Objective To understand the determinants of troponin release in patients with stable coronary artery disease (CAD) by comparing high sensitive troponin T (hsTnT) levels with computed tomography angiography (CTA) characteristics of atherosclerotic plaque.
Methods hsTnT was determined in 124 consecutive patients with stable angina, who underwent clinically indicated 256-slice CTA for suspected CAD. CTA was used to assess (1) coronary calcification; (2) stenosis severity; (3) non-calcified plaque volume; (4) plaque composition (soft or mixed, described as ‘non-calcified’ versus calcified) and (5) the presence of vascular remodeling in areas of non-calcified plaque.
Results All CT scans were performed without adverse events, and diagnostic image quality was achieved in 1830/1848 available coronary segments (99.0%). In 29/124 patients, hsTnT was ≥14 pg/ml (range 14.0–34.4). Weak, albeit significant, correlations were found between hsTnT and calcium scoring (r=0.45, p<0.001), while a stronger correlation was found between hsTnT and the total non-calcified plaque burden (r=0.79, p<0.001). Patients with non-calcified plaque (n=44) yielded significantly higher hsTnT values than those with normal vessels (n=46) or those with only calcified lesions (n=26), (12.6±5.2 vs 8.3±2.6 and 8.8±3.0 pg/ml, respectively, p<0.001). Furthermore, those with remodeled non-calcified plaque (n=8) showed even higher hsTnT values of 26.3±6.5 pg/ml than all other groups (p<0.001). This allowed the identification of patients with remodeled non-calcified plaque by hsTnT with high accuracy (area under the curve=0.90, SE=0.07, 95% CI 0.84 to 0.95).
Conclusions Chronic clinically silent rupture of non-calcified plaque with subsequent microembolisation may be a potential source of troponin elevation. In light of recent imaging studies, in which patients with positively remodeled non-calcified plaque were shown to be at high risk for developing acute coronary syndromes, hsTnT may serve as a biomarker for such ‘vulnerable’ coronary lesions even in presumably stable CAD.
- High sensitive troponin T, atherosclerotic plaque
- plaque composition, vascular remodeling
- cardiac CT
- CT scanning
Statistics from Altmetric.com
- High sensitive troponin T, atherosclerotic plaque
- plaque composition, vascular remodeling
- cardiac CT
- CT scanning
Despite advances in the medical and interventional treatment of coronary artery disease (CAD), it remains the leading underlying cause of myocardial infarction and sudden cardiac death.1 2 In patients with acute coronary syndromes (ACS), the rupture of coronary plaques with initiation of thrombus formation and potential embolisation of atherosclerotic debris results in myocardial cell necrosis. In these patients troponin leakage leads to detectable levels of troponin in blood, aiding their diagnostic classification and risk stratification.3–5
Minor increases of troponins (below the 99th centile reference value of currently established assays), however, are detectable in ACS but have also been reported in stabilised patients after ACS,6 in congestive heart failure,7 in presumably healthy elderly subjects8–10 and in patients with preserved left ventricular (LV) function and stable CAD.11 Although minimally increased troponin levels are considered as a surrogate marker for highly probable irreversible injury, the precise causes of this phenomenon in non-ACS patients remain unclear.9 12 13 However, evidence from angioscopic studies demonstrates that plaque rupture and thrombosis are present in ∼20% of patients with stable CAD,14 and it is conceivable that coronary plaque rupture, with subsequent microinfarctions,15 16 may be a potential source of chronic troponin leakage.
To date, 256-slice computed tomography angiography (CTA) with submillimetre spatial and high temporal resolution allows for the non-invasive assessment of (1) coronary calcification; (2) stenosis severity; (3) atherosclerotic plaque size; (4) plaque composition and (5) vascular remodeling. In this study, we sought to understand the determinants of troponin release in patients with stable CAD by comparing high sensitive troponin T (hsTnT) levels with CTA characteristics of atherosclerotic plaque. Furthermore, in order to investigate the potential role of inflammation in patients with hsTnT elevation, high sensitive C-reactive protein (hsCRP), a sensitive biomarker of systemic low-grade inflammation, was measured.
