|
||||||||||||||||||||
|
|
MGMT gene promoter methylation and its correlation with clinicopathological parameters in glioblastomas
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.236974
Keywords: ATRX protein, EGFR gene amplification, glioblastoma, IDH1/2 mutation, MGMT promoter methylation, necrosis
Glioblastoma (GBM) is the most common, and unfortunately, the most aggressive primary malignant brain tumor. The median survival of GBM patients with treatment is 12–18 months.[1],[2] Traditionally, GBS are classified as primary (de novo), or secondary, arising from a low-grade glioma. Currently, the common molecular markers for adult diffuse gliomas in clinical practice are IDH1/2 mutations, 1p19q co-deletion, telomerase reverse transcriptase (TERT promoter mutations), MGMT methylation, ATRX expression, TP53 mutations, epidermal growth factor receptor (EGFR) amplification (including EGFRvIII mutant), and phosphatase and tensin homolog (PTEN) alterations.[3],[4],[5] Evaluation for IDH mutations [initially by immunohistochemistry (IHC) evaluation for IDH1R132H, followed by sequencing for IDH1/2 mutations only in IHC negative cases], 1p19q co-deletion by fluorescent in-situ hybridization (FISH), TERT promoter mutations by sequencing, MGMT methylation by methylation-specific polymerase chain reaction [PCR] (MSP), quantitative MSP (qMSP) and pyrosequencing, ATRX expression and p53 over-expression by IHC and EGFR amplification by FISH, are the current preferred standard methods of detection in routine clinical practice. However, two large trials from European Organization for Research and Treatment of Cancer (EORTC) and National Cancer Institute of Canada (NCIC) groups demonstrated the prognostic and predictive role of MGMT promoter methylation by MSP methodology in high-grade glioma (HGG) patients for alkylating agent-based chemotherapy.[1] In this study, we evaluated adult GBMs for MGMT methylation by MSP and correlated its presence with other molecular (IDH1R132H mutations, ATRX expression, p53 expression, EGFR amplification), histological, and clinical parameters.
Sample One hundred and thirty-four cases of non-brainstem GBMs retrieved from adult (>18 years) patients and diagnosed at our centre from June 2014 to June 2015, which were evaluated for MGMT methylation as a part of the routine clinical management, formed the study sample. The relevant clinical data (age, sex, location of tumor) were retrieved from the hospital electronic medical records. Histology All cases were reviewed and re-confirmed based upon the 2007 4th World Health Organization (WHO) classification of tumors of the nervous system. These were subtyped into different established histomorphological patterns as follows:
Immunohistochemistry (IHC) IHC for p53 protein (Dako, D07, 1:50 dilution), ATRX (Sigma, polyclonal, 1:750 dilution), and IDH1R132H (Dianova H06; 1:100 dilution) were done for all cases on 4-μm formalin-fixed paraffin-embedded tissue (FFPE) sections using the Ventana automated stainer by the polymer detection kit.
