Cardiac magnetic resonance for evaluating nonculprit lesions after myocardial infarction comparison with fractional flow reserve

OBJECTIVES This study sought to determine the agreement between cardiac magnetic resonance (CMR) imaging and invasive measurements of fractional ﬂ ow reserve (FFR) in the evaluation of nonculprit lesions after ST-segment elevation myocardial infarction (STEMI). In addition, we investigated whether fully quantitative analysis of myocardial perfusion is superior to semiquantitative and visual analysis. BACKGROUND The agreement between CMR and FFR in the evaluation of nonculprit lesions in patients with STEMI with multivessel disease is unknown. METHODS Seventy-seven patients with STEMI with at least 1 intermediate (diameter stenosis 50% to 90%) nonculprit lesion underwent CMR and invasive coronary angiography in conjunction with FFR measurements at 1 month after primary intervention. The imaging protocol included stress and rest perfusion, cine imaging, and late gadolinium enhancement. Fully quantitative, semiquantitative, and visual analysis of myocardial perfusion were compared against a reference of FFR. Hemodynamically obstructive was de ﬁ ned as FFR # 0.80. RESULTS Hemodynamically obstructive nonculprit lesions were present in 31 (40%) patients. Visual analysis displayed an area under the curve (AUC) of 0.74 (95% con ﬁ dence interval [CI]

O ver the past few decades, cardiac magnetic resonance (CMR) imaging has emerged as a robust tool to evaluate patients with suspected obstructive coronary artery disease (CAD).Among the myriad of imaging techniques, CMR has the advantages of being cost-effective (1), widely available, and multiparametric (i.e., it allows to evaluate several pathological features during a single session).In addition, CMR does not expose patients to ionizing radiation while permitting imaging at high spatial resolution.Finally, patients with normal CMR results have a reassuring low risk of future cardiovascular events (2).
In the clinical setting, perfusion images are typically assessed visually; however, CMR can also quantify myocardial blood flow (MBF) in relative (semiquantitative) (3) and even absolute terms (fully quantitative) (4).
Conflicting evidence exists as to whether quantification improves the diagnostic accuracy of perfusion CMR.Although some studies reported superiority of fully quantitative analysis over semiquantitative and visual analysis (5), others found no benefit of quantification in the diagnosis of obstructive CAD (6)(7)(8).In patients with stable CAD, invasive measurements of fractional flow reserve (FFR) are considered the optimal index for diagnosing hemodynamically obstructive CAD and guiding revascularization (9).Although CMR and FFR are reported to have a high concordance in the assessment of patients with stable CAD (10), the agreement between CMR and FFR in the evaluation of nonculprit lesions after ST-segment elevation myocardial infarction (STEMI) is still unknown.

Nonculprit lesions are present in approximately 50%
of patients with STEMI and, if left unattended, carry a poor prognosis (11,12).In these patients, CMR is of special interest as it allows for simultaneous assessment of left ventricular function, infarction size, and perfusion.CMR thus not only provides valuable prognostic information, but also identifies patients with ischemia in nonculprit vascular territories who may benefit from revascularization.
The aim of the present study was to determine the agreement between CMR and invasive measurements of FFR in the evaluation of nonculprit lesions in patients with STEMI with multivessel disease.Visual assessment of myocardial perfusion images was compared to semi-and fully quantitative analysis.