The study population consisted of 180 consecutive patients scheduled to undergo clinically indicated cardiac CTA. Exclusion criteria were non-sinus rhythm (n=6), ACS (n=12), significant valvular disease (n=13), elevated serum creatinine (>1.5 mg/dl) (n=6), known CAD (n=9), history or ECG signs of previous infarction (n=8), or a contraindication to the administration of contrast agent (n=2). All patients underwent two-dimensional echocardiography before enrolment and patients with impaired systolic ejection fraction (<55%) or presence of regional wall motion abnormalities (n=4) were also excluded from analysis. Thus, our final patient population comprised124 patients with preserved LV function, who underwent clinically indicated CTA for suspected CAD.
Traditional risk factors for CAD, including arterial hypertension (blood pressure≥140/90 mm Hg or antihypertensive therapy), hyperlipidaemia (low-density lipoprotein cholesterol (LDL-C) ≥130 mg/dl or statin therapy), current or prior smoking, diabetes mellitus and a family history of CAD were recorded at the time of the CT scans. Furthermore, the clinical presentation of chest pain (typical angina, atypical angina or non-cardiac chest pain)17 was registered, and the Duke Clinical Score, which incorporates chest pain characteristics, age, gender, and atherogenic risk factors,18 was calculated. Patients were categorised as a low (<30%), intermediate (30–70%), or high (>70%) pre-test probability group for CAD, as reported previously.19 In addition, chest pain was classified into four classes of severity according to the criteria of the Canadian Cardiovascular Society (CCS).20
The CTA protocol included the intravenous administration of incremental doses of 2.5 mg of metoprolol (range 2.5–25.0 mg), (Lopressor, Novartis, Pharma GmbH, Nuernberg, Germany) starting 10–20 min before CTA in patients with heart rates ≥65 beats/min. If the heart rate remained ≥65 beats/min despite the administration of metoprolol, a retrospective scan was performed. If the heart rate decreased to <65 beats/min, prospective CTA scans were acquired. Furthermore, sublingual glyceryl nitrate was administrated before CTA for coronary vasodilatation in all patients. All procedures complied with the Declaration of Helsinki, were approved by our local ethic committee and all patients gave written informed consent.
256-Slice CT scanning technique
CT scans were performed using a 256-slice Brilliance iCT scanner (Philips Healthcare Best, The Netherlands) that features a gantry rotation time of 270 ms, resulting in a temporal resolution of 36–135 ms, depending on the heart rate of the patient and the reconstruction mode, and an isotropic submillimetre spatial resolution of 0.67×0.67×0.67 mm3.
Coronary calcium scoring
For the assessment of coronary calcification prospective ECG-gated non-contrast scans were performed at 75% of the cardiac cycle, using 120 kV tube voltage and 80 mA tube current, and resultant images with a 3 mm slice thickness were used for the calculation of the Agatston score.
