March 26, 2008

Gliomas in Neurofibromatosis Type 1: A Clinicopathologic Study of 100 Patients

By Rodriguez, Fausto J Perry, Arie; Gutmann, David H; O'Neill, Brian Patrick; Leonard, Jeffrey; Bryant, Sandra; Giannini, Caterina

Abstract There are few pathologic studies of gliomas in patients with neurofibromatosis type 1. We analyzed clinical and pathologic features of gliomas from 100 neurofibromatosis type 1 patients (57 men; 43 women). The median age at tumor diagnosis was 13 years (range, 4 months to 68 years). Most tumors were typical pilocytic astrocytoma (PA) (49%) or diffusely infiltrating astrocytoma (DA) (27%) that included World Health Organization Grades II (5%), III (15%), and IV (7%); others were designated as low-grade astrocytoma, subtype indeterminate (LGSI; 17%). Two pilomyxoid astrocytomas, 1 desmoplastic infantile ganglioglioma and 1 conventional ganglioglioma, were also identified. The tumors in 24 cases arose in the optic pathways and included PA (n = 14), LGSI (n = 4), DA (n = 4), pilomyxoid astrocytoma (n = 1), and ganglioglioma (n = 1). The prognoses of the PA and LGSI gliomas overall were generally favorable; there were no survival differences between PA and LGSI groups based on site, tumor size, mitotic activity, or MIB-1 labeling index. In the combined PA and LGSI group, age younger than 10 years and gross total resection were associated with an increased overall survival rate (p = 0.047 and 0.002, respectively). Compared with the combined group (PA + LGSI), patients with DA at all sites had decreased overall and recurrence-free survival times (p

Key Words: Astrocytoma, Brain tumor, Central nervous system, Glioma, Neurofibromatosis, Pilocytic astrocytoma.


Neurofibromatosis type 1 (NF1) is an inherited tumor predisposition syndrome, and those affected with NF1 are prone to the development of both peripheral and central nervous system (CNS) tumors. The most common primary CNS tumors are gliomas that usually involve the optic pathways especially in children. In this regard, bilateral optic gliomas are nearly pathognomonic for NF1 (1). On pathologic examination, most are pilocytic astrocytomas (PAs), World Health Organization (WHO) Grade I (2, 3). However, patients with NF1 may also develop diffusely infiltrating astrocytomas (DAs) (WHO grades II-IV), particularly with an onset later in life (4). In some instances, NF1-associated astrocytomas show indistinct or overlapping features that prevent a precise histopathologic classification (5), thus precluding an accurate prediction of their biologic behavior.

Tumors involving the optic pathways in patients with NF1 have characteristic features on neuroimaging, exhibit largely indolent behavior, and may even regress without treatment (6). Therefore, pathologic specimens are generally not available for study. Even clinically progressive optic gliomas are often treated without a tissue diagnosis. Consequently, most modern reports of gliomas in patients with NF1 either do not focus on their histopathologic features (6-10) or are limited to small subsets and/or series (11, 12). Conversely, older series and reviews (13, 14) were written before the most recent WHO grading and classification schemes and predate the formulation of diagnostic criteria for NF1.

The purpose of this study was to provide a detailed clinicopathologic description of the spectrum of gliomas in patients with NF1 using pathologic material available from 2 large referral centers with dedicated NF1 clinics and to determine whether specific pathologic or clinical features predict clinical behavior.


