337 Diagnostic Performance of MRI and CSF Cytology in Detecting Leptomeningeal Metastasis and Recurrence in Medulloblastoma: Implications for Surveillance Strategies

Le résumé fourni par la source

INTRODUCTION: Postoperative surveillance in medulloblastoma is not standardized. Current data support obtaining one CSF sample 14–21 days after surgery and more frequent MRIs during the first year; however, the value of additional CSF cytologies and optimal MRI timing is unclear. Advances in MRI technology also warrant reevaluation of its diagnostic performance relative to CSF for detecting leptomeningeal dissemination (LMD). METHODS: 10,781 MRI and CSF reports from 407 medulloblastoma patients across two institutions were retrospectively reviewed. CSF and MRI reports obtained within 30 days of each other were compared for LMD detection. For sensitivity calculations, the gold standard was LMD or recurrence diagnosed by the oncologist. RESULTS: Mean age was 10.6 [10] years. Most were male (n = 248, 60.9%). Molecular subtypes were: Non-SHH/WNT (n = 160, 39.3%), SHH (n = 78, 19.2%), WNT (n = 39, 9.6%), unknown (n = 130, 31.9%). LMD was diagnosed in 31.4% (n = 128), including 18.7% (n = 76) postoperatively. Recurrence occurred in 29.7% (n = 121). Overall CSF-MRI concordance was 74.0%, highest in WNT (91.1%), vs. SHH (77.7%) and Non-WNT/SHH (70.4%) (p < 0.001). MRI more frequently identified positive or equivocal findings not seen on CSF at 3T (21.1%) vs. 1.5T (16.9%) (p = 0.02). Sensitivities for LMD detection were 85.5% (spine MRI), 49.4% (brain MRI), and 27.1% (CSF). For recurrence detection, sensitivity was 88.2% (spine MRI) and 82.8% (brain MRI). No difference in overall (p = 0.073) or LMD-free survival (p = 0.5) was found between patients with single vs. multiple postoperative CSF cytologies. CONCLUSIONS: Spine MRI is the most sensitive modality for detecting LMD and should be prioritized in surveillance. Multiple CSF cytologies offer limited value, and diagnostic sensitivity varies by subtype and MRI parameters. Current surveillance guidelines are based on limited data; this study provides a robust foundation for data-driven, individualized surveillance in medulloblastoma.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.