Diagnostic accuracy of neuropathological assessment across different diagnostic settings for high-grade gliomas: the added value of DNA methylation profiling

Abstract Background Accurate histopathological diagnosis is essential for the management of patients undergoing surgery for suspected high-grade glioma (HGG). We evaluated the diagnostic accuracy of pathological assessment performed in a general pathology department and a specialist neuropathology department, using EPIC-based DNA methylation classification as the operational reference standard. Methods This retrospective diagnostic accuracy study included 130 patients with complete results from three diagnostic assessments (selected from 173 screened records) with intracranial space-occupying lesions who underwent surgical biopsy or resection between May 2019 and February 2025. Tissue specimens were assessed in a general pathology department and subsequently evaluated independently in a specialist neuropathology department, followed by DNA methylation profiling using the Illumina Infinium MethylationEPIC BeadChip. HGGs (CNS WHO grades 3–4) were classified as positive and all other lesions as negative. Results Assessment in the general pathology department yielded 20 HGG classifications discordant with the EPIC reference (false negatives), including 13 tumors initially classified as low-grade gliomas, two as other intracranial lesions, and five as non-neoplastic brain tissue. Sensitivity was 79%, specificity 100%, positive predictive value 100%, negative predictive value 62%, and overall accuracy 84%. Specialist neuropathological assessment achieved a sensitivity of 100%, specificity of 97%, positive predictive value of 99%, negative predictive value of 100%, and overall accuracy of 99%. Three cases differed in specific classification or binary status: two HGG diagnoses were further refined as glioblastoma, IDH-wildtype, CNS WHO grade 4, while one specimen interpreted histopathologically as showing proliferative and mitotic activity was classified as tumor-free brain tissue by DNA methylation profiling. Conclusions Specialist neuropathological assessment showed high agreement with the EPIC-based reference classification for HGGs, whereas assessment outside a specialized neuropathology setting showed lower sensitivity for HGG identification in this selected cohort. DNA methylation profiling provides complementary diagnostic value by refining tumour classification.