Using variant allelic fraction of driver mutations to determine the reliability of MGMT promoter methylation testing in adult-type diffuse gliomas.

Abstract

e14034 Background: Methylation of the O6-methylguanine-DNA methyltransferase ( MGMT) promoter predicts response to temozolomide (TMZ) in adult-type diffuse gliomas. MGMT testing is thus standard practice. However, accurate results depend on adequate tumor cellularity in the analyzed sample. If the sample contains too many admixed nonneoplastic cells (which are consistently MGMT unmethylated), a false-negative result may be generated. Currently, ̃70% tumor cells are considered the minimum for reliable testing, but that is based on visual estimation by light microscopy; such estimation varies between observers and is often inaccurate. Since next-generation sequencing (NGS) is also frequently done on adult-type diffuse gliomas, and such tumors usually have heterozygous driver mutations known to be present in nearly 100% of tumor cells (IDH mutations in astrocytomas and oligodendrogliomas; TERT promoter mutations in ̃85% of IDH wild-type GBM), we sought to determine whether the variant allelic fraction (VAF) of these mutations could improve quality assurance in MGMT promoter methylation testing. Methods: We used an in-house cohort of 175 WHO gliomas (112 IDHwt, TERTmut GBM, 34 WHO grade 2-4 IDHmut astrocytomas, 29 WHO grade 2-3 IDHmut oligodendrogliomas), all with MGMT pyrosequencing and NGS data obtained from the same tissue blocks. Gliomas with average 10% methylation across all four CpG sites in the MGMT promoter were considered positive, as per standard cutoffs for that assay. Results: Overall, there was a positive correlation between driver VAF and amount of MGMT promoter methylation (slope = 0.25, R2= 0.33, P = 0.016). The largest dropoff in average methylation was when VAF was below 13% (7.4% average CpG methylation when VAF < 13% vs. 19.0% when VAF≥13%, P= 0.029). Based on our prior published study (PMID: 33830235), 36.5% of IDHwt GBMs, 86.7% of IDHmut astrocytomas, and 98.6% of IDHmut oligodendrogliomas should have MGMT promoter methylation. Among our GBMs, only 1/7 (14.0%) below a TERT mutant VAF of 13% reached the cutoff for MGMT promoter methylation, whereas 39/105 (37.0%) of GBMs VAF≥13% did. However, this did not reach statistical significance, likely because our cohort was underpowered for GBMs with TERTmut VAF < 13%. Among 34 IDHmut astrocytomas, only 1 had a VAF below 13%, and was reported as MGMT promoter unmethylated; all others had VAF between 21-56%, and showed consistent rates of MGMT methylation positivity across that entire range. All 29 oligodendrogliomas had VAFs of at least 26%; 2/29 were MGMT-negative, and both had VAFs at or above 36%. Conclusions: Additional cases are being prospectively collected to boost statistical power. Still, our existing data suggest that, when driver mutation VAF falls below ̃13% (i.e., tumor cellularity ̃26%), the risk of false-negative MGMT methylation results increases, and caution is warranted in such cases.