CT angiography and estimation of the radiation dosage
For CTA a bolus of 80 ml of contrast agent (Ultravist 370, Bayer Healthcare, Berlin, Germany) was injected intravenously (6 ml/s). As soon as the signal in the descending aorta reached a predefined threshold of 100 HU, the scan started automatically and the entire volume of the heart was acquired during one breath-hold in 4–7 s with simultaneous ECG recording. The detector collimation was 2×128×0.625 mm, with 256 overlapping slices of 0.625 mm thickness and dynamic z-focal spot. The tube voltage was 120 kV and the gantry rotation time was 0.27 s. A current of 800–1050 mA (depending on patient habitus) was used for retrospective acquisitions and a current of 200 mA for prospective acquisitions. With retrospective acquisitions reconstructions were routinely performed at 40%, 70%, 75% and 80% of the cardiac cycle. With prospective acquisitions reconstructions were available at 75% of the cardiac cycle. The effective dose was calculated for all CTA scans, based on the dose–length product and an organ weighting factor for the chest as the investigated anatomical region (k=0.014 mSv×(mGy×cm)−1) averaged between male and female models.21
Assessment of stenosis severity and plaque volume and composition
CTA datasets were anonymised and were analysed in random order using commercially available software (Philips Extended Brilliance Workspace 4.0). The composition of atherosclerotic plaques was performed using the Plaque SW version 4.0.2, as described previously.22 Briefly, for each coronary artery the vessel lumen and wall were automatically registered, and after identification of each lesion the boundaries were manually edited if necessary. Subsequently, the identified plaques were marked, and the validity of the proposed lesion areas was evaluated in adjacent cross-sectional multiplanar reconstructed images of the coronaries. Care was taken to correctly discriminate between iodinated blood (300–600 HU) and calcified plaque, and Gaussian algorithms were used to distinguish between components of low to intermediate attenuation (0–150 HU) and calcified plaque components with higher attenuation values. By this model, components with different signal intensity within the plaque are separated, using a mixture Gaussian model, into a linear combination which includes 1–3 Gaussians curves. For each lesion the following were assessed: non-calcified plaque volume; coronary lumen narrowing; plaque composition; vascular remodeling.
Non-calcified plaque volume
The non-calcified plaque volume for each individual lesion and each patient was obtained by summing the individual volumes of soft or mixed plaques in all three coronary vessels.
Coronary lumen narrowing
Coronary lumen narrowing (maximum diameter reduction) was determined in curved multiplanar reformatted reconstructions by dividing the minimal diameter in the diseased segment through the diameter in the adjacent proximal disease-free section. If stenosis was present at the ostium of a coronary artery or a vessel bifurcation point, the distal or the proximal reference vessel point were used, respectively. When multiple lesions were present in a segment, the most severe lesion was considered.
Plaque composition was measured by percentage of calcified content. According to the calcium content, plaques were classified into (i) soft (calcium content <20%); (ii) mixed (calcium content 20–80%) and (iii) calcified (calcium content >80%). Soft and mixed plaques are expected to contain substantial amount of lipid cores or fibrotic tissues23 apart from calcified tissue, and will be referred to as ‘non-calcified’ plaques throughout our manuscript.
Vascular remodeling was defined as a change in the vessel diameter at the plaque site in comparison with the reference segment proximal to the lesion (reference diameter). Quantification of the vessel diameters in longitudinal reconstructions was used to assess the remodeling index (lesion diameter/reference diameter), which was considered positive when the diameter at the plaque site was ≥10% larger than that measured in the reference segment.23
Measurement of hsTnT and hsCRP
We recently developed and validated an hsTnT assay, which allows the detection of TnT at the 99th centile of a healthy reference cohort with <10% variability, further enhancing the accuracy of fourth-generation assays.24 Blood samples were collected from all patients within 2 h before the CTA scans, centrifuged and stored at −80°C until analysis. For hsTnT quantification an ELECSYS 2010 automated analyser was used (Roche Diagnostics, Mannheim, Germany). The diagnostic range of this assay is 3–10 000 pg/ml with an interassay coefficient of variation of 8% at 10 pg/ml and 2.5% at 100 pg/ml. The intra-assay coefficient of variation is 5% at 10 pg/ml and 1% at 100 pg/ml. Thus, based on a healthy reference population, an upper reference limit of 14 pg/ml (99th centile for TnT) is recommended.
hsCRP was quantified by nephelometry, using polystyrene bead-coupled antibodies (Siemens Healthcare Diagnostics, Eschborn, Germany).
