The Impact of MGMT Promoter Methylation and Temozolomide Treatment in Serbian Patients with Primary Glioblastoma

Abstract

Background and objective: Despite recent advances in treatment, glioblastoma (GBM) remains the most lethal and aggressive brain tumor. A continuous search for a reliable molecular marker establishes the methylation status of the O6-methylguanine-DNA methyltransferase (MGMT) gene promoter as a key prognostic factor in primary glioblastoma. The aim of our study was to screen Serbian patients with primary glioblastoma for an MGMT promoter hypermethylation and to evaluate its associations with overall survival (OS) and sensitivity to temozolomide (TMZ) treatment. Materials and methods: A cohort of 30 Serbian primary glioblastoma patients treated with radiation therapy and chemotherapy were analyzed for MGMT promoter methylation and correlated with clinical data. Results: MGMT methylation status was determined in 25 out of 30 primary glioblastomas by methylation-specific PCR (MSP). MGMT promoter hypermethylation was detected in 12 out of 25 patients (48%). The level of MGMT promoter methylation did not correlate with patients’ gender (p = 0.409), age (p = 0.536), and OS (p = 0.394). Treatment with TMZ significantly prolonged the median survival of a patient (from 5 to 15 months; p < 0.001). Conclusions: Due to a small cohort of primary GBM patients, our study is not sufficient for definitive conclusions regarding the prognostic value of MGMT methylation for the Serbian population. Our preliminary data suggest a lack of association between MGMT promoter methylation and overall survival and a significant correlation of TMZ treatment with overall survival. Further population-based studies are needed to assess the prognostic value of the MGMT promoter methylation status for patients with primary glioblastoma.

1. Introduction

Glioblastoma (GBM)—World Health Organization (WHO) grade IV diffuse glioma—represents the highly invasive and infiltrative type of primary brain tumor associated with poor prognosis and a 5.6% five-year survival rate [1,2,3]. GBM is the most common type of malignant central nervous system tumor in adults (47.7%–49%) that accounts for the majority of gliomas (56.6%) according to a recent Central Brain Tumor Registry of the United States (CBTRUS) Statistical Report and EUROCARE-5 study [4]. Comprehensive genomic characterization studies revealed an underlying complex network of different molecular aberrations which provoke GBM development through changes in major signaling pathways [5,6]. These studies also contributed toward defining the methylation status of the O⁶-methylguanine-DNA methyltransferase (MGMT) gene promoter as one of the most relevant prognostic markers in GBM patients [7,8,9,10,11].

The MGMT gene encodes a DNA-repair protein that removes cytotoxic alkyl adducts from O⁶-guanine [12]. This protein inhibits the effect of cancer treatment with alkylating agents such as nitrosoureas, tetrazines, and procarbazine that induce apoptosis in cancer cells [12,13,14]. The alkylating agent Temozolomide (TMZ) was approved in 2005 by the US Food and Drug Administration (FDA) for use in the treatment of GBM [15,16]. TMZ is an imidazotetrazine derivative of decarbazin that induces cell cycle arrest at G2/M. In Serbia, a GBM treatment protocol that includes TMZ as adjuvant therapy was introduced in 2011 [17,18]. Although it was demonstrated that TMZ improves the overall survival (OS) and progression-free survival (PFS) of GBM patients, at least 50% of them do not benefit from TMZ due to treatment resistance caused by over-expression of MGMT in GBM cells [19,20]. To date, the bulk of evidence suggests that epigenetic silencing of the MGMT gene through hypermethylation of the cytidine phosphate guanosinedinucleotides (CpG) in the promoter region is associated with greater response to the TMZ treatment of GBM patients [15,21,22,23,24].

A methylation-specific polymerase chain reaction (MSP) is one of the most commonly used methods for assessing the MGMT methylation status in either snap-frozen GBM tissue samples or formalin-fixed, paraffin-embedded (FFPE) tissue [25,26,27,28]. This method is based on sodium bisulfite treatment of isolated DNA samples which results in the conversion of unmethylated cytosines into uracil, leaving methylated cytosines unchanged. Bisulfite conversion of template DNA is followed by PCR reactions using two primer sets for both an unmethylated and methylated MGMT promoter variant, which allow for the evaluation of the methylation status at six to nine CpG sites [28,29]. The difference in amplicon lengths after conducting PCR reactions with primer sets for each variant of MGMT promoter provides easy-to-interpret results that can be visualized by agarose gel electrophoresis. Since MSP was established, this method has evolved as the “gold standard” that enables a cost-efficient non-quantitative method of MGMT methylation analysis suitable for routine clinical diagnostics with low sample numbers [27].

