Article Text
PDF
Abstract
Objective To investigate the ability of cardiac troponin I (cTnI) to predict functional recovery and left ventricular remodelling following primary percutaneous coronary intervention (pPCI) in ST-elevation myocardial infarction (STEMI).
Design Post hoc study extending from randomised controlled trial.
Patients 132 patients with STEMI receiving pPCI.
Main outcome measures Left ventricular ejection fraction (LVEF), end-diastolic and end-systolic volume index (EDVI and ESVI) and changes in these parameters from day 5 to 4 months after the index event.
Methods Cardiac magnetic resonance examination performed at 5 days and 4 months for evaluation of LVEF, EDVI and ESVI. cTnI was sampled at 24 and 48 h.
Results In linear regression models adjusted for early (5 days) assessment of LVEF, ESVI and EDVI, single-point cTnI at either 24 or 48 h were independent and strong predictors of changes in LVEF (p<0.01), EDVI (p<0.01) and ESVI (p<0.01) during the follow-up period. In a logistic regression analysis for prediction of an LVEF below 40% at 4 months, single-point cTnI significantly improved the prognostic strength of the model (area under the curve = 0.94, p<0.01) in comparison with the combination of clinical variables and LVEF at 5 days.
Conclusion Single-point sampling of cTnI after pPCI for STEMI provides important prognostic information on the time-dependent evolution of left ventricular function and volumes.
Trial Reg No The study was part of a randomised controlled trial http://www.clinicaltrials.gov; Unique identifier: NTC00326976.
- MRI
- coronary angioplasty (PCI)
- STEMI
- stunned myocardium
Statistics from Altmetric.com
Introduction
In ST-elevation myocardial infarction (STEMI), timely reperfusion therapy reduces infarct size, preserves left ventricular function and improves clinical outcomes.1 After restoration of tissue perfusion, some degree of reversible myocardial dysfunction (stunning) is seen in most patients.2 A major clinical challenge in the peri-myocardial infarction period is to discriminate between patients with reversible dysfunction and those who will experience little functional recovery accompanied by left ventricular (LV) volume expansion.3 4 Such risk stratification is important as chronic LV dysfunction and adverse remodelling are the strongest predictors of subsequent mortality.5 6 Early identification of high-risk patients allows for early intervention with targeted, aggressive treatment strategies and close follow-up. Infarct size is the most important determinant of LV function and volumes after myocardial infarction,3 4 and studies have shown that a single measurement of the myocardial necrosis biomarker cardiac troponin 24–96 h after myocardial infarction correlates with infarct size and early left ventricular function in patients with STEMI.7–12 However, it remains largely unknown whether cardiac troponin measurements provide clinically useful information on the potential for recovery of LV function and the degree of LV volume expansion, even when functional and volumetric parameters have been obtained before discharge. Potentially, cardiac troponin represents a cost-effective and simple tool for risk stratification early after STEMI that is complementary to echocardiographic assessment. In this post hoc analysis of a prospective, multicentre, randomised study of patients with STEMI undergoing primary percutaneous coronary intervention (pPCI), we aimed to assess the ability of a single-point cardiac troponin I (cTnI) sample to predict functional recovery and extent of LV remodelling, even taking into account early assessment of LV function and volumes.
Patients and methods
Patient population
The F.I.R.E. trial (http://clinicaltrials.gov NCT00326976) investigated the cardioprotective properties of the novel compound FX06 by randomising 234 patients with STEMI (first time myocardial infarction) receiving pPCI within 6 h from onset of symptoms to active drug or placebo.13 Follow-up was 4 months and all patients were specified according to protocol to undergo a complete cardiac magnetic resonance (CMR) examination at 5 days and 4 months in addition to cTnI sampling at 24 and 48 h. No significant effects on the primary outcome measure could be attributed to FX06.13
For this analysis, we excluded 95 patients from the intention-to-treat group (n=227) because they had missed one CMR examination or cTnI measurement. No clinically or statistically significant differences were observed between the substudy cohort and the rest of the total study population. The inclusion and exclusion criteria of F.I.R.E. and other details of the protocol have been reported previously.13 14 All local ethics committees approved the protocol, all patients gave informed consent before enrolment and the study was performed in accordance with the declaration of Helsinki.
