Aberrant Hypermethylation of p14^ARF and O⁶‐methylguanine‐DNA Methyltransferase Genes in Astrocytoma Progression

Abstract

The aim of the present study was to elucidate genetic alterations that are critically involved in astrocytoma progression. We characterized 27 World Health Organization grade II fibrillary astrocytomas which later underwent recurrence or progression, paying specific attention to the CpG island methylation status of critical growth regulatory genes. p14^ARF and O⁶‐methylguanine‐DNA methyltransferase (MGMT) hypermethylation represented frequent events (26% and 63%, respectively), which were mutually exclusive except in one case, with alternate or simultaneous methylation of these two genes occurring in 85% of our tumor series. Seventeen tumors (63%) contained TP53 mutations, which were closely related to the presence of MGMT methylation. Methylation of the p21^Waf1/Cip1, p27^Kip1 and p73 genes and homozygous deletion of the p16^INK4a, p15^INK4b and p14^ARF genes were not detected in any of the primary low‐grade tumors. The presence of p14^ARF methylation at first biopsy was associated with shorter patient survival, whereas the presence of MGMT methylation carried a better clinical outcome after salvage therapy. Examination of 20 cases whose histological data for recurrent tumors were available revealed that malignant progression occurred in all of the tumors with p14^ARF methylation but less frequently (50%) in the lesions with MGMT methylation. On analysis of their respective recurrent tumors, five of six patients whose primary low‐grade tumors carried p14^ARF methylation exhibited homozygous co‐deletions of the p14^ARF, p15^INK4b and p16^INK4a genes, which were restricted to glioblastoma as the most malignant end point. Our findings suggest that p14^ARF hypermethylation and MGMT hypermethylation constitute distinct molecular pathways of astrocytoma progression, which could differ in biological behavior and clinical outcome.

INTRODUCTION

Low‐grade diffuse astrocytomas [World Health Organization (WHO) grade II)] are well‐differentiated tumors that grow slowly but display an intrinsic tendency to progress to a more malignant phenotype, that is, anaplastic astrocytomas (WHO grade III) and eventually glioblastomas (WHO grade IV). The gemistocytic subtype appears to be particularly prone to rapid recurrence and malignant transformation, whereas the fibrillary subtype, the most frequent histological variant of diffuse astrocytoma, exhibits a clinical and biological diversity, which makes it difficult to predict its prognosis accurately (10).

Over the past decade, our understanding of the molecular genetic abnormalities involved in the pathogenesis and progression of diffuse astrocytomas has advanced significantly. Mutation of the TP53 tumor suppressor gene represents the most frequent genetic alteration in diffuse astrocytomas and appears to be of significance for the initiation of malignant transformation (9, 10, 17, 18). Several studies have demonstrated a tendency toward malignant progression and poor prognosis in patients with WHO grade II astrocytic tumors harboring TP53 mutations (6, 20), although the prognostic impact of the TP53 status appears to be closely related to the influence of the gemistocytic subtype (20). In previous studies examining exclusively supratentorial fibrillary astrocytomas in adults, we found that promoter hypermethylation of the DNA repair gene O⁶‐methylguanine‐DNA methyltransferase (MGMT) was the most powerful determinant of a shortened recurrence‐free survival among various genetic alterations (8, 25). As a lack of MGMT repair capacity can confer cancer cells with additional mutability of other critical growth regulatory genes such as TP53 (2, 14), transcriptional silencing of MGMT by promoter hypermethylation could contribute to the genesis and progression of diffuse astrocytomas. However, the presence of MGMT methylation was found not to correlate directly with the occurrence of malignant progression (8), suggesting that an additional, not yet unequivocally identified genetic alteration may be critically involved in astrocytoma progression.

The present study focuses on low‐grade diffuse astrocytomas with evidence of recurrence or progression. We profiled abnormalities of the RB1, p16^INK4a, p15^INK4b, TP53, p14^ARF, p21^Waf1/Cip1, p27^Kip1, p73 and MGMT genes in 27 patients with WHO grade II fibrillary astrocytomas. In addition, the sequence of genetic alterations during astrocytoma progression was assessed by examining their recurrent tumors.

