Multidisciplinary review outperforms qualitative MR perfusion in distinguishing post-radiation brain metastasis...

Distinguishing recurrent brain metastases (rBrM) from radiation necrosis (RN) is of key clinical importance but is challenging given radiographic similarities. Magnetic resonance perfusion (MRP) imaging is a noninvasive tool used to differentiate these lesions post-stereotactic radiosurgery (SRS). In melanoma, melanocytic and hemorrhagic signal characteristics may confound MRP by yielding atypically low plasma volume. Recently, multidisciplinary teams (MDT) have been employed to guide care of complex brain metastasis cases. We sought to define the predictive values of MRP imaging and MDT consensus in distinguishing these pathologies in melanoma and a control arm of non-small cell lung cancer (NSCLC) brain metastases.

Patients with post-SRS enlarging lesions from melanoma and NSCLC that underwent surgical resection between September 2013 and March 2025 were identified. Radiographic analysis was performed by blinded neuroradiologists, and prospective MDT consensus opinions were compared to pathologic diagnoses.

Of 1,583 patients screened, 145 lesions from 131 patients met inclusion criteria. Among melanoma lesions (n = 76), 56 were rBrM and 20 RN; in non-small cell lung cancer (NSCLC; n = 69), 45 were rBrM and 24 RN. MRP demonstrated limited diagnostic performance in melanoma (sensitivity = 60%, specificity = 30%) and NSCLC (sensitivity = 73%, specificity = 33%). Quantitative perfusion parameters did not significantly differ between rBrM and RN in either group. MDT consensus improved performance over MRP alone, reaching sensitivity and specificity of 97% and 33% in the melanoma cohort, and 89% and 36% in NSCLC.

MRP has limited standalone diagnostic performance in distinguishing rBrM versus RN. Multidisciplinary consensus improves accuracy when integrated with MRP, highlighting the value of MDT input in guiding care of this complex population.