Characterizing benefit from temozolomide in MGMT promoter unmethylated and methylated glioblastoma: a systematic review and meta-analysis.

BACKGROUND: The current standard of care for the management of patients with newly diagnosed glioblastoma (GBM) includes maximal safe resection followed by radiotherapy (RT) with concurrent and adjuvant temozolomide (TMZ). While it is well established that TMZ has better efficacy in patients with MGMT promoter methylation, it remains an area of debate whether TMZ should be omitted when treating GBM patients with unmethylated MGMT.
METHODS: We conducted a systematic review and meta-analysis to provide separate estimates of median overall survival (OS) and progression-free survival (PFS) for patients with methylated and unmethylated GBM treated with RT with or without TMZ. We searched multiple databases from inception to January 13, 2020.
RESULTS: The median OS for patients with unmethylated GBM treated with RT/TMZ pooled from 5 phase III studies (N = 655) was 14.11 months (95% confidence interval [CI], 13.18-15.04) with a median PFS of 4.99 months (95% CI, 4.25-5.72). In contrast, the median OS for patients with methylated GBM pooled from 6 studies (N = 753) was 24.59 months (95% CI, 22.19-26.99) with a median PFS pooled from 7 studies (N = 805) of 9.51 months (95% CI, 7.41-11.61). There is a paucity of prospective data pertaining to OS/PFS in unmethylated patients treated with RT only and therefore a direct comparison was not possible.
CONCLUSIONS: This meta-analysis provides estimates of survival for patients with MGMT methylated or unmethylated GBM treated with RT/TMZ. Further research is needed to delineate whether TMZ should be withheld for patients with unmethylated GBM outside of the setting of clinical trials.

This meta-analysis provides pooled survival estimates for GBM based on MGMT status.

Pooled median OS: 14.11 and 24.59 months for unmethylated and methylated GBM, respectively.

Whether TMZ can be omitted in unmethylated GBM cannot be answered definitively.

While it is well established that TMZ has better efficacy in patients with MGMT promoter methylation, it remains an area of debate whether temozolomide should be omitted when treating glioblastoma patients with unmethylated MGMT. Median survival differs vastly between MGMT methylated and unmethylated GBM and estimates from this analysis should be cited when discussing prognosis with patients.

The EORTC-26981-22981 has established the current standard of care for the management of newly diagnosed glioblastoma (GBM) which includes maximal safe resection followed by concomitant radiotherapy (RT) and temozolomide (TMZ) followed by adjuvant TMZ.[1]  O-Methylguanine-DNA methyltransferase (MGMT) gene promoter methylation has been demonstrated to be a prognostic and predictive factor of response to TMZ in patients with GBM.[2] This clinical benefit has been attributed to the fact that MGMT can remove the damaging alkyl groups from the O6 position of guanine and repairs the DNA damage caused by alkylating agents; MGMT promoter methylation therefore results in compromised DNA repair and promotes tumor cell death. However, it remains an area of debate whether TMZ should be used for MGMT unmethylated patients. While some strongly believe that TMZ is ineffective in this subgroup of GBM patients and advocate for omitting TMZ from the treatment of patients with MGMT unmethylated GBM, especially in the elderly population,[3] others have argued that there is insufficient evidence to withhold an approved treatment from those with the poorer prognosis.[4] Several published clinical trials (eg, RTOG 0525 and RTOG 0825) included preplanned analyses and stratifications based on MGMT methylation status and confirmed the prognostic value of MGMT.[5,6] We performed this systematic review and meta-analysis to provide cumulative estimates of survival for unmethylated and methylated GBM patients treated with TMZ.

Methods

This systematic review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.

Eligibility Criteria and Search Strategy

Specific vocabulary supplemented with keywords was used to search for phase III trials that provided survival data separately for methylated and unmethylated GBM patients treated with RT alone or RT plus TMZ (as a control or intervention arm). The search strategy was designed and conducted by an experienced librarian with input from the study’s investigators. The search was limited to the English language only and excluded animal studies. The comprehensive search included several databases from inception to January 13, 2020 (the earliest study was from 1999).

