Fax: (011) 43-1-40400-4566
O6-methylguanine DNA methyltransferase immunoexpression in nonfunctioning pituitary adenomas†
Are progressive tumors potential candidates for temozolomide treatment?
Article first published online: 20 JAN 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 5, pages 1070–1080, 1 March 2009
How to Cite
Widhalm, G., Wolfsberger, S., Preusser, M., Woehrer, A., Kotter, M. R., Czech, T., Marosi, C. and Knosp, E. (2009), O6-methylguanine DNA methyltransferase immunoexpression in nonfunctioning pituitary adenomas. Cancer, 115: 1070–1080. doi: 10.1002/cncr.24053
We thank Brigitte Hammer-Schmiedel and Irene Leisser for technical assistance with preparation of tissue specimens.
- Issue published online: 18 FEB 2009
- Article first published online: 20 JAN 2009
- Manuscript Accepted: 27 AUG 2008
- Manuscript Revised: 25 AUG 2008
- Manuscript Received: 9 JUL 2008
- progressive nonfunctioning pituitary adenomas;
- O6-methylguanine DNA methyltransferase;
Currently, no effective alternative treatment exists for progressive, regrowing, nonfunctioning pituitary adenomas (NFPA) that are resistant to conventional multimodality therapy. Temozolomide (TMZ) was proposed as a treatment option for pituitary carcinomas and aggressive pituitary adenomas. Recently, it was suggested that the responsiveness of pituitary tumors to TMZ depends on the immunoexpression of O6-methylguanine DNA methyltransferase (MGMT). Therefore, the authors of this report assessed MGMT expression in a series of patients with progressive, regrowing NFPAs to evaluate whether TMZ may serve as alternative treatment option.
On the basis of postoperative magnetic resonance imaging, 45 patients with NFPAs were allocated to either a group with progressive, regrowing tumors (n = 24) or a tumor-free group (n = 21), which served as a control. MGMT expression was assessed semiquantitatively by immunohistochemistry (low expression was defined as ≤50% immunostained adenoma cells, and high expression was defined as >50% immunostained adenoma cells) and was compared between the 2 groups.
At the time of initial surgery, low MGMT expression was observed in 12 of 24 patients (50%) in the study group with progressive, regrowing NFPAs. In the control group of tumor-free patients, only 5 of 21 patients (24%) exhibited low MGMT expression. A comparable distribution of MGMT expression was observed in the specimens from repeat surgeries. A shorter interval to second surgery was observed in patients who had low MGMT expression.
The current data has suggested that half of the patients with progressive, regrowing NFPAs exhibit low MGMT expression and are potential candidates for treatment with TMZ. These findings provide a rationale for the use of TMZ as an alternative treatment approach in this subgroup if conventional therapy, including reoperation, radiosurgery, and radiotherapy, fails. Cancer 2009. © 2009 American Cancer Society.
Transsphenoidal surgery currently is considered the standard treatment of nonfunctioning pituitary adenomas (NFPA). Because of their frequently large size at the time of diagnosis and their potentially invasive growth into perisellar structures, incomplete resection is not uncommon1-16 despite advancing technologies in surgery, such as endoscopy, neuronavigation, and intraoperative magnetic resonance imaging (MRI). Because regrowth from adenoma remnants is observed frequently (in 38%-95% of patients who undergo incomplete resection4, 6, 12, 14, 17, 18), additional treatments, such as reoperation, stereotactic radiosurgery, and fractionated radiotherapy, may be required. Despite multimodality treatment, however, NFPAs can progress in rare cases and pose a significant therapeutic dilemma; because, currently, there are no pharmacologic treatment options available. For adenomas that escape standard therapeutic regimens, the establishment of alternative, clinically applicable therapeutic options is important.
