Age-Dependent Attenuation of the MGMT Methylation Survival Advantage in IDH-Wildtype Glioblastoma
O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation is an established prognostic and predictive biomarker in glioblastoma (GBM), yet its clinical utility may not be uniform across age groups. We analyzed 264 IDH-wildtype GBM patients from The Cancer Genome Atlas with complete data on MGMT promoter methylation status, transcriptomic subtype, and overall survival. Using Cox proportional hazards regression, we tested whether patient age modifies the prognostic effect of MGMT methylation. In a multivariable model adjusting for transcriptomic subtype, the interaction between MGMT methylation and age >=65 was statistically significant (HR = 2.14, 95% CI: 1.23-3.72, p = 0.01), indicating that the protective association of MGMT methylation diminishes substantially in elderly patients. Stratified analyses confirmed this pattern: MGMT methylation was associated with a significant survival advantage in patients aged 55-64 (HR = 0.46, 95% CI: 0.26-0.81, p = 0.007) but conferred no detectable benefit in patients >=65 (HR = 1.00, 95% CI: 0.60-1.67, p = 0.997). The age-dependent attenuation was consistent across Classical, Mesenchymal, and Proneural transcriptomic subtypes. These findings suggest that MGMT methylation status should be interpreted in the context of patient age, with implications for clinical decision-making in elderly GBM patients.
Reviews
This study uses TCGA (n=264) IDH-wildtype glioblastomas to test whether age modifies the survival association of MGMT promoter methylation, using Cox regression with an MGMT×(age≥65) interaction and adjustment for transcriptomic subtype. The reported interaction (HR~2.14, p=0.01) and age-stratified estimates—clear benefit in ages 55–64 (HR~0.46) but none in ≥65 (HR~1.00)—are coherent and clinically plausible, and the attempt to check consistency across subtypes is a strength. As presented, the main conclusion (“the MGMT survival advantage attenuates in the elderly”) is directionally supported by the reported model outputs. However, the excerpt provides too little methodological detail to judge whether the interaction is robust rather than an artifact of confounding, selection, or modeling choices. Key omissions include treatment variables (temozolomide/radiation intensity, hypofractionation, trial vs non-trial care), performance status, extent of resection, and comorbidity—factors strongly correlated with age and survival and potentially with MGMT testing/availability in TCGA. The dichotomization at 65 and selective reporting of a 55–64 stratum raise concerns about multiple testing and sensitivity to cutpoints; proportional hazards and functional form of age (continuous vs categorical) are not addressed. Without a pre-specified analysis plan, missing-data handling, and external validation (e.g., CGGA/clinical cohorts), the evidence supports a hypothesis of effect modification but is not yet sufficient for practice-changing claims about interpreting MGMT differently in older patients.