Molecular subgrouping of ependymoma across three anatomic sites and their prognostic implications

The 2021 WHO classification stratifies ependymoma (EPN) into nine molecular subgroups according to the anatomic locations which outperforms histological grading. We aimed at molecularly reclassifying 200 EPN using immunohistochemistry (IHC) and sequencing for ZFTA fusions in supratentorial (ST) EPN. Further, we assessed the utility of L1CAM, cyclinD1, and p65 markers in identifying ZFTA fusion. Demographic profiles, histologic features, molecular subgroups and clinical outcome were retrospectively analyzed. IHC for L1CAM, cyclinD1, p65, H3K27me3, and H3K27M and sequencing for ZFTA fusion were performed. ZFTA fusions were identified in 44.8% ST EPN. p65 displayed the highest specificity (93.8%), while L1CAM had the highest sensitivity (92.3%) in detecting ZFTA fusions. The negative predictive value approached 96.6% and sensitivity improved to 96.2% with combinatorial IHC (L1CAM, cyclinD1, p65). H3K27me3 loss (PF-A) was noted in 65% PF EPN. Our results provide evidence that a combination of two of three (L1CAM, p65, and cyclinD1) can be used as surrogate markers for predicting fusion. ZFTA fusion, and its surrogate markers in ST, and H3K27me3 and younger age (< 5 years) in PF showed significant correlation with PFS and OS on univariate and Kaplan-Meier analysis. On multivariate analysis, H3K27me3 loss and younger age group are associated with poor clinical outcome.