Abstract
The power and widespread use of next-generation sequencing (NGS) in surgical neuropathology has raised questions as to whether NGS might someday fully supplant histologic-based examination. We therefore sought to determine the feasibility of relying on NGS alone for diagnosing infiltrating gliomas. A total of 171 brain lesions in adults, all of which had been analyzed by GlioSeq NGS, comprised the study cohort. Each case was separately diagnosed by 6 reviewers, based solely on age, sex, tumor location, and NGS results. Results were compared with the final integrated diagnoses and scored on the following scale: 0 = either wrong tumor type or correct tumor type but off by 2+ grades; 1 = off by 1 grade; 2 = exactly correct. Histology alone was treated as a seventh reviewer. Overall reviewer accuracy ranged from 81.6% to 94.2%, while histology alone scored 87.1%. For glioblastomas, NGS was more accurate than histology alone (93.8%-97.9% vs 87.5%). The NGS accuracy for grade II and III astrocytoma and oligodendroglioma was only 54.3%-84.8% and 34.4%-87.5%, respectively. Most uncommon gliomas, including BRAF-driven tumors, could not be accurately classified just by NGS. These data indicate that, even in this era of advanced molecular diagnostics, histologic evaluation is still an essential part of optimal patient care.