The main goal of our study was to determine MGMT promoter methylation and its relevance for the prediction and prognosis of clinical outcomes of the Serbian population with glioblastoma. The study was designed to investigate the effect of novel therapeutic treatment (TMZ) on overall survival. Also, the potential use of MSP as a semi-quantitative method for assessing MGMT methylation status in snap-frozen GBM samples was investigated.

2. Materials and Methods

2.1. Patients and Tumor Specimens

GBM patients operated on the Neurosurgery Clinic (The Clinical Centre of Niš, Serbia) between 2013 and 2017 were included in this study. All patients underwent total resection of the tumor and had a Karnofsky score ≥80%. Tumor specimens were snap frozen and stored at −80 °C. All samples were confirmed with glioblastoma WHO grade IV by an expert neuropathologist (N.V. and M.K.). The study protocol and informed consent form were approved by the Ethics Committee of the Faculty of Medicine, Niš, Serbia (01-2113-10). Written informed consent was obtained from all study participants. All patients received combined radiotherapy and chemotherapy. Patients were irradiated with 3D conformal radiotherapy at a dosage of 60 Gy in 30 fractions (2 Gy per day, 5 days a week) (radiotherapy (RT)). Patients were classified into three groups based on chemotherapy administered in 6 cycles:

• Group 1 (n = 10 patients): temozolomide (TMZ)—the first cycle at a dose of 150 mg/m² for 5 days; the next 5 cycles at a dose of 200 mg/m². Cycles were repeated every 3 weeks.
• Group 2 (n = 10 patients): procarbazine, lomustine (1-[2-chloroethyl]-3-cyclohexyl-1-chloroethylnitrosourea (CCNU)) and vincristine (PCV regimen): CCNU 110 mg/m² p.o. day 1; procarbazine 60 mg/m² per os (p.o.) days 8–21; vincristine 1.4 mg/m² (maximum 2 mg), i.e., days 8 and 21. Cycles were repeated every 6–8 weeks.
• Group 3 (n = 10 patients): carmustine (BCNU) 200 mg/m², i.e., day 1. Cycles were repeated every 8 weeks.

Although 30 patients were enrolled in this study, DNA was successfully obtained for only 25 samples, see

Table 1. List of patients and treatments involved in DNA analysis.

Type of Therapy	Patient Mark
RT + TMZ (Group 1)	1, 2, 5, 8, 9, 15, 16, 17, 19 and 25
RT + PCV (Group 2)	4, 7, 10, 12, 18, 22, 23 and 24
RT + BCNU (Group 3)	3, 6, 11, 13, 14, 20 and 21

RT: radiotherapy; TMZ: temozolomide; PCV: procarbazine, lomustine, and vincristine; BCNU: carmustine.

. For 5 patients, we did not have sufficient tissue specimen for DNA analysis (2 from Group 2 and 3 from Group 3 of treatment). During DNA isolation and PCR analysis, we conducted blind-experiments without knowledge of patients’ diagnosis and treatment (tumor specimens were coded).

2.2. DNA Isolation and Bisulfite Conversion

Genomic DNA was extracted from frozen tumor tissues by QIAamp^® DNA Mini Kit (Qiagen, Hilden, Germany) [27]. Quantity and quality of isolated DNA was determined by a BioSpec–nano UV–Vis Spectrophotometer (Shimadzu, Kyoto, Japan). A total of 2 µg of genomic DNA was modified by sodium bisulfite using EpiTect^® Bisulfite Kit (Qiagen, Hilden, Germany).

