Can Simple MRI Characteristics Predict Molecular Subgroups in Paediatric Medulloblastoma?

INTRODUCTION: Medulloblastoma is the most common malignant paediatric central nervous system (CNS) tumour and is classified using molecular subgroups: WNT, SHH and non-WNT/non-SHH. Subgroup identification guides prognosis and treatment, but requires molecular testing, which is not universally available. MRI is routine in assessment of CNS tumours, and the MRI characteristics may serve as surrogate markers for medulloblastoma subgrouping. This study aims to use simple MRI surrogate markers identified by Perreault et al., against local medulloblastoma cases and their final molecular outcome. METHOD: Retrospective cohort study of paediatric medulloblastomas at a single tertiary children's hospital between January 2015 and October 2025. MRI studies were reviewed by a paediatric neuroradiologist and a radiology registrar in consensus. Tumours were classified by location (cerebellar hemisphere, cerebellopontine angle/cerebellar peduncle (CP/CPA), or midline/fourth ventricle) and enhancement pattern (minimal/none = < 10%, heterogeneous = 10%-90%, solid = > 90%). Associations between MRI features and molecular subgroup were assessed. RESULTS: Twenty-two cases were analysed: SHH 22.7% (5/22), WNT 13.6% (3/22) and non-WNT/non-SHH 63.7% (14/22). Most tumours were midline (90.9%), with two hemispheric lesions and no CP/CPA. Enhancement patterns were heterogeneous (54.5%), solid (36.4%) and minimal/none (9.1%). Hemispheric location was associated with SHH tumours (p = 0.043), and all WNT tumours demonstrated solid enhancement (p = 0.036). No other statistically significant associations were identified. CONCLUSION: MRI characteristics offer promising surrogate markers for medulloblastoma subtype. The only strong predictor we found corresponding with Perreault et al. is that hemispheric lesions are correlated with SHH-activated medulloblastoma. This supports further investigation of MRI markers as non-invasive and preoperative medulloblastoma subgroup prediction.