Temozolomide use in elderly patients with MGMT promoter unmethylated glioblastoma: Is it finally time to dismount a dead horse?

Temozolomide (TMZ) remains the only chemotherapy agent associated with a survival benefit in glioblastoma (GBM). In the landmark “Stupp” study, the median overall survival (mOS) was 14.6 months for patients treated with radiation therapy (RT) and TMZ compared to 12.1 months for patients treated with RT alone.¹ Epigenetic silencing via methylation within the CpG island of the promoter region of the O6-methylguanine DNA methyltransferase (MGMT) gene was associated with significantly improved outcomes in the RT/TMZ arm. Patients with MGMT promoter methylated tumors treated with RT/TMZ had an mOS of 21.7 months as compared with 15.6 months for those treated with RT alone, while those with MGMT promoter unmethylated tumors had a marginal survival benefit (12.7 vs. 11.8 months, respectively; P = .06).² However, in this trial, 10% of patients with MGMT unmethylated tumors treated with RT/TMZ were alive at 3 years. Coupled with the mild toxicities and absence of compelling alternatives, these data have driven the routine use of TMZ in clinical practice regardless of MGMT status.

Treatment of elderly and frail GBM patients represents a special clinical challenge where toxicity must be balanced with potential benefits. Due to poor tolerance of standard RT (60 Gy in 30 fractions) among elderly patients, hypofractionated RT (34 Gy in 3.4 fractions) was evaluated in the Nordic study, in which hypofractionated RT was associated with improved mOS compared to standard RT (7.0 vs. 5.2 months, respectively).³ TMZ alone compared to radiotherapy alone was studied among elderly patients in the German Neuro-oncology Working Group NOA-08 study. TMZ was noninferior to RT alone with an mOS of 8.6 versus 9.6 months, respectively.⁴ Combinatorial therapy was studied in the Canadian CE.6 study. mOS was longer with the addition of TMZ compared to hypofractionated RT (40 Gy in 15 fractions) alone (9.3 vs. 7.6 months, respectively), and patients with MGMT methylated tumors treated with TMZ had the greatest benefit (13.5 vs. 7.6 months).⁵ mOS also was modestly prolonged among patients with MGMT promoter unmethylated tumors who received TMZ (10.0 vs. 7.9 months).

In a prior analysis, Hegi et al. used quantitative MGMT methylation PCR data from 4 randomized clinical trials to refine the cutoffs used to define unmethylated GBM.⁶ In this issue of Neuro-Oncology, Hegi et al. expand their work to elderly patients with GBM treated as part of the Nordic, NOA-08, and CE.6 trials.⁷ They report a lack of benefit from TMZ among elderly GBM patients with “truly” unmethylated MGMT promoter. The retrospective analysis evaluates survival data from elderly patients with available quantitative MGMT methylation results, who are then classified as “methylated,” “gray zone,” or “truly unmethylated.” An important limitation of this approach is that the quantitative MGMT data was only available for about half of the patients treated in the clinical trials. Acknowledging this caveat, in the Nordic/NOA-08 data, patients treated with TMZ with “truly” unmethylated tumors have the poorest mOS (truly unmethylated: 6.7; gray zone: 7.5; MGMT methylated: 11.8 months). Similarly, within the CE.6 data, patients “truly” unmethylated tumors treated with RT and TMZ had the worst mOS (truly unmethylated: 9.5; gray zone: 11.4; MGMT methylated: 13.3months). The authors conclude from their analysis that patients’ TMZ nonresponders can be identified using their modified cutoff values. However, cautious interpretation is needed since mOS curves for “truly” unmethylated and gray zone patients overlap in the Nordic/NOA-08 trial and similar overlap was seen for gray zone and methylated patients from the CE.6 trial. These differences exemplify the difficulty in setting a cutoff value for any clinical assay and highlight ambiguity in treatment recommendations for gray zone patients.

A major challenge for the implementation of a 3-tiered definition of MGMT status is the variety of molecular assays used to define MGMT promoter methylation. The MGMT CpG island contains 98 CpG sites,⁸ and there is a lack of consensus regarding which CpG sites are most critical and what cutoff level is sufficient to delineate a tumor as MGMT promoter methylated or unmethylated.⁹ The quantitative methylation-specific polymerase chain reaction (qMS-PCR) assay from the present study uses PCR primers that specifically hybridize across 9 CpG sites, with the methylation signal derived only from these sites. With many variations of MGMT qMS-PCR assays using different primers that may not query the same CpGs, the signal returned from each assay is a complex integration of methylation across the sites being queried and is not interchangeable between assays. Pyrosequencing assays are more complex with the return of discrete methylation values for each CpG site queried within the island and an assay-specific algorithm used to define clinical methylation cutoffs. Even more complicated are DNA methylation microarrays, which specifically query one or more CpG sites within the MGMT promoter, but may also use methylation signals outside of the CpG island to “tune” the assay for accurate prediction of MGMT methylation status. In this context, implementation of a 3-tiered assay as proposed by Hegi et al. will require re-calibration and validation for each MGMT assay.

While unlikely to immediately change clinical practice, these results provide some clear calls to action. First, in the context of randomized clinical trials, the use of highly validated, central MGMT testing would enable the use of a 3-tiered MGMT methylation assessment, and this could be critically important for clinical trials enrolling MGMT unmethylated patients. Second, these data provide a clear justification for reporting and collection of the quantitative data used to define MGMT status. Prospective collection of these data across large cohorts of patients will be very useful in validating this important finding from Hegi et al. Moreover, such quantitative reporting could permit more nuanced decision-making compared to traditional reports and allow clinicians to have more informed conversations with their patients with borderline MGMT methylation results.

Conflict of interest statement

None.