Author + information
- S0735109716007221-8b73413378ea301325c4d69f2da4ae74Raymond J. Gibbons, MD∗ ( and )
- S0735109716007221-02e0170818f5ad1ac790026e72bf1003Philip Araoz, MD
- ↵∗Reprint requests and correspondence:
Dr. Raymond J. Gibbons, Mayo Clinic, Cardiovascular Diseases, 200 First Street SW, Gonda 5-412, Rochester, Minnesota 55905.
There is very extensive published data employing clinical parameters, angiographic variables, assessment of left ventricular function, and stress testing to identify patients who are at higher risk for cardiac events after ST-segment elevation myocardial infarction (STEMI). Myocardial infarct size by single-photon emission computed tomography (SPECT) and cardiac magnetic resonance (CMR) are relative newcomers that may be useful for post-discharge risk stratification. In this issue of the Journal, Stone et al. (1) report a patient-level pooled analysis from 10 randomized trials utilizing primary percutaneous coronary intervention (PCI) in 2,632 patients to examine the value of infarct size for predicting subsequent patient outcome.
Most patients (71.8%) had infarct size assessed by CMR; the remainder had infarct size assessed by SPECT. CMR or SPECT was performed at a median of 4 days after STEMI, although SPECT measurements were done at a minimum of 5 days after STEMI in 2 of the trials. Twenty-five percent of the patients had infarct size measurements >10 days after STEMI. By Kaplan-Meier analysis, 1-year rates of all-cause mortality, reinfarction, and hospitalization for heart failure were all relatively modest (2.2%, 2.5%, and 2.6%, respectively). After adjustment for multiple clinical, angiographic, and treatment parameters, infarct size remained strongly associated (p < 0.0001) with both all-cause mortality and hospitalization for heart failure, but not with reinfarction.
Although the authors suggest that these findings have potential significance for the design of randomized trials, we have chosen not to address the controversial topic of surrogate endpoints. We will instead focus on the strengths and weaknesses of this study and offer our thoughts on the measurement and application of infarct size in clinical decision making.
This study has a number of strengths. The analysis was conducted at a patient-level. Although meta-analyses of overall trial results can provide important insights, patient-level analyses are required to fully account for individual patient characteristics, which are critical here.
The patient-level data are drawn from randomized trials of primary PCI, which offer many advantages. The entry criteria are well-defined in previous publications, and the overall level of care is excellent (Table 2 in Stone et al. ). The endpoints of mortality, reinfarction, and hospitalization for heart failure were all adjudicated by each trial, although the definitions might differ slightly between trials. Most importantly, this incremental analysis was done in a very rigorous fashion, with consideration of clinical (age, sex, risk factors), angiographic (left anterior descending or non–left anterior descending infarct vessel, baseline TIMI [Thrombolysis In Myocardial Infarction] flow) and treatment (symptom-to-first device time) variables that have been demonstrated in previous studies to be of prognostic importance. In a single study, where the number of events is usually modest (which it is here) these individual variables may not necessarily be statistically significant (which is evident from Table 4 in Stone et al. ), so they must be “forced into” the models to ensure that they are part of the adjustment process. The authors report the area under the curve (AUC) for their models with and without infarct size, as well as a reclassification analysis.
This study also has potential weaknesses, many of which are acknowledged by the authors. A majority of the patients had anterior infarcts, which are larger, increasing the value of infarct size measurements. Although the authors considered many important variables, they were unable to consider 2—collateral flow and ejection fraction (EF)—which are important. The importance of collateral flow has long been recognized in animal models, and confirmed in clinical studies of SPECT infarct size (2). Many clinicians regard the EF as a measure of left ventricular damage, although myocardial stunning, ventricular loading, other myocardial conditions, and variability in the measurement in normal patients may all reduce its utility. Despite the large number of patients considered in the study, the number of endpoints in the analysis was relatively modest—54 deaths, 60 reinfarctions, and 58 hospitalizations for heart failure. Models that incorporate more than 6 variables (which is true for all of the models shown in Table 4 in Stone et al. ) may risk “over-fitting,” as they exceed the commonly accepted guideline of 1 variable for each 10 events. Patient entry criteria varied across the trials. Although the hospitalizations for heart failure were adjudicated, clinicians managing the patients were not blinded to their infarct size measurements, and it is possible that clinicians were more likely to suspect heart failure in patients with larger infarct sizes and then perform a chest x-ray, thereby contributing to ascertainment bias.
