Population-Based Studies on Incidence, Survival Rates, and Genetic Alterations in Astrocytic and Oligodendroglial Gliomas (original) (raw)

Abstract

Published data on prognostic and predictive factors in patients with gliomas are largely based on clinical trials and hospital-based studies. This review summarizes data on incidence rates, survival, and genetic alterations from population-based studies of astrocytic and oligodendrogliomas that were carried out in the Canton of Zurich, Switzerland (approximately 1.16 million inhabitants). A total of 987 cases were diagnosed between 1980 and 1994 and patients were followed up at least until 1999. While survival rates for pilocytic astrocytomas were excellent (96% at 10 years), the prognosis of diffusely infiltrating gliomas was poorer, with median survival times (MST) of 5.6 years for low-grade astrocytoma WHO grade II, 1.6 years for anaplastic astrocytoma grade III, and 0.4 years for glioblastoma. For oligodendrogliomas the MST was 11.6 years for grade II and 3.5 years for grade III. TP53 mutations were most frequent in gemistocytic astrocytomas (88%), followed by fibrillary astrocytomas (53%) and oligoastrocytomas (44%), but infrequent (13%) in oligodendrogliomas. LOH 1p/19q typically occurred in tumors without TP53 mutations and were most frequent in oligodendrogliomas (69%), followed by oligoastrocytomas (45%), but were rare in fibrillary astrocytomas (7%) and absent in gemistocytic astrocytomas. Glioblastomas were most frequent (3.55 cases per 100,000 persons per year) adjusted to the European Standard Population, amounting to 69% of total incident cases. Observed survival rates were 42.4% at 6 months, 17.7% at one year, and 3.3% at 2 years. For all age groups, survival was inversely correlated with age, ranging from an MST of 8.8 months (<50 years) to 1.6 months (>80 years). In glioblastomas, LOH 10q was the most frequent genetic alteration (69%), followed by EGFR amplification (34%), TP53 mutations (31%), p16INK4a deletion (31%), and PTEN mutations (24%). LOH 10q occurred in association with any of the other genetic alterations, and was the only alteration associated with shorter survival of glioblastoma patients. Primary (de novo) glioblastomas prevailed (95%), while secondary glioblastomas that progressed from low-grade or anaplastic gliomas were rare (5%). Secondary glioblastomas were characterized by frequent LOH 10q (63%) and TP53 mutations (65%). Of the TP53 mutations in secondary glioblastomas, 57% were in hot-spot codons 248 and 273, while in primary glioblastomas, mutations were more evenly distributed. G:C→A:T mutations at CpG sites were more frequent in secondary than primary glioblastomas, suggesting that the acquisition of TP53 mutations in these glioblastoma subtypes may occur through different mechanisms.

Astrocytoma, EGFR amplification, Glioblastoma, Oligodendroglioma, LOH 10, LOH 1p, LOH 19q, Population-based study, PTEN mutation, P16INK4a deletion, TP53 mutation

Introduction

Gliomas of astrocytic, oligodendroglial and ependymal origin account for more than 70% of all brain tumors, but current knowledge on the survival of glioma patients and on factors that are predictive of outcome is based largely on clinical trials on patients with malignant gliomas (anaplastic astrocytoma and glioblastoma). However, clinical trials have a strong bias towards the recruitment of patients with better prognosis, i.e. high preoperative Karnofsky performance score and younger age (1, 2). Therapeutic trials have less frequently addressed the outcome of patients with low-grade diffuse astrocytoma or oligodendroglioma, since in many centers these patients are not subjected to adjuvant radio- and chemotherapy. Similarly, studies on genetic alterations and how they influence response to therapy and survival are usually based on small numbers of patients, often contradictory and difficult to validate. In recent years, it has been established that primary (de novo) glioblastomas and secondary glioblastomas derived from low-grade or anaplastic gliomas develop through different genetic pathways (3-5). However, these studies, too, were based on small cohorts of selected cases and the relative frequencies of these glioblastoma subtypes remained unclear. In order to overcome these problems, we conducted a large, population-based study on close to 1,000 patients (6-8).

Patient Population and Review of Histology

We carried out a population-based study of patients who were diagnosed with an astrocytic or oligodendroglial tumor (total 987 cases) in the Canton of Zurich, Switzerland (population approximately 1.16 million) during the 15-year period from 1980 to 1994 (6, 8). Patients were followed up until death or at least until the end of 1999. The mean patient follow-up periods were 12 years for pilocytic astrocytoma (6), 90.2 ± 57.8 months for low-grade diffuse glioma (7), 37.6 ± 45.2 months for anaplastic glioma, and 7.2 ± 7.6 months for glioblastoma (8). Of 987 cases, 80% were histologically diagnosed after surgical intervention (611 cases, 62%) or at autopsy (180 cases, 18%); the remaining 20% were diagnosed by CT or MRI without histopathologic verification. Histologic diagnoses were reevaluated according to the new WHO classification of tumors of the nervous system (9, 10).

Incidence Rates

Incidence rates for histologic subtypes of CNS tumors are not readily available, since most cancer registries give combined data for all brain tumors. More detailed information is collected in certain regions; in the United States, registration of all brain neoplasms, including benign lesions, in the SEER database has recently become mandatory (11, 12).

Incident cases included in this population-based study are summarized in Table 1. With 3.55 new cases per 100,000 persons per year, adjusted to the European Standard Population, the glioblastoma is the most frequent histologic type, and accounts for 69% of all incident cases of astrocytic and oligodendroglial tumors. The incidence rate of glioblastomas in the USA, adjusted to the US Standard Population, is 29.6 new cases per million population per year (11). These incident cases do not include secondary glioblastomas that progressed from low-grade or anaplastic gliomas, since only the first diagnosis is considered as an incident case. Incidence rates of all astrocytic and oligodendrogliomas combined, adjusted to the European and US Standard Populations were 5.27 and 5.17 per 100,000 persons per year, respectively (Table 1).