DNA extraction Genomic DNA was isolated from five 10-μM paraffin sections using a genomic DNA extraction kit (QIAamp DNA Mini Kits, Qiagen.), quantified on Nanodrop Lite Spectrophotometer, and used for both Sanger sequencing and MS-PCR. MSP for MGMT gene promoter methylation Genomic DNA (500 ng) from each sample was modified by sodium bisulfite treatment (MethylEasy Xceed Rapid DNA modification kit, Human Genetics signature, Australia). The modified DNA was subjected to MSP using a two-step approach with HOTSTAR TAQ Mastermix (QIAGEN India Pvt. Ltd) and nested primers for methylated MGMT (Forward 5'-TTTGTGTTTTGATGTTTGTAGGTTTTTGT; reverse AACTCCACACTCTTCCAAAAACAAAACA) and unmethylated MGMT (Forward 5'-TTTC GACGTTCGTAGGTTTTCGC; Reverse 5'-GCACTCT TCCGAAAACGAAACG). MGMT methylation pattern was graded as:
Fluorescence in-situ hybridization EGFR gene amplification was done using the dual color probe comprising locus-specific identifier probes for EGFR Spectrum Orange/CEP7 Spectrum Green Probe from Vysis (Abbott Laboratories, Downers Grove, IL, USA) on 3μ representative tumor fomalin fixed paraffin embedded (FFPE) section mounted on double poly-L lysine coated slides. Slides were evaluated using an Olympus BX53F upright fluorescence microscope equipped with appropriate excitation and emission filters for visualization of the orange and green signals, QIcam (Q34130) Olympus camera, and Qcapture pro 7.0 image analyser software.[6] For all cases, at least 100 non-overlapping intact tumor cell nuclei were evaluated, and the numbers of EGFR (red) and CEP7 (green) signals/nuclei were recorded. The mean signal number for the EGFR gene (red) and CEP7 (green) was calculated for each case followed by the calculation of EGFR gene/CEP7 ratio. The EGFR gene was quantified as:
Statistical analysis Comparison of MGMT promoter methylation status with age, sex, location, p53 expression, ATRX expression, IDH1R132H mutation EGFR amplification, confluent necrosis, pseudo-palisading necrosis, and calcification was done by applying chi-square and Fisher's exact tests using IBM Statistical Package for the Social Sciences (SPSS) Software 22 version. Non-contributory/uninterpretable results were excluded from the analysis. A P - value of <0.05 was considered statistically significant. The association of MGMT methylation status as a dependent variable with other clinicopathological parameters was evaluated by logistic regression analysis. A univariable logistic regression analysis was performed followed by a multivariable regression analysis. Variables with a P value <0.20 on univariable model were selected for multivariable regression analysis. Odds ratio (OR) and 95% confidence intervals (CI) were also calculated for each parameter in the study.
One hundred and thirty-four cases of histologically diagnosed cases of GBM (molecularly unaided) with interpretable MGMT formed the study population. Clinical The predominant location was the cerebral hemisphere (n: 130; frontal: 50; occipital: 6; parietal: 22; temporal: 52); others included thalamus (n: 2), cerebellum (n: 1), and corpus callosum (n: 1). The male (M) to female (F) ratio was 3.3:1 (M: 103; F: 31). The median age was 51 years, ranging from 20–75 years (≤30 years: 15; 31–40 years: 24; 41–50 years: 22; 51–60 years: 41; >60 years: 32). The follow-up data was available in 93 patients with a median follow-up of 8 months (range, 0.10–148 months). Sixteen patients died of the disease, with a median overall survival of 8.6 months, ranging from 0.10 to 148 months. Histology The predominant histological type was GBM-NOS [Figure 1]a (n = 105; ≤30 years: 11; 31–40 years: 19; 41–50 years: 19; 51–60 years: 31; >60 years: 25). Others include GBM-O (n = 20; ≤30 years: 1; 31–40 years: 3; 41–50 years: 3; 51–60 years: 8; >60 years: 5), GBM-S (n = 1; >60 years), GBM-G (n = 3; ≤30 years: 2; 51–60 years: 1), GBM-E (n = 2; ≤30 years: 1; >60 years: 1), GS (n = 3; 31–40 years: 2; 51–60 years: 1) [Figure 1]b.