This study was approved by the institutional review boards at both the Mayo Clinic and at the Washington University School of Medicine. A search was performed of the clinical and pathologic records of both medical centers for patients with a clinical diagnosis of NF1 and a glioma between 1950 and 2006. The patients for whom there was pathologic material available for review and who satisfied established current clinical criteria for NF1 (1) were included in this study. Clinical data were abstracted from retrospective chart review. The extent of surgery was determined predominantly from operative reports and postoperative imaging reports. All available hematoxylin and eosin-stained slides were reviewed by at least 2 neuropathologists (C.G., A.P., F.J.R.), and the tumors were classified and graded based on the current WHO classification (2). Nuclear atypia was graded in a 3-tiered scale (mild, moderate, and severe) based on chromatin and nuclear irregularities independent of degenerative changes. Cellularity was rated as low, moderate, or high. Cellularity was high when there was frequent nuclear crowding (nuclei touching each other), moderate when intercellular spaces equivalent to 1 to 2 nuclear diameters were present in most fields, and low when there was significant acellular stroma/processes in between cells. Degree of invasiveness was classified as absent, partial, or diffuse based on neurofilament protein immunostains demonstrating underlying axons, when available, in combination with hematoxylin and eosin identification of entrapped neurons and/or axons when visible. A mitotic count was performed in 10 consecutive high-power fields (HPF) in the areas of maximum proliferation. The following histologic features were evaluated as being present or absent: biphasic pattern, multinucleated cells, oligodendroglial-like cells, Rosenthal fibers, eosinophilic granular bodies, microcysts, chronic inflammation, hemosiderin-laden macrophages, calcification, glomeruloid vessels, vascular hyalinization, necrosis, perivascular pseudorosettes, fascicular pattern, and leptomeningeal extension.

Immunohistochemical stains were performed using antibodies directed against S100 protein (Dako, Carpinteria, CA; polyclonal; dilution, 1:1600), glial fibrillary acidic protein (Dako; polyclonal; 1:4000), p53 protein (Dako; clone DO7; 1:2000), neurofilament protein (Dako; clone 2F11; 1:800), synaptophysin (ICN, Costa Mesa, CA; clone SY38; 1:40), chromogranin (Chemicon, Temecula, CA; clone LK2H10; 1:500), Neu-N (Chemicon; 1:10000), and Ki-67 (MIB- 1; Dako; 1:300) in selected cases. MIB-1 labeling indices were evaluated in 51 cases at first resection by examining 20 consecutive tumor fields using the CAS200 computer imaging system (CAS 200; Bacus Laboratories, Lombard, IL). Nuclear p53 immunostaining was scored on a 4-tiered scale as follows: no staining (0); focal to less than 10% of cells (1+); 10% to 50% of cells or weak staining greater than 50% of cells (2+); and strong staining of greater than 50% of cells (3+).

Statistical Methods

Medians, interquartile ranges (IQRs), ranges, and frequencies were used to describe patient and tumor characteristics as appropriate. Only the first primary tumor for each patient with surgically obtained material that was classified as PA WHO Grade I, low-grade astrocytoma, subtype indeterminate (LGSI), or DA WHO Grades II to IV was included for survival analyses (n = 91). Overall and disease-specific survivals, as well as recurrence, were evaluated using the Kaplan-Meier method. The Fisher exact test was used to compare categorical variables. In all analyses, a p value less than 0.05 was considered statistically significant. All statistical analyses were performed using SAS software (SAS Institute Inc., Cary, NC).



The initial search revealed approximately 300 patients. From these, 100 NF1 patients were identified. From these patients, there were 121 pathologic specimens available for review that included biopsies, resections, and autopsies. There were 57 male and 43 female patients with a median age at tumor diagnosis of 13 years (range, 4 months to 68 years). A family history of NF1 was documented in 37 patients. Clinical features of NF1 included cafe au lait spots (97%), freckling (56%), neurofibromas (45%), Lisch nodules (35%), skeletal abnormalities (29%), intellectual impairment (25%), plexiform neurofibroma (23%), and dysmorphic features (16%). A total of 98 pathologic specimens of primary tumors were identified. Three patients had second primary tumors, 1 of which may have been induced by radiation (see succeeding sentences); 14 patients had recurrent tumors with available pathologic material. Surgical procedures included biopsy (36%), subtotal resection (35%), gross total resection (24%), or they were not specified (9%). Autopsies were performed in 4 cases, 2 of which did not have prior diagnostic surgical procedures.

