The biology behind the measurement is central to why this matters. MGMT encodes O6-methylguanine–DNA methyltransferase, a repair enzyme that directly reverses the kind of DNA damage temozolomide inflicts. When the gene’s promoter is silenced by methylation — the addition of methyl groups to CpG dinucleotides in the regulatory region — tumor cells produce little of the repair protein and remain vulnerable to the drug. Methylation of the promoter is therefore a predictive biomarker: patients whose tumors are hypermethylated derive greater benefit from temozolomide, showing improved progression-free and overall survival, while those with unmethylated tumors respond poorly. In the study’s large molecular cohort, tumors classified as hypermethylated at the first time point carried a mean methylation percentage of 35.6 percent, ranging up to 89 percent, whereas unmethylated tumors averaged just 2.1 percent — a wide biological gulf that the binary labels tend to conceal.

To interrogate how that gulf behaves over the course of disease, the researchers, led by Henry Noren and Morana Vojnic of Rutgers Cancer Institute and collaborating institutions, turned to a large commercial molecular profiling database maintained by Caris Life Sciences. They identified patients with recurrent glioma who had undergone molecular testing at two distinct time points and, after excluding cases with insufficient tissue, assembled 426 paired samples. MGMT promoter methylation was assessed by pyrosequencing of CpG sites 74 through 78, following bisulfite conversion of DNA extracted from formalin-fixed paraffin-embedded tissue with at least 50 percent tumor nuclei. The laboratory reported both a quantitative methylation percentage and a categorical status: hypermethylated at 9 percent or above, equivocal between 7 and under 9 percent, and unmethylated below 7 percent. Equivocal cases were assigned to the unmethylated group for analysis, in line with prior studies.

The headline finding is strikingly simple. In 84 percent of tumors — 358 of 426 — the categorical MGMT status was identical at diagnosis and at recurrence, with 52 percent remaining unmethylated and 32 percent remaining hypermethylated. Only 16 percent changed status, and the direction of change was uneven: 11 percent shifted from hypermethylated to unmethylated, while just 5 percent gained hypermethylation. More importantly, when the researchers examined the underlying quantitative values, the average change in methylation percentage across the whole cohort was a modest minus 1.4 percentage points, with a standard deviation of 16.0. Tumors that kept their original classification barely moved at all, drifting by only minus 0.3 points in the consistently unmethylated group and plus 0.6 points in the consistently hypermethylated group. The epigenetic landscape, in other words, is largely frozen between the two surgeries.

The apparent transitions told a different and more instructive story. Tumors that flipped from hypermethylated to unmethylated had baseline methylation levels on average 13.5 percentage points lower than tumors that stayed hypermethylated, a difference that was highly statistically significant. Tumors that gained hypermethylation at recurrence had recurrence values on average 16.3 points lower than consistently hypermethylated tumors. In plain terms, the tumors that changed categories were clustered near the classification thresholds, where a small quantitative shift — whether from intratumoral heterogeneity, differences in sampling, assay variability, or modest epigenetic drift — is enough to move a tumor across the line. The authors conclude that most categorical status changes reflect variability at the margins of classification rather than true epigenetic reprogramming under therapeutic pressure, a distinction with real consequences for how clinicians should read a changed report.

The study also mapped methylation dynamics onto the temporal arc of the disease, using the interval between the two tissue collections as a surrogate for disease course. Among patients whose samples were collected at least three months apart, tumors that remained hypermethylated, lost hypermethylation, or gained hypermethylation all showed longer intervals between samples than tumors that stayed unmethylated. Consistently hypermethylated tumors had intervals 6.7 months longer than consistently unmethylated ones, and tumors that gained hypermethylation showed an 8.2-month difference, with both comparisons reaching statistical significance. The researchers interpret this descriptively: it mirrors the established association between MGMT hypermethylation and better response to alkylating chemotherapy, rather than implying that methylation changes themselves alter the disease trajectory. An exploratory geographic analysis added a curious footnote — frontal lobe tumors were more likely to remain hypermethylated at recurrence, at 41 percent, and less likely to remain unmethylated, at 45 percent, compared with tumors elsewhere in the brain.