Calcification was seen in 15 cases (GBM-NOS: 12; GBM-O: 3). Confluent necrosis was seen in 92 (GBM-NOS: 69; GBM-G: 1; GBM-E: 1; GS: 3, GBM-O: 18), and pseudo-palisading necrosis in 22 cases (GBM-NOS: 13; GBM-G: 1; GBM-O: 7; GS: 1). MGMT promoter methylation [Figure 2]
MGMT promoter was methylated in 66 cases [M: 52; LM: 14] (GBM-NOS: 48; GBM-S: 1; GBM-G: 2; GBM-E: 1; GS: 2, GBM-O: 12). The median ages for the MGMT methylated and the unmethylated cases were 51 years (range, 21–75 years; <=30 years: 8; 31–40 years: 11; 41–50 years: 11; 51–60 years: 19; >60 years: 17) and 51.5 years (range, 20–74; ≤30 years: 7; 31–40 years: 13; 41–50 years: 11; 51–60 years: 22; >60 years: 15), respectively. The male: female ratio of the MGMT methylated cases was 3.4:1 (51/15). Of the 16 patients who died of the disease, 10 had a methylated and 6 had an unmethylated GBM. EGFR amplification EGFR was interpretable in 119 cases; 27 were amplified [Figure 1]g (GBM-NOS: 19; GBM-O: 6, GBM-S: 1; GBM-E: 1), and 92 cases were non-amplified [Figure 1]h. The median age for EGFR amplified and non-amplified cases was 55 years (≤30 years: 1; 31–40 years: 4; 41–50 years: 3; 51–60 years: 10; >60 years: 9) and 51 years (≤30 years: 12; 31–40 years: 17; 41–50 years: 15; 51–60 years: 27; >60 years: 21), respectively. The male: female ratio of EGFR amplified cases was 1.8:1 (M: 18; F: 10). IDH1R132H mutation and ATRX expression All cases were evaluated for IDH1R132H mutation by IHC, of which 18 were positive (GBM-NOS: 17; GBM-O: 1) [Figure 1]c, and 116 (GBM-NOS: 88; GBM-O: 19; GBM-S: 1; GBM-E: 2; GBM-G: 3; GS: 3) were negative. ATRX expression was retained in 98/134 cases [Figure 1]e (GBM-NOS: 75; GBM-O: 16; GBM-S: 1; GBM-E: 2; GBM-G: 1; GS: 3), lost in 20 [Figure 1]f (GBM-NOS: 18; GBM-G: 2), equivocal in 5, and noncontributory in 11 cases. Three (of 98) ATRX retained cases, 13 (of 20) cases with ATRX loss, and one case each of ATRX equivocal (of 5) and non-contributory (of 11), respectively, were positive for IDH1R132H mutation on IHC. Of the 102 IHC IDH1R132H negative cases with interpretable ATRX results, 95 showed retained expression and 7 showed a loss for ATRX protein. Of the 16 IHC IDH1R132H positive cases, 3 showed retained expression and 13 showed loss of ATRX expression. Additional Sanger sequencing for IDH1/2 was done in 56/116 IHC IDH1R132H negative cases (of these 56 cases, the ATRX results were – loss: 6; retained: 42; equivocal: 4, non-contributory: 4). One of these 56 IDH1R132H IHC negative cases showed IDHR132S mutation (which showed ATRX loss and was EGFR non-amplified). However, 5 other ATRX loss cases did not show any IDH1/2 mutations. p53 protein expression On IHC, p53 was positive (over-expressed) in 55 cases [Figure 1]d (GBM-NOS: 45; GBM-G: 1; GBM-E: 1; GS: 1; GBM-S: 1; GBM-O: 6; age range, ≤30 years: 10; 31–40 years: 11; 41–50 years: 13; 51–60 years: 7; >60 years: 14), focal positive in 34 cases (31–40 years: 7; 41–50 years: 4; 51–60 years: 12; >60 years: 11) and negative in 43 cases (≤30 years: 4; 31–40 years: 6; 41–50 years: 5; 51–60 years: 21; >60 years: 7). Two cases were interpreted as non-contributory as there were no cells with any amount of nuclear staining. No significant correlation was seen between p53 expression and ATRX expression (P - value, 0.192). A statistically significant correlation was seen between p53 over-expression (after grouping focal positive and negatives as non-positive group) and EGFR amplification status (P value 0.020). No significant correlation of p53 expression with ATRX was found even after combining the positives and focal positives as one group of p53 overexpression (P value, 0.598). The frequencies of histological and molecular parameters investigated in relation to morphological patterns are shown in [Table 1].