TABLE 1.

Incident Cases of Astrocytic and Oligodendroglial Tumors in Population-Based Study in the Canton of Zurich, Switzerland

TABLE 1.

Incident Cases of Astrocytic and Oligodendroglial Tumors in Population-Based Study in the Canton of Zurich, Switzerland

Age and Sex Distribution

For pilocytic astrocytomas there was a clear correlation between the age of patients and the location of tumors (6): the cerebellum was the most frequent tumor site in children (67%); supratentorial examples were not observed in patients younger than 12 years of age. There was no significant difference between cerebellar and supratentorial locations in patients older than 35 years of age (6). Low-grade and anaplastic gliomas developed in middle-aged patients (Table 1).

Among low-grade astrocytomas, fibrillary astrocytomas were diagnosed at a mean age of 39 years, while the gemistocytic variant developed in older patients (mean, 50 years; p = 0.0072) (7).

The mean age of all glioblastoma patients (including secondary glioblastomas) was 61.3 ± 14.0 years. Their age distribution at the population level (including secondary glioblastomas) was as follows: 6.9% < 39 years, 12.5% 40 to 49 years, 21.1% 50 to 59 years, 29.9% 60 to 69 years, 22.1% 70 to 79 years, and 7.6% > 80 years. It is notable that the age of glioblastoma patients is significantly higher than that of patients recruited in clinical trials (1, 2). Older patients are often not eligible in the setting of a therapeutic trial due to a low Karnofsky performance score, and they are less likely to be treated by surgery and radiotherapy. Most histologic types of astrocytic and oligodendroglial gliomas developed more frequently in males than in females, with an overall male female ratio of 1.26 (Table 1).

Treatment

All the pilocytic astrocytoma patients underwent microsurgery, which aimed at maximum tumor removal. Of the 55 patients, only 7 (13%) underwent postoperative radiotherapy. Irrespective of radiotherapy, the prognosis of patients with pilocytic astrocytoma was excellent (6). Most patients (116/122, 95%) with low-grade gliomas also underwent surgical intervention. The management policy of treatment for low-grade gliomas at the University of Zurich at that time was surgical intervention aiming at maximum tumor removal without adjuvant radio-chemotherapy. Radiotherapy was standard procedure for anaplastic astrocytomas and anaplastic oligodendrogliomas (WHO grade III) and for glioblastomas (WHO grade IV). For the latter, it typically consisted of 2 Gy fractions and a total dose of 60 Gy.

We were surprised to see that at the population level, only 54% of glioblastoma patients underwent surgical intervention, either partial or complete resection (8). The mean age of these patients was significantly younger (56.1 years) than those who did not receive an operation (67.5 years; p < 0.0001) and their survival times were longer (median, 7.9 months) than those of patients without surgery (2.5 months; p < 0.001). Similarly, the mean age of patients who received radiotherapy (54.6 years) was significantly lower than that of patients who did not receive radiotherapy (68.4 years; p < 0.0001) and their survival time was longer (median, 10 months vs. 2 months; p < 0.001) (8). However, these differences probably reflect the better prognosis of younger patients more than the efficacy of treatment.

Histologic Types and Survival

Except for patients with pilocytic astrocytoma, who had excellent outcome irrespective of radiotherapy (6), survival of patients with an astrocytic tumor is still very poor. The median survival time (MST) of patients with a low-grade astrocytoma was 5.6 years, with anaplastic astrocytoma 1.6 years and with glioblastoma only 4.9 months (Fig. 1). The observed survival rates of glioblastoma patients at the population level were 42.4% at 6 months, 17.7% at one year, 3.3% at 2 years, and 1.2% at 3 years (8). This is consistent with population-based data from Canada, which showed that only 15 out of 689 glioblastoma patients (2.2%) diagnosed during 1975-1991 survived 3 years or longer (13).

Survival of patients with pilocytic astrocytoma (WHO grade I), low-grade astrocytoma (WHO grade II), anaplastic astrocytoma (WHO grade III), and glioblastoma (WHO grade IV). Except for patients with pilocytic astrocytoma, who have excellent survival, outcome of other diffuse astrocytomas (in particular glioblastomas) is poor (6-8).

FIGURE 1.

For low-grade astrocytomas, survival of patients with gemistocytic astrocytoma (MST = 3.8 years; 16% survival at 5 years) was significantly shorter than that of fibrillary astrocytoma (MST = 5.9 years; 31% survival at 10 years; p = 0.007) (7). The survival of patients with gemistocytic astrocytoma was longer than that of anaplastic astrocytoma, but the difference was not significant (p = 0.2). This population-based finding is consistent with previous observations in several hospital-based studies, which indicate a tendency towards rapid recurrence and malignant progression in patients with gemistocytic astrocytoma (9, 14). It may, therefore, be justified to assign the grade III to the gemistocytic variant in future editions of the WHO classification. The biologic basis of the poor outcome in patients with gemistocytic astrocytoma has not been elucidated; gemistocytes themselves show very low proliferative activity (14, 15), although they are truly neoplastic cells, typically with a TP53 mutation (16, 17).

The presence of oligodendroglial components in low-grade gliomas is associated with longer survival: oligodendroglioma patients (MST = 11.6 years; 51% at 10 years) survived longer than patients with oligoastrocytoma (MST = 6.6 years; 49% at 10 years) or fibrillary astrocytoma (MST = 5.9 years; 31% at 10 years) (7). For oligodendrogliomas the MST was 11.6 years for grade II (7) and 3.5 years for grade III (unpublished data).