Cardiac magnetic resonance
Details of the CMR examination have been elaborated in the main report from the F.I.R.E. trial.13 In short, CMR was performed at 5 days and at 4 months and analysed at the central MR core laboratory, University Hospital Basel, by a single experienced CMR reader who was blinded to the study groups. This was followed by a blinded review by a level III CMR expert. Intraobserver variability was assessed for the primary reader in a subset of 40 randomly chosen studies and the intraclass correlation was 0.85 for the studies acquired on day 5. The CMR examination involved an electrocardiogram-triggered acquisition of short-axis slices of the left ventricle from the base to the apex using a steady-state free precession pulse sequence. LV end-systolic and LV end-diastolic volumes (ESVI and EDVI, respectively), LV ejection fraction (LVEF) and myocardial mass were calculated in a standard fashion.15
Cardiac troponin I
cTnI was sampled at 24 (cTnI24) and 48 h (cTnI48) after admission. All samples were analysed in a blinded core laboratory (Spranger Laboratories, Ingolstadt, Germany). cTnI was measured on the Abbott AxSym System (Abbott Diagnostics, Abbott Park, Illinois, US) using the second-generation AxSYM Troponin-I ADV assay. The analytical sensitivity of the assay is 0.02 ng/ml with a 10% coefficient of variation at 0.16 ng/ml, and the 99th percentile of a reference population is 0.04 ng/ml.
End points and stratification of patients into subgroups
Outcome measures were LVEF, EDVI and ESVI; and change in LVEF (ΔEF=LVEF2−LVEF1, change in ESVI (ΔESVI=ESVI2−ESVI1) and change in EDVI (ΔEDVI=EDVI2−EDVI1) from 5 days to 4 months. For some analyses, patients were stratified in three groups on the basis of their initial LVEF with cut-off points at 40% and 50%. There is no consensus about the exact demarcation between compromised and preserved LVEF,16 so the definition of these subgroups was not data driven, but rather was thought to represent a clinically reasonable stratification.
Statistics
The normal distribution of outcome variables was confirmed by Kolmogorov–Smirnov tests. Differences between groups were assessed by Student's t-test. Associations between cTnI and end points were explored by Pearson correlations, and linear and logistic regression analyses. For linear regression analyses of changes in LV function and volumes (ΔVLEF, ΔEDVI and ΔESVI), all variables (for details, see online supplementary appendix) were initially evaluated separately only adjusting for the baseline measurement of LVEF, EDVI, or ESVI, as appropriate. All variables with a p value <0.25 in these initial analyses were included in the multivariable models. In logistic regression analysis for prediction of LVEF <40% at 4 months we aimed to explore if addition of cTnI24 or cTnI48 to early LVEF would improve the prognostic power of the model. In addition, adjustments were made for age, gender, randomisation, body mass index and infarct location. Discrimination and calibration of the logistic regression models were evaluated by the c-statistic and Hosmer–Lemeshow tests.17 18 For the c-statistic (area under the curve) a value of 0.5 indicates no better discriminatory ability than chance alone, and a value of 1.0 indicates perfect discrimination for the binary outcome. All analyses were performed with SPSS version 16.0 (SPSS Inc).
Baseline characteristics (n=132)
Results
Demographics and procedural parameters are presented in table 1.
