Although direct comparisons are not always warranted because of a degree of heterogeneity in the population characteristics or in the different anatomical and functional parameters examined, natriuretic peptides are generally better able to detect reduced LVEF [35-38]. Therefore, the ability of circulating cardiac biomarkers to detect abnormal MFS remains to be determined in other elderly populations.


This epidemiological study has several limitations. First, because of the cross-sectional design, causality between circulating biomarkers and cardiac phenotype cannot be inferred. Secondly, the study was limited to subjects within a preset age range (65–84 years), thus the findings may not be valid outside this range. Thirdly, the circulating biomarker levels were measured only once so, because of inherent biological variability, particularly for natriuretic peptides, they may not truly reflect the homeostatic state. Fourthly, although we used a careful protocol for centralized reading of echocardiograms by two independent observers blind to the clinical data, systematic errors cannot be completely excluded. Finally, the analytical accuracy of hs-cTnT in the very low range of concentrations (ng L−1) may be another potential source of error, although less important than the variations in echocardiographic measurements.


Very low circulating levels of troponin measured with a highly sensitive assay may reveal subclinical chronic myocardial abnormalities in elderly community-dwelling individuals, and help identify an intermediate phenotype in the pathway to HF.

Conflict of interests and disclosures

SM and RL have received institutional research support from Roche Diagnostics. U-HW-T, DB and CZ are employees of Roche Diagnostics. Reagents for measuring high-sensitivity cardiac troponin T, high-sensitivity C-reactive protein and cystatin C were kindly donated by Roche Diagnostics, Rotkreuz, Switzerland. There are no other potential conflicts of interest relevant to this article.

Funding sources

The PREDICTOR study was funded by Takeda Italia Farmaceutici (TIF) which had no role in the analysis or interpretation of the data or the writing of the report.