DISP-36. Maintaining diagnostic accuracy in oncologic neuropathology in resource-limited settings: strategy for addressing an emerging major health disparity challenge

Abstract BACKGROUND The World Health Organization (WHO) Classification of Central Nervous System (CNS) tumors 5th edition (CNS5) and recent Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy-Not Official WHO (cIMPACT-NOW) updates (2024, 2025) incorporate recurrent molecular alterations into CNS tumor diagnostic and grading criteria. This tissue-based approach increases diagnostic accuracy and prognostic value. However, the escalating need for sophisticated time-consuming and costly molecular testing poses a challenge in resource-limited settings, contributing to a widening health disparity. An emerging critical need is for a cost-effective strategy for clinically-actionable brain tumor stratification in settings with limited availability of comprehensive molecular testing. METHODS We reviewed the WHO CNS5 (2021) and current cIMPACT-NOW guidelines (2024, 2025) for molecular parsing of primary CNS tumors, with the aim of identifying efficacious, pragmatic diagnostic strategies that emphasize utilization of informative clinical features, surrogate immunophenotyping, and high positive predictive value (PPV) routine imaging-derived molecular signatures (e.g., T2/FLAIR Mismatch; PPV approaching 100% for adult cerebral IDH-mutant diffuse astrocytic disease) into a single fully integrated diagnostic approach. RESULTS We identified essential minimal cost-effective testing, in sequential steps, for achieving reliable brain tumor diagnosis in low-resource settings. Our proposed algorithm provides a much-needed practical approach for low-resource settings, and is also applicable in high-resource settings when urgent treatment initiation is required, with no time for next generation sequencing (NGS) or DNA methylation profiling testing. CONCLUSIONS Identifying cost-effective surrogates to address the rapid advancement in the increasingly complex molecular landscape of CNS tumor classification is essential to reduce burgeoning cancer diagnosis inequities in low-resource settings. Affordable and easily accessible resources can be sufficient to attain this goal in a majority of cases. As a future direction, artificial intelligence (AI)-augmented analysis of H&E-stained tissue sections/routine MR imaging studies offers promise as a rapid and inexpensive surrogate for NGS/methylation profiling.