Temozolomide (TMZ), an oral alkylating agent, is part of the routine therapy for glioblastoma multiforme and recently also has been applied successfully for the treatment of endocrine neoplasms.19 TMZ also has been proposed for the treatment of prolactin-producing and nonfunctioning, aggressive pituitary adenomas and carcinomas (Table 1).20-27 The few case studies that have been published reported clinical improvement, tumor shrinkage, and reduction of hormone secretory activity in most (but not all) patients who were treated. Consequently, the assessment of O6-methylguanine DNA methyltransferase (MGMT) expression as a potential marker for adenoma responsiveness to TMZ treatment has been proposed on the basis of 2 clinical cases: In the first case, an MGMT-immunonegative tumor was identified as responsive to TMZ; whereas, in the second case, high MGMT expression was associated with TMZ resistance.23
|Proliferation Rate (Ki-67)||TMZ Treatment|
|Patient No.||Reference||WHO Classification||Before TMZ||After TMZ||Dopamine Agonist Resistance||Radiotherapy||Duration, mo||Response||MGMT Expression|
|1||Kovacs 200823||NF (silent subtype II)||ND||ND||—||ND||10*||No||Positive|
|3||Kovacs 2007,22 Kovacs 2008,23 Syro 200626||PRL||40-60||5||Yes||Yes||7||Yes||Negative|
The objective of the current study was to assess MGMT immunoexpression in a series of recurrent NFPAs that required further treatment. Our data may serve as a basis for considering whether TMZ could serve as a potential alternative treatment for the rare cases of NFPAs that are resistant to conventional therapy.
MATERIALS AND METHODS
Definition of the Study Cohort
For the current study, the database of the Department of Neurosurgery of the Medical University Vienna was screened for NFPAs (Table 2). Patients were selected according to the following criteria:
- 2)Availability of adequate tissue specimens for histologic typing and immunohistochemistry (therefore, apoplectic pituitary adenomas without vital tumor areas were excluded);
- 3)Postoperative radiographic follow-up: Only patients who had MRI follow-up performed at a minimum of 3 months, 6 months, and 12 months postoperatively and at 12-months intervals thereafter were included. In patients who had large residual tumor mass, critical location (eg, vicinity of the optic nerve), or accelerated tumor growth, reassessments were carried out at shorter time intervals.
|No. of Patients (%)|
|Characteristic (n=45)||Study Group (Progressive, Regrowing)||Control Group (Tumor Free)|
|Median no. of surgeries [range]||2 [1-6]||1|
|Median follow-up, y||9.3 [5.4-16.3]|
|Median time to second surgery [range], y||4.3 [1.3-10.3]|
|Median age [range], y||53 [17-83]||57 [27-73]|
|Men||14 (58)||15 (71)|
|Women||10 (52)||6 (29)|
|Median tumor size [range], mm||30 [15-55]||25 [10-50]|
|Approach at first surgery|
|Transsphenoidal||19 (79)||21 (100)|
|Transcranial||5 (21)||0 (0)|
|Invasiveness (surgical)||17 (71)||5 (24)|
|Null cell adenoma||15 (63)||5 (24)|
|Null cell adenoma, oncocytic variant||1 (4)||1 (5)|
|Gonadotroph adenoma||2 (8)||6 (29)|
|Plurihormonal adenoma||3 (12)||9 (43)|
|Silent corticotroph adenoma||1 (4)||—|
On the basis of postoperative MRI, patients were distributed into the following 2 groups.
Study group (progressive, regrowing tumors)
This group was defined with regard to the possibility of future treatment with TMZ. Patients were included in this group if a postoperative tumor remnant was detected on imaging studies (median time to detection of residual tumor, 6 months; range, 3-24 months) that changed in size and configuration over time and necessitated further interventions, as verified by at least 2 sequential MRI examinations.
This group included only postoperative patients who were tumor-free: If a postoperative residual tumor definitely was excluded by means of MRI, then patients were categorized as tumor-free. The minimum follow-up was 5 years (median, 9.3 years).
Clinical, endocrinologic, and radiologic data were retrieved from the hospital archives. In 5 patients, a transcranial approach was used at first surgery. For all other patients, the transsphenoidal route was used. All tumors fulfilled the criteria of macroadenomas (>1 cm in greatest dimension).
Tumor invasiveness was assessed according to surgical reports: Adenomas were considered invasive if intraoperative signs of dural, bone, and/or cavernous sinus invasion were noted. Preoperative MRI data also were used to evaluate invasiveness only if the surgical report was inconclusive.
