Abstract
Detecting homozygous deletion (HD) of CDKN2A is critical in BRAF-altered gliomas, as this molecular alteration has both diagnostic and prognostic significance. It is predominantly associated with BRAF-altered high-grade gliomas and has been associated with poorer prognosis in certain BRAF-altered low-grade glioma tumor types. The 2021 WHO classification of central nervous system tumors therefore recommends screening for this alteration in most BRAF-altered gliomas, but it does not recommend one specific technique over another. Here, we compare the performance of several detection methods, including p16 immunohistochemistry, fluorescence in situ hybridization (FISH), droplet digital PCR, next-generation sequencing and DNA methylation profiling-derived copy-number variation (CNV) analysis, in a retrospective cohort of 25 BRAF-altered gliomas. Ten cases showed diffuse p16 immunohistochemical expression (10/25) with no associated CDKN2A HD, whereas 15 cases had complete absence of p16 expression (15/25). In the latter group, a high level of discrepancy in CDKN2A HD detection when considering FISH versus other techniques was observed, suggesting a high false-negative rate with FISH. Using an original bioinformatic pipeline leveraging genome alignment of routinely available CNV raw data, we identified among most false-negative cases (4/5) a large and undeleted region encompassing MTAP, which is targeted by most commercial CDKN2A FISH probes. This is likely due to non-specific probe hybridization. Our finding suggests that FISH probes targeting the entire 9p21 locus may have lower sensitivity than anticipated among BRAF-altered gliomas and emphasizes the critical need for appropriate probe selection.