Age and Survival

Age of patients is a significant predictive factor of survival among patients with low-grade glioma or glioblastoma at the population level (7, 8). In patients with glioblastoma, older age was significantly associated with shorter survival at the population level in both univariate (Fig. 2) and multivariate analyses (Table 2). Older patients are less frequently treated by surgical intervention and radiotherapy, and those who are treated show better survival than those who are not (see above) (8). Thus, the poor survival of many older patients could be due to lack of treatment. However, this is not likely; even among treated patients older age remained a significant predictor for poorer survival (unpublished data).

TABLE 2.

Multivariate Analysis for the Effect of Age, Sex, and Genetic Alterations on Survival of Glioblastoma Patients

TABLE 2.

Multivariate Analysis for the Effect of Age, Sex, and Genetic Alterations on Survival of Glioblastoma Patients

Kaplan-Meier curves showing that younger age of patients with glioblastoma is a significant predictive factor for longer survival. On the basis of these data, formulae were established to calculate the mean and median survival time from the date of glioblastoma diagnosis (upper right). Modified from Ohgaki et al (8).

FIGURE 2.

Recurrence and Progression

Five of 55 patients (9%) with pilocytic astrocytoma underwent a second operation because of recurrence 1 to 12 years after the first intervention. In all cases, the diagnosis was again of a pilocytic astrocytoma, and no case progressed from pilocytic astrocytoma (WHO grade I) to anaplastic pilocytic astrocytoma (WHO grade III) (8).

During the study period, at least one recurrence was observed in 43% patients with low-grade glioma (low-grade astrocytoma, oligoastrocytoma, and oligodendroglioma), and in 68% of these cases this was associated with progression to a more malignant histologic grade. At least one recurrence was observed in 42% patients with anaplastic glioma (anaplastic astrocytoma, anaplastic oligoastrocytoma, and anaplastic oligodendroglioma), and in 66% of these cases this was associated with progression to glioblastoma. The mean time from progression from low-grade astrocytoma or oligoastrocytoma WHO grade II) to anaplastic astrocytoma or anaplastic oligoastrocytoma (WHO grade III) was 5.9 + 3.0 years, from anaplastic astrocytoma or anaplastic oligoastrocytomas to glioblastoma (WHO grade IV) 1.4 + 1.0 years, and that from low-grade astrocytoma or oligoastrocytoma to glioblastoma 5.3 + 5.1 years. The mean time till progression from oligodendroglioma to anaplastic oligodendroglioma was 6.6 + 4.2 years.

Genetic Alterations and Their Predictive Value in Low-Grade Gliomas

TP53 mutations were most frequent in gemistocytic astrocytomas (88%), followed by fibrillary astrocytomas and oligoastrocytomas, but infrequent (13%) in oligodendrogliomas (7). Conversely, LOH 1p/19q was frequent in oligodendrogliomas (69%), less frequent in oligoastrocytomas (44%), rare in fibrillary astrocytomas and absent in gemistocytic astrocytomas (Fig. 3) (7). LOH 1p + 19q occurred in 19/35 (54%) of oligodendrogliomas. Except for 2 cases, TP53 mutation and LOH 19q were mutually exclusive in low-grade gliomas and there was not a single case with both TP53 mutation and LOH 1p (7). These findings indicate that LOH 1p/19q and TP53 mutations are genetic alterations that clearly distinguish 2 pathways leading to oligodendrogliomas and to low-grade astrocytomas, respectively. Although oligoastrocytomas may carry LOH 1p/19q and TP53 mutations, it is extremely rare that both LOH 1p/19q and TP53 are detected in the same tumor (7, 18).

FIGURE 3.

TP53 mutations and LOH 1p/19q in low-grade diffuse gliomas. TP53 mutations are most frequent in gemistocytic astrocytomas, followed by fibrillary astrocytomas, and lowest in oligodendrogliomas. LOH 1p/19q is most frequent in oligodendrogliomas, but rare in low-grade diffuse astrocytomas. Oligoastrocytomas show a genetic status between those of low-grade diffuse astrocytomas and oligodendrogliomas (7).

Multivariate analysis adjusting for histologic type, age, sex, and genetic alterations revealed that the presence of neither TP53 mutations nor LOH 1p/19q was significantly predictive of survival in low-grade gliomas (7). These findings suggest that although TP53 mutations and LOH 1p/19q may be key events in the development of low-grade gliomas, their presence or absence alone does not predict clinical outcome. Combining histologic diagnosis and genetic alterations with gene expression patterns may yield data of higher predictive value (19).

Primary and Secondary Glioblastomas

Tumors were considered primary (de novo) if a glioblastoma diagnosis was made at the first biopsy, without clinical or histopathologic evidence of a less malignant precursor lesion. The diagnosis of secondary glioblastoma was made only in cases with histopathologic evidence of preceding low-grade or anaplastic glioma. Using these criteria, secondary glioblastomas were rare at the population level, amounting to only 5% of all glioblastomas (8). This is consistent with the finding by Dropcho et al at the University of Alabama (20), who observed that 19 of 392 patients (5%) with glioblastomas had histologically proven prior low-grade gliomas. However, in population-based data in the USA, low-grade and anaplastic astrocytomas and mixed gliomas comprised approximately 25% of incident glioblastoma cases (11, 21). Similarly, in the present population-based study, the incidence rate of low-grade and anaplastic gliomas is approximately 2 to 3 times higher than that of secondary glioblastoma (7). The higher frequency of precursor lesions may be explained at least in part by the fact that a significant fraction of patients with low-grade or anaplastic astrocytoma die before progression to glioblastoma occurs. However, some cases with very rapid progression from low-grade or anaplastic astrocytoma may have been misclassified as primary glioblastoma. Even taking into account this possibility, at the population level secondary glioblastomas constitute a rare disease compared to primary glioblastoma.

