By Michael H. Crawford, MD, Editor

SYNOPSIS: Among patients with suspected cardiac tumors, cardiac MRI was highly accurate at distinguishing tumor from thrombus and benign from malignant tumors, using subsequent clinical data over five years of follow-up as the diagnostic standard.

SOURCE: Shenoy C, Grizzard JD, Shah DJ, et al. Cardiovascular magnetic resonance imaging in suspected cardiac tumor: A multicenter outcomes study. Eur Heart J 2021; Sep 21:ehab635. doi: 10.1093/eurheartj/ehab635. [Online ahead of print].

Although believed to be the gold standard for diagnosing the etiology of cardiac masses, there is a paucity of data that correlates cardiac MRI findings with outcomes. Investigators from four large U.S. academic medical centers prospectively enrolled 935 patients referred to cardiac MRI for suspected cardiac tumors between 2003 and 2014. The authors collected clinical information and patients for the primary endpoint of all-cause mortality. All four sites used the Society for Cardiovascular Magnetic Resonance protocol, which included late gadolinium enhancement (LGE) imaging.

Cardiac MRI interpretations were conducted blinded to the clinical data and were categorized into five categories: no mass, pseudomass, thrombus, benign tumor, and malignant tumor. Pseudomasses were prominent normal structures, such as the crista terminalis in the right atrium. In 32 patients, the masses did not fit into the five categories (e.g., vegetations), and they were excluded. The median age of the remaining 903 patients was 60 years, and 36% were men. The most common preceding imaging study was echocardiography (78%).

The cardiac MRI diagnosis was no mass in 25%, pseudomass in 16%, thrombus in 16%, benign tumor in 17%, and malignant tumor in 23%. In comparison to the final clinical diagnosis, the cardiac MRI diagnosis was correct 98% of the time. One of the 236 patients categorized as no mass exhibited a small mobile mass on the mitral chordal apparatus that was diagnosed as a papillary fibroelastoma. All 149 patients categorized as pseudomass proved to be correct. Four of the 146 patients categorized as thrombus turned out to have benign tumors, three of which were myxomas. In the five patients incorrectly diagnosed as benign tumors, two that were called myxomas included thrombi and two were malignant tumors. The four of 213 called malignant tumors turned out to be benign.

During the almost five-year median follow-up, 376 patients died. In comparison to the no mass group, the pseudomass group recorded a similar mortality rate (HR, 1.03; 95% CI, 0.70-1.51), as did those with a benign tumor diagnosis (HR, 0.77; 95% CI, 0.50-1.17). Those with thrombus (HR, 1.46; 95% CI, 1.00-2.11; P = 0.48) and those with a malignant tumor (HR, 3.31; 95% CI, 2.40-4.57; P < 0.001) recorded higher mortality rates. Also, the cardiac MRI diagnosis provided incremental prognostic value vs. other clinical parameters, such as left ventricular ejection fraction and extracardiac malignancy. The authors concluded cardiac MRI carries a high rate of diagnostic accuracy for cardiac masses and is a strong independent predictor of mortality.

COMMENTARY

The persistent problem with new cardiac imaging techniques that tout better diagnostic accuracy and improved patient outcomes is deciding what gold standard to use for comparison. Often, there is not such a standard, but because the pictures look better, some might believe it must be a superior technique. This is certainly the case for cardiac MRI for the diagnosis of suspected cardiac tumors.

Although believed to be the best current imaging technique for this purpose, there is a lack of solid clinical research to back this up. Thus, this multicenter study, which included an analysis of subsequent clinical data (biopsies and a five-year follow-up to validate the imaging diagnoses of masses suspected to be tumors), is of interest. Not surprisingly, diagnostic accuracy was high, but examining the errors is instructive. One of the 236 patient masses categorized as no mass was incorrect and was caused by a mobile mass attached to the mitral chordae, which was shown to be a papillary fibroelastoma at surgery. This mobile mass appeared on echocardiography but not on the cardiac MRI, probably because of motion averaging. Thus, echo might be a better tool to identify mobile masses.

None of the 149 patients categorized as pseudomasses were incorrect. The most common pseudomasses were hypertrophied interatrial septum, prominent epicardial fat pad, prominent Eustachian valve, prominent crista terminalis, and hiatal hernia. Perhaps a more robust echo lab would have prevented a cardiac MRI for these patients. Four patients diagnosed as thrombus were incorrect; four were benign tumors, of which three were myxomas. Five of 159 patients diagnosed as benign tumor were incorrect; two were thrombi diagnosed as myxomas, and two were malignant tumors. Therefore, the distinction between myxomas and thrombus may be a weak area. This is important because the treatment of thrombi and myxomas is different. However, the investigators excluded patients with obvious thrombi to prevent overloading the study population with the more common thrombi, which makes the distinction between tumor and thrombus more challenging. For example, the post-myocardial infarction patient with an apical aneurysm and a mass would have been excluded as obvious thrombus. Four of the 213 patients diagnosed as malignant tumor all had benign tumors, mainly myxomas. Although most myomas are benign, they can be friable and result in systemic emboli, the work up for which is often how they are discovered.

All that said, the authors did not systematically compare echo vs. cardiac MRI. Only patients referred for cardiac MRI were included, so there is a selection bias for suspected tumors. There are no data on primary vs. metastatic malignant tumors. Also, researchers only examined all-cause mortality, not diagnosis-specific mortality. In addition, T1 and T2 mapping were not used, as this technique was not available for the entire study period and may have helped better examine the tissue characteristics of the masses, but probably would not have altered the basic distinction between tumor and thrombus much. Overall, the study demonstrated a 99% accuracy for distinguishing tumor from thrombus and a 98% accuracy for distinguishing benign from malignant tumors. Thus, for discriminating between cardiac masses, cardiac MRI clearly is robust.