The hormone expression profile of all pituitary adenomas was reviewed according to the current World Health Organization (WHO) classification 200430:
- 1)Null cell adenoma: tumor tissue that is immunonegative for anterior pituitary hormones or that contains only scattered cells or groups of cells that are immunopositive for 1 or more anterior pituitary hormone(s); also, there is an oncocytic variant, which is similar to ‘null cell adenoma’ but composed of acidophilic cells that have abundant cytoplasm (oncocytes) in hematoxylin and eosin stain;
- 2)Gonadotroph adenoma: tumor tissue that is immunopositive for follicle-stimulating hormone (FSH) and/or luteinizing hormone (LH) (the presence of a few scattered cells that are positive for FSH or LH is considered insufficient evidence of hormone expression);
- 3)Plurihormonal adenoma: tumor tissue that is immunopositive for more than 1 anterior pituitary hormone (the presence of a few scattered cells that are positive for any 1 hormone is considered insufficient evidence for plurihormonality);
- 4)Silent corticotroph adenoma: no clinical or biochemical evidence of hormone excess and tumor tissue that is immunopositive for adrenocorticotropic hormone (ACTH) (the presence of a few scattered cells that are positive for ACTH is considered insufficient evidence of hormone expression); and
- 5)Unclassified: all pituitary adenomas that did not meet the above criteria.
For the evaluation of hormone expression, sections of formalin-fixed, paraffin-embedded specimens of pituitary adenomas were incubated with primary antibody to pituitary hormones: antimouse ACTH (1:75 dilution; DAKO, Glostrup, Denmark), antihuman growth hormone (1:1000 dilution; DAKO), antimouse LH (1:1 dilution; BioGenex, San Ramon, Calif), antimouse thyroid-stimulating hormone (1:50 dilution; BioGenex), antimouse FSH (dilution; 1:200; BioGenex), antihuman prolactin (1:2000 dilution; DAKO), and antihuman α subunit of pituitary hormone (1:400 dilution; Immunotech, Marseille, France).
MGMT Immunoexpression and Analysis
Sections were cut at a thickness of 3 μm to 5 μm from routinely formalin-fixed, paraffin-embedded tumor tissue blocks. Slides underwent heat-induced antigen retrieval for 20 minutes in citrate buffer, pH 6.0, in a steamer. Sections were incubated overnight with a mouse monoclonal anti-MGMT antibody (clone MT3.1; Neo Markers, Fremont, Calif) at 1:50 dilution. Negative controls were performed by omission of the primary antibody. For a positive control of the staining reaction, a tumor section from a glioblastoma with previously verified MGMT expression was stained in parallel to all cases in our study cohort.
MGMT immunostaining was evaluated jointly by 3 observers (A.W., G.W., M.P.) on a multiheaded microscope. The amount of anti-MGMT-immunostained tumor cells was scored semiquantitatively by visual impression as follows (Fig. 1A-C): 1) none, few endothelial cells immunostained and <10% of tumor cells immunostained; 2) few, 10% to 50% of tumor cells immunostained; and 3) many, >50% of tumor cells immunostained.
According to the hypothesis that low MGMT expression predicts responsiveness to TMZ treatment,23 tumor species with none (<10%) and few (10%-50%) MGMT-positive tumor cells were subsumed into a low MGMT expression group. High MGMT expression was defined as the group in which >50% of tumor cells were positive for MGMT.
For statistical analyses, the SPSS software package (version 15.0; SPSS Inc., Chicago, Ill) was used. The chi-square test was used to compare qualitative variables between subgroups. Kaplan-Meier plots with log-rank tests were used to compare the MGMT subgroups with regard to the percentage of patients who underwent reoperation. A P value <.05 was considered significant.
MGMT immunoexpression yielded interpretable results in tumor tissue specimens from all patients.
Total Study Cohort
MGMT expression was determined in primary surgery specimens of all patients (n = 45) (Table 3). Low MGMT expression was observed in 17 patients (38%), and the remaining 28 patients (62%) had high MGMT expression.