The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations and Daily, Ovid Embase, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. The actual strategy listing all search terms used and how they are combined is available in the Supplementary Appendix.

Only relevant papers that included median survival data along with confidence intervals (CIs) and stratified by MGMT status were included. Studies focusing on the elderly population were analyzed separately. Separate analyses were also done for overall survival (OS) and progression-free survival (PFS).

Study Selection and Data Extraction

Two investigators independently screened the papers for inclusion and were in agreement on more than 98% of the included studies. Any disagreements were resolved through discussion and consensus. Then, for each study, the OS and PFS data pertaining to RT or RT plus TMZ stratified by MGMT status were collected. When not available, these data were solicited from corresponding authors.

Risk of Bias Assessment of Included Studies

The methodological quality of the included studies was evaluated using the revised Cochrane risk of bias tool for randomized trials.[7]

Statistical Analysis

Due to the heterogeneity of study settings and populations, we used the random-effect model to perform a meta-analysis of median survival data and 95% CIs. Heterogeneity among studies was evaluated by the I2 index. Statistical significance was reported using P < .05. Stata 15 software (StataCorp) was used to conduct the analyses.

Results

Study Selection

A total of 385 references were screened, 369 of which were excluded after review of the study details provided in the abstracts. Subsequently, 16 full-text references were assessed for inclusion. Six studies were then excluded for the following reasons: 3 studies lacked survival data stratified by MGMT methylation status and treatment arm,[8-10] 2 studies did not test MGMT methylation status,[11,12] and one study had a mixed control arm of patients who received or did not receive TMZ based on treating center.[13]  Figure 1 shows the PRISMA flow diagram of study inclusion and exclusion for this systematic review and meta-analysis.[14]

Figure 1.

PRISMA flow diagram of study inclusion and exclusion. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses.

Methodologic Quality

The methodological quality of the included studies was evaluated using the revised Cochrane risk of bias tool for randomized trials (ROB 2).[7] The methodological quality assessment is summarized in Supplementary Appendix. The majority of the studies had a moderate risk of bias; most of the studies were not blinded, did not report the number of patients lost to follow-up, or did not adhere to treatment (Figure 2).

Figure 2.

Risk of bias assessment for studies included in the meta-analysis.

Study Characteristics

Table 1 summarizes the characteristics of the 10 trials included in the meta-analysis (including 3 elderly trials).

Table 1.

Characteristics of Phase III Studies Included in This Systematic Review and Meta-Analysis

Study	Reference	Age inclusion (years)	Control arm	Treatment arm
Stupp’05	1	≥18	RT	RT + TMZ
Gilbert’13 (RTOG 0525)	5	18–70	RT + TMZ	RT + dose-dense TMZ
Gilbert’14 (RTOG 0825)	6	≥18	RT + TMZ	RT + TMZ + bevacizumab
Stupp’14	17	≥18	RT + TMZ	RT + TMZ + cilengitide
Westphal’15	15	18–70	RT + TMZ	RT + TMZ + nimotuzumab
Stupp’17 (EF-14)	16	≥18	RT + TMZ	RT + TMZ + TTF
Herrlinger’19 (NOA-09)	18	18–70	RT + TMZ	RT + TMZ + CCNU
Wick’12	19	>65	RT (60 Gy)	Dose-dense TMZ
Malmström’12	20	>60	RT (34–60 Gy)	TMZ
Perry’17	21	≥65	RT (40 Gy)	RT + TMZ

TTF, tumor-treating fields (RT + TMZ refers to concomitant RT/TMZ followed by adjuvant TMZ); CCNU, lomustine.

MGMT Unmethylated GBM

Five published phase III clinical trials were included in the meta-analysis for the unmethylated group. These included 5 phase III trials: Stupp’05,[1] Gilbert’13 (RTOG 0525),[5] Gilbert’14 (RTOG 0825),[6] Westphal’15,[15] and Stupp’17 (EF-14).[16]

OS in the unmethylated group.