There was a striking difference in the age distribution (Fig. 4). The mean age of primary glioblastoma patients was 62 years, while secondary glioblastomas developed in younger patients (45 years). The median survival of secondary glioblastoma patients was 7.8 months, significantly longer than that of primary glioblastoma patients (4.7 months; p = 0.003) (Fig. 5). However, this was largely due to the younger age of secondary glioblastoma patients; after age adjustment, there was no significant difference in survival.

Secondary glioblastomas develop significantly in younger patients than primary glioblastomas (p < 0.0001). TP53 mutations occur in patients of any age group, while EGFR amplification occurs in older patients. Note that there is no single case of glioblastoma with EGFR amplification that developed in a patient younger than 35 years of age. Modified from Ohgaki et al (8).

FIGURE 4.

Secondary glioblastomas develop significantly in younger patients than primary glioblastomas (p < 0.0001). TP53 mutations occur in patients of any age group, while EGFR amplification occurs in older patients. Note that there is no single case of glioblastoma with EGFR amplification that developed in a patient younger than 35 years of age. Modified from Ohgaki et al (8).

Univariate analysis shows that survival of secondary glioblastoma patients is significantly longer than that of primary glioblastoma patients (logrank test; p = 0.003). Modified from Ohgaki et al (8).

FIGURE 5.

Primary glioblastomas developed more frequently in men (M/F ratio, 1.33), while secondary glioblastomas developed more frequently in women (M/F, 0.65). This corroborates a previous observation that glioblastomas with TP53 mutations (a genetic hallmark of secondary glioblastomas) are more common in women (22). This is surprising, since in previous hospital-based studies (9, 23) and population-based studies (7, 11), the incidence of low-grade or anaplastic gliomas in males was similar to or higher than that in females. The possibility exists that in female patients, gliomas progress more frequently or more rapidly to glioblastoma.

Genetic Pathways to Glioblastomas

At the population level, the most frequent genetic alteration in glioblastomas was LOH 10q (69%), followed by EGFR amplification (34%), TP53 mutations (31%), p16INK4a homozygous deletion (31%), and PTEN mutations (24%) (8). Similar frequencies of genetic alterations were observed in primary glioblastomas (Fig. 6). In secondary glioblastomas, LOH 10q and TP53 mutations were frequent (63% vs. 65%), but other genetic alterations were infrequent (4%-19%) (Fig. 6).

FIGURE 6.

Genetic pathways to primary (de novo) and secondary glioblastomas. LOH 10q is frequent in both primary and secondary glioblastomas. TP53 mutations are early and frequent genetic alterations in the pathway leading to secondary glioblastomas. Asterisks (*) indicate genetic alterations that are significantly different in frequency between primary and secondary glioblastomas. Modified from Ohgaki et al (8).

Genetic alterations that are significantly more frequent in primary than in secondary glioblastomas are EGFR amplification and PTEN mutations, whereas TP53 mutations are significantly more frequent in secondary than primary glioblastomas (Fig. 6). In the pathway leading to secondary glioblastomas, TP53 mutations are early genetic events, since they are already present in low-grade and anaplastic gliomas at similar frequencies (Fig. 6), while LOH 10q is a late genetic event. In the absence of identifiable precursor lesions, the sequence of genetic alterations in primary glioblastomas remains enigmatic.

Some patients with glioblastomas have an extended preoperative clinical history but without radiologic or histopathologic evidence of less malignant precursor lesions. Thus, the possibility exists that these glioblastomas are clinically and genetically intermediate between primary and secondary glioblastomas. However, the pattern of genetic alterations was similar in primary glioblastomas with long and short preoperative histories (unpublished results), making the existence of hybrid subtypes unlikely. Even among primary glioblastomas, some tumors grow more rapidly than others, possibly due to different combinations of genetic alterations and other factors, such as the genetic background of the individual.

Age and Genetic Alterations in Glioblastomas

Glioblastomas with a TP53 mutation were observed in younger patients (mean, 53 years; Table 3), in particular in patients younger than 35 years (Fig. 7). The mean age of patients with glioblastomas carrying a PTEN mutation and LOH 10q were older than those without these alterations (Table 3). EGFR amplification closely reflects the age distribution of primary glioblastomas; most interestingly, EGFR amplification was never observed in any glioblastoma that developed in patients below 35 years of age (Figs. 4, 7). Among patients older than 35 years, the overall frequencies of key genetic alterations in glioblastoma were similar among different age groups (Fig. 7). Thus, the poor prognosis of older patients cannot be explained by the frequency of specific genetic alterations or the combination. Although there may be as yet unidentified transformation-associated genes that are more frequently altered in glioblastomas of older patients and may affect susceptibility to therapy, it is also possible that the sum of all changes, i.e. the level of genomic instability, is more relevant.

TABLE 3.

Univariate Analysis for the Effect of Genetic Alterations on Survival and Mean Age of Glioblastoma Patients

TABLE 3.

Univariate Analysis for the Effect of Genetic Alterations on Survival and Mean Age of Glioblastoma Patients

Genetic alterations versus age of patients with all glioblastoma. Glioblastomas developed in patients younger than 35 years are genetically characterized by frequent TP53 mutations and absence of EGFR amplification, while there is no significant difference in genetic alterations between different age groups of patients older than 35 years.