Pathologic review of primary tumors revealed that WHO Grade I PAs were the predominant histologic tumor type (49%), followed by infiltrating astrocytomas (27%) and indeterminate astrocytomas (19%). Although the latter were easily identifiable as having an astrocytic phenotype, they did not have specific features that permitted a definitive classification as PA or DA. Most (17/20) were low-grade gliomas based on low mitotic and MIB-1 labeling indices and a predominant solid architecture and were classified as LGSI. Two cases were high-grade, and an additional spinal cord tumor was also classified as an indeterminate astrocytoma because of limited biopsy material. The remaining diagnoses were desmoplastic ganglioglioma (n = 1), ganglioglioma (n = 1), pilomyxoid astrocytoma (n = 1), and glial atypia suspicious but not diagnostic for glioma (n = 2). Pathologic Features by Group PA and Variants

The histologic features are summarized in Table 1 and illustrated in Figure 1. When the tumors involved the optic pathways, they tended to spread to the adjacent leptomeninges and surround the optic nerve in a circumferential manner. Some tumors exhibited prominent infiltration of optic nerve fascicles (Fig. 1A). Proliferative activity (median [IQR] was low as assessed by mitotic count [0 {0-1.0}] or MIB-1 labeling index [1.76 {0.4-5.3}]). The median (IQR) p53 immunopositivity score was 1 (0-2).

Perivascular pseudorosettes were identified in 3 of these tumors. However, only 1 case had all the classic features of pilomyxoid astrocytoma at initial biopsy. A second case showed features of a LGSI at initial spinal cord biopsy (Fig. 2A; see succeeding sentences), likely because of the very small size of the biopsy. The tumor, however, had diagnostic features of pilomyxoid astrocytoma at autopsy with spread to supratentorial leptomeninges and cauda equina (Fig. 2B, C).

TABLE 1. Pathologic Features of PAs and LGSI at Diagnosis

Low-Grade Astrocytoma, Subtype Indeterminate

Histologic features are summarized in Table 1, and representative features are illustrated in Figure 3. Although the tumors were easily recognized as astrocytic, classic features of PA, including biphasic growth pattern, Rosenthal fibers, eosinophilic granular bodies, and glomeruloid vessels, were either absent or present in isolation, not allowing a confident diagnosis of PA. Most tumors had less specific features, including refractile conspicuous cell processes, low to moderate cellularity, and mild nuclear atypia (Fig. 3). Median (IQR) mitotic activity and MIB-1 labeling indices were 0 (0-1) and 3.7 (1.7-6.7). The median (IQR) p53 immunopositivity score was 1 (0.5-1.5).

Pathologic material was available from a second time point in 4 cases, including the pilomyxoid astrocytoma mentioned previously. One of the tumors showed definite features of PA at resection for recurrence. In the other 2 cases, the histologic findings revealed LGSI.

Astrocytomas With Broad Cytoplasmic Processes and Prominent Nucleoli

Six tumors that were classified eimer as PA (n = 4) or LGSI (n = 2) showed distinctive morphology (Fig. 4). They were composed of medium-sized cells with plump eosinophilic, nonglassy cytoplasm, broad processes, and prominent nucleoli (Fig. 4B, C) that often stained cherry-red. Rosenthal fibers were absent. Eosinophilic granular bodies were present in 2 cases, although they were not conspicuous. The mitotic index in these tumors ranged from none to 4 per 10 HPF (median 1).

