PTHP-15. Enhancing Medulloblastoma Classification: Integrating IHC and DNA Methylation Data from the SJMB12 Clinical Trial

Abstract The SJMB12 trial used immunohistochemistry (IHC)-based molecular grouping to separate medulloblastoma (MB) into WNT, SHH, and non-WNT/non-SHH (NWNS) groups for risk and treatment stratification. IHC was selected due to its rapid turnaround, cost-effectiveness, and low-tech nature, which allowed its immediate implementation in a prospective clinical trial. Nevertheless, the accuracy and reproducibility of IHC on a prospective cohort were unknown. Here, we compared IHC-based MB molecular group assignment to DNA methylation profiling for 634 evaluable patients from the SJMB12 trial. Robust concordance was observed. All 88 IHC-defined WNT MBs were classified as WNT by methylation. Of 107 IHC-defined SHH MBs, 106 were identified as SHH. Furthermore, 390 (98%) of 398 IHC-defined NWNS cases were identified as Group 3 or Group 4 MBs. Discordant cases were minimal: one IHC-defined SHH tumor was reclassified as glioblastoma by methylation, and 8 IHC-defined NWNS cases were reclassified (SHH (n=4), Pineoblastoma (n=2), WNT (n=1), or unclassified (n=1)). The performance metrics for IHC across MB groups were robust. For WNT, sensitivity was 93.6%, specificity 100%, PPV 100%, and NPV 98.9%. For SHH, sensitivity was 89.1%, specificity 99.8%, PPV 99.1%, and NPV 97.5%. For NWNS (proxy for Group 3/4), sensitivity was 94.9%, specificity 96.4%, PPV 98.0%, and NPV 91.1%. Notably, 41 (6%) tumors could not be grouped by IHC and were placed into an “indeterminate” category. These were assigned by methylation profiling to WNT (n=5), SHH (n=9), Group 3 (n=20), Group 4 (n=1), medullomyoblastoma (n=5), and GBM (n=1). Intriguingly, these cases were not random; they were enriched in MBs belonging to Group 3 subgroup 2, displaying myogenic and/or melanotic differentiation, and TP53-mutated SHH tumors. In conclusion, these findings demonstrate that IHC is a valuable tool for molecular grouping in medulloblastoma. Despite some limitations, it provides accurate results and, even when indeterminate, can reveal intriguing disease characteristics.