Correlation of MGMT status with clinico-pathological parameters [Table 2], [Table 3], [Table 4]
The MGMT promoter methylation group (the combined group of methylation and low methylation) did not show statistically significant correlation with age (P - value, 0.969), sex (P - value, 0.91), or location (P - value, 0.56) [Table 2]. Confluent necrosis was noted in 55 of the 68 (80%) unmethylated cases vs 37 of the 66 (56%) methylated cases. The odds of MGMT unmethylation were three times more in the presence of confluent necrosis (OR, 3.3; CI, 1.5–7.2), with statistical significance (P - value, 0.002). Pseudo-palisading necrosis was seen in 8/66 (11.4%) of methylated and 14/68 (26.4%) of unmethylated cases and was not statistically significant (P - value, 0.18). Six of 66 (9%) methylated and 9/68 (13%) of unmethylated cases showed calcification with no significant correlation between them (P - value, 0.45) [Table 3]. MGMT methylation status was significantly associated with IDH1R132H mutation (OR, 4.3; CI, 1.3–13.8; P - value, 0.01). Seventy-four percent (14/18) of IDH1R132H mutated cases were methylated. The frequency of IDH1R132H mutation was 21% (14/66) and 5.8% (4/68) in the methylated and unmethylated groups, respectively. ATRX loss was seen in 15/66 (23%) of the methylated and 5/68 (7.3%) of the unmethylated cases. There was a statistical significant correlation between MGMT methylation and ATRX loss (OR, 4.0; CI, 1.3–11.8; P - value 0.01). The frequency of p53 protein over-expression was 40% in both methylated and unmethylated groups with no correlation between them (P - value, 0.86). EGFR amplification was seen in 13 of 66 (20%) of methylated vs 14/68 (20.5%) of unmethylated cases. There was no significant correlation between MGMT methylation status and EGFR amplification (OR, 0.85; CI, 0.3–2.0; P - value, 0.71) [Table 4]. Statistical analysis to test the association between the MGMT methylation status with clinico-pathological parameters was also carried out after excluding the 14 LM cases from the methylated group. A significant association of MGMT unmethylation was seen with confluent necrosis (P - value, 0.001); while methylation was observed for IDHR132H mutation (P - value, 0.01), and ATRX loss (P - value, 0.03). However, no significant association was seen with age, sex, location, pseudo-palisading necrosis, calcification, p53 expression, or EGFR amplification. Logistic regression analysis On univariable logistic regression analysis, only four variables – IDHR132H mutation, ATRX expression, confluent necrosis, and pseudo-palisading necrosis had a P - value of <0.20, and hence, were included as covariates in multivariable logistic regression analysis [Table 5].
Multivariable logistic regression revealed only the presence of confluent necrosis as an independent predictor of MGMT unmethylation status (OR, 2.5; CI, 1.0–5.8; P - value, 0.04) [Table 5].
MGMT promoter methylation in GBM is one of the prognostic and clinical trial proven predictive biomarkers for chemoradiation.[1],[7] The MGMT gene is located at chromosome 10q26 and encodes a DNA repair enzyme, which is regulated by methylation-dependent epigenetic silencing mechanisms. Its promoter contains TATA-less and CAAT-less 777-bp 97 CpG (short 300-3000bp stretches of C-G rich regions) sites, which are usually unmethylated in normal tissues. Methyl CpG binding proteins bind to aberrantly methylated sequences on CpG islands, leading to condensation of chromatin and prevent the binding of transcriptional machinery, thus inhibiting the process of gene transcription (gene silencing).[7],[8] Nakagawachi et al., and Everhard et al., in their work on identification of regions correlating MGMT promoter methylation and gene expression in GBM, have demonstrated two regions in the promoter that are prone to high levels of methylation, one of which is the enhancer region. Methylation occurring at this region is more critical for the loss of MGMT gene expression; therefore, most methylation-specific tests are designed to target this region.[9],[10] In the present study also, the MGMT promoter methylation evaluation was based on primers targeting the same region. Several methods have been used for the assessment of MGMT methylation. MSP, the most commonly used technology, to date is the only test that has repeatedly been shown to be of predictive or prognostic value in different clinical trials.