|No. of Patients (%)|
|Variable||Overall||Study Group (Progressive Regrowing)||Control Group (Tumor Free)|
|All patients||45 (100)||24 (53)||21 (47)|
|Low MGMT expression, % immunostained adenoma cells|
|<10||7 (16)||6 (25)||1 (5)|
|10-50||10 (22)||6 (25)||4 (19)|
|Total||17 (38)||12 (50)||5 (24)|
|High MGMT expression, % immunostained adenoma cells|
|>50%||28 (62)||12 (50)||16 (76)|
|Low MGMT expression, % immunostained adenoma cells|
|High MGMT expression, % immunostained adenoma cells|
|Changes in MGMT expression in first versus last surgery|
|High to low||—||5 (25)||—|
|Low to high||—||2 (10)||—|
Study Group (Progressive, Regrowing) Versus Control Group (Tumor-free)
We compared MGMT expression in patients who had progressive, regrowing tumors (n = 24) with MGMT expression in tumors from patients who remained tumor-free after the first operation (n = 21): In the group with progressive, regrowing tumors, low MGMT expression was observed more frequently compared with the tumor-free group (50% vs 24%). Conversely, high MGMT expression was more common in the tumor-free group than in the group with progressive, regrowing tumors (76% vs 50%). However, these differences did not reach statistical significance (P = .071).
MGMT expression was assessed in tissue samples from repeat surgery in the group with progressive, regrowing tumors. Four patients had to be excluded for the following reasons: 1) The poor condition of 1 elderly patient did not allow for reoperation, and the patient died of aggressive tumor progression; 2) in 2 patients, reoperation currently is scheduled; and 3) in 1 patient, no adequate vital tumor tissue was available for the assessment of MGMT expression after an apoplectic event before repeat surgery. Thus, 20 patients were included in the subsequent analysis.
We observed that MGMT expression in tissue specimens from repeat surgery was comparable to the level of MGMT expression in the primary tumor (50% high MGMT expression, 50% low MGMT expression). However, the proportion of patients who had tumors that displayed ‘no MGMT expression’ (<10% MGMT-positive tumor cells) increased compared with the proportion of patients who had tumors with no MGMT expression at primary surgery (P value nonsignificant.).
To assess changes in MGMT expression within a regrowing tumor over time, a pairwise analysis of tissue samples from primary surgery and last surgery was conducted. In the majority of patients (13 patients, 65%), MGMT expression remained unchanged. However, in 5 patients (25%), changes from high to low MGMT expression and, in 2 patients (10%), changes from low to high MGMT expression were detected.
MGMT Expression and Time to Reoperation
Finally, the time to second operation with respect to low versus high MGMT expression in all patients was studied (Fig. 2). We observed a shorter interval to second surgery in patients who had low MGMT expression (6.2 years for patients with low MGMT expression vs 10.7 years for patients with high MGMT expression; log-rank P = .057).
Because of their frequently large size at the time of diagnosis and their frequently invasive growth into perisellar structures, the complete surgical removal of NFPAs often is not possible. Regrowth has been reported in 38% to 95% of patients.4, 6, 12, 14, 17, 18 In these patients, current standard treatments include: 1) reoperation for patients with visual deterioration for opticochiasmatic decompression; 2) stereotactic radiosurgery, which is effective in patients with small adenoma remnants distant to the opticochiasmatic system31-37; and 3) fractionated radiotherapy, which is reserved for the rare cases that are considered inoperable because of their fibrous consistency and unsuitable for radiosurgery because of large tumor size, location, and extension.14, 15, 31-33, 38-40 Although tumor control is achieved in up to 93% of patients,40 long-term side effects of fractionated radiotherapy are common and include hypopituitarism, cerebrovascular disease, cognitive impairment, optic neuropathy, and the formation of secondary tumors as well as increased mortality.15, 31, 33, 38, 39, 41, 42
Unlike prolactinomas, growth hormone-producing adenomas, and thyrotropin-producing adenomas, to our knowledge there currently is no routinely applicable pharmacologic therapy available for NFPAs.17, 43-46 In addition, for the rare cases of NFPA that remain progressive despite multimodality treatment, the establishment of alternative clinically applicable therapeutic options is desirable.
TMZ, an alkylating agent, rapidly penetrates the blood-brain barrier after oral administration and generally is well tolerated. Currently, TMZ is an essential component of the standard treatment regimen for glioblastoma multiforme, and it has been determined that TMZ significantly improves the median survival of patients harboring these tumors.47-53 In endocrine tumor entities, patients with pheochromocytoma, pancreatic endocrine neoplasms, and carcinoid tumors had biochemical and radiologic improvement after oral treatment with TMZ.19
The potential advantages of alkylating drugs such as TMZ are that they are not cell cycle-specific, they can inhibit all stages of tumor cell growth and, thus, they are suitable for slow-growing tumors.24 Consequently, TMZ was examined as a potential treatment alternative for aggressive, prolactin-producing, and nonfunctioning pituitary carcinomas and adenomas.