Correlation coefficients
Correlation coefficients
Correlation coefficients between cTnI and LVEF, EDVI and ESVI at 5 days and 4 months are shown in table 2. We have earlier reported strong correlation coefficients between cTnI and CMR-determined infarct size.8
Evolution of left ventricular function and volumes stratified by early LVEF and cTnI24
Distribution of LV function and volumes at 5 days and 4 months
Patients were categorised into three groups according to their LVEF at 5 days (group 1, <40%; group 2, 40% to <50% and group 3, ≥50%) and relationships between cTnI24 and ΔLVEF, ΔESVI and ΔEDVI for each group are presented in scatter plots (figure 1). Subsequently, these three groups were then divided into two subgroups depending on whether the cTnI24 measurement was below or above the median (for group 1, 91.00 ng/ml; group 2, 49.76 ng/ml; group 3, 20.16 ng/ml) (table 3). Overall, LVEF improved during follow-up among patients with early LVEF <50% and remained stable in those with an early LVEF ≥50%. In the two groups with LVEF <50% at 5 days, stratification by cTnI24 separated patients with little functional improvement accompanied by volume expansion (high cTnI24) from patients with substantial improvement of function with no volume expansion (low cTnI24). Among those with an initial LVEF ≥50%, only small differences were seen, though LV volume expansion mainly occurred in the high cTnI24 subgroup. The limited number of patients in each subgroup did not allow for robust exploration of potential differences in these dynamics between anterior and non-anterior infarction. However, exploratory analyses showed that the trends in the time-dependent changes were the same for both non-anterior and anterior locations, but more pronounced for the anterior infarctions.
Scatter plots (with Pearson's correlation coefficients) showing the relation between cardiac troponin I (cTnI) at 24 h and changes in left ventricular ejection fraction (LVEF), end-systolic volume index (ESVI) and end-diastolic volume index (EDVI) in patients stratified by ejection fraction at 5 days.
Linear regression analyses for predictors of change in left ventricular function and volumes between 5 days and 4 months
Regression analysis for predictors of functional recovery and extent of remodelling
In linear regression analyses, cTnI24 and cTnI48 (p<0.01 for all models, data not shown) were independent predictors of changes in LVEF and LV volumes. All independent predictors in the multivariable models are shown in table 4 (complete univariable and multivariable models are provided in the supplementary appendix). For prediction of chronic LVEF <40%, addition of cTnI24 or cTnI48 to early LVEF significantly increased the global fit of the model as evaluated by the change in the χ2 statistic and improved the discriminatory ability as evaluated by the c-statistic (table 5).
Logistic regression models for prediction of an ejection fraction < 40% at 4 months
Although all regression analyses also included infarct location as a covariate, we also explored the issue by stratifying patients by infarct locations. We found that cTnI was a stronger predictor of LV volume changes for anterior infarctions, while functional recovery was more closely associated with cTnI samples among non-anterior infarctions. However, the trend of changes was in the same direction and cTnI remained statistically significant across all subgroups.
Discussion
In a contemporary, homogeneous and well-defined cohort of patients with STEMI undergoing pPCI we demonstrate that a single measurement of cTnI, obtained at 24 or 48 h after interventional management, provides useful prognostic information on changes in LV function and volumes during the first 4 months after the acute event, even taking into account early measurement of LV function and volumes. As cTnI release reflects myocardial necrosis, our data are consistent with the concept that after myocardial infarction, the degree of functional recovery and the accompanying adaptive volumetric changes occur in response to the amount of myocardial destruction.3 4 These observations suggest that cTnI measurements may facilitate risk stratification of patients during the critical early stage after STEMI.
Several previous studies have shown that cardiac troponin predicts LV function after myocardial infarction,9 11 12 19 20 but some of these studies also included non-STEMI patients; and most patients with STEMI were reperfused pharmacologically. As the release dynamics of cTnI are different for patients with or without STEMI and also dependent on the reperfusion modality, the usefulness of drawing inferences from those observations to the contemporary context of mechanical reperfusion for STEMI may be limited. We describe a population where all patients underwent pPCI and epicardial blood flow was re-established in >92% (99% had thrombolysis in myocardial infarction score ≥2). The correlation coefficients between early LVEF and cTnI reported here are in line with a recent study extending from a comparable pPCI population employing the same troponin assay.7
The major novel aspect of our study lies in the use of both early and late (4 months) measurements of LV function and volumes, allowing the time-dependent changes of these parameters and their association with cTnI values to be explored. In patients with mild to severe global LV dysfunction at discharge (LVEF <50%), we found that there were large individual differences in the degree of subsequent functional recovery and adaptive volumetric changes; and that the evolution of these processes occurred in relation to cTnI values at 24 or 48 h. The same pattern was also present in patients with an initial LVEF >50%, although changes in LV volumes and function were less pronounced in this group. The association of cTnI with LV functional recovery and remodelling was confirmed in linear regression models after adjustments for clinical and procedural covariates. Additionally, cTnI was an independent predictor of an LVEF <40% at 4 months after adjustment for pre-discharge LVEF. Taken together, these findings underline the significant imprecision of early estimates of LV function and volumes after STEMI, and demonstrate the value of adding a marker of myocardial necrosis for assessing the risk of long-term LV dysfunction and adverse remodelling.
