Noninvasive MRI Spectroscopy Classifies Pediatric Medulloblastoma Molecular Subgroups

Medulloblastoma, the most common malignant pediatric brain tumor, requires precise molecular classification to guide therapy and predict outcomes. Current subgrouping typically depends on surgical sampling and specialized assays, which can delay treatment planning and add procedural risk. Faster, noninvasive methods that deliver subgroup information before surgery could streamline care. Researchers have now shown that proton magnetic resonance spectroscopy integrated with routine MRI can identify medulloblastoma molecular groups presurgically.

Investigators at Children’s Hospital Los Angeles evaluated in vivo proton magnetic resonance spectroscopy (1H-MRS) performed as part of standard MRI to distinguish the four molecular subgroups—WNT-activated, SHH-activated, group 3, and group 4—before surgery. The method targets tumor metabolism to derive subgroup-specific spectral signatures. The findings are reported in a study published in Neuro-Oncology.

The approach analyzes metabolite patterns captured during 1H-MRS to infer molecular identity. Distinct profiles were observed across subgroups: group 3 tumors showed elevated taurine and higher creatine-to-choline ratios, whereas group 4 tumors demonstrated increased choline and glycine/myo-inositol signals. WNT-activated tumors were marked by elevated choline and γ-aminobutyric acid (GABA) with low taurine, while SHH-activated tumors exhibited low creatine, low or absent taurine, and elevated glutamate-related metabolites.

In a retrospective design, the team analyzed pretreatment spectra from 95 pediatric patients with molecularly characterized medulloblastoma. Spectroscopy was integrated into routine MRI and required approximately five additional minutes of imaging. The MRS-derived metabolic profiles were consistent with prior ex vivo metabolomic findings, supporting their biological relevance as imaging biomarkers.

To quantify diagnostic performance, machine learning was applied to MRS features for subgroup classification. An ensemble model achieved a mean cross-validated area under the curve (AUC) of 0.94. The results indicate that in vivo 1H-MRS can provide a rapid, noninvasive means of presurgical molecular stratification. By integrating seamlessly into existing MRI protocols, the technique may complement tissue-based diagnostics and inform risk-adapted decisions in pediatric neuro-oncology.

Related Links
Children’s Hospital Los Angeles