—The median OS for patients with unmethylated GBM pooled from phase III studies (N = 655) was 14.11 months (95% CI, 13.18–15.04; I2 = 28.2%; Figure 3).

Figure 3.

Forest plot showing pooled OS for patients with unmethylated GBM.

PFS in the unmethylated group.

—The median PFS for patients with unmethylated GBM pooled from phase III studies (N = 655) was 4.99 months (95% CI, 4.25–5.72; I2 = 64.6%; Figure 4).

Figure 4.

Forest plot showing pooled PFS for patients with unmethylated GBM.

MGMT Methylated GBM

Seven phase III studies were included in the meta-analysis for the methylated group: Stupp’05,[1] Gilbert’13 (RTOG 0525),[5] Gilbert’14 (RTOG 0825),[6] Stupp’14,[17] Westphal’15,[15] Stupp’17 (EF-14),[16] and Herrlinger’19.[18] Stupp’14[17] and Herrlinger’19[18] only included patients with MGMT methylated status and therefore were not included in the previous unmethylated analysis.

OS in the methylated group.

—One study (Westphal’15)[15] was not included in this analysis given the upper limit of the CI was not reached. The median OS for patients with methylated GBM pooled from 6 studies (N = 753) was 24.59 months (95% CI, 22.19–26.99; I2 = 22.3%; Figure 5).

Figure 5.

Forest plot showing pooled OS or patients with methylated GBM.

PFS in the methylated group.

—The median PFS for patients with methylated GBM pooled from 7 studies (N = 805) was 9.51 months (95% CI, 7.41–11.61; I2 = 54.3%; Figure 6).

Figure 6.

Forest plot showing pooled PFS for patients with methylated GBM.

Unmethylated GBM in the Elderly

Three phase III studies were included in the meta-analysis for the unmethylated elderly group: Wick’12 (NOA-08),[19] Malmström’12 (Nordic),[20] and Perry’17.[21] Results of the RT alone arms were combined.

The median OS for elderly patients with unmethylated GBM treated with RT alone pooled from 3 studies (N = 223) was 8.35 months (95% CI, 6.46–10.25; I2 = 78%; Supplementary Figure 1).

Discussion

We conducted a systematic review and meta-analysis to provide separate estimates of median OS and PFS for patients with MGMT methylated and unmethylated GBM treated with RT with or without TMZ. We decided to include only phase III trials to ensure better quality meta-analysis and limit heterogeneity. Multiple agents in neuro-oncology have succeeded in phase II trials but failed in phase III trials due to intrinsic limitations in phase II trial designs.

A meta-analysis by Zhao et al.[22], among many other studies, has confirmed the prognostic value of MGMT in newly diagnosed and recurrent GBM. While it is well established that TMZ has better efficacy in the MGMT methylated group, no study has prospectively compared RT versus RT plus TMZ in the MGMT unmethylated GBM subgroup. There is a paucity of historic data from previous trials examining this question as well. An additional challenge is that MGMT promoter methylation assays have not been standardized. Most trials included in this meta-analysis used methylation-specific PCR (MSP).[23] While MSP is a reliable test, cutoff values for differentiating between MGMT methylated and MGMT unmethylated gliomas have not been well defined.[24] It has been recommended to use a lower safety margin when interpreting the MSP results in clinical trials to avoid withholding TMZ from “low MGMT methylation” patients who may still benefit from this treatment.[25] Similar limitations apply to pyrosequencing: A recent study suggested that 17% achieved the highest precision for correlation with clinical outcomes,[26] whereas 8–10% has been widely used in clinical practice.