METHODS
STUDY POPULATION.This is a substudy of data from patients enrolled in the REDUCE-MVI (Reducing Micro Vascular Dysfunction in Acute Myocardial Infarction by Ticagrelor) trial in the Amsterdam UMC, location VUmc.The main results of this trial have been published previously (13).Briefly, patients with a first STEMI and at least 1 intermediate lesion in a nonculprit vessel were enrolled after successful revascularization of the culprit.Intermediate was defined as 50% to 90% diameter stenosis.The main exclusion criteria were cardiogenic shock, decompensated heart failure, left main disease, and chronic total occlusion.Acute STEMI management and subsequent medical care followed contemporary guidelines (14).Follow-up CMR and invasive assessment were scheduled 1 month after the index event.CMR always preceded invasive assessment.Patients were instructed to refrain from products containing caffeine or xanthine for 24 h before the follow-up visit.To evaluate angina burden, patients completed the Seattle Angina Questionnaire (SAQ) upon arrival at the hospital (15).The study protocol was approved by the institutional review committee of the Amsterdam UMC, location VUmc.All patients gave written informed consent.
CMR IMAGE ACQUISITION.All images were obtained on a 1.5-T clinical scanner (Magnetom Avanto, Siemens Healthineers, Erlangen, Germany).Perfusion imaging was performed using a dual-sequence technique (16).High-resolution images of myocardial perfusion were acquired using an echo planar imaging sequence in 3 parallel short-axis slices planned at the basal, mid and apical levels.To assess the arterial input function, low-resolution turboFLASH images were acquired at the basal level using a sequence optimized for the high gadolinium concentration.Perfusion images were obtained every heartbeat for 50 to 70 cardiac cycles following intravenous injection of a 0.075 mmol/kg bolus of a gadoliniumbased contrast agent (DOTAREM, Guerbet, Villepinte, France).In-plane respiratory motion of the heart was corrected using a nonrigid registration (17).
Perfusion images were corrected for surface coil induced inhomogeneities through a separate prescan normalization (18).Typical in-plane resolution of the myocardial perfusion images was 2. LGE images were analyzed using commercially available software (QMASS version 7.6, Medis, Leiden, the Netherlands).The myocardium was divided into 16 segments (true apex not included), which were allocated to vascular territories using anatomic information from the invasive angiogram.In addition to this modified segmentation, myocardial segments were allocated to vascular territories according to the standard segmentation model of the American Heart Association (19).Visual analysis of perfusion images was performed by consensus of 2 expert observers (R.N. and R.L.).First, quality of the images was graded as 0 (not assessable), 1 (poor), 2 (moderate), or 3 (good).Next, the occurrence of splenic switch-off was visually assessed by comparing the rest and stress perfusion images.Thereafter, perfusion was scored per segment as 0 (normal), 1 (mildly abnormal), or 2 (severely abnormal).LGE images were reviewed alongside perfusion images to evaluate hyperenhancement.The transmural extent of hyperenhancement was scored per segment as: 0 (absent), 1 (1% to 25%), 2 (26% to 50%), 3 (51% to 75%), and 4 (>75%).Summation scores were calculated on a per-vessel and per-patient basis by adding the perfusion scores of the individual segments.Because the region of hyperenhancement resulting from the recent STEMI may exceed the segments allocated to the culprit (20), all hyperenhanced segments were excluded provided that the hyperenhancement was clearly related to the index event.Segments with hyperenhancement according to a nonischemic pattern were also excluded (21); thus, segments in nonculprit vascular territories with incidental hyperenhancement were only included if the hyperenhancement was clearly unrelated to the index event and demonstrated an ischemic pattern.Semiand fully quantitative analysis of perfusion were performed by a single observer (H.E.), supervised by a second observer (R.N.).Endocardial and epicardial contours were manually drawn on the basal, mid, and apical slices of both the stress and rest perfusion series.A region of interest was placed in the blood pool of the perfusion series obtained for the arterial input were not subjected to physiological interrogation and were also regarded as hemodynamically obstructive.
All 5 patients with absent splenic switch-off had negative CMR results, although FFR was #0.80 in 1 (20%) patient.The agreement between of CMR and FFR was unaltered by including only patients who demonstrated splenic switch-off (Supplemental Figure 1).Stress MBF and MFR in myocardium supplied by stenotic non-culprit vessels and in myocardium supplied by vessels without significant angiographic stenosis (remote).Stenotic vessels are further stratified according to FFR.NC ¼ nonculprit vessel; other abbreviations as in Figures 1 and 3. ered to be the gold standard for myocardial perfusion imaging, is reported to be concordant with FFR in 86% of cases (27).When only vessels with angiographic stenosis are included, the agreement drops to 72% (28), similar to our results.Second, hyperemic perfusion and FFR show physiological discordance in approximately 25% of cases (27), mainly because of varying degrees of microvascular dysfunction (29).
After STEMI, microvascular dysfunction can be present either concomitantly or as a result of acute ischemic injury.Several studies have indicated that in STEMI acute microvascular dysfunction is not confined to the culprit area but also affects remote myocardium (30,31).Given a certain geometric ste- area under the curve CAD = coronary artery disease CI = confidence interval CFRthermo = thermodilution derived coronary flow reserve CMR = cardiac magnetic resonance FFR = fractional flow reserve IMR = index of microcirculatory resistance LGE = late gadolinium enhancement MBF = myocardial blood flow MFR = myocardial flow reserve MFR rel upslope = myocardial flow reserve relative upslope derived flow reserve ROC = receiver-operator characteristic stress rel upslope = stress signal intensity time curve SAQ = Seattle angina questionnaire STEMI = ST-segment elevation myocardial infarction made available by the Top Sector Life Sciences & Health to stimulate public-private partnerships.The funding sources did not have any role in the study design; collection, analysis, or interpretation of data; preparation of the manuscript; or decision to submit it for publication.Dr. van Royen reports research grants from AstraZeneca, Abbott, Philips, Biotronik, and Top Sector Life Sciences & Health during the conduct of the study.Dr. van Leeuwen has received research grants from AstraZeneca.Dr. Nijveldt has received research grants from Philips and Biotronik; and financial support from the Netherlands Organization for Health Research and Development (grant 9071544).Dr. Demirkiran has received a research grant from the postdoctoral international research fellowship program of the Scientific and Technological Research Council of Turkey (ref: 53325897-115.02-170549).All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.Manuscript received May 28, 2019; accepted July 10, 2019.
Abbott, Chicago, Illinois) was placed in the distal part of the stenotic nonculprit vessel and connected to the RadiAnalyzer interface (Abbott).Next, hyperemia was induced through intravenous infusion of adenosine using the same protocol as applied during CMR.FFR was obtained after at least 2 min of infusion.Lesions with FFR #0.80 were regarded as hemodynamically obstructive.Lesions with $90% diameter stenosis