Recent studies have highlighted the prognostic value of contrast-enhanced CMR delineation of infarct size and microvascular obstruction for risk stratification before discharge in STEMI populations.3 4 21–24 However, these imaging techniques are expensive, logistically challenging, and hence not available for the majority of patients. Our data suggest that a single cTnI measurement may provide a convenient and useful alternative to CMR estimation of infarct size.
Several tentative clinical implications are suggested by these data. Our findings indicate that a single cTnI measurement after STEMI is instructive for early assessment of subsequent risk and provides complementary information to assessment of function and volumes before discharge. In particular, this would be useful in patients presenting with an initial LVEF <40%, where cTnI sampling could help define target populations for comprehensive and aggressive anti-remodelling strategies or other therapeutic interventions, such as implantable cardioverter-defibrillator placement. By aiding in the selection of such target populations, our data are also relevant for design of clinical trials aimed at improving postmyocardial infarction outcomes further; as these studies are highly reliant upon simple and accurate identification of patients who will benefit the most from novel interventions.
Specific strengths and limitations of our study should be noted. The study is particularly reinforced by the fact that it extends from a well-defined, contemporary patient cohort where all variables included were prospectively defined and all analyses were performed in designated core laboratories. Nevertheless, this remains a post hoc study and before the full clinical usefulness of the prognostic strategy presented here can be determined, prospective studies are needed to validate our findings; and appropriate cut-off values for risk stratification need to be established and standardised across the wide range of troponin assays currently in use. Sampling of cTnI at other time points would have allowed calculation of derived variables (eg, peak) and more extensive comparisons of different time points. However, there is a growing body of evidence indicating that single-point sampling at 24 or 48 h is comparable and even better than other sampling strategies.9 25 26 In any case, early sampling increases the clinical utility. Our observations are based on patients with first-time myocardial infarction receiving pPCI within 6 h from onset of symptoms, and exclusion criteria included a number of existing comorbidities. Hence, the usefulness of single-point cTnI sampling might be more restricted in different or more complex clinical situations. Last, we acknowledge that LVEF at best correlates modestly with functional clinical outcomes such as exercise tolerance and quality of life.
In conclusion, we find that a single-point measurement of cTnI24 or cTnI48 after pPCI provides substantial prognostic information on the evolution of LV function and the risk of LV expansion. Single-point cTnI sampling at 24 or 48 h after STEMI can potentially facilitate identification of individuals at particularly high risk who might benefit from closer follow-up and targeted treatment.
Acknowledgments
The authors thank the doctors who contributed to patient enrolment at each participating centre. The authors would also like to thank Professor Peter Buser, MD PhD, and Jens Bremerich, MD (University Hospital Basel, Switzerland) for reading cardiac magnetic resonance images.
References
Supplementary materials
Web Only Data hrt.2009.190819
Files in this Data Supplement:
Footnotes
Funding JH was supported by grants from the South-Eastern Norway Regional Health Authority, Aker University Hospital Research Foundation and Centre for Heart Failure Research, Oslo, Norway.
Competing interests ASJ has consulted over time for most of the major diagnostic companies.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of all local ethics committees.
Provenance and peer review Not commissioned; externally peer reviewed.