[11],[12],[13],[14],[15],[16],[17] It allows the evaluation of methylation status at six to nine CpG sites.[13],[14],[18] When well-optimized, the method delivers easy-to-interpret results, and the gel-based detection does not require expensive equipment. However, the disadvantages include the conduction of a non-quantitative technique, a lower number of assessable CpG dinucleotides, and a relative lack of or no control for bisulfite conversion, and hence, incomplete conversion of unmethylated cytosines may be interpreted as methylation, leading to false-positive results.[8] However, in the present study, all reactions were run with appropriate positive and negative controls, with many of them run in duplicates with reproducible results. Other quantitative and semiquantitative techniques with unique advantages have also been used. qPCR, a sensitive and specific technique, allows higher levels of standardization and lacks definition of cut-offs for methylation, as the percentage of contaminating stromal cells cannot be accurately assessed.[19] Pyrosequencing and bisulfite sequencing are currently regarded as the gold standard as they provide single base pair resolution and quantitative methylation information but are cost and labor intensive.[20] However, the clinical significance of detection of such low levels of methylation was always questionable. Combined bisulfite restriction analysis (COBRA) utilizes restriction enzymes that differentiate between methylated and unmethylated sequences.[20] Methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA) is fast gaining popularity as it does not require bisulfate modification.[21] Methyl-specific high-resolution melting analysis (MS-HRM) evaluates large regions. Bead arrays, mass spectroscopy, and denaturing high-performance liquid chromatography are some of the other available techniques.[22],[23],[24],[25] No concordance between MGMT IHC and MGMT methylation (by MSP) and patient survival has been reported.[26],[27],[28] The reasons include lack of reproducibility among neuropathologists, lack of cut-off values, tumor heterogeneity, and admixture of MGMT expressing non-neoplastic cells.[29] At present, IHC is not the method of choice for the detection of MGMT methylation and should be discouraged in clinical practice. The frequency rates of MGMT methylation by MSP varies from as low as 35% to as high as 68% in various studies.[11],[12],[15],[16],[30],[31],[32],[33],[34] The frequency of methylation detected by other techniques ranges from 26% to 80% (pyrosequencing 53–67%; qMSP 24–69%; MS-MLPA 58–80%, COBRA 41%, bead arrays 38–60%).[9],[17],[22],[27],[34],[35],[36],[37],[38],[39] The frequency of methylation in our study by MS-PCR was 49.2%. There was no significant difference in the median age (51.5 years; range 21–75 years vs 51.5 years; range 20–74 years) between the methylated and unmethylated groups in our study and there was also no significant correlation with gender (P - value, 0.91). Eoli et al., in their study of 86 GBMs, have shown a correlation between the MGMT promoter methylation status and tumor location. GBMs with MGMT methylation were preferentially located in the parietal and occipital lobes and unmethylated cases were located in the temporal lobe.[30] In our study, 44% of the methylated cases were located in the frontal lobe, and 47% of unmethylated cases were located in the temporal lobe. Compared to the frontal lobe (taken as the baseline for comparison of odds ratio (OR) of other sites in the study relative to the baseline), MGMT unmethylated cases were 2.2 times more commonly located in the temporal lobe (OR, 2.2; confidence interval [CI], 1.0–4.8), however, with borderline statistical significance (P - value, 0.05). Necrosis is the hallmark of GBM and indicates rapid growth and malignant behavior. It is a significant independent prognostic factor that negatively affects progression-free survival. Eoli et al., in their study of 86 GBMs, observed that necrosis was present in 67.5% of unmethylated cases and 32% of methylated cases, with a significant association between MGMT unmethylation and necrosis.[30] In our study also, a significant association was seen between MGMT unmethylation and confluent necrosis (P - value, 0.002; CI, 1.5–7.2). On multivariable logistic regression, confluent necrosis was found to be an independent predictor of MGMT promoter unmethylation status (OR, 2.5; CI, 1.0–5.8; P - value, 0.04). However, no significant correlation was seen with pseudo-palisading necrosis or calcification. Mutational analysis of GBM has revealed somatic mutations of IDH1 gene in 12% of GBMs.