Novel Molecular Subgroups Improve Clinical Classification and Outcome Prediction for Childhood Medulloblastoma

Background - International consensus recognises four medulloblastoma molecular subgroups - WNT (MBWNT), SHH (MBSHH), Group 3 (MBGrp3) and Group 4 (MBGrp4) - each defined by their characteristic genome-wide transcriptomic and DNA methylomic profiles. Subgroups harbor distinct clinico-pathological and molecular features, underpin current disease sub-classification and initial subgroup-directed therapies are underway in clinical trials (i.e. reduced risk-adapted treatments for favorable-risk MBWNT patients; SMO inhibitors for MBSHH patients). However, significant biological heterogeneity and differences in survival are apparent within each subgroup, which remain to be resolved. Methods - We undertook comprehensive molecular profiling and unsupervised class discovery (non-negative matrix factorization, t-SNE) of test and validation cohorts (n=704 in total), to identify consensus primary molecular subgroups within childhood medulloblastoma (3.0 years). Findings - Seven robust and reproducible primary molecular subgroups of childhood medulloblastoma were identified, characterized by distinct biological/clinical features. For instance, MBSHH comprised two age-dependent subgroups, while MBGrp3 and MBGrp4 each split into two subgroups with significantly different survival rates. Survival analysis identified secondary features predictive of outcome. Cross-validated subgroup-dependent models incorporating these novel subgroups along with secondary features and established disease risk-factors, outperformed current disease risk-stratification schemes. These schema stratified patients into four clinical risk-groups - favorable-risk (91% 5-year survival, 25% of patients), standard-risk (81%, 23%), high-risk (42%, 38%) and very high-risk (28%, 13%) - to be considered for treatment reduction, intensification or novel therapies respectively. Interpretation - The discovery of seven novel, clinically-significant, subgroups significantly improves disease risk-stratification and provides a new foundation for future research and clinical investigations.