In 2004, Fadul et al described 2 patients with progressive pituitary carcinomas despite multimodality treatment who responded to TMZ with persistent clinical improvement.20 In the same year, Zhu et al treated a prolactin-producing carcinoma with TMZ, resulting in clinical improvement, tumor shrinkage, and a decrease in prolactin levels.27 Two later case studies confirmed the responsiveness of pituitary carcinomas to TMZ.21, 24
After the initial success in carcinomas, TMZ treatment consequently was expanded to patients with aggressive pituitary adenomas. In 2006, Kovacs et al and Syro et al treated a patient who had a large, invasive, prolactin-producing pituitary neoplasm with TMZ after several unsuccessful surgical procedures.22, 23, 26 Although clinical improvement and near normalization of prolactin levels was observed, MRI studies revealed intratumoral hemorrhage, necrosis, and tumor shrinkage. After treatment with TMZ, the residual tumor was easily resectable because of its now friable and soft consistency. Neuropathologic examination confirmed necrosis and hemorrhage and revealed fibrosis and neuronal transformation. Finally, mitotic activity and the Ki-67 labeling index decreased after TMZ treatment. Neff et al reported another patient with an invasive prolactinoma that was resistant to dopamine agonists and was treated successfully with TMZ.25 Taken together, these results indicate that TMZ has a positive effect on tumor shrinkage and hormone activity in prolactin-producing and nonfunctioning pituitary adenomas and carcinomas that are unresponsive to standard treatment after a median treatment duration of 11 months (range, 7-26 months).
The cytotoxic effect of TMZ is caused mainly by methylation of the O6 position of guanine in DNA. The DNA repair enzyme MGMT plays a crucial role in the removal of alkylating lesions induced by TMZ. Therefore, it is believed that high expression of MGMT induces resistance to TMZ. Epigenetic silencing of the MGMT gene by methylation of the MGMT promoter (with subsequent low MGMT expression) turned out to be the strongest predictive marker of a favorable outcome in patients with glioblastomas treated with TMZ.53-56
On the basis of observations in 2 clinical cases, Kovacs et al recently hypothesized that MGMT immunoexpression also may predict the responsiveness of pituitary tumors to treatment with TMZ.23 In their study they observed that, in 1 patient, a prolactinoma that was responsive to TMZ was immunonegative for MGMT; whereas, in the other patient, who had a tumor resistant to TMZ (a silent subtype II corticotroph adenoma), high MGMT expression was detected. In light of these findings, we designed the current study to assess MGMT immunoexpression in a series of patients who possibly were suitable candidates for TMZ treatment, ie, patients who had progressive, regrowing NFPAs after surgical treatment.
For routine clinical purposes, MGMT immunohistochemistry has the advantages of being available in most laboratories, easier, and more quickly accessible, and less expensive than methylation-specific polymerase chain reaction (MSP). In high-grade gliomas, the clinical usefulness of immunohistochemistry for the assessment of MGMT expression as an alternative to MSP has been controversially discussed.57-59 In gliomas, non-neoplastic cell elements entrapped within the diffusely infiltrating tumor tissue (eg, reactive astrocytes, microglial cells) impede the reliability of immunohistochemical analysis of MGMT expression. Conversely, pituitary adenomas have homogenous tumor cell cytology and lack significant non-neoplastic cell populations in the tumor tissue, providing favorable features for the reliable assessment of MGMT expression. Indeed, in most cases in our present series, MGMT expression could be assigned clearly to the 3-tiered system of none/few/many immunostained tumor cells, suggesting that the assessment of MGMT expression by immunohistochemistry can be considered a reliable technique for evaluating pituitary adenomas. Although a previous study detected MGMT gene promoter methylation in 23% of pituitary adenomas, we observed low MGMT protein expression in 38% of cases.60 Further studies are needed to assess the correlation of MGMT promoter methylation status with MGMT protein expression in pituitary adenomas.
We observed low MGMT expression more frequently in the group with progressive, regrowing NFPAs compared with the tumor-free control group (50% vs 24%, respectively; P = .071). It is noteworthy that samples from repeat surgery had comparable low MGMT expression in 50% of patients. Thus, based on results from other tumor entities and on case reports of pituitary tumors treated with TMZ, half of patients who have progressive, regrowing NFPAs are potential candidates for TMZ treatment even after they undergo multiple surgical procedures.