2.3. Methylation-Specific Polymerase Chain Reaction (MSP)

The MSP was conducted in a total volume of 20 µL containing 1 × PCR buffer with 1.5 mM MgCl₂ (Qiagen, Hilden, Germany), 10 pM of appropriate forward and reverse primer, 0.2 μM dNTP mix, 1U HotStar Taq polymerase (Qiagen, Hilden, Germany), and 100 ng of bisulfite-converted template DNA. Primers used for amplification of MGMT promoter and control ALU–C4 sequences are shown in

Table 2. Primer sequences and amplification.

Gene	Primer Sequence (5’–3’)	Amplicon Size (bp)	References
MGMT unmethylated (U)	F: TTTGTGTTTTGATGTTTGTAGGTTTTTGT	93	[26]
	R: AACTCCACACTCTTCCAAAAACAAAACA
MGMT methylated (M)	F: TTTCGACGTTCGTAGGTTTTCGC	81	[26]
	R: GCACTCTTCCGAAAACGAAACG
ALU–C4	F: GGTTAGGTATAGTGGTTTATATTTGTAATTTTAGTA	98	[31]
	R: ATTAACTAAACTAATCTTAAACTCCTAACCTCA

MGMT: O⁶-methylguanine-DNA methyltransferase; F: forward primer; R: reverse primer.

. The amplification reaction was carried out in a Mastercycler Gradient (Eppendorf) using the following program: 95 °C for 15 min, then 35 cycles of 95 °C for 50 s, 59 °C for 50 s and 72 °C for 50 s, and final extension at 72 °C for 10 min. Control PCR reactions were performed using EpiTect PCR Control DNA set (Qiagen, Hilden, Germany) consisting of:

-

unmethylated and unconverted human DNA (genomic DNA purified from a human colorectal cancer cell line HCT116 DKO with double knockouts of both DNA methyltransferases (DNMT1 (-/-) and DNMT3b (-/-)) (K1 in Figure 1);

-

unmethylated and bisulfite-converted human DNA (genomic DNA originated from the same HCT116 DKO cell line as K1 DNA, but modified by sodium bisulfite upon isolation; as a result of bisulfite conversion non-methylated cytosines were turned to uracils) (K2 in Figure 1);

-

methylated and bisulfite-converted human DNA (genomic DNA derived from HCT116 DKO cell line which was in vitro methylated at all cytosine positions comprising CpG dinucleotides by M.SssI methyltransferase and then treated with sodium bisulfite; the final outcome of the bisulfite treatment was that 5-methylcytosines were left unaffected) (K3 in Figure 1).

The function of these control DNAs in MSP were as follows; while K1 served as negative control in MSP with M or U primers (independently) and for assessment of the efficiency of bisulfite-mediated conversion of DNA, K2 was used as a positive control in MSP with U primers specified for non-methylated cytosines, and K3 was used as a positive control in MSP with M primers specified for 5-methylated cytosines in CpG dinucleotides of MGMT promoter.

Also, a non-template PCR reaction was included as a negative (water) control of PCR (K− in Figure 1 and Figure 2).

ALU–based control reaction was used as a control reaction to measure input DNA levels and normalized the signal for each methylation reaction (ALU–C4 in Table 2 and Figure 2).

All PCR reactions were performed in duplicate.

Amplified PCR products were detected by ultraviolet (UV) light on a 2% agarose gel stained with ethidium bromide. A visible M primer band of MGMT indicated a positive MGMT methylation status, while the absence of an M primer PCR product was considered as a negative methylation status of MGMT. A visible U primer band of MGMT indicated the presence of unmethylated MGMT promoter [26]. Primer dimerization was noticed in PCR reactions with U primer (PD in Figure 1).

Gel images were subject to ImageJ software analysis (National Institute of Health, Bethesda, MD, USA) [30].

2.4. Quantification of Methylation Data

The level of methylated DNA (percentage of methylated reference (PMR)) was calculated by three different approaches. The first approach compared the intensity of methylated (M) and unmethylated (U) MSP bands on agarose gel using the following formula [27]:

$$PMR= M/U$$

(1)

The other two approaches for MSP quantification included two control PCR products: ALU–C4 (ALU) as a DNA input normalizer and commercial methylated bisulfite-converted human DNA (Qiagen) as a fully methylated control [32,33]. Equations used for these two approaches were:

$$PMR=\frac{ \mathrm{M}/\mathrm{U} / ALU \mathrm{for} \mathrm{sample}}{ \mathrm{M}/\mathrm{U} / ALU \mathrm{for} \mathrm{methylated} \mathrm{control}}$$