The infarct size measurements reported here were performed by CMR in most patients and by SPECT in a minority of the patients. CMR infarct size was performed by manually tracing the infarct on myocardial delayed enhancement images, which can be highly reproducible if high-quality CMR images are obtained and if performed by experienced observers. One of the CMR core laboratories (University of Leipzig-Heart Center), which analyzed 56.6% of the total patients in this study, has reported its CMR infarct size interobserver variability to be ± 2.2% (3).
This study adds to the previous published data showing the prognostic value of SPECT infarct size measurement at discharge. In a previous review (4), we summarized the single-center data from Mayo, as well as a multicenter experience from the Mayo core laboratory. These studies were smaller than the current study, with fewer events, and were therefore unable to adjust for all of the important patient and angiographic variables included in the current study.
Similarly for CMR infarct sizing, a prior single-center study (5), a multicenter study (including a substudy of 1 of the studies pooled in the current study) (6), and a recent meta-analysis (7) all showed an association between infarct size and poor outcomes. However, they all had fewer patients and events than the current study and had limited power for multivariate analysis.
What is the overall clinical effect of these results on the basis of the AUC and reclassification analysis? Although both the difference in AUC with and without infarct size and the integrated discrimination improvement were statistically significant for the combined endpoint of death and heart failure hospitalization, neither was significant for mortality alone. We find it difficult to interpret the reclassification analysis without tables showing different risk groups. For most patients, we do not think that medical therapy and decisions regarding PCI/coronary artery bypass grafting are likely to change on the basis of a slightly higher risk of death or heart failure hospitalization. However, because infarct size is strongly associated with late (6 weeks and 1 year) EF (4), we would suggest that patients in the top quartile of infarct size in this study (30% of the left ventricle) should undergo late EF measurement to determine their eligibility for implantable cardioverter-defibrillator (ICD) placement, as ICDs are currently implanted in <1 in 10 eligible patients (8).
Before using CMR infarct size measurements, clinicians should carefully review the methodology used in their hospital or practice. If the tracings are traced manually, is the interobserver variability as low as that reported by Leipzig? The timing of the images is important, and is quite variable in the studies reported here. It is important to note that patients with arrhythmia, renal failure, and claustrophobia were excluded from CMR studies. For that reason, clinicians will not have CMR infarct size measurements available in all patients, and will have to decide whether to use SPECT imaging as a substitute.
In summary, this patient-level analysis adds to the body of scientific evidence confirming the prognostic significance of measurements of infarct size by CMR or SPECT prior to discharge. Evidence-based clinicians should consider whether infarct size will improve their management of STEMI patients by better identifying those with large infarcts who should have late EF assessment to determine their eligibility for ICD placement.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Gibbons has served as a consultant for Lantheus Medical Imaging, Astellas Pharm US, and Stealth Pepides. Dr. Araoz has reported that he has no relationships relevant to the contents of this paper to disclose. Alan S. Jaffe, MD, served as Guest Editor for this paper.
- American College of Cardiology Foundation
- Stone G.W.,
- Selker H.P.,
- Thiele H.,
- et al.
- Christian T.F.,
- Schwartz R.,
- Gibbons R.J.
- Thiele H.,
- Kappl M.J.,
- Conradi S.,
- et al.
- Gibbons R.J.,
- Valeti U.S.,
- Araoz P.A.,
- Jaffe A.S.
- Izquierdo M.,
- Ruiz-Granell R.,
- Bonanad C.,
- et al.
- Eitel I.,
- de Waha S.,
- Wohrle J.,
- et al.