Analysis was performed using commercially available software MedCalc9.3 (MedCalc software, Mariakerke, Belgium) and data are presented as mean±SD. The relation between Agatston score and total non-calcified plaque volume with hsTnT and hsCRP was assessed using linear regression analysis. Differences in hsTnT and hsCRP levels by stenosis severity and by plaque composition with or without vascular remodeling were assessed using analysis of variance with Bonferroni's adjustment for multiple comparisons. Receiver operating characteristics were used to assess the ability of hsTnT and hsCRP to predict CAD by stenosis severity and by plaque composition. Cox proportional-hazards models were used to estimate the risk of CAD risk factors, hsTnT and hsCRP for the presence of non-calcified and of remodeled plaque, and the calculated HRs with their corresponding 95% CIs. Furthermore, CTA findings for lumen narrowing and plaque composition were analysed in patient tertiles based on hsTnT values. Intra- and interobserver variability for quantification of (1) plaque volume; (2) plaque subtype categorisation; (3) vascular remodeling and (4) coronary lumen narrowing were calculated by repeated analysis of 40 randomly selected cases. Differences were considered statistically significant at p<0.05.
Clinical characteristics and CTA results
Clinical and demographic data of our 124 patients including clinical presentation of chest pain and CCS classifications are illustrated in table 1. Furthermore, the mean pre-test probability was 47±19% (range 4–86%), and the majority of our patients (81/124, 65%) had an intermediate pre-test probability for CAD by predefined criteria (online supplementary figure 1).
Forty-six patients had normal coronary arteries without plaque or lumen narrowing, 58 had <50% stenosis and 20 showed ≥50% lesions (nine with single-vessel, seven with two-vessel and four with three-vessel CAD). Of 78 patients with plaque, 26 patients showed only calcified lesions, 44 showed both non-calcified (soft or mixed) and calcified plaque and eight showed remodeled lesions (table 2). Overall 54 soft, 154 mixed and 160 calcified lesions were identified. Their mean volume, signal intensity and calcium content are summarised in supplementary table 1.
All CT scans were performed without adverse events, and diagnostic image quality was achieved in 1830/1848 available coronary segments (99.0%). In 88/124 patients (71%) prospective CTA scans were acquired (mean heart rate=59±6/min) with a mean effective dose of 3.1±0.4 mSv. In 36/124 patients (29%) with heart rates>65/min despite β-blocker administration (mean heart rate=69±9/min), retrospective scans were acquired, resulting in a mean effective dose of 13.4±1.7 mSv (p<0.001 for radiation exposure).
Association of hsTnT and hsCRP with calcified and non-calcified plaque
All patients yielded hsTnT levels <50 pg/ml, while in 29/78 patients with plaque (37%) hsTnT was ≥14 pg/ml (range 14.0–34.4). Moderate and close correlations were found between hsTnT and total calcium scoring and non-calcified plaque volume and, respectively (figure 1A,B). Weak associations on the other hand, were found between the latter parameters and hsCRP (figure 1C,D).
Relation of plaque characteristics and CAD severity to hsTnT and hsCRP levels
Patients with non-calcified and with remodeled plaque showed higher hsTnT levels than those with normal vessels or with only calcified plaque. Consequently, high accuracies were achieved for the differentiation of patients by plaque composition (figure 2A–C). Conversely, only patients with remodeled non-calcified plaque yielded increased hsCRP values (figure 2D–F). No associations were seen between hsTnT and clinical presentation of chest pain and chest pain severity. Furthermore, no associations were found between the number of the diseased vessels and coronary stenosis severity with hsTnT or hsCRP (figure 3).
CTA characteristics of plaque in patient tertiles based on their hsTnT values are presented in table 3. The corresponding sensitivity, specificity and predictive values of hsCRP and hsTnT for plaque composition, using cut-off values to provide an optimal trade-off between sensitivity and specificity and using the recommended cut-off value of hsTnT=14 pg/ml, are summarised in table 4.
Representative images of patients with (A) normal vessels; (B) calcified plaque; (C) non-calcified plaque and (D) remodeled non-calcified plaque and their corresponding hsTnT values are shown in figure 4.