To connect the molecular findings to actual clinical behavior, the team analyzed a separate retrospective cohort of 27 adult patients from two academic institutions, Rutgers Cancer Institute and Lenox Hill Hospital, with detailed treatment records. These patients were predominantly diagnosed with glioblastoma, which accounted for 74.1 percent of cases, and nearly all — 96.3 percent — received temozolomide after their initial surgery regardless of MGMT status, consistent with standard of care. At recurrence, 44.4 percent of the cohort was re-treated with the drug. The pattern of who got re-treated tracked the biomarker: temozolomide rechallenge was pursued in 54.5 percent of patients whose tumors remained hypermethylated, compared with 41.7 percent of those whose tumors stayed unmethylated, though the difference did not reach statistical significance in this small sample. Re-irradiation showed the inverse pattern, being more common among patients with consistently unmethylated tumors.

That treatment asymmetry echoes findings from larger clinical trials, including the DIRECTOR trial, which demonstrated that MGMT promoter methylation is a strong predictor of benefit from dose-intensified temozolomide rechallenge in progressive glioblastoma. Retrospective analyses have likewise shown that patients with retained hypermethylation gain more from a second course of the drug, while those with unmethylated tumors are more often directed toward alternative systemic therapies, clinical trials, re-irradiation, or supportive care. What the new study adds is real-world evidence that clinicians are, in fact, weighing MGMT status at recurrence when making these calls — and a quantitative argument that the status they weigh is usually the same one measured months or years earlier at diagnosis.

The implications for laboratory practice may prove to be the study’s most durable contribution. Because binary classification remains the norm in clinical reporting, tumors with borderline methylation — those hovering near the 7 and 9 percent cutoffs — can be labeled in ways that obscure their true position on what is increasingly understood as a quantitative continuum. Prior work has shown that intermediate or low-level methylation is associated with less predictable response to temozolomide and diminished prognostic value compared with clearly unmethylated or robustly hypermethylated tumors. The new data suggest these borderline tumors form a biologically less stable subgroup, vulnerable to reclassification between surgeries for reasons that may have little to do with genuine epigenetic evolution. Routine quantitative reporting, the authors argue, would give oncologists the context needed to interpret a changed status appropriately — and to recognize when a change is more artifact than biology.

The authors are careful to acknowledge the limits of their analysis. Both cohorts were retrospective, introducing potential selection and sampling biases; the interval between surgeries is an imperfect proxy for progression-free survival; the commercial database lacked treatment details, precluding adjustment for therapy-related confounders; and the 27-patient clinical cohort was too small and underpowered for survival analyses, making its findings exploratory. The molecular data also cannot fully disentangle the contributions of intratumoral heterogeneity, sampling variability, and selective pressure from chemotherapy. Still, with one of the largest paired cohorts of recurrent glioma assembled to date and the integration of quantitative methylation percentages alongside categorical labels, the study makes a compelling case that MGMT assessment at diagnosis remains clinically reliable — and that reassessing it at recurrence, with numbers rather than just labels, could sharpen one of the most difficult decisions in brain tumor medicine.

Subject of Research: Stability of MGMT promoter methylation in recurrent glioma and its implications for temozolomide rechallenge

Article Title: Quantitative stability of MGMT promoter methylation in recurrent glioma and implications for temozolomide rechallenge

Article References: Noren, H., Pelofsky, G., Suffren, B., Mittelman, L., Yohn, C., Twyman, A., Febres-Aldana, C. A., Abramyan, A., Sertil, A., D’Amico, R. S., Sherman, J. H., & Vojnic, M. (2026). Quantitative stability of MGMT promoter methylation in recurrent glioma and implications for temozolomide rechallenge. Journal of Neuro-Oncology, 179(3), Article 85. https://doi.org/10.1007/s11060-026-05778-y

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05778-y

Keywords: glioma, glioblastoma, MGMT promoter methylation, temozolomide, epigenetics, DNA methylation, biomarker, pyrosequencing, recurrent brain tumor, chemotherapy rechallenge, precision oncology, neuro-oncology