A retrospective analysis of EORTC-26981-22981 first reported the lack of statistically significant difference between OS in MGMT unmethylated patients treated with RT only (N = 54; 11.8 months [95% CI, 9.7–14.1]) and with RT plus TMZ (N = 60; 12.7 months [95% CI, 11.6–14.4]).[2] PFS was 5.9 months (5.3–7.7) in the RT group (N = 46) and 10.3 months (6.5–14) in the RT plus TMZ group (N = 46). Another retrospective study of 225 GBM patients reported improved PFS and OS of patients with MGMT methylated GBM treated with RT alone compared to those with MGMT unmethylated tumors (31 vs 15 weeks and 63 vs 51 weeks, respectively), suggesting that MGMT promoter methylation could fundamentally be a prognostic factor for GBM.[27] No CIs were reported and therefore these results could not be combined with the EORTC-26981-22981 retrospective analysis. Finally, a prospective study of 301 patients of the German Glioma Network estimated median OS of 7.14 months for unmethylated patients treated with RT alone.[28] Beyond this, there is a paucity of data pertaining to OS in unmethylated patients treated with RT only in adult patients.

In our meta-analysis, the median OS of adult patients with unmethylated GBM pooled from 5 phase III studies (N = 655) was 14.11 months (95% CI, 12.85–17.97). The median PFS for patients with unmethylated GBM pooled from the same studies was 4.99 months (95% CI, 4.25–5.72). The CIs of our analyses still overlap with the large CIs from the RT only group in the retrospective analysis of EORTC-26981-22981. However, the published literature does not yield more robust survival data for unmethylated patients treated with RT alone. On the other hand, the median OS for patients with methylated GBM pooled from 6 studies (N = 753) was 24.59 months (95% CI, 22.19–26.99). The median PFS for patients with methylated GBM pooled from 5 studies (N = 805) was 9.51 months (95% CI, 7.41–11.61). One limitation of this analysis was that some major phase III trials had to be excluded from this meta-analysis, Weller’17[8] and Chinot’14,[9] because of lack of survival data stratified by MGMT methylation status and treatment arm. Additionally, the majority of the studies had a moderate risk of bias as most of the studies were not blinded and did not report the number of patients lost to follow-up. Moreover, the initial time point of randomization was different among trials as patients were randomized either before or after the concomitant RT/TMZ phase.

The argument against using TMZ for unmethylated patients in the elderly population is even more intense. The lack of the definition of the elderly population further adds to the heterogeneity of the inclusion age of the various studies. The NOA-08 trial[19] compared RT alone versus TMZ alone in elderly patients with GBM. Event-free survival (EFS) was longer in patients with MGMT promoter methylation who received TMZ than in those who underwent RT, whereas the opposite was true for patients with no methylation of the MGMT. On the other hand, Perry et al.[21] compared short-course RT alone to short-course RT plus TMZ, in a randomized phase III study, and reported benefit of adding TMZ even in the unmethylated group, albeit not statistically significant (10 vs 7.9 months; hazard ratio 0.75; P = .08).[24] In our meta-analysis, the pooled median OS for unmethylated elderly GBM patients treated with RT alone was 8.35 months (6.46–10.25), compared to 10 months (8.3–10.7) in the RT + TMZ arm in the Perry trial. In a recent network meta-analysis for elderly patients with GBM, the pooled analysis suggested that the addition of TMZ to RT had the greatest probability of being ranked as the optimal treatment.[29]

In summary, we provide estimates for survival for patients with methylated and unmethylated GBM treated with RT and TMZ. These numbers should be cited when discussing prognosis with patients as the unmethylated and methylated groups vary vastly in terms of median OS (14.11 vs 24.59 months, respectively). There is paucity in historic data to drive any conclusions regarding withholding TMZ in unmethylated GBM, including the elderly with good functional status. The exclusion of TMZ in patients with an unmethylated MGMT gene promoter in favor of an experimental therapy that holds great promise based on phase II trial results and/or mechanism of action appears reasonable at this time. Such exclusion also appears reasonable in elderly patients with an unmethylated MGMT gene promoter with significant comorbidity and perceived higher risk from chemotherapy. However, we have otherwise continued to treat “low-risk”, non-elderly patients with unmethylated MGMT gene promoter GBM with RT and TMZ based on a small improvement in median survival over RT alone noted in our analysis. We anticipate further refinements in our position as the true significance of MGMT gene promoter methylation assays (cutoffs, best assays) as well as more data become available.

Click here for additional data file.

Click here for additional data file.