Left Ventricular Thrombus Detection by Routine Echocardiography: Insights Into Performance Characteristics Using Delayed Enhancement CMR
What is the diagnostic accuracy of routine transthoracic echocardiography (echo) for detection of left ventricular thrombus (LVT) compared to delayed-enhancement cardiac magnetic resonance imaging (DE-CMR)?
Both cine-CMR and DE-CMR were performed in 243 patients for detection of LVT. All patients had also undergone routine clinically indicated echocardiography with 1 week of CMR. All studies were read by two experienced independent blinded readers, and a third reader adjudicated discrepancies between the initial two readers. Thrombus was defined on DE-CMR as an avascular tissue density mass within the LV cavity appearing as a low signal attenuation mass surrounded by high signal intensity structures such as blood or hyper-enhanced myocardial infarction. For cine-CMR and echo, thrombus was defined as a mass within the LV cavity with distinct margins separating it from the ventricular endocardium papillary muscles, trabeculae, etc. Echoes considered positive for LVT were given a diagnostic quality score of low, medium, and high. Six-month follow-up was available for 216 patients for endpoints consistent with an embolic event, including stroke.
Average patient age was 60 years, and 63% were male. Ischemic cardiomyopathy was present in 63%, and atrial fibrillation in 19%; only 14% were on warfarin. Average LV ejection fraction was 28 ± 11%, and 60% of patients had an aneurysm. LVT was present in 24 (10%) on DE-CMR. Readings for LVT were discordant between the two initial readers in 1% of cine-CMR, 4% of DE-CMR, and 11% of echoes. Echocardiograms were interpretative if consistent with thrombus in 26 patients, only eight of whom had thrombus on DE-CMR. Compared with DE-CMR, the sensitivity and specificity were 33% and 91% with positive and negative predictive values of 29% and 93%. Echoes were performed for a clinical indication of evaluation of possible LVT in only 13 patients, in whom sensitivity and specificity were 60% and 88%. Cine-CMR had an overall sensitivity of 58% and specificity of 99% when compared with DE-CMR. Echo contrast for LV opacification was used in only 10 patients (4%). Of these 10, two had LVT on DE-CMR; echo was negative for thrombus in all 10 patients. Echo was more sensitive for detection of pertuberant than mural thrombus, with only one of seven mural thrombi being detected. The average size of thrombus detected by echocardiography was 6.6 cm3 compared to 2.0 cm3 for missed thrombus (p = 0.054). Echoes interpreted as showing LVT and concordant with DE-CMR had high confidence scores compared to those with discordant results. On follow-up, clinical events consistent with an embolic phenomenon occurred within 16.7% of those with LVT on DE-CMR, and in only 3.1% of those without LVT on DE-CMR (p = 0.02). Corresponding event rates for echo were 7.7% and 4.2% (p = 0.34).
Clinically indicated routine transthoracic echocardiography has a low yield for detection of LVT using DE-CMR as a standard for its detection. It is most accurate for identification of large pertuberant thrombus and less accurate for mural thrombus.
Multiple previous smaller studies have demonstrated the diagnostic accuracy of CMR imaging for identification of LV thrombi. This study directly compared routine, clinically indicated transthoracic echocardiography for detection of LVT in 243 patients and suggested a definite superiority of CMR for this purpose. Certainly the multi-modality imaging aspect of CMR, including delayed enhancement as well as high resolution visualization of the entire endocardial surface area, would confer an advantage to CMR for detection of thrombi, although the poor performance of echocardiography in this study is somewhat surprising. It should be emphasized that the ultrasound platforms utilized here were one and two generations beyond current state of the art, and that contrast for LV opacification was used in only 4% of patients. Additionally, the overwhelming majority of studies were not done for specific evaluation of thrombus, and the degree to which a more targeted echocardiogram for thrombus would have resulted in a higher yield remains conjectural. The doubling of sensitivity in the very small number of patients for whom LVT was the clinical indication suggests a role for more dedicated imaging. In addition to comparing the diagnostic accuracy of the two techniques, the authors have provided follow-up for clinically relevant events suggesting embolization from a cardiac source. The fivefold greater event rate in patients with thrombus on DE-CMR compared to those without certainly suggests an enhanced accuracy and prognostic role for the technique. This compares to the only twofold increase of event rates when echocardiography was used to predict presence of thrombus. While the performance of echocardiography may be understated in this study related to concerns expressed above, the results of this work should certainly give a clinician concern for absolute reliance on echocardiography for diagnosis or exclusion of LVT.
Keywords: Heart Diseases, Myocardial Infarction, Follow-Up Studies, Thrombosis, Cardiomyopathies, Magnetic Resonance Imaging, Echocardiography
< Back to Listings