MATERIALS AND METHODS

Patients and tumor samples.

Between 1981 and 1998, a total of 64 adult patients (≥18 years of age) with a new histological diagnosis of supratentorial fibrillary astrocytoma (grade II) classified according to the WHO criteria (10, 11) were treated at our institution. Among these cases, 46 tumor samples were available for various genetic analyses as reported previously (25). Up to the time of the last follow‐up (April 2006), 27 patients eventually displayed tumor recurrence that was confirmed by imaging studies. In total, a series of 27 primary tumors and 18 recurring tumors were investigated. Genomic DNA was extracted from paraffin sections as described previously (8, 25).

Methylation‐specific polymerase chain reaction (PCR).

Promoter hypermethylation of the RB1, p16^INK4a, p15^INK4b, p14^ARF, p21^Waf1/Cip1, p27^Kip1, p73 and MGMT genes was determined by the methylation‐specific PCR (5). Sodium bisulfite modification was performed with a CpGenome^TM DNA Modification Kit (Intergen, Oxford, UK) as described previously (22, 23). The methods used for the methylation‐specific PCR of these genes were as described previously (8, 16, 19, 24, 25). The amplified products were electrophoresed on 3% agarose gels and visualized with ethidium bromide. CpGenome Universal Methylated DNA (Intergen) and normal blood DNA were included in each PCR set as methylated and unmethylated controls, respectively.

Differential PCR.

To assess homozygous deletions of the p14^ARF (exon 1β), p15^INK4b and p16^INK4a (exon 1α) genes, differential PCR was carried out as described previously (16, 23, 25). As a reference, the β‐actin sequence was used for p15^INK4b and p16^INK4a deletions, and the glyceraldehyde‐3‐phosphate dehydrogenase sequence for the p14^ARF deletion. Samples showing less than 20% of the control signal were considered homozygous deletions (16, 23, 25).

Statistical analysis.

The Kaplan and Meier method was used to calculate the recurrence‐free survival, post‐recurrence survival and overall survival. We designated the recurrence‐free survival as the time period from the first operation to the point when tumor recurrence or regrowth was confirmed by an imaging study. The post‐recurrence survival was defined as the time interval between tumor recurrence or regrowth and death or the date of the most recent evaluation. The overall survival was determined as the sum of the recurrence‐free survival and the postrecurrence survival. The log‐rank test was used to assess the degree of significance of differences among the different subgroups. The relationships between the various parameters were analyzed statistically by Fisher’s exact test. The significance level chosen was P < 0.05, and all tests were two‐sided. The statistics were analyzed using a personal computer running Stat View J‐5.0 software (Abacus Concepts, Berkeley, CA, USA).

RESULTS

Table 1 summarizes the clinicopathological characteristics and genetic alterations observed in the 27 patients. At the initial treatment, eight of the patients (cases no. 1–8) were administered immediate radiation therapy, nine patients (cases no. 9–17) received surgery alone and 10 patients (cases no. 18–27) were treated with human fibroblast interferon without radiation therapy as reported previously (26). None of the patients were subjected to chemotherapy as the first adjuvant modality. The RB1 methylation, p16^INK4a methylation and homozygous deletion, p15^INK4b methylation and homozygous deletion, TP53 mutation, p14^ARF methylation and homozygous deletion, and MGMT methylation in all primary tumors, and MGMT methylation in 14 recurrent tumors had been examined previously (8, 25). In all of the recurrent cases, TP53 mutation could not be evaluated because of limitations in the DNA available for analysis.

Table 1.

Summary of data for 27 diffuse astrocytomas with evidence of recurrence or progression. Abbreviations: Me = methylation; Mu = mutation; Ho = homozygous deletion; MGMT = O⁶‐methylguanine‐DNA methyltransferase gene; ACNU = 1‐(4‐amino‐2‐methyl‐5‐pyrimidinyl)methyl‐3‐2(2‐chloroethyl)‐3‐nitrosourea; PCZ = procarbazine; VCR = vincristine; rad = radiation therapy.