(2)

and

$$PMR=\frac{ \mathrm{M} / ALU \mathrm{for} \mathrm{sample} }{ \mathrm{M}/ ALU \mathrm{for} \mathrm{methylated} \mathrm{control}}$$

(3)

where in all three approaches for the quantification of MSP: PMR > 1 indicates a strong MGMT promoter methylation (hypermethylated), PMR = 0 (no M primer MSP product detectable) indicates an unmethylated MGMT promoter and PMR < 1 indicates weak MGMT promoter methylation.

2.5. Statistical Analysis

Statistical analyses were performed using the SPSS 16.0 software package (IBM Corp., Armonk, NY, USA) with p < 0.05 considered significant. Continuous data were presented as mean ± standard variation while categorical data were shown as frequencies (%). Fisher’s exact test was used to test the association between categorical variables and a Student’s t-test was used to compare continuous variables.

The patient analysis included gender, age, Karnofsky performance status, methylation status, treatment with TMZ, and survival. Overall survival (OS) was measured from the date of surgery to the date of death or last follow-up. OS curves were estimated by the Kaplan–Meier method and their comparison was performed with the use of a univariate log-rank test. In order to compare the three variants of PMR for assessment of the MGMT methylation status, the interclass correlation coefficient (ICC) was determined.

3. Results

3.1. Methylation Status of the MGMT Promoter and Clinical Parameters

DNA obtained from 25 patients with primary glioblastoma was subjected to MSP with specific primers for methylated (M) and unmethylated (U) template detection. Methylation data were successfully determined for all GBM samples, see Figure 1. Control PCR reactions with ALU primers for every specimen were done simultaneously, see Figure 2.

Characteristics of patients within the study group (6 females, 19 males; age 59.6 ± 13.07; 29 to 80 years old) and their methylation status are shown in

Table 3. Clinical characteristics and methylation status of primary glioblastoma (GBM) patients.

		Unmethylated (n = 13)	Methylated (n = 12)
Sex, n (%)	male	9 (69.2)	10 (83.3)
	female	4 (30.8)	2 (16.7)
Age, mean ± SD, years		58.00 ± 12.85	61.33 ± 13.65
Age, n (%)	<50 yr	4 (30.8)	2 (16.7)
	>50 yr	9 (69.2)	10 (83.3)
Preoperative KPS, (%)		81.64 ± 12.01	81.64 ± 12.01
Postoperative KPS, (%)		80.00 ± 12.06	80.00 ± 12.06

KPS: Karnofsky performance status.

. A positive methylation status was detected in 12 patients (48%). Statistical analysis did not find a significant correlation between MGMT promoter methylation and gender (χ² = 0.680; p = 0.409) or the age of patients with primary GBM (t = 0.629; p = 0.536).

3.2. Different Approaches in MSP Data Quantification

Methylation levels were estimated by three different approaches. The first assessed MGMT promoter methylation by a simple M/U ratio for each tumor specimen (PMR (I)). The other two approaches allowed better discrimination between MGMT methylation levels in different samples by the inclusion of the PCR signal of a commercial fully methylated control and ALU DNA input control (PMR (II) and (III), respectively). Results are shown in

Table 4. Semi-quantitative evaluation of the level of methylated MGMT promoter using different approaches ((I) to (III)); see Material and Methods).

PMR	(I)	(II)	(III)
<1, n (%)	5 (20.0)	6 (24.0)	5 (20.0)
>1, n (%)	7 (28.0)	6 (24.0)	7 (28.0)
0, n (%)	13 (52.0)	13 (52.0)	13 (52.0)

.

MGMT promoter methylation status evaluated as PMR (I) and (III) showed identical distribution among patients (five patients with M/U ratio <1 and seven patients with M/U ratio >1), while PMR (II) had different pattern (six patients with M/U ratio <1 and six patients with M/U ratio >1).

Levels of coincidence between various PMR approaches are shown in

Table 5. Correlation of MSP data obtained by various percentage of methylated reference (PMR) approaches.