Diffusely Infiltrating Astrocytoma

The pathologic features of mese tumors are summarized in Table 2, and representative features are shown in Figure 5. Most were fibrillary astrocytomas (96%). Two tumors had a gemistocytic component, and 1 had small cell undifferentiated features. One Grade III gemistocytic astrocytoma developed a small cell undifferentiated component and progressed to glioblastoma. Four cases involved the optic pathways (see succeeding sentences) and demonstrated nuclear atypia and conspicuous mitotic activity together with a diffusely infiltrating growth, rather than expanding optic nerve septae and leptomeninges (Fig. 6). A Grade IV fibrillary astrocytoma in a 21- year-old female patient with a history of colon cancer with microsatellite instability demonstrated preserved immunoexpression of the mismatch repair enzymes MLH1, MSH2, MSH6, and PMS2 (data not shown). The median (IQR) mitotic activity, MIB-1 labeling indices, and p53-immunopositive scores were 2 (1-5), 4.5 (2.8-12.9), and 2 (1- 3), respectively.

FIGURE 1. Histologic features of pilocytic astrocytoma (PA). Cross-sectional view of a PA that expands optic nerve fascicles (A) (hematoxylin and eosin; 100 x ). Intermediate power view highlighting numerous Rosenthal fibers (B) (hematoxylin and eosin; 200 x ). Loose architecture with microcysts and bipolar astrocytes is another characteristic feature of PA (C) (hematoxylin and eosin; 400 x ). Areas with increased cellularity may be present in some PAs of the optic nerve (D) (hematoxylin and eosin; 400 x ).

FIGURE 2. Low-grade astrocytoma, subtype indeterminate, in a small biopsy. High-power view of a small spinal cord biopsy demonstrates neoplastic astrocytes with eosinophilic processes but lacking other specific features (hematoxylin and eosin; 400 x ) (A). Cross section of the thoracic spinal cord at autopsy shows a distinct tumor (left) with overlying leptomenlngeal spread (B) (Luxol fast blue/cresyl violet). Perivascular pseudorosettes characteristic of pilomyxoid astrocytomas were observed in this case (C) (hematoxylin and eosin; 200 x ).

Glioneuronal Tumors

One ganglioglioma (WHO Grade I) and a desmoplastic infantile ganglioglioma (Fig. 7A) each were identified. The ganglioglioma was composed of clusters of dysmorphic neurons, some with binucleation. The desmoplastic infantile ganglioglioma had prominent desmoplasia (Fig. 7B) labeled strongly with antibodies directed against glial fibrillary acidic protein (Fig. 7C) and S100 in the predominant astrocytic component. Scattered small neurons showing immunoreactivity for synaptophysin (Fig. 7D) and chromogranin were observed. A biopsy of an enlarging cyst at the operative site 2 years later revealed gliotic cortex and microdysgenesis without evidence of recurrent tumor.

FIGURE 3. Low-grade astrocytoma, subtype indeterminate. In some low-grade astrocytomas, infiltration was evident in the form of entrapped axons (arrows) (A) (hematoxylin and eosin; 400 x ). Rare Rosenthal fibers were present in this tumor (arrow) but not in sufficient numbers for the diagnosis of pilocytic astrocytoma (PA) (B) (hematoxylin and eosin; 400 x ). Another tumor had moderate nuclear atypia, no mitotic activity, and retractile eosinophilic processes but no specific features of PA (C) (hematoxylin and eosin; 400 x ). Some perivascular aggregation of tumor cells was also present (D) (hematoxylin and eosin; 400 x ).

FIGURE 4. Low-grade astrocytoma, subtype indeterminate, with broad cytoplasmic processes and prominent nucleoli. Axial T2- weighted magnetic resonance imaging demonstrating a well- circumscribed tumor in the right insular region (A). The tumor showed uniform enhancement on postcontrast T1-weighted magnetic resonance imaging (B). Large cells with thick processes and prominent nucleoli in an intraoperative smear (phloxine- hematoxylin; 600 x ) (C). Another LGSI with ample eosinophilic cytoplasm and prominent nucleoli (hematoxylin and eosin; 400 x ) (D).