Prediction of non-calcified plaque by baseline parameters, hsTnT and hsCRP
Using univariate analysis, age, arterial hypertension, the total number of atherogenic risk factors, hsCRP and hsTnT were associated with plaque composition. By multivariate analysis hsTnT was the only independent predictor of non-calcified plaque and of remodeled plaque (table 5).
Inter- and intraobserver variability was 13% and 8% for the assessment of total plaque burden, and 9% and 7% for coronary lumen narrowing measures, respectively. For differentiation between soft, mixed and calcified plaque (by predefined criteria) and for the assessment of vascular remodeling, agreement between observers was 92% (κ=0.86) and 93% (κ=0.82), respectively.
Our study demonstrates for the first time that hsTnT elevations are associated with CTA plaque characteristics in patients with stable CAD. hsTnT is increased above the 99th centile for cTnT in ∼35% of patients with stable disease and hsTnT elevation is not associated with stenosis severity but rather with plaque composition. Thus, patients with non-calcified coronary lesions and especially those with remodeled lesions exhibit increased hsTnT levels compared with patients with normal vessels and with those with pure coronary calcification. hsCRP, a marker of systemic low-grade inflammation, on the other hand, is weakly associated with plaque composition. Only patients with atherosclerotic lesions which have progressed (remodeled non-calcified plaque) exhibit increased concentrations of this marker in blood.
Cardiac troponin for the detection of myocardial cell injury
During more than two decades after our first development of the troponin T assay,25 cardiac troponins (T and I) have become the preferred biomarkers for the diagnosis of acute myocardial infarction. The incremental diagnostic value of troponin as compared with other enzymes is due to its cardio-specific isoform expression of troponin T and I, which allows for differentiation of cardiac and skeletal muscle injury and provides a high myocardium-to-blood concentration gradient. This is associated with a high signal-to-noise diagnostic ratio following troponin membrane leakage. Owing to their high sensitivity and specificity, troponins are critical for the detection of suspected myocardial injury in patients with inconclusive ECG findings. In many randomised trials testing treatment options in patients with non-ST-segment elevation ACS, increases in TnT were associated with a higher cardiovascular event rate than for troponin-negative patients, and the former derived a marked benefit from more aggressive medical and interventional treatment strategies.26 27 Despite the presence of a risk gradient depending on the TnT levels in blood, even patients with only minor TnT elevations close to the detection limits of the assays had significantly more events than marker-negative subjects. Driven by this observation, and in order to meet quality requirements of analytical and pre-analytical precision, we have recently improved the analytical sensitivity of fourth-generation TnT assays.24
Potential pathophysiological determinants of chronic troponin leakage in stable CAD
In patients with ACS, rupture of atherosclerotic plaque in epicardial arteries with subsequent occlusive thrombosis causes prolonged myocardial ischaemia. This results in irreversible damage of the cell membrane of the cardiomyocytes, causing degradation of the cell membrane and, consequently, the release of myofibril-bound cytosolic troponin complexes into the serum.28 Milder forms of plaque rupture with subsequent microembolisation of atherothrombotic burden into the coronary microcirculation have also been recognised previously.14–16 However, their frequency in patients with stable disease and the clinical relevance of this phenomenon remain unclear.
In our study, and using a highly sensitive troponin assay, increased hsTnT was observed in ∼35% of patients with coronary lesions, which is higher than that previously reported in similar cohorts, using older fourth-generation assays.29 Furthermore, hsTnT elevation was associated with calcium scoring (as previously described12), but in contrast to previous reports no association was found between hsTnT elevation and stenosis severity.30 Conversely, increased hsTnT levels were associated with the composition and size of the atherosclerotic burden. Thus, patients with non-calcified (soft or mixed) atherosclerotic lesions and higher total non-calcified plaque volume showed higher hsTnT levels than those with only calcified plaque or with normal coronaries. Furthermore, patients with remodeled non-calcified plaque yielded even higher hsTnT levels than all other groups.
