Differentiating benign from malignant cardiac tumors with cardiac magnetic resonance imaging

Background: The purpose of this analysis is to describe the differences in cardiac magnetic resonance characteristics between benign and malignant tumors, which would be helpful for surgical planning. Methods: This was a prospective cohort study of 130 patients who underwent cardiac magnetic resonance imaging for evaluation of a suspected cardiac mass. After excluding thrombi and tumors without definitive diagnosis, 66 tumors were evaluated for morphologic features and tissue composition. Results: Of the 66 patients, 39 (59.0%) had malignant tumors and 27 (41.0%) had benign tumors. Patients with malignant tumors were younger when compared with those with benign tumors (age 51 years [42.8-60.0] vs 65 years [60.0-71.0] median). Malignant tumors more often demonstrated tumor invasion (69% vs 0% P <.001) and were more often associated with pericardial effusion (41% vs 7.4% P =.004). Presence of first-pass perfusion (100% vs 33% P <.001) and late gadolinium enhancement (100% vs 59.2%, P <.001) were significantly higher in malignant tumors. In logistic regression modeling, tumor invasion (P <.001) and first-pass perfusion (P <.001) were independently associated with malignancy. Furthermore, using classification and regression tree analysis, we developed a decision tree algorithm to help differentiate benign from malignant tumors (diagnostic accuracy ∼90%). The algorithm-weighted cost of misclassifying a malignant tumor as benign was twice that of classifying a benign tumor as malignant. Conclusions: Our study demonstrates that cardiac magnetic resonance imaging is a useful noninvasive method for differentiating malignant from benign cardiac tumors. Tumor size, invasion, and first-pass perfusion were useful imaging characteristics in differentiating benign from malignant tumors.