FIGURE 7.

Genetic alterations versus age of patients with all glioblastoma. Glioblastomas developed in patients younger than 35 years are genetically characterized by frequent TP53 mutations and absence of EGFR amplification, while there is no significant difference in genetic alterations between different age groups of patients older than 35 years.

Predictive Value of Genetic Alterations in Glioblastomas

LOH 10 is the most frequent genetic alteration in glioblastomas and occurs in 60% to 80% of cases (24-26). Many glioblastomas appear to have lost one entire copy of chromosome 10. LOH occurs most frequently at 3 common loci (i.e. 10p14-p15, 10q23-24 [PTEN], and 10q25-pter), suggesting the presence of several tumor suppressor genes (24-27). LOH 10q was the only genetic alteration associated with shorter survival of glioblastoma patients at the population level (Tables 2 and 3) (8), in accordance with results from several previous hospital-based studies (28-30).

The PTEN gene encodes a protein that plays important roles in the regulation of cell proliferation, apoptosis, and tumor invasion (31, 32). PTEN mutations have been reported in 15% to 40% of glioblastomas (32, 33). In several hospital-based studies, PTEN mutations were not associated with prognosis of glioblastoma patients (28, 34, 35), and this was confirmed at the population level (8). Most missense mutations were located in exons 1 to 6, the region homologous to tensin, auxilin, and dual-specificity phosphatases, whereas nonsense mutations and deletions or insertions leading to stop codons and protein truncation were located more evenly throughout the gene, suggesting that cells with PTEN truncation at any site or PTEN missense mutations in the region homologous to tensin/auxilin and dual-specificity phosphatases acquire a transformed phenotype (8).

The TP53 gene plays important roles in several cellular processes including the cell cycle, response to DNA damage, apoptosis, cell differentiation, and neovascularization (36). Data on the predictive value of TP53 mutations in glioblastomas have been contradictory. While some hospital-based studies showed no association between TP53 status and outcome of glioblastoma patients (34, 37), one study showed that the presence of TP53 mutations was a favorable prognostic factor (28). At the population level, univariate analysis revealed that the presence of TP53 mutations was predictive of longer survival (8). However, age-adjusted multivariate analysis revealed no difference in survival between patients with and without TP53 mutations (8).

Epidermal growth factor receptor (EGFR) is a transmembrane receptor that binds to extracellular ligands such as EGF and transforming growth factor alpha (TGFα), and transduces a mitotic signal (38). The predictive value of EGFR amplification has been unclear. In earlier hospital-based studies (<40 cases), EGFR amplification was associated with poorer survival of glioblastoma patients (39, 40). In contrast, a meta-analysis of 7 previous studies (total 395 glioblastoma cases) did not reveal a significant predictive value of EGFR amplification (41). Several studies have suggested that EGFR amplification is predictive for certain age groups of glioblastoma patients (34, 37, 42, 43). Shinojima et al (42) reported that EGFR amplification was a significant predictor of poorer overall survival in glioblastoma patients and that the EGFR gene status was a more significant prognostic factor in younger patients (<60 years). Simmons et al (37) reported that EGFR overexpression was associated with poorer survival of glioblastoma patients younger than the median age, and that EGFR overexpression was negatively associated with survival in cases without TP53 mutation. Other studies found EGFR amplification to be a predictor of longer survival only in older glioblastoma patients (34, 43). At the population level, EGFR amplification did not affect survival of glioblastoma patients at any age (8).

p16INK4a binds to cyclin-dependent kinase 4 (CDK4) and inhibits the CDK4/cyclin D1 complex (44, 45). This complex phosphorylates the RB1 protein (pRb), thereby inducing release of the E2F transcription factor that activates genes involved in the late G1 and S phases (44, 45). In glioblastomas, disruption of the p16INK4a gene occurs through homozygous deletion or promoter methylation (9, 46). Findings regarding the predictive value of p16INK4a homozygous deletion in glioblastomas have been inconsistent. One study showed that homozygous p16INK4a deletion was a significantly unfavorable predictor for survival of glioblastoma patients (47), while another study showed that homozygous p16INK4a deletion was associated with shorter survival only in glioblastoma patients older than 50 years of age (48). At the population level, both univariate and multivariate analyses failed to show any predictive value of homozygous p16INK4a deletion (8).

Copresence of Genetic Alterations in Glioblastomas

LOH 10q is not only most the frequent genetic alteration in both primary and secondary glioblastomas, but is also frequently copresent with additional genetic alterations, i.e. with EGFR amplification (25.8%), TP53 mutations (23.7%), p16INK4a deletion (23.3%), and PTEN mutations (16.2%) (8). Other combinations of genetic alterations were less frequent; TP53 mutations showed significant inverse correlations with p16INK4a deletion and EGFR amplification (8). There were many different combinations of genetic alterations (Fig. 8), the most frequent being 2 alterations (36.1%), followed by one alteration (26.2%) and 3 alterations (24.1%), while more than 4 were rarely observed (4.8%) (Fig. 8). Taken together, LOH 10q plus at least one or 2 other genetic alterations appear to be operative in the development of the majority of glioblastomas. LOH 10q25-qter distal to the PTEN appears to be associated with acquisition of the glioblastoma phenotype (49), suggesting that a tumor suppressor gene in this region may be crucial in the development of glioblastomas. Candidate genes include DMBT1 (29, 50) and FGFR2 (29). Copresence of EGFR amplification and p16INK4a deletion was also frequent (17%), with a significant positive association in glioblastomas (8), in agreement with the findings of previous hospital-based studies based on small numbers of cases (51, 52).