TABLE 2. Pathologic Features of Diffusely Infiltrating Astrocytomas at Diagnosis

Other Astrocytomas

One of the 2 tumors classified as high-grade astrocytoma, subtype indeterminate, developed 4 years after radiation therapy for a right thalamic PA. It was characterized by a relatively solid architecture and the presence of granular bodies but, in addition, had extensive necrosis, 6 mitoses per 10 HPF, and an MIB-1 labeling index of 11.5%. The second tumor was composed of plump cells with a partial fascicular arrangement, but lacking obvious permeation. Increased mitotic activity (7/10 HPF) and pseudopalisading necrosis were present.

Another astrocytoma with a diffusely infiltrating pattern and strong and widespread p53 immunostaining had a high mitotic index (13/10 HPF) despite the presence of palisading around hyalinized vessels and a loose stroma reminiscent of pilomyxoid astrocytoma (Fig. 5C).

Clinical Features

The clinical features of the different astrocytoma subtypes are summarized in Table 3.

Pilocytic Astrocytoma

The median age at pathologic diagnosis was 13 years (IQR, 9-21), with 31% younger than 10 years, 43% between 10 and 20 years, 10% between 20 and 30 years, 10% between 30 and 40 years, and 4% between 40 and 50 years. A single patient was older than 60 years at the time of diagnosis. Radiologic studies included magnetic resonance imaging (27/50), computed tomographic scan (12/50), plain x-ray with or without optic canal views (10/50), ventriculography/ pneumoencephalography (5/50), and myelogram (2/50). When visualized by computed tomographic or magnetic resonance imaging after contrast, the tumors were usually well circumscribed, enhancing, and exhibited both cystic and solid components. Radiographic progression was noted in 15 tumors after surgery. Postoperative treatment modalities included observation (36%), radiation (28%), chemotherapy (12%), both radiation and chemotherapy (5%), or unknown (19%). Multifocality was noted on imaging in 16 patients. Of the 2 pilomyxoid astrocytomas identified, 1 was located in the hypothalamus of a 5-year-old boy (see succeeding sentences), and the second was located in the spinal cord of a 61-year-old man originally diagnosed as LGSI. The latter demonstrated extensive leptomeningeal spread at autopsy, the patient having died soon after surgery for a pheochromocytoma. No other cases demonstrated clinical or pathologic evidence of leptomeningeal spread.

Low-Grade Astrocytoma, Subtype Indeterminate

Presenting symptoms and radiographic findings were also comparable to those of the PA group and are summarized in Table 3. Progression was noted on follow-up imaging in 6 of 17 patients.

FIGURE 5. Diffusely infiltrating astrocytomas. Numerous neoplastic gemistocytic astrocytes in an anaplastic astrocytoma (A; hematoxylin and eosin; 400 x ). Subsequent material obtained during progression to glioblastoma demonstrated a proliferative component characterized by small hyperchromatic undifferentiated cells. Incipient necrosis is evident on the right side (arrows) (B; hematoxylin and eosin; 400 x ). Extensive areas of geographic necrosis were present elsewhere in the tumor (not shown). An example of anaplastic astrocytoma with a focal perivascular arrangement of neoplastic cells around hyalinized vessels (C; hematoxylin and eosin; 200 x ). p53 immunostaining strongly labeled most nuclei (D; 400 x ). FIGURE 6. Anaplastic astrocytoma involving the optic nerve. Low-power photograph demonstrates neoplastic glial cells diffusely Infiltrating optic nerve fascicles (A; hematoxylin and eosin; 100 x ). On high power, the tumor cells demonstrate hyperchromasia and marked nuclear atypia (B; hematoxylin and eosin; 400 x ). Perivascular inflammation was present (C; hematoxylin and eosin; 400 x ). Mitotic figures were frequent (arrow) (D; hematoxylin and eosin).

Astrocytomas With Broad Cytoplasmic Processes and Prominent Nucleoli

These tumors were all in hemispheric, extraventricular locations; 4 of 6 occurred in the temporal lobe.

