Tumor heterogeneity poses a major challenge to the precision treatment of medulloblastoma (MB). Rapid and accurate subtyping tools are urgently needed for informed clinical decision-making. Herein, we demonstrate the utility of Raman spectroscopy (RS) as a label-free, noninvasive approach for molecular characterization and subtype discrimination of MB at the single-cell level. Five representative MB cell lines from group 3 and sonic hedgehog (SHH) subtypes, along with microglial controls, were profiled by RS and validated using mass spectrometry-based metabolomics. RS-derived single-cell metabolic fingerprints revealed subtype-specific variations in nucleic acids, amino acids, lipids, and proteins. Group 3 cells, exemplified by metastatic D283, exhibited elevated levels of unsaturated lipids compared with most SHH cells. Notably, Daoy cells from the SHH subtype displayed unsaturation levels comparable to D283, reflecting intragroup heterogeneity and membrane remodeling in MB. Machine learning classifiers achieved high diagnostic performance, with an average area under the curve of 0.994 and a Matthews correlation coefficient of 0.935. Furthermore, cisplatin-treated D283 cells showed increased unsaturated lipid content relative to Daoy cells, revealing subtype-dependent metabolic shifts in drug response. These findings underscore the metabolic diversity of MB subtypes and the role of lipid metabolism in tumor progression and therapy. Collectively, our results exemplify the power of RS combined with metabolomics for molecule-resolved, label-free MB subtyping and therapeutic stratification.
Peng et al. (Fri,) studied this question.