“Hyper-Hypermethylated Status” of the MGMT Gene Promoter Confers Additional Survival Benets in IDH-Wild-Type Glioblastoma Patients

Methylation status of the O(6)-Methylguanine-DNA methyltransferase (MGMT) promoter plays key role in glioblastoma (GBM) with respect to the patient’s responses to temozolomide chemotherapy and disease prognosis. Although the cut-off value of MGMT methylation ( ≥ 10%) is currently widely used to dichotomize the MGMT status as “methylated” or “unmethylated” in pyrosequencing (PSQ) analysis, it is still unclear whether it reects the actual MGMT methylation status of the patients. Thus, we investigated if there is a so called “hyper-hypermethylation cut-off value” that confers additional survival benets in IDH-wild-type GBM patients. Methods


Introduction
Glioblastoma (GBM) is the most commonly occurring malignant brain tumor in adults that manifests hostile clinical behavior [1]. Despite the aggressive therapy, its prognosis continues to be dismal, with an overall survival (OS) of 15-17 months [2]. Stupp protocol is the standard of care for GBM treatment; it consists of neurosurgical resection followed by concomitant chemoradiation and adjuvant chemotherapy with temozolomide (TMZ) [3,4]. TMZ is a cytotoxic alkylating agent, which causes double-strand DNA breaks in the cells. Chemosensitivity to TMZ varies greatly, owing to the methylation status of O(6)-Methylguanine-DNA methyltransferase (MGMT) promoter [3][4][5][6]. Previous randomized trials have shown that MGMT promoter methylation is signi cantly associated with higher survival rates in TMZ-treated Page 3/12 GBM patients. The DIRECTOR trial indicated TMZ rechallenge as a treatment possibility for recurrent GBM with methylated MGMT promoter [7]. Based on an NOA-09 trial, it was suggested that more intense rst-line treatment therapy with combined lomustine/TMZ can improve prognosis compared to the conventional TMZ therapy [5]. All these clinical trials emphasized the importance of MGMT promoter methylation status for prognosis. Thus, precise measurement of the MGMT promoter methylation status is essential for clinical planning in GBM patients. Pyrosequencing (PSQ) is a widely used technique for determining the MGMT promoter methylation status [7][8][9][10]. PSQ quantitatively assesses this status, thereby rendering advantages over methylation-speci c polymerase chain reaction (MSP) [11][12][13].
Currently, whether the currently widely accepted MGMT methylation cut-off value (≥ 10%) used to dichotomize the MGMT promoter methylation status as "methylated" or "unmethylated" can reliably re ect the actual MGMT methylation status of GBM patients remains unclear [14][15][16][17][18]. Several studies have suggested that there exists a "gray zone" where the methylation level is close to the cut-off value, its actual methylation status remains unclear [9,11,16,[19][20][21]. Thus, the further subclassi cation of the MGMT methylation status may be needed to confer survival bene ts in glioma patients. In addition, although the association of MGMT promoter methylation status with GBM prognosis has been established, only a few studies have focused particularly on isocitrate dehydrogenase (IDH)-wild-type GBM [22].
In this study, we investigated if there is a "hyper-hypermethylated" MGMT status that could further bene t GBM patients in terms of the survival and progression of IDH-wild-type GBM.

Patient selection and clinicopathologic evaluation
We retrieved 317 GBM cases from computerized les of surgical pathology diagnosis. These cases underwent surgically resections from May 2016 to December 2019. Of these cases, 110 consecutive cases of IDH-wild-type GBM that: 1) had undergone gross total tumor resection and 2) completed the Stupp protocol with 3) Karnofsky performance status ≥70 were selected. Postoperative MRI was obtained within 48 h after surgery and gross total resection was de ned when no enhancing lesion was observed. The decision of gross total resection was carried out by two neuro-radiologists. GBM diagnosis was determined according to the 2016 WHO classi cation of CNS tumors. All cases were reviewed and con rmed by two specialized neuro-oncopathologists (authors, MK and SHK). 1p/19q status of the cases was evaluated using uorescence in situ hybridization. IDH mutational status was assessed using IDH1 R132H immunohistochemistry and Sanger sequencing of IDH1 and IDH2 for all cases. Clinicopathologic data including age at diagnosis, sex, Karnofsky performance status, extent of tumor resection and residual tumor volume, types and duration of adjuvant therapy, follow-up time, progression-free survival, and OS were thoroughly investigated from the electronic medical records of the hospital. The study plan was reviewed and accepted by the Institutional Review Board of Severance Hospital (4-2020-0547).
Areas with > 80% tumor cells were chosen for MGMT promoter methylation analysis. Genomic DNA was extracted from freshly frozen tumor tissues and formalin-xed para n-embedded samples "Hyper-hypermethylation" cut-off points in MGMT promoter methylation status We then investigated the additional cut-off points of the MGMT promoter methylation status that can maximize the difference in OS among the methylated groups. The estimated hyper-hypermethylation MGMT cut-off value was ≥40%.

Discussion
In this well-established IDH-wild-type GBM cohort, we investigated the association between the MGMT promoter methylation status and GBM prognosis. Out of all the reports on MGMT promoter methylation status, only a few have strictly focused on patient cohorts limited to IDH-wild-type GBM [15]. Currently, PSQ is the accepted 'gold standard' to evaluate MGMT promoter methylation status [12,23,24]. In addition, the cutoff value of ≥10% is widely used to determine MGMT promoter status as "methylated" and "unmethylated" in PSQ [12,19,21]. Although previous studies have shown that MGMT promoter methylation confers survival bene t in glioma patients [25][26][27][28], there were some issues that needed to be clari ed. Studies have reported the presence of a "gray zone" in the methylation status and questioned whether the current dichotomizing method could re ect the actual methylation status of GBM patients [14,15,24]. In addition, some studies have suggested the possibility of a "hypermethylated" MGMT methylation status, which may provide additional prognostic bene ts to GBM patients [15,26,29].
In this study, we demonstrated that MGMT "hyper-hypermethylation" can confer additional survival bene ts to GBM patients showing MGMT methylation. Because individual CpG sites show methylation heterogeneity and may have different impacts on the prognostic value [24], low-methylated GBMs could be partly methylated tumors that cannot fall into methylated or unmethylated category. The extent of MGMT promoter methylation or protein expression level may have affected the distinct biologic behavior between hyper-hypermethylated group and unmethylated group [26]. In the meanwhile, the hyperhypermethylation status did not demonstrate signi cant differences in PFS compared with the lowmethylated group in the present study. Taken together, we suggest that even low levels of MGMT methylation can bene t the patient in terms of disease progression; a hyper-hypermethylated status is more likely to affect the additional survival bene ts of the patient; however, studies using larger cohorts will be needed to con rm the results of this study.
Although the present study has major ndings, there are some associated limitations as well. Clinicopathologic characteristics [30-32], such as age, mutational status of TERT promoter or TP53, and gliomatosis, may have affected the results of this study. In addition, this study strictly focused on IDHwild-type GBM. Further studies thus will be needed to con rm that the hypermethylation status will also have prognostic value in other types of gliomas.
In conclusion, we suggest that further dividing the MGMT methylation status according to the hypehyperrmethylated status can have prognostic implications for IDH-wild-type GBM patients. Subsequent studies in larger patient cohorts should be used to validate the ndings of our study. Tables Table 1. Clinicopathologic characteristics of the patient cohort