Assessment of cardiac masses by magnetic resonance imaging: prognostic value and agreement with histopathology

Background

Cardiovascular magnetic resonance (CMR) is a highly valuable tool for evaluating cardiac masses (CM). However, data on its ability to predict patient outcomes remain limited. Therefore, our objective was to assess the accuracy of CMR in determining malignancy, its association with mortality, and its agreement with histopathological analysis.

Methods

This was an observational, retrospective study. We examined patients who underwent CMR due to suspected CM from January 2004 to June 2023 at an university hospital. Patients with suspected infarction-related thrombosis were excluded. Data were collected from electronic medical records. Images were reanalyzed in a blinded manner by two cardiac imaging specialists, documenting predefined imaging characteristics to classify masses as malignant or non-malignant (including cysts, thrombi, and normal variants), leading to a presumptive diagnosis. Mortality rates across groups were compared using survival analysis and Cox regression. In cases with histological confirmation, agreement between the presumptive CMR diagnosis and the final histological diagnosis was evaluated using the Cohen’s Kappa coefficient.

Results

We identified 75 patients with suspected CM, of which 24 (32 %) were classified as malignant and 51 (68 %) as non-malignant. Imaging variables most strongly associated with malignancy included the presence of multiple masses, involvement of multiple chambers, signs of infiltration, pericardial effusion, perfusion abnormalities, and late gadolinium enhancement. In contrast, mass mobility was associated with a non-malignant diagnosis.

The median follow-up was 30 months [IQR 4-67.5]. Malignant masses identified by CMR were associated with higher mortality: (HR: 3.72; 95 % CI: 1.8–7.72, p < 0.001). Histopathological studies were performed in 34 patients (45 %) and compared with the presumptive etiological diagnosis obtained by CMR. The level of agreement was excellent for malignancy (k = 0.88, p < 0.001) and good for etiological diagnosis (k = 0.63, p < 0.001).

Conclusions

Malignancy of a CM, as determined by CMR, was associated with higher mortality. There was good agreement between the presumptive diagnosis by CMR and the histopathological findings.