	ICC	95% CI	p
(I) vs. (II)	0.763	0.532–0.888	<0.001
(I) vs. (III)	0.493	0.139–0.739	0.005
(II) vs. (III)	0.844	0.678–0.928	<0.001

ICC: interclass correlation coefficient; CI: Confidence interval.

. PMR (II) and PMR (III) variants of MSP data demonstrated the highest level of coincidence (ICC = 0.844), while the lowest level of coincidence was between PMR (I) and PMR (III).

3.3. MGMT Status, TMZ Therapy, and Survival

Univariate analyses showed that TMZ-treated patients had a statistically significant improvement in overall survival (median survival 15 months) in comparison with patients without TMZ treatment (median survival five months) (p < 0.001), see

Table 6. Associations between overall survival, TMZ treatment, gender and MGMT methylation assessed by univariate analyses (log-rank test).

	$\mathbf{Overall} \mathbf{Survival} \left(\mathbf{Months}\right) \overline{\mathit{x}}$	SE	95% CI		Log-Rank	p
			Lower Limit	Upper Limit
No TMZ	5.000	0.781	3.469	6.531	15.721	<0.001
TMZ	15.000	1.799	11.473	18.527
Male	8.167	1.462	5.300	11.033	0.221	0.638
Female	7.167	2.151	2.950	11.383
Unmethylated MGMT	7.083	1.574	3.999	10.168	0.727	0.394
Methylated MGMT	8.750	1.855	5.114	12.386

$\overline{x}$: mean value; SE: standard error; CI: confidence interval; p: probability value calculated using log-rank test; p values < 0.05 were considered statistically significant.

. It was found that this improvement was not associated with the methylation status of the MGMT promoter or gender. Kaplan–Meier OS curves are shown in Figure 3.

4. Discussion

There is ongoing debate concerning the most suitable technique for the determination of the MGMT promoter methylation and the prognostic importance of the obtained methylation status for patients with GBM [28,34]. MGMT testing in our study is performed by MSP as one of the oldest and the most widely used techniques [25,26,27,28]. Notably, MSP is cost-effective, gel-based, and the most appropriate method for resource-limited settings and routine diagnostics with low sample numbers. However, this technique is especially prone to producing false-positive results when performed on low quality/quantity DNA, partially bisulfite-converted DNA, or tumor specimens with irregular mosaic methylation patterns [28]. Generally, only vital (non-necrotic) tumor specimens should be used for MSP analysis to avoid false-negative results [28].

In order to improve MSP semi-quantitative potentials, we performed additional normalization of the methylation signal by ALU control and universal positive methylation control [32,33]. Therefore, we compensate PMR for variations in copy number due to differences in sample handling, DNA isolation and tumor heterogeneity. Optimally standardized and easy-to-interpret MSP data were used in our study for evaluation of the clinical importance of the methylation status of the MGMT promoter.

Further, numerous GBM clinical trials with TMZ have established a positive methylation status of the MGMT promoter as the strongest predictor for OS and progression-free survival (PFS) benefit [13,19,23,27,35]. However, our study showed no significant impact of the MGMT promoter methylation on the survival outcome and TMZ treatment benefit. Although, we should emphasize that these are only preliminary data based on low sample quantity. Nevertheless, the same observation was made in the above-mentioned study of 110 GBM patients from Serbia; although, the methylation status was assessed in only 62 patients (56.4%) of the cohort [17].

Controversial observations about the predictive and prognostic value of MGMT promoter methylation were noted in several studies [14,36] and in meta-analysis [13]. Jesien-Lewandowicz et al. (2009) detected a positive methylation status in 23 out of 32 (72%) primary GBM patients from Poland treated with surgery and radiotherapy [14]. In univariate analysis, the presence of MGMT promoter methylation was not associated with the patient’s gender and longer survival. Kalkan and colleagues (2015) assessed MGMT promoter methylation status on 40 primary glioblastoma from Turkish patients [36]. They found positive methylation in 13 samples (32.5%) and no statistical significance between MGMT methylation and gender and overall survival.