[40] Sanson et al., found a statistical significant correlation between MGMT methylation and IDH1 mutation in 194 gliomas.[41] However Jha et al., and Suri et al., did not find any statistical correlation between MGMT methylation and IDH1 mutation in GBMs.[33],[34] Jha et al., found the frequency of IDH 1 mutation in 4.8% and 6.7% tumors in methylated and unmethylated GBM groups, respectively. Felsberg et al., and Weller et al., also did not find any significant correlation between the two in their series.[16],[42] The frequency of IDH1R132H mutation in our study was 13.4% and was significantly associated with MGMT methylation status (OR, 4.3; CI, 1.3–13.8; P - value, 0.01). However, the present study is limited by lack of data on other IDH1/2 mutations, which would have been more relevant. Three cases of GBM (two frontal and one parietal) with IDHR132H mutation retained ATRX expression and non-amplified EGFR were re-reviewed and histologically they were unequivocal GBMs. No oligodendroglial morphology was identified. However, the characteristics of these tumors are limited by the lack of 1p19q deletion data. The German NOA-04 trial (NCT00717210) correlating IDH mutations and MGMT promoter methylation with ATRX expression found that ATRX loss occurred almost exclusively in tumors harboring IDH mutations (42/98 IDH mutated tumors). Furthermore, no association between MGMT promoter methylation and ATRX loss was observed.[43] In our study, ATRX loss was seen in 15% of GBMs and correlated significantly with MGMT promoter methylation status (OR, 4.0; CI, 1.3–11.8; P - value, 0.01). The incidence of p53 protein accumulation is more frequently seen than p53 mutations, and the over-expression of p53 in GBM is often associated with the presence of p53 mutations.[44] Studies on the association of MGMT promoter methylation with EGFR amplification and p53 mutation are highly contradicting. Shamshara et al., and Lofti et al., reported a statistically significant correlation between MGMT promoter methylation and p53 mutations in GBMs.[45] However, Jesien-Lewandowicz et al., in their study of 32 primary GBMs did not find any significant correlation.[46] In our study, p53 positivity was the same (40%) in the methylated and unmethylated groups with no statistical significant association between p53 expression and MGMT methylation (P - value, 0.86). Eoli et al., found EGFR amplification to be more frequent in unmethylated than in methylated GBMs (40% vs. 28%) though the difference was not statistically significant.[30] Jha et al., found EGFR amplification in 30.4% of methylated GBMs and in 42.1% of unmethylated GBMs with no significant association.[33] In our study, the frequency of EGFR amplification in methylated and unmethylated groups was relatively similar (20% and 20.5%, respectively) with no significant association between them (P - value, 0.71). Hegi et al., in the EORTC/NCIC phase III trial established that MGMT methylated patients lived longer, irrespective of the treatment. The median overall survival among patients with methylation was 18.2 months compared with 12.2 months among those without methylation. A 5-year analysis of the same study demonstrated that MGMT methylation is the strongest predictor for outcome and benefit from temozolomide (TMZ) chemotherapy.[47] The Radiation Therapy Oncology Group (RTOG) 0525 study in GBM showed that MGMT methylation confers benefit in terms of progression-free survival and overall survival. Other studies have also validated the same findings. In the present study, as the follow-up data available is too short, statistical correlation of MGMT methylation with survival has not been done. To summarize, the frequency of MGMT promoter methylation in GBMs in our study was 49.2%, which was significantly associated with IDHR132H mutation and ATRX loss. In addition, the presence of confluent necrosis was significantly associated with MGMT unmethylation and was found to be an independent predictor of the same. The frequency of methylation in the frontal location is higher compared to other cerebral hemispheric locations. Acknowledgments The authors acknowledge Dr. Salma Ayis, King's College London, for the statistical advice. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]
|
|
|||||