Diffusely Infiltrating Astrocytoma

This group encompassed approximately a third of the cases (27%), with a median age of 28 (IQR, 9-38) and the following age distributions: 28% between 0 and 9 years, 12% between 10 and 19 years, 12% between 20 and 29 years, 24% between 30 and 39 years, 8% between 40 and 49 years, 4% between 50 and 59 years, and 12% older than 60 years. On neuroimaging studies, contrast enhancement was seen in more than half of the tumors (16/28), and an infiltrative growth pattern was found in one fourth (7/28). Radiographic progression was documented in 14 of 28 tumors, 1 of which was consistent with subependymal spread.

Glioneuronal Tumors

The patient with a desmoplastic infantile ganglioglioma was 4 months old at the time of diagnosis. The tumor arose in the left temporal lobe, and no further treatment was given. Cystic enlargement at the operative site was noted 2 years after surgery on magnetic resonance imaging, but a biopsy was negative. The tumor had not recurred 18 years after the first surgery, although the patient underwent treatment (i.e. radiation and chemotherapy) for a rhabdomyosarcoma of the paranasal sinuses at 18 years of age. The patient is alive without evidence of disease 1 year later.

The 1 ganglioglioma developed in the hypothalamic region of a 34- year-old male patient. Subsequent growth required gamma-knife treatment 13 months after resection. The patient developed progressive neurologic deterioration and died 3.5 years after surgery.

Other CNS Tumors

One high-grade astrocytoma, subtype indeterminate, arose in a 50- year-old man 4 years after radiation of a recurrent thalamic PA (pathology discussed in previous sentences). Magnetic resonance imaging demonstrated ring enhancement and marked edema. Progression was noted, and the patient died 10 months after surgery. The second such tumor arose de novo in the right frontal lobe of a 49-year-old woman who was treated with radiation therapy and developed progressive neurologic deterioration. She died 15 years after diagnosis, most likely secondary to sequelae of radiation.

One right frontal anaplastic astrocytoma with perivascular palisades developed in a 10-year-old boy. This patient received postoperative radiation and temozolomide and is stable with no evidence of disease 20 months after surgery.

One female patient who had undergone resection of a cerebellar PA included in this study developed a pleomorphic spindle cell sarcoma after 12 years, at age 17. The tumor was superficially located in the left frontal lobe but recurred 1 year later and led to her death.

Optic Pathway/Hypothalamic Gliomas

Twenty-four cases involved the optic pathway/hypothalamic region; of these, 14 were classified histologically as PA. Of the remainder, 4 were classified as LGSI, 4 as anaplastic astrocytomas (WHO Grade III), 1 as pilomyxoid astrocytoma (WHO Grade II), and 1 as ganglioglioma. Among the anaplastic astrocytomas, 3 are alive without disease 8 to 16 years after surgery, whereas a fourth patient died of unknown causes 9 years after. Conversely, the patient with the single hypothalamic pilomyxoid astrocytoma underwent a recurrence 1.5 years after surgery but remains stable 15 years after the second surgery.

FIGURE 7. Desmoplastic infantile ganglioglioma. Highpower view shows cells with variable morphology, including elongation and cells with round nuclei and fine chromatin (A; hematoxylin and eosin; 400 x ). The marked desmoplasia, in some areas with a characteristic storiform pattern, is highlighted with a reticulln stain (B; 200 x ). Most cells demonstrate consistent glial fibrillary acidic protein immunoreactivity (C; 400 x ). Scattered small neurons showing synaptophysln immunoreactivity (D; 400 x ).

TABLE 3. Clinical Features of PAs, LGSI, and DAs at Diagnosis

CNS Tumors Arising After Radiation Treatment and Additional Non- CNS Tumors

Eight patients who had received irradiation, 7 for optic pathway gliomas and 1 for a thalamic PA (age range, 2-31 years; mean, 13 years) developed a second tumor. The subsequent tumors included 4 PAs involving the lateral ventricles (n = 2), midbrain (n = 1), and temporoparietal lobe (n = 1), as well as fibrillary astrocytomas WHO Grades II and III in the brainstem (1 each), an LGSI also arising in the brainstem, and a large high-grade astrocytoma, subtype indeterminate, of the right frontal lobe. It is uncertain whether all these tumors arose in the prior radiation fields; therefore, it is unclear if some could be postirradiation gliomas or, rather, second tumors secondary to the NF1 status. An additional patient developed an aggressive enhancing brainstem mass 5 years after irradiation treatment for a recurrent hypomalamic PA that was not biopsied.