Combinations of genetic alterations in glioblastomas, including secondary glioblastomas. There are many combinations of genetic alterations, but it is clear that the majority of glioblastomas (62.9%) show LOH 10q alone and LOH 10q plus one or 2 additional genetic alterations.

FIGURE 8.

TP53 Mutations as Molecular Fingerprints

Several familial cancer syndromes are associated with tumors of the nervous system, the most important ones being the Li-Fraumeni syndrome (TP53 germline mutations), neurofibromatosis 1 (NF1), neurofibromatosis 2 (NF2) and Turcot syndrome (APC and hMLH1/hPSM2). However, such syndromes resulting from mutations in high-penetrance genes are rare (9, 21). The etiology of sporadic gliomas is largely unknown. Several occupations, environmental carcinogens, and diet (_N_-nitroso compounds) have been implicated, but the only environmental factor unequivocally associated with an increased risk of brain tumors, including gliomas, is therapeutic X-irradiation. In particular, children treated with X-irradiation for acute lymphoblastic leukemia show a significantly elevated risk of developing gliomas and primitive neuroectodermal tumors (PNET), often within 10 years after therapy (21, 53, 54).

Mutations induced in transformation-associated genes by environmental carcinogens (55) may give clues to the etiology of the respective tumor type, e.g. G:C→T:A mutations in codon 249 of TP53 in hepatocellular carcinomas induced by aflatoxin B1 (56) and G:C→T:A transversions in lung carcinomas of smokers (57). Among TP53 mutations identified in astrocytic brain tumors, G:C→A:T transitions are the most frequent and are located predominantly at CpG sites (8, 58). A similar pattern was seen in colon cancer, sarcomas and lymphomas and among germline TP53 mutations (58).

At the population level, the type and distribution of TP53 mutations were different in the pathways leading to primary and secondary glioblastomas. A total of 47 mutations were detected in 40 patients with low-grade diffuse glioma and these were most frequently located in the 2 hot-spot codons 248 and 273, corresponding to 47% of all mutations. Of all TP53 mutations identified in low-grade diffuse gliomas, 64% were G:C→A:T transitions and of these, 90% were located at CpG sites (Table 4) (7). A similar pattern was seen in secondary glioblastomas (Table 4) (8). In contrast, TP53 mutations were distributed more evenly among exons 5 to 8 of the TP53 gene (only 17% at codons 248 and 273) and only 30% of TP53 mutations in primary glioblastomas were G:C→A:T transitions at CpG sites (Table 4) (8). These observations suggest that the mechanisms of acquisition of TP53 mutations in these glioblastoma subtypes may differ (8).

TABLE 4.

Nature of TP53 Mutations in Primary Glioblastomas, Secondary Glioblastomas, and Low-Grade Gliomas

Nature of TP53 Mutations in Primary Glioblastomas, Secondary Glioblastomas, and Low-Grade Gliomas

TABLE 4.

Nature of TP53 Mutations in Primary Glioblastomas, Secondary Glioblastomas, and Low-Grade Gliomas

Nature of TP53 Mutations in Primary Glioblastomas, Secondary Glioblastomas, and Low-Grade Gliomas

The best-characterized mechanism of G:C→A:T transitions at CpG sites is deamination of 5-methylcytosine at a CpG site, resulting in substitution of 5-methylcytosine by thymine. This occurs spontaneously or is factor-mediated, for example, through the action of oxygen radicals or nitric oxide produced by nitric oxide synthase under conditions of chronic inflammation (59). Since all CpG sites in exons 5 to 8 of the TP53 gene have been reported to be methylated in normal human tissues (60), factors that affect the rate of deamination may be critical for the acquisition of G:C→A:T mutations.

In addition, the involvement of promutagenic alkyl groups in the O6 position of guanine in DNA cannot be ruled out, since in low-grade astrocytomas there is a significant correlation between the presence of TP53 mutations and promoter methylation of the _O6_-methylguanine-DNA methyltransferase (MGMT) gene (61). MGMT specifically removes promutagenic alkyl groups from the O6 position of guanine in DNA (62). Lack of MGMT in histologically normal perifocal brain is frequent and is age-dependent in brain tumor patients, while lack of MGMT in normal brain is infrequent in nontumor patients (63). DNA methylation may be caused by _N_-nitroso compounds in food or tobacco (21, 64), and through nonenzymatic methylation by _S_-adenosylmethionine (SAM) (65). Kang et al reported that _O6_-methylguanine is present in leukocytes (0.7-4.6 adducts/108 guanine) and in liver (1.1-6.7 adducts/107 guanine) from healthy volunteers (66). The levels of _O6_-methylguanine in normal human brain tissue remain to be assessed.

Conclusions

Population-based analyses provide epidemiological data without the selection bias typically associated with clinical cancer registration. Based on a large number of patients diagnosed with astrocytic or oligodendroglial tumors in the Canton of Zurich during 1980-1994, the following conclusions emerged:

Patients with pilocytic astrocytoma (WHO grade I) have an excellent prognosis after surgical intervention (96% 5-year survival), suggesting that adjuvant radio- or chemotherapy is unnecessary in the management of this generally benign neoplasm.
In low-grade gliomas (WHO grade II), age and histologic type are significant predictors of clinical outcome. TP53 mutations and LOH 1p/19q are genetic hallmarks of astrocytomas and oligodendrogliomas, respectively, and these alterations are mutually exclusive. With respect to gene status and survival, oligoastrocytomas were intermediate between low-grade diffuse astrocytomas and oligodendrogliomas.
The prognosis of glioblastoma patients at the population level is even worse than generally assumed from clinical trials, with less 2% surviving more than 3 years. Through the entire age range, higher age is the strongest predictor of poor outcome. Patients older than 70 years amount to 30% of all GBM patients and are under-represented in clinical trials. Older patients are also less likely to receive surgical intervention and radiotherapy but there is little evidence to suggest that this is a major factor for their low survival rates.
Secondary glioblastomas originating from low-grade or anaplastic astrocytoma are rare, accounting for only about 5% of all glioblastomas. However, they are a distinct entity, based on their occurrence in younger patients, a higher proportion of females and a genetic pathway characterized by frequent and early TP53 mutations. When corrected for their different age distribution, there was no difference in survival after glioblastoma diagnosis.
EGFR amplification, TP53 mutations, p16INK4a homozygous deletion, and PTEN mutations are considered key genetic events in the evolution of glioblastomas, but the presence or absence of any of these changes does not affect survival.
LOH 10q is the most frequent genetic alteration in both pathways to primary and secondary glioblastomas and the only one associated with poorer survival of glioblastoma patients. LOH 10q plus any of one or 2 additional alterations are the most frequent combination, suggesting that LOH 10q plus at least one or 2 additional genetic alterations is a major player in the development of glioblastomas.
In several studies, including our population-based survey, LOH 10q has been associated with poorer survival of glioblastoma patients (8, 28-30), while PTEN mutations were not associated with prognosis (8, 28, 34, 35). This suggests that an additional, not yet unequivocally unidentified, tumor suppressor on chromosome 10q25-qter may be critically involved in the evolution of the glioblastoma phenotype. Identification and validation of such a gene could greatly advance in our understanding of the pathogenesis of glioblastomas, and in the development novel strategies for the treatment of this most malignant brain tumor.
G:C→A:T TP53 mutations at CpG sites, particularly in the hotspot codons 248 and 273, constitute an early genetic event associated with malignant transformation in the pathway to secondary glioblastoma. In primary glioblastomas they are less frequent and may, at least in part, reflect increased genomic instability during tumor progression.

References

Chemotherapy in adult high-grade glioma: A systematic review and meta-analysis of individual patient data from 12 randomised trials

Lancet

2002

;

359

1011

–

Neuro-Oncology Working Group 01 trial of nimustine plus teniposide versus nimustine plus cytarabine chemotherapy in addition to involved-field radiotherapy in the first-line treatment of malignant glioma

J Clin Oncol

2003

;

3276

–

Overexpression of the EGF receptor and p53 mutations are mutually exclusive in the evolution of primary and secondary glioblastomas

Brain Pathol

1996

;

217

–

Phenotype vs genotype in the evolution of astrocytic brain tumors

Toxicol Pathol

2000

;

164

–

Genetic basis of gliomas progression

Proc Jpn Acad

2003

;

–

et al. .

A population-based study on the incidence and survival of patients with pilocytic astrocytoma

J Neurosurg

2003

;

1170

–

et al. .

Population-based study on incidence, survival rates, and genetic alterations of low-grade astrocytomas and oligodendrogliomas

Acta Neuropathol

2004

;

108

–

et al. .

Genetic pathways to glioblastoma: A population-based study

Cancer Res

2004

;

6892

–

(eds).

WHO Classification of Tumours. Pathology and Genetics of Tumours of the Nervous System

Lyon

IARC Press

2000

et al. .

The WHO classification of tumors of the nervous system

J Neuropathol Exp Neurol

2002

;

215

–

Consensus conference on brain tumor definition for registration. November 10;2000

Neuro-oncol

2002

;

134

–

et al. .

Which glioblastoma multiforme patient will become a long-term survivor? A population-based study

Ann Neurol

1999

;

183

–

Role of gemistocytes in astrocytoma progression

Lab Invest

1997

;

277

–

Regional differences in bromodeoxyuridine uptake, expression of Ki-67 protein, and nucleolar organizer region counts in glioblastoma multiforme

Acta Neuropathol (Berl)

1994

;

586

–

Genetic evidence of the neoplastic nature of gemistocytes in astrocytomas

Acta Neuropathol

2001

;

102

422

–

p53 and PTEN gene mutations in gemistocytic astrocytomas

Acta Neuropathol

1998

;

559

–

et al. .

Molecular genetic evidence for subtypes of oligoastrocytomas

J Neuropathol Exp Neurol

1997

;

1098

–

1104

et al. .

Gene Expr profiling and subgroup identification of oligodendrogliomas

Oncogene

2004

;

6012

–

The prognostic impact of prior low-grade histology in patients with anaplastic gliomas: A case-control study

Neurology

1996

;

684

–

Epidemiology and etiology of gliomas

Acta Neuropathol

2005

;

109

–

108

et al. .

Comparative study of p53 gene and protein alterations in human astrocytic tumors

J Neuropathol Exp Neurol

1993

;

–

Clinical outcome of supratentorial astrocytoma WHO grade II

Acta Neurochir (Wien)

1998

;

140

1213

–

et al. .

Chromosome 10 deletion mapping in human gliomas: A common deletion region in 10q25

Oncogene

1995

;

2243

–

et al. .

Loss of heterozygosity in malignant gliomas involves at least three distinct regions on chromosome 10

Hum Genet

1993

;

169

–

Distinct patterns of deletion on 10p and 10q suggest involvement of multiple tumor suppressor genes in the development of astrocytic gliomas of different malignancy grades

Genes Chromosomes Cancer

1998

;

–

et al. .

Microsatellite deletion mapping on chromosome 10q and mutation analysis of MMAC1, FAS, and MXI1 in human glioblastoma multiforme

Int J Oncol

1998

;

905

–

et al. .

Impact of genotype and morphology on the prognosis of glioblastoma

J Neuropathol Exp Neurol

2002

;

321

–

et al. .