Case no.	Age/sex	Genetic alterations at first biopsy	Recurrence‐free survival (months)	Genetic alterations at second biopsy	Histological grade	Salvage adjuvant therapy	Post‐recurrence survival (months)
1	47/F	p14ARF Me	29	p14ARF/p15INK4b/p16INK4a Ho	IV	ACNU	14
2	35/M	MGMT Me, TP53 Mu	54	—	IV	ACNU	9
3	52/F	MGMT Me, TP53 Mu	47	—	—	ACNU	9
4	49/M	p15INK4b/p16INK4a/MGMT Me	39	—	—	ACNU	2
5	22/F	No aberrations were detected	126	—	—	ACNU	9
6	33/F	MGMT Me, TP53 Mu	87	—	—	ACNU	4
7	45/F	p14ARF Me	61	p14ARF/p15INK4b/p16INK4a Ho	IV	ACNU	17
8	65/M	p16INK4a Me, TP53 Mu	13	—	—	ACNU	14
9	43/F	MGMT Me	67	MGMT Me	II	—	172+
10	37/M	MGMT Me, TP53 Mu	27	MGMT Me	II	ACNU, rad	58
11	42/F	MGMT Me, TP53 Mu	32	MGMT Me	III	ACNU, rad	35
12	21/M	MGMT Me, TP53 Mu	44	p73/MGMT Me	IV	ACNU, rad	35
13	24/M	MGMT Me, TP53 Mu	51	MGMT Me	II	ACNU, rad	93
14	46/M	p14ARF/MGMT Me, TP53 Mu	23	—	—	ACNU, rad	31
15	42/F	MGMT Me, TP53 Mu	56	—	—	ACNU, rad	23
16	19/M	p15INK4b/MGMT Me	16	—	II	ACNU, rad	31
17	27/F	p14ARF Me	58	p14ARF/MGMT Me	IV	ACNU, rad	24
18	48/M	RB1 Me, MGMT Me	43	RB1/MGMT Me	II	—	84
19	22/M	MGMT Me, TP53 Mu	56	MGMT Me	III	ACNU, PCZ, VCR, rad	62
20	48/F	No aberrations were detected	88	p73/MGMT Me	IV	ACNU, PCZ, VCR, rad	18
21	44/F	MGMT Me, TP53 Mu	38	MGMT Me	III	ACNU, PCZ, VCR, rad	49
22	18/F	p14ARF Me, TP53 Mu	14	p73 Me, p14ARF/p15INK4b/p16INK4a Ho	IV	ACNU, PCZ, VCR, rad	24
23	49/M	p53 Mu	39	MGMT Me	II	ACNU, PCZ, VCR, rad	38
24	22/F	MGMT Me, TP53 Mu	47	MGMT Me	II	ACNU, PCZ, VCR, rad	51
25	30/F	p14ARF Me	47	p14ARF/p15INK4b/p16INK4a Ho	IV	ACNU, PCZ, VCR, rad	15
26	50/F	MGMT Me, TP53 Mu	66	MGMT Me	III	ACNU, PCZ, VCR, rad	25+
27	33/M	p14ARF Me	39	p14ARF/p15INK4b/p16INK4a Ho	IV	ACNU, PCZ, VCR, rad	11

Among the 27 primary tumors investigated, promoter hypermethylation of the MGMT, RB1, p14^ARF, p16^INK4a and p15^INK4b genes was detected in 17 (63%), one (4%), seven (26%), two (7%) and two samples (7%), respectively. Methylation of the p21^Waf1/Cip1, p27^Kip1 and p73 genes and homozygous deletion of the p16^INK4a, p15^INK4b and p14^ARF genes were not detected in any of the tumors investigated. Seventeen tumors (63%) contained TP53 mutations. The patterns of these abnormalities in the individual tumors were then compared. The presence of MGMT methylation was positively associated with TP53 mutation (P = 0.0402). Conversely, MGMT methylation and p14^ARF methylation were mutually exclusive, except in one case (P = 0.0042). p14^ARF methylation was inversely associated with TP53 mutation, although the relationship did not reach statistical significance (P = 0.0840).

Of the 27 patients, 20 were given a second operation, and 13 (65%) of these cases were proven to progress to anaplastic astrocytoma or glioblastoma. All of the patients whose primary tumors contained p14^ARF methylation showed a higher grade of malignancy on recurrence, whereas seven (50%) of the 14 tumors without p14^ARF methylation maintained their original grade (P = 0.0515). In contrast, patients whose primary tumors carried MGMT methylation showed malignant progression less frequently. Malignant transformation was identified in six (50%) of the 12 tumors with MGMT methylation as compared to seven (88%) of the eight tumors without MGMT methylation (P = 0.1577). There was no evident difference in the occurrence of malignant progression between the patients who did and those who did not receive radiation therapy as the first adjuvant modality (three of three patients treated with immediate radiation therapy vs. 10 of 17 patients treated without radiation therapy; P = 0.5211).

At the time of recurrence, alkylating agent‐based chemotherapy was administered to 25 patients. All patients treated without radiation therapy at the first treatment, except for two who underwent total removal at re‐operation and maintained their original grade (cases no. 9 and 18), received radiation therapy in combination with chemotherapy. Eventually, the case no. 18 patient underwent radiation therapy at second recurrence. Although the sampling numbers were too small to perform a rigorous statistical analysis, the presence of p14^ARF methylation at the time of first biopsy (Figure 1) was associated with a significant decrease in post‐recurrence survival (Figure 2). The median post‐recurrence survival among patients with methylated tumors vs. those with unmethylated tumors was estimated to be 17 and 35 months, respectively (P = 0.0410). On the other hand, a positive relationship existed between MGMT methylation and post‐recurrence survival (Figure 3). Patients whose primary tumors carried MGMT methylation had a median postrecurrence survival of 35 months, as opposed to a median postrecurrence survival of 15 months in patients with unmethylated tumors (P = 0.0061). We also examined the effects of these epigenetic events on the overall survival. While MGMT methylation had no statistically significant effect (median overall survival: 85 months for MGMT‐methylated tumors vs. 62 months for MGMT‐unmethylated tumors; P = 0.1336; Figure 3), the presence of p14^ARF methylation at the time of first biopsy emerged as a significant predictor of a shorter overall survival (Figure 2). The median overall survival for patients with p14^ARF‐methylated tumors was 54 months, while that for p14^ARF‐unmethylated tumors was 85 months (P = 0.0031). In terms of the recurrence‐free survival, neither p14^ARF methylation nor MGMT methylation demonstrated a prognostic utility. The median recurrence‐free survival among patients with p14^ARF‐methylated tumors vs. those with p14^ARF‐unmethylated tumors was estimated to be 39 and 47 months, respectively (P = 0.2421). The median recurrence‐free survival among patients with MGMT‐methylated tumors vs. those with MGMT‐unmethylated tumors was estimated to be 47 and 39 months, respectively (P = 0.4258). As regards the TP53 mutation, its presence did not correlate with either the recurrence‐free survival, postrecurrence survival or overall survival (data not shown).

Figure 1.

Methylation‐specific polymerase chain reaction (PCR) of p14^ARF promoter in primary and recurrent tumors. The primary low‐grade tumor of the case no. 5 patient was unmethylated, and the primary low‐grade tumor of the case no. 7 patient showed methylation. In the case no. 17 patient, hypermethylation was present in both the primary low‐grade and recurrent high‐grade tumors. S = molecular size marker; U = PCR product amplified by unmethylated‐specific primers; M = PCR product amplified by methylated‐specific primers; NC = normal control; PC = positive control.

Figure 2.

Post‐recurrence survival (PRS) and overall survival (OS) of 27 patients according to methylation status of the p14^ARF promoter.

Figure 3.

Post‐recurrence survival (PRS) and overall survival (OS) of 27 patients according to methylation status of the MGMT promoter.

We also undertook examinations of 18 recurrent tumors that were available for genetic analysis. In contrast to primary low‐grade tumors, methylation of the p73 gene and homozygous deletion of the p16^INK4a, p15^INK4b and p14^ARF genes were detected in some of the high‐grade recurrent tumors. The p16^INK4a, p15^INK4b and p14^ARF genes were simultaneously deleted in five recurrent tumors (Figure 4), which were restricted to glioblastoma as the most malignant end point. None of the tumors carried one or two deleted loci of the 9p21 gene cluster. Notably, five of six patients whose primary low‐grade tumors carried p14^ARF methylation exhibited concordant homozygous deletion of these three genes at the stage of high‐grade tumors. Homozygous co‐deletion of the 9p21 genes was also observed more frequently in patients who received immediate radiation therapy (100%: two of two) than those who did not (19%: three of 16), although this tendency did not reach statistical significance (P = 0.0654).

Figure 4.

Differential polymerase chain reaction to assess p14^ARF, p15^INK4b and p16^INK4a homozygous deletion in recurrent tumors. The case no. 7, 22, 25 and 27 patients showed concordant homozygous deletion of the p14^ARF, p15^INK4b and p16^INK4a genes at the stage of glioblastoma. As demonstrated in Figure 1, the primary low‐grade tumor of the case no. 7 patient carried p14^ARF methylation. The case no. 20 patient had a normal gene status. GAPDH = glyceraldehyde‐3‐phosphate dehydrogenase; S = molecular size marker; NC = normal control.

DISCUSSION

The present investigation was undertaken to characterize 27 primary WHO grade II fibrillary astrocytomas that eventually showed tumor recurrence during the observation period, with specific attention to the methylation status of critical growth regulatory genes. Although all patients had been included in previous studies (8, 25), the present report analyzes additional genetic parameters (ie, methylation of the p21^Waf1/Cip1, p27^Kip1 and p73 genes in primary and recurrent tumors, and homozygous deletion of the p14^ARF, p15^INK4b and p16^INK4a genes in recurrent tumors) with longer follow‐up periods. Our data demonstrated that p14^ARF and MGMT methylation represented frequent events in our series (26% and 63%, respectively) and that they were negatively associated with one another. In accordance with previous studies (2, 14), the presence of MGMT methylation was closely related to TP53 mutation. We further provided evidence that patients whose primary tumors harbored p14^ARF methylation invariably underwent malignant progression and displayed a shorter survival, whereas patients with tumors carrying MGMT methylation were less likely to progress and had a favorable clinical outcome. Although these results need to be confirmed in larger series, our observations do suggest the existence of two independent genetic pathways underlying astrocytoma progression, which could reflect differential clinical and biological behaviors.

The p14^ARF, p15^INK4b and p16^INK4a genes cluster together on chromosome 9p21. The frequency of promoter hypermethylation in diffuse gliomas varies considerably among different genes and among different study groups. Our own data from primary low‐grade astrocytomas demonstrated values of 26%, 7% and 7% methylation for the p14^ARF, p15^INK4b and p16^INK4a genes, respectively. Such preferential methylation of p14^ARF rather than p15^INK4b or p16^INK4a is in agreement with data given in some previous reports (13, 22, 23), suggesting that the p14^ARF locus may represent a hotspot of aberrant hypermethylation of the 9p21 gene cluster in diffuse gliomas.

p14^ARF plays a major role in the TP53 tumor suppressor pathway by antagonizing MDM2‐mediated degradation of TP53. Thus, theoretically, p14^ARF alteration would be predicted to reduce the frequency of concomitant TP53 mutations (25). In the present study, the presence of p14^ARF methylation tended to be inversely associated with TP53 mutation, although the relationship failed to achieve statistical significance, possibly because of small patient number and selected patient population.

In the present study, five of six patients whose primary low‐grade tumors carried p14^ARF methylation exhibited homozygous deletion of p14^ARF together with p15^INK4b and p16^INK4a on recurrence. One possible reason for an unfavorable prognostic influence of p14^ARF methylation is that such sequential acquisition of homozygous co‐deletions of the 9p21 genes could participate in a progression to a more malignant phenotype. In support of this idea, p14^ARF, p15^INK4b and/or p16^INK4a homozygous deletion has frequently been found in high‐grade gliomas but is either absent or extremely rare in low‐grade lesions (13, 22, 23). Furthermore, several clinical studies have demonstrated a significant association between the presence of p14^ARF or p16^INK4a homozygous deletion and a poorer survival in a subset of glioblastomas (7, 15), although this correlation was not consistently sustained among larger series (18). In the study of Nakamura et al (13), who analyzed multiple biopsies from the same patients during astrocytoma progression, all glioblastomas that had subsequently progressed from low‐grade astrocytomas with p14^ARF methylation displayed either p14^ARF hypermethylation or homozygous deletion. A similar observation has been reported for some renal cancers, in which hypermethylation at locus D17S5 of chromosome 17p was already present in the early‐stage tumors and was followed by allelic loss of 17p in the late‐stage tumors (12). These findings suggest that regional hypermethylation of a certain locus could facilitate genetic instability, leading to subsequent chromosomal structural aberrations at the relevant loci.