Excluding neurofibromas, 13 patients developed 1 or more additional tumors outside of the CNS, including pheochromocytoma (3), malignant peripheral nerve sheath tumors (3), breast carcinoma (2), rhabdomyosarcoma (1), juvenile myelomonocytic leukemia (1), Burkitt lymphoma (1), colonic adenocarcinoma (1), pleomorphic spindle cell sarcoma (1), gastrointestinal stromal tumor (1), and prostate adenocarcinoma (1).

Statistical Analysis

Overall survival times at 5 and 10 years were 85% and 68% in the PA group, 74% for both in the LGSI cases, and 39% and 24% in DA, respectively. Recurrence-free rates at 5 and 10 years were 84% and 72% in the PA group, 64% for both in the LGSI cases, and 40% and 32% in the DA, respectively. The DAs had a significantly decreased overall survival rate and an increased risk of recurrence when compared with the other groups (p

Pilocytic astrocytoma versus LGSI proliferative indices (mitoses/ 10 HPF; MIB-1), p53 labeling, and degree of microscopic infiltration were not significantly different between the PA and LGSI cases. Moreover, there were no statistical significant differences between these 2 groups for radiographic progression, time to recurrence, or overall and disease-specific survival times (p > 0.18). Prognostic variables in the combined PA/LGSI groups are summarized in Table 4.


Overall survival rate at 5 and 10 years was significantly better in patients younger than 10 years at diagnosis in the combined PA and LGSI groups (93% vs 69% and 93% vs 53%, respectively) (p = 0.047) (Fig. 8A). Similar, although not significant, results were seen with disease-specific survival (p = 0.07), with 5- and 10-year disease-specific survival rates in patients younger than 10 years of 93% and 5- and 10-year disease-specific survival rates in older patients of 84% and 72%, respectively. Increasing age was associated with radiographic progression in the PA-only group (p = 0.014) and marginally in the combined PA and LGSI groups (p = 0.058). Patients with optic pathway tumors were significantly more likely to be younger than 10 years of age (66%) than those with either infratentorial (32%) or extra-optic supratentorial tumors (10%) (p = 0.001).

FIGURE 8. Survival in NF1 astrocytomas. Kaplan-Meier curves illustrate decreased overall (A) and recurrence-free (B) survival rates in diffusely infiltrating astrocytomas (DAs), compared with pilocytic astrocytomas (PAs), and low-grade astrocytoma, subtype indeterminate (LGSI), at 5 and 10 years.

Gross Total Resection

After subtotal resection or biopsy in the PA/LGSI combined group, the patients had a significantly worse overall survival rate than those who had undergone gross total resection (Table 4). The extent of resection was not, however, significantly associated with either recurrence or radiographic progression. The extent of surgery was not significantly related to tumor location (p = 0.09).

TABLE 4. Summary of Prognostic Relationships in PA and LGSI Patients

Postoperative radiation was associated with significantly decreased overall survival time in the PA and LGSI combined groups (p = 0.009). No obvious prognostic differences were noted among PA alone or combined PA and LGSI groups according to tumor size, mitotic activity, cellularity, nuclear atypia, biphasic pattern, necrosis, or MIB-1 labeling index (p > 0.07).


Although the predisposition for CNS tumor development in patients with NF1 is well recognized, and there are numerous studies focusing on the clinical features (4, 6-10), detailed pathologic studies on the histopathology of these brain tumors are lacking in the literature. In 1985, Ilgren et al (14) reported a series of 87 gