Role of Cardiovascular Magnetic Resonance as a Gatekeeper to Invasive Coronary Angiography in Patients Presenting With Heart Failure of Unknown Etiology

Study Questions:

How does cardiovascular magnetic resonance imaging (CMR) compare to conventional coronary arteriography (CA) for identifying the etiology of new-onset congestive heart failure (CHF)?


CMR and CA were performed in 120 consecutive patients presenting with new-onset CHF of uncertain etiology. All were ≥35 years of age, were New York Heart Association class I-III, and no patient had a prior history of coronary artery disease (CAD), chest pain, or significant valvular heart disease. CA and CMR were separately assessed by two independent consensus groups of three experienced cardiologists. CMR was analyzed for late gadolinium enhancement (LGE) as well as magnetic resonance CA (MRCA). For the CMR subgroup, a diagnostic algorithm of initial review for LGE, followed by a recommendation for CA if LGE was present, was followed. If subendocardial LGE was not present, the MRCA was reviewed, and if it was thought to be consistent with CAD, a recommendation for CA was made; and in the absence of CAD on MRCA, the patient was given a diagnosis of nonischemic dilated cardiomyopathy (DCM) and CA was not suggested. A third independent and separate consensus group of three cardiologists reviewed both the CA and CMR to arrive at a “gold standard diagnosis.” Each of the three analysis groups characterized each patient into one of six groups defined as: 1) nonischemic DCM: no LGE on CMR and no obstructive CAD on CA; 2) CHF due to CAD: defined as ischemic pattern LGE in ≥3 segments with >1 coronary stenosis >75% in the affected area; 3) nonischemic DCM with bystander infarct: defined as LGE affecting ≤2 of 17 segments in a globally hypokinetic left ventricle (LV) with no obstructive CAD; 4) nonischemic DCM with bystander CAD: defined as no LGE with CAD insufficient to explain magnitude of LV dysfunction; 5) CHF due to ischemic heart disease with unobstructive coronary arteries: defined as ≥3 segments of LGE with associated hypokinesia, but unobstructive CAD; 6) CHF resulting from CAD: no LGE, but severe proximal three-vessel disease. For cost-analysis purposes, the cost of CA was set at £1.255 and the cost of CMR at £600.


The “gold standard consensus” diagnosis was true DCM in 83 (69%), true CAD in 27 (23%), DMC with bystander infarct in four (3%), DCM with bystander CAD in four (3%), MI with unobstructed CAD in two (2%), and no patient was classified as severe proximal CAD without MI. In the CA as a first-line strategy subgroup, the initial diagnosis was DCM in 89, which was supported in 87 by the gold standard group, whereas two were thought to be related to CAD. A CAD etiology was ascribed to 31 patients, 27 of whom were given this diagnosis by the gold standard group. In the CMR as an initial strategy subgroup, 89 patients were diagnosed as DCM and the diagnosis was supported in all by the gold standard group. Thirty-three were diagnosed as CAD and hypothetically referred for CA. Of these, a CAD etiology was identified in 27 and a final diagnosis was supported in all by the gold standard group, and six CAs indicated a DCM etiology, four of whom were deemed to have true DCM by the gold standard group. Sensitivity, specificity, and positive and negative predictive values for CMR were 100%, 96%, 88%, and 100%; and for CA, 93%, 96%, 87%, and 98%. Cost analysis suggested an average cost of £1255 per patient for CA as an initial screening tool, compared to an average cost of £945 per patient for CMR as an initial screening tool (p = 0.001).


CMR is a clinically effective and potentially economically beneficial initial method for screening of patients presenting with new-onset CHF of uncertain etiology to distinguish nonischemic DCM from a coronary etiology.


The precise etiology of LV systolic dysfunction in a globally hypokinetic LV in patients with new CHF is often difficult to ascertain. Obviously in patients with a remarkably low pretest of likely CAD such as the very young, a nonischemic etiology can safely be ascribed in the majority of instances. For the more usual patient presenting such as those evaluated here, an ischemic etiology is often under consideration, and may result in a change of management strategy with respect to revascularization and risk assessment. Noninvasive strategies to identify true nonischemic DCM from ischemic CHF have been made in the past with variable degrees of success, and many institutions resort to standard CA as a routine part of the workup. This study nicely demonstrates that CMR analysis, looking for LGE in a pattern consistent with an ischemic etiology, provides excellent diagnostic accuracy for establishing the diagnosis of ischemic and nonischemic cardiomyopathy. This study also nicely demonstrates that these patients do not neatly fall into two subgroups of ischemic and nonischemic, but often have a combination of factors with bystander CAD or bystander infarct in patients where the extent of CAD is well under that required to result in substantial degrees of LV diastolic function. It is interesting to note in this study that when an expert panel has access to CMR data for identification of areas of fibrosis in a pattern consistent with coronary disease, that the diagnoses of ischemic and nonischemic cardiomyopathy often become somewhat blurred when compared to standard CA alone. As CMR provides information regarding myocardial scar and may alter management strategies with respect to implantable defibrillators, it may represent an ideal first screening tool for assessment of the etiology of onset HF in select patients.

Clinical Topics: Arrhythmias and Clinical EP, Heart Failure and Cardiomyopathies, Invasive Cardiovascular Angiography and Intervention, Noninvasive Imaging, Valvular Heart Disease, Atherosclerotic Disease (CAD/PAD), Implantable Devices, SCD/Ventricular Arrhythmias, Acute Heart Failure, Interventions and Coronary Artery Disease, Interventions and Imaging, Interventions and Structural Heart Disease, Angiography, Magnetic Resonance Imaging, Nuclear Imaging

Keywords: Coronary Artery Disease, Hypokinesia, Sensitivity and Specificity, Gadolinium, New York, Magnetic Resonance Imaging, Cicatrix, Coronary Stenosis, Cardiomyopathies, Risk Assessment, Magnetic Resonance Spectroscopy, Fibrosis, Uncertainty, Myocardial Ischemia, Defibrillators, Ventricular Dysfunction, Coronary Angiography, Chest Pain, Heart Failure, Heart Valve Diseases, Diastole, Heart Ventricles, Magnetic Resonance Angiography

< Back to Listings