FIGURE 1
FIGURE 1 Flowchart of Study Population SEMIQUANTITATIVE ANALYSIS.The relative upslope of the stress signal intensity time curve (stress rel upslope) and relative upslope derived flow reserve (MFR rel upslope) were significantly lower in myocardium supplied by stenotic nonculprit vessels with FFR #0.80 compared with stenotic nonculprit vessels with FFR >0.80 (10.0 AE 3% vs. 8.6 AE 2%; p ¼ 0.04 for stress rel upslope and 1.7 AE 0.6 vs. 1.4 AE 0.5; p ¼ 0.009 for MFR rel upslope).ROC curve analysis revealed an AUC of 0.66 (95% CI: 0.54 to

FIGURE 2
FIGURE 2 Case Example

FIGURE 3
FIGURE 3 Diagnostic Performance of Cardiac Magnetic Resonance

Figure 4
Figure 4 displays the diagnostic performance of CMR for diagnosing hemodynamically obstructive nonculprit when only vessels with CFR thermo >2 or IMR <25 U are included.The agreement between CMR

3 , 2 0 2 0 FIGURE 4 CFR
FIGURE 4 Diagnostic Performance Of CMR In Patients With Normal Microvascular Function

Table 1
CFR thermo ¼ thermodilution derived coronary flow reserve; CMR ¼ cardiac magnetic resonance; FFR ¼ fractional flow reserve; IMR ¼ index of microcirculatory resistance.Everaars et al.STATISTICAL ANALYSIS.Continuous variables are presented as mean AE SD or median (interquartile range), whereas categorical variables are expressed as frequency with percentage.Rest and stress perfusion were compared using the paired sample Student's ttest.Means of perfusion indexes were compared between vessels using a mixed linear model with a random effect for patient.Receiver-operatoring characteristic (ROC) curve analysis and the Youden index were used to define optimal cutoff values.RESULTSPATIENT CHARACTERISTICS.Stress perfusion CMR was successfully performed in 77 of the 90 patients enrolled in the Amsterdam UMC, location VUmc (Figure1).Baseline characteristics of the study cohort are listed in Table1.Table2displays the angiographic and CMR characteristics.Follow-up CMR and invasive assessment were performed on the same day in 53 (69%) patients.Figure2depicts a case of concordance between CMR and FFR.In the 77 patients, 94 nonculprit vessels had $50% diameter stenosis and were used for analysis.Hemodynamically obstructive nonculprit lesions were present in 36 (38%) vessels of 31 (40%) patients.Collaterals pathways were identified in 5 (6%) patients (details listed in Supplemental VISUAL ANALYSIS.The calculated summed stress scores ranged from 0 to 14. ROC curve analysis revealed an area under the curve (AUC) of 0.74 (95% confidence interval [CI]: 0.62 to 0.83) for visual analysis to detect hemodynamically obstructive
of CMR using the average myocardial perfusion in the vascular territory of the nonculprit vessel.In both models, semi-and fully quantitative analysis demonstrated similar diagnostic performance (all p > 0.05).INFLUENCE OF MICROVASCULAR FUNCTION ONDIAGNOSTIC PERFORMANCE.CFR thermo and IMR were obtained in 68 (72%) vessels of 66 (86%)
caution.Finally, collateral connections from and to stenotic nonculprit vessels may have influenced FFR measurements.In the present study, visual collateral vessels were identified in 5 patients.Collateral pathways of smaller size that are not visualized on coronary angiography may nonetheless have influenced the results.CONCLUSIONS CMR and FFR have moderate-good agreement in the evaluation of nonculprit lesions in reperfused patients with STEMI with multivessel disease.Fully quantitative analysis of myocardial perfusion is not superior to semiquantitative or visual analysis.ADDRESS FOR CORRESPONDENCE: Dr. Robin Nijveldt, Amsterdam University Medical Center, location VUmc, Department of Cardiology, De Boelelaan 1117, 1081 HV, Amsterdam, the Netherlands.E-mail: robin@nijveldt.net. in silent ischemia after myocardial infarction: the SWISSI II randomized controlled trial.JAMA 2007;297:1985-91.39. iFR Guided Multi-vessel Revascularization During Percutaneous Coronary Intervention for Acute Myocardial Infarction (iMODERN).KEY WORDS acute myocardial infarction, cardiac magnetic resonance, fractional flow reserve, non-culprit lesions, quantitative myocardial perfusion APPENDIX For supplemental tables and figures, please see the online version of this paper.
make the present data unique.Second, invasive measurements of FFR were used as reference.Although FFR is considered the gold standard for guiding revascularization in patients with stable CAD, it has not been established as such for the management of nonculprit lesions after STEMI.The results of the present study should therefore be interpreted with consuming process, in contrast to the more straightforward visual analysis.To implement perfusion quantification into the clinic, image acquisition and post-processing have to be accelerated and automated.Everaars et al.interventions