In light of recent imaging studies, which elegantly demonstrated, that non-ACS patients with positively remodeled coronaries in areas of non-calcified plaque are at high risk for developing ACS,23 31 our current findings implicate chronic clinically silent coronary plaque rupture and microembolisation as a potential pathophysiological source of chronic troponin leakage in patients with coronary plaque and stable disease. Such microembolisation may occur in the absence of flow-limiting stenosis, so that the dissociation between troponin leakage and lumen narrowing or chest pain severity, as observed in our study is not surprising. Similarly, previous studies demonstrated that hsTnT elevation is rather a consequence of irreversible myocyte death than inducible myocardial ischaemia during exercise or pharmacological stress testing.32 On the other hand, increased cardiovascular risk in patients and in apparently healthy subjects with elevations of cardiac troponins,9 10 12 suggests that clinically silent plaque rupture causing repetitive microembolisation may precede the clinical manifestation of myocardial infarction or sudden cardiac death.16 This is in agreement with recent clinical findings that increased hsTnT concentrations are independently associated with the incidence of cardiovascular mortality in patients with stable CAD.11
Increased demand ischaemia, myocardial strain because of volume or pressure overload, and impaired cell membrane integrity due to systemic inflammatory response or apoptosis, however, have also been discussed as possible causes for chronic troponin leakage (reviewed by Jeremias and Gibson13). Although in our cohort clinical conditions such as sepsis, myocarditis, chemotherapy and congestive heart failure were not present clinically and preserved ejection fraction without regional wall motion abnormalities was documented by echocardiography in all patients, the possibility of a contribution of such variables to hsTnT elevation cannot be completely excluded.
hsCRP, a marker of systemic low grade inflammation, was weakly associated with plaque composition in our study. Thus, a weak association between hsCRP and coronary calcification was seen, which is in agreement with previous observations.33 34 Furthermore, in contrast to hsTnT, hsCRP was not an independent predictor of non-calcified plaque or of remodeled plaque after adjustment for age and conventional atherogenic factors. This is in line which previous studies, where this biomarker was found to be a weak independent predictor of cardiovascular disease with limited value for clinical risk profiling.35 36 On the other hand, patients with lesions which had progressed and vascular remodeling, who also had increased hsTnT levels, yielded higher hsCRP values than all other patient subgroups. The reason for this association remains unclear. However, it is conceivable that plaque microembolisation in some patients with atherosclerosis which has progressed may subsequently cause inflammatory reactions at the myocardial level, resulting in measurable systemic inflammation.
The main limitations of our study are the relatively small number of patients included and the lack of outcome data at this time point. Thus, longitudinal studies will be necessary to investigate the prognostic significance of increased troponin levels in the presence of non-calcified plaque in patients with stable CAD. Furthermore, the value of stress ECG or echocardiography was not systematically investigated in our study, which is another limitation. In addition, the ability of CTA to differentiate between lipid-rich and fibrous plaque content is limited, owing to substantial overlap of the corresponding attenuation values.37 In this context intravascular ultrasound measurements may be helpful in future studies. Furthermore, the cut-off values selected for hsTnT to assess plaque composition were determined within the same group of patients, and using a predefined cut-off value of hsTnT=14 pg/ml resulted in low sensitivities, except for the detection of remodeled plaque. Thus, the cut-off values determined in this study for patients with stable CAD merit prospective validation in larger patient cohorts in future studies.
Our study demonstrates an association between CTA plaque characteristics and small troponin increases in patients with stable CAD. Although an explanation of causality cannot be supported by these data, non-calcified plaques, which have been reported as a predictor for future events in previous imaging studies, may contribute to the ongoing, clinically silent troponin leakage possibly owing to repetitive microembolisation of atherosclerotic debris. The notion that hsTnT may serve as a biomarker for such ‘vulnerable’ coronary lesions even in presumably stable CAD is in agreement with the previously shown prognostic value of troponin elevation for adverse coronary events in community-based trials and registries.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.