REPLY: Prognostic Role of CMR Imaging After Myocardial Infarction We thank Drs. Eitel and Thiele for the interest in our work (1). They highlight some possible limitations of our approach on several points that merit discussion. Cardiovascular magnetic resonance imaging (CMR) is a continuously evolving technique, which leads to new opportunities and the expansion of indications. In our systematic review, we summarized all pub- lished papers regarding the prognostic value of CMR ndings in patients with a recent myocardial infarc- tion (MI) or suspected or known coronary artery disease. As mentioned in our study, if the CMR nding was studied in <1,000 patients, it was classied as not enough evidence to make inference about its prog- nostic value(1). Although this value is arbitrarily chosen, lowering this threshold would lead to more CMR ndings incorrectly being classied as an independent prognostic marker. In the majority of studies, the number of events/variable in multi- variable analyses was less than 10, leading to a possible overestimation or underestimation of the reported hazard ratios in those studies (2). Although there is no established threshold, we considered the value of 1,000 patients to be appropriate. We concluded that none of the CMR ndings, including left ventricular ejection fraction (LVEF), was studied in more than 1,000 patients with an MI, and therefore no inferences could be made about its independent prognostic value to predict hard clinical events. The prognostic value of late microvascular obstruction (MVO) was assessed in 2 papers, in- cluding 624 patients, and 1 of the studies showed a signicant result. We found that LVEF is an independent prognostic CMR nding to predict major adverse cardiovascular events in patients with a recent MI. As for late MVO, all 3 studies including 668 patients showed a signi- cant result. This shows that late MVO is a promising CMR nding to predict future events but is not studied in enough patients yet. Due to the difference between studies in CMR ndings included in the multivariable analyses, we were not able to directly compare the prognostic value of the CMR ndings. Therefore, we can conclude that LVEF is an inde- pendent CMR nding, but its relative value with respect to (other) CMR ndings, such as late MVO, could not be studied. We agree that heterogeneity in reporting and sta- tistical analyses is an important limitation of the re- sults presented in systematic reviews. We therefore could not aggregate the results of the individual studies. As discussed in our paper (1) and the associated editorial (3), a large individual patient data meta-analysis can provide more information on the relative prognostic value of CMR ndings. We recently started the PROMISE collaboration (4) to collect data from published studies assessing the prognostic value of CMR. By analyzing data from previously published studies that might be underpowered themselves on the patient level and by accounting for differences in patient and study characteristics, the incremental value of the CMR ndings, in addition to already known risk factors, can be assessed. In this way, readily available data can be used, and it might not be necessary to spend costly resources to perform multicenter studies that require a large number of patients and long-term follow-up, as suggested by Drs. Eitel and Thiele. *Hamza El Aidi, MD Arthur Adams, MD, PhD Pieter Doevendans, MD, PhD Eike Nagel, PhD Tim Leiner, MD, PhD *Department of Cardiology University Medical Center Utrecht Heidelberglaan 100 P.O. Box 85500 HP E01.132 3508 GA Utrecht the Netherlands E-mail: h.elaidi@umcutrecht.nl http://dx.doi.org/10.1016/j.jacc.2014.08.032 REFERENCES 1. El Aidi H, Adams A, Moons KG, et al. Cardiac magnetic resonance imaging ndings and the risk of cardiovascular events in patients with recent myocardial infarction or suspected or known coronary artery disease: a sys- tematic review of prognostic studies. J Am Coll Cardiol 2014;63:103145. 2. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:13739. 3. Gottlieb I, Camargo G. Is cardiac magnetic resonance one of cardiologys magic crystal balls? J Am Coll Cardiol 2014;63:10467. 4. PROMISE. Available at: http://www.promise-ipd.com. Accessed September 9, 2014. Letters JACC VOL. 64, NO. 19, 2014 NOVEMBER 11, 2014:2069 70 2070