Analysis of loss of heterozygosity on chromosome 10 in patients with malignant astrocytic tumors: Correlation with patient age and survival

J Neurosurg

2001

;

651

–

et al. .

Prognostic value of loss of heterozygosity around three candidate tumor suppressor genes on chromosome 10q in astrocytomas

J Neurooncol

2002

;

107

–

PTEN: Life as a tumor suppressor

Exp Cell Res

2001

;

264

–

PTEN, a unique tumor suppressor gene

Endocr Relat Cancer

2000

;

115

–

Pten signaling in gliomas

Neuro-oncol

2002

;

196

–

211

et al. .

PTEN mutation, EGFR amplification, and outcome in patients with anaplastic astrocytoma and glioblastoma multiforme

J Natl Cancer Inst

2001

;

1246

–

et al. .

Short postoperative survival for glioblastoma patients with a dysfunctional Rb1 pathway in combination with no wild-type PTEN

Clin Cancer Res

2003

;

4151

–

The p53 gene and its role in human brain tumors

Glia

1995

;

308

–

et al. .

Analysis of complex relationships between age, p53, epidermal growth factor receptor, and survival in glioblastoma patients

Cancer Res

2001

;

1122

–

Epidermal growth factor receptor dependence in human tumors: More than just expression?

Oncologist

2002

;

Suppl

–

Amplification of epidermal growth factor receptor gene in gliomas: Histopathology and prognosis

J Neuropathol Exp Neurol

1992

;

–

Relationships between Ki-67 labelling index, amplification of the epidermal growth factor receptor gene, and prognosis in human glioblastomas

Acta Neurochir (Wien)

1992

;

117

182

–

Epidermal growth factor receptor gene amplification as a prognostic marker in glioblastoma multiforme: results of a meta-analysis

Oncol Res

2000

;

107

–

et al. .

Prognostic value of epidermal growth factor receptor in patients with glioblastoma multiforme

Cancer Res

2003

;

6962

–

et al. .

Age-dependent prognostic effects of genetic alterations in glioblastoma

Clin Cancer Res

2004

;

228

–

p16INK4a and the control of cellular proliferative life span

Carcinogenesis

1999

;

921

–

Tumor suppression by Ink4a-Arf: Progress and puzzles

Curr Opin Genet Dev

2003

;

–

et al. .

P14Arf deletion and methylation in genetic pathways to glioblastomas

Brain Pathol

2001

;

159

–

et al. .

Analysis of homozygous deletion of the p16 gene and correlation with survival in patients with glioblastoma multiforme

J Neurosurg

2002

;

815

–

et al. .

Quantitative real-time PCR does not show selective targeting of p14(ARF) but concomitant inactivation of both p16(INK4A) and p14(ARF) in 105 human primary gliomas

Oncogene

2001

;

1103

–

Acquisition of the glioblastoma phenotype during astrocytoma progression is associated with LOH on chromosome 10q25-qter

Am J Pathol

1999

;

155

387

–

et al. .

DMBT1, a regulator of mucosal homeostasis through the linking of mucosal defense and regeneration?

FEBS Lett

2003

;

540

–

Hemizygous or homozygous deletion of the chromosomal region containing the p16INK4a gene is associated with amplification of the EGF receptor gene in glioblastomas

Int J Cancer

1997

;

–

Association of EGFR gene amplification and CDKN2 (p16/MTS1) gene deletion in glioblastoma multiforme

Brain Pathol

1997

;

871

–

et al. .

High incidence of secondary brain tumours after radiotherapy and antimetabolites

Lancet

1999

;

354

–

et al. .

Second malignant neoplasms in patients treated for childhood leukemia. A population-based cohort study from the Nordic countries

The Nordic Society of Pediatric Oncology and Hematology (NOPHO). Acta Paediatr Scand

1991

;

1220

–

p53 mutation spectrum and load: The generation of hypotheses linking the exposure of endogenous or exogenous carcinogens to human cancer

Mutat Res

1999

;

428

–

Mutational hotspot in the p53 gene in human hepatocellular carcinomas

Nature

1991

;

350

427

–

On the origin of G→T transversions in lung cancer

Mutat Res

2003

;

526

–

Li-Fraumeni syndrome and TP53 germline mutations

. In: , eds.

WHO Classification of Tumours. Pathology and Genetics of Tumours of the Nervous System

Lyon

IARC Press

2000

231

Chronic infections and inflammatory processes as cancer risk factors: Possible role of nitric oxide in carcinogenesis

Mutat Res

1994

;

305

253

–

Complete and tissue-independent methylation of CpG sites in the p53 gene: Implications for mutations in human cancers

Oncogene

1995

;

1493

–

Promoter hypermethylation of the DNA repair gene MGMT in astrocytomas is frequently associated with G:C→A:T mutations of the TP53 tumor suppressor gene

Carcinogenesis

2001

;

1715

–

Repair of O6-alkylguanine by alkyltransferases

Mutat Res

2000

;

462

–

100

et al. .

Lack of the DNA repair protein O6-methylguanine-DNA methyltransferase in histologically normal brain adjacent to primary human brain tumors

Proc Natl Acad Sci USA

1996

;

6941

–

Chem Biol of N-nitroso Compounds

New York

Cambridge University Press

1992

Endogenous DNA damage in humans: A review of quantitative data

Mutagenesis

2004

;

169

–

Detection of O6-methylguanine, O4-methylthymine and O4-ethylthymine in human liver and peripheral blood leukocyte DNA

Carcinogenesis

1995

;

1277

–

ICD-O International Classification of Diseases for Oncology

, 3rd Edition,

Geneva

World Health Organization

2000