It has been observed that loss of MGMT expression occurs during tumorigenesis and results in impaired capability of DNA repair.61-65 In contrast, it has been demonstrated that MGMT overexpression prevents carcinogenesis in several mouse models.61 Previous studies reported a significant decrease in MGMT expression as astrocytomas progressed from WHO grade 2 to glioblastoma.66, 67 Accordingly, negative MGMT expression was considered a reliable marker for malignant transformation of low-grade astrocytomas.68 In pituitary adenomas, promoter methylation69-72 with subsequent MGMT inactivation60 is a frequent event and is believed to play a key role in adenoma development and progression.
In this sense, the higher proportion of patients who had tumors that lacked MGMT expression (<10% MGMT-immunostained cells) in tumor specimens from the last surgery could be interpreted as a change of biologic behavior into a more dedifferentiated adenoma. Furthermore, a reduction in MGMT expression between the first and last surgeries was reported in 25% of patients in our study and may reflect a step in adenoma progression. Two patients, however, displayed an increase in MGMT expression, and 1 of those was the only patient in our current series who was assessed after radiotherapy. Therefore, the assessment of MGMT expression status from a tissue sample of the current tumor maybe of interest in the future before initiating TMZ treatment.
In untreated, low-grade astrocytomas, another study demonstrated a shorter progression-free survival in patients who had methylated MGMT promoter (with subsequent reduction of MGMT expression) compared with patients who had unmethylated MGMT promoter.73, 74 Likewise, a negative trend in survival was reported recently for patients who had untreated glioblastomas with methylated MGMT promoter.75
Similarly, we observed a shorter time to reoperation in our NFPA patients who had low MGMT expression (6.2 years vs 10.7 years; log-rank P = .057) using survival analysis. Furthermore, we observed a higher proportion of adenomas that had low MGMT expression in the group with progressive, regrowing tumors. Because MGMT expression was not correlated significantly with the routinely used prognostic marker MIB-1 labeling index76 in our series, low MGMT expression may serve as an additional, independent prognostic factor before more aggressive therapy is initiated.
In conclusion, TMZ has been proposed as a possible alternative in the rare cases of progressive and regrowing NFPAs that are refractory to standard treatment regimens. According to our results, a positive response to TMZ may be expected in 50% of patients because of low MGMT immunoexpression. Because MGMT expression can change with regrowth and after radiotherapy, the reassessment of recurrent tumors may be of interest before initiating treatment with TMZ. However, further clinical studies are needed to clarify the relation between MGMT expression and responsiveness to TMZ in aggressive NFPAs.
The assessment of MGMT expression in NFPAs may serve as an additional prognostic factor with respect to tumor dedifferentiation and progression. MGMT expression, therefore, may influence decisions regarding the postoperative management of individual patients, such as the frequency of radiographic surveillance and the consideration of further therapeutic interventions.
Conflict of Interest Disclosures
The authors made no disclosures.
- 20Pituitary carcinomas respond to temozolomide [abstract]. Neuro Oncol. 2004; 6: 374., , , et al.
- 27Temodar: novel treatment for pituitary carcinoma [abstract]. US Endocr Soc. 2007; 138: 43-45., , , , , .
- 30Pathology and Genetics of Tumors of Endocrine Organs. World Health Organization Classification of Tumours, vol 1. Lyon, France: IARC Press; 2004., , .
- 32Radiotherapy of nonfunctioning and gonadotroph adenomas. Pituitary. 2008 Feb 20. [Epub ahead of print], , .
- 41Mortality following pituitary radiotherapy. Pituitary. 2008 Jan 4. [Epub ahead of print], .
- 44Hormone levels and tumour size response to quinagolide and cabergoline in patients with prolactin-secreting and clinically non-functioning pituitary adenomas: predictive value of pituitary scintigraphy with 123I-methoxybenzamide. Clin Endocrinol (Oxf). 2000; 52: 437-445., , , et al.
- 74Promoter hypermethylation of the DNA repair gene O6-methylguanine-DNA methyltransferase is an independent predictor of shortened progression free survival in patients with low-grade diffuse astrocytomas. Brain Pathol. 2003; 13: 176-184., , , , , .