Intratumoral and temporal heterogeneity may underlie the described discrepancies in our and other studies with negative prognostic values of the MGMT status [37]. Alternatively, negative conclusions in MGMT studies with Polish, Turkish, and Serbian GBM patients may reflect population molecular differences in gliomagenesis. Although, we should mention that these are small size studies which require confirmation in larger-scale, prospective controlled trials. Previously, Wiencke et al. (2005) showed a substantial ethnic specificity of molecular features (MGMT, TP53 and EGFR) in 556 glioma samples in the San Francisco Bay Area [38].

Our study has several limitations. First, it was conducted on small cohorts of patients from a single Clinical Centre in Serbia and the obtained results should be interpreted with care. Therefore, we could not definitively rule out the prognostic value of the MGMT promoter methylation status in the Serbian GBM population. Second, only the independent prognostic value of MGMT methylation was considered. Although the study was carefully performed, the complexity of gliomagenesis and the latest WHO classification of glioma 2016, suggested that the combination of MGMT, IDH1, and/or TP53 analysis is more relevant for the prediction of survival of patients with GBM [2].

The significance of the combination of predictive biomarkers rather than their individual status for survival prediction in patients with GBM was demonstrated widely [39,40,41,42]. Meta-analysis of Zou and colleagues suggested that IDH mutations were tightly associated with MGMT promoter hypermethylation (p < 0.001) and TP53 gene mutation (p < 0.001) [39]. They indicated that the IDH mutation rate was linked to the glioma’s genomic profile. Higher rates of G to A transitions in IDH1 codon 132 and TP53 codons 248 and 273 were explained by higher levels of methylation of the MGMT promoter CpG islands [39,40]. These mutational events were considered as early events in gliomagenesis which might affect a common stem glial precursor cell population. They were linked with a low proliferation tumor phenotype and a favorable prognosis in glioma patients. Similarly, Shamsara et al. (2009) detected hypermethylation of the MGMT promoter in 24 out of 50 patients (48%) and mutation of TP53 gene in 26 out of 50 patients (52%) with primary glioblastoma in Iran [41]. A significant association between MGMT methylation status and TP53 mutation status was found (p < 0.05). TP53 mutations were observed in 17 out of 26 patients (65.4%) with MGMT-hypermethylated glioblastoma. Likewise, in the previously mentioned study of Jesien-Lewandowicz and associates, the frequency of TP53 G:C to A:T mutations were higher in patients with MGMT promoter methylation (6 out of 23 patients (26%), p = 0.376) [14]. Further, Wang et al. (2014) investigated the predictive value of the combination of MGMT methylation status and TP53 and IDH1 mutation status in 78 patients with GBM from China [42]. For patients with IDH1 mutation, MGMT hypermethylation was correlated with better overall survival (p = 0.013), while for the patients without IDH1 mutation, the presence of TP53 mutation was associated with improved survival (p = 0.029).

A remarkable improvement in the overall survival of GBM patients is recorded from 2005 since the approval of TMZ for concomitant treatment with radiotherapy (RT) and adjuvant treatment for newly diagnosed GBM [15,22,24]. Meta-analysis of survival outcomes of newly diagnosed GBM patients revealed that the RT + TMZ-treated group of patients had a significantly higher median survival (13.41–19 months) in comparison with RT-alone group (7.7–17.1 months) [22].

In Serbia, TMZ was introduced in 2011. Recent studies suggested that TMZ treatment had a favorable impact on the overall survival of GBM patients in Serbia [17,18]. In comparison with RT + BCNU/CCNU treatment, the overall survival of TMZ treated patients was significantly higher (the first study 19 months vs. 13 months; the second study 14.79 months vs. 9. 91 months) [17,18]. Our study confirmed previous findings regarding the favorable impact of TMZ treatment on OS of GBM patients in Serbia (15 months vs. 5 months).

5. Conclusions

In contrast to the generally accepted attitude of the prognostic significance of MGMT promoter methylation in GBM patients, our study failed to show its prognostic value. Our preliminary data suggest the absence of a prognostic implication of MGMT promoter methylation and confirm TMZ treatment benefit on the survival outcome of patients with primary GBM in Serbia. The present small cohort study cannot be used for definitive conclusions and demands independent confirmation in larger population-based studies. Furthermore, elucidation of the true importance of MGMT methylation status in primary GBM requires its association with other markers (IDH1, TP53, etc.)