Background: Radiolabeled antibody specific for cardiac myosin administered intravenously has been used to define noninvasively regions of myocardial necrosis. Inflammatory heart disorders such as myocarditis and heart transplant rejection demonstrate diffuse and often faint myocardial uptake of antimyosin antibody. This study was undertaken to evaluate the reproducibility and diagnostic accuracy of antimyosin antibody imaging for the detection of patients with suspected myocarditis.
Methods and results: Fifty antimyosin scans, performed consecutively in patients with suspected myocarditis, were evaluated by one independent observer and two panels of observers. Antimyosin scan interpretations were compared with endomyocardial biopsy results and also with serial changes in left ventricular function. An independent observer (A) and a panel of five observers (A through E) interpreted the antimyosin scans as positive or negative on the basis of both planar images and tomographic reconstructions. Three of the five observers (A through C) again interpreted the scans but based interpretation only on planar images. Blinded random sequence evaluation of antimyosin scans based on the planar and tomographic interpretations revealed moderate agreement between the independent observer (A) and the group of observers (A through E) (kappa = 0.58). There was also moderate agreement between interpretations based on planar images alone and interpretations based on both planar and tomographic images (kappa [A through E]/[A through C] = 0.57; kappa [A through C]/A = 0.48). Comparison of antimyosin scan results with histologic evidence of myocarditis in endomyocardial biopsy specimens demonstrated that all scan results obtained from the individual or the panels of observers had a very high sensitivity (91% to 100%) and a high negative predictive value (93% to 100%). The specificity (31% to 44%) and positive predictive value (28% to 33%) were less impressive. We also compared the scan and biopsy results with the composite clinical standard of significant left ventricular functional improvement. Endomyocardial biopsy demonstrated poor sensitivity (35%) compared with antimyosin scans (82% to 94%) but had superior specificity (endomyocardial biopsy, 79%; antimyosin scan, 25% to 42%). The specificity of interpretations based on planar and tomographic interpretations (38% to 42%) was better than the planar images alone (25%). If reversible left ventricular dysfunction is considered clinical evidence of myocarditis, this study suggests that a negative endomyocardial biopsy significantly misses the presence of the disease. On the other hand, a negative antimyosin scan almost invariably excludes myocarditis.
Conclusions: This study demonstrates a high degree of interobserver reproducibility of antimyosin interpretation. Comparison of the scintigraphic results with histologic and clinical standards indicates a high sensitivity of antimyosin scans for the detection of myocarditis. The antimyosin scan is also not likely to miss clinically or pathologically diagnosed myocarditis, in contrast to the endomyocardial biopsy, which missed clinically validated myocarditis 65% of time. The combination of high sensitivity and negative predictive value suggests that antimyosin scintigraphy may be an effective screening procedure for obviating biopsies in patients with suspected myocarditis.