In a previous cohort of 49 newly diagnosed WHO grade II astrocytomas (8), MGMT methylation represented a significant adverse indicator of the recurrence‐free survival. This study failed to demonstrate a significant correlation between the occurrence of malignant transformation and the presence of MGMT methylation, indicating that MGMT methylation is associated with recurrence of low‐grade astrocytomas rather than malignant progression. In the present study, MGMT methylation did not influence the recurrence‐free survival significantly. In terms of the post‐recurrence survival, the presence of MGMT methylation at the time of first biopsy emerged as a significant positive predictor. The most likely explanation for these conflicting results is the difference in the selected patient populations: the current study included only a series of patients with evidence of tumor recurrence or regrowth, which made it inherently difficult to assess the actual predictor of the recurrence‐free survival. As both the recurrence‐free survival and overall survival of patients with MGMT‐methylated tumors were similar to those of patients with MGMT‐unmethylated tumors, the difference in the postrecurrence survival might reflect the effect of the treatment at the time of recurrence rather than the intrinsic biological aggressiveness of the tumor. MGMT plays a major role in conferring glioma cells with a resistance to alkylating chemotherapeutic agents (3). Recent clinical studies have shown that epigenetic inactivation of MGMT by promoter hypermethylation can increase the sensitivity of malignant gliomas to alkylating agent chemotherapy (1, 4, 27). Given the fact that none of our patients received chemotherapy as the first adjuvant modality, salvage chemotherapy could be responsible for the prolonged postrecurrence survival among patients with MGMT‐methylated tumors, although the responsiveness to chemotherapy could not be evaluated because of the small number of assessable patients and nonuniformity in the therapeutic management.

In a recent report by Tso et al (21), who used a DNA microarray, secondary glioblastomas that developed from lower grade astrocytomas displayed aberrant expressions of mitotic cell cycle‐associated genes more frequently as compared with primary glioblastomas that rapidly developed de novo without clinical, radiological or morphological evidence of a less malignant precursor lesion. This suggests that disruption of cell cycle regulation may potentially play a crucial role in the pathway leading to glioblastomas, and so prompted us to assess the methylation status of the key cell cycle regulatory genes (RB1, p16^INK4a, p15^INK4b, p14^ARF, p21^Waf1/Cip1, p27^Kip1 and p73). Except for p14^ARF, each gene methylation was either absent or rare in both primary and recurrent tumors, suggesting a minimal role for epigenetic silencing of these regulatory genes in astrocytoma progression.

The therapeutic contribution of radiation therapy for patients with low‐grade astrocytomas still remains a matter of discussion. In the present study, administration of immediate radiation therapy did not directly correlate with the occurrence of malignant transformation but appeared to be closely related to the presence of homozygous co‐deletions of the p14^ARF, p15^INK4b and p16^INK4a genes at the time of recurrence. These observations raise the possibility that irradiation could invoke a certain kind of genetic alteration, although such an assumption requires careful evaluation in further in vitro and in vivo investigations.

In conclusion, we have provided evidence to suggest the existence of two independent genetic pathways in astrocytoma progression. One pathway was characterized by p14^ARF methylation, which led to homozygous co‐deletions of the 9p21 genes during progression and carried a worse prognosis. The other pathway was characterized by MGMT methylation, which was positively correlated with TP53 mutation and yielded favorable clinical results following salvage adjuvant therapy. These two subtypes accounted for 85% of all cases in our series, and it seems likely therefore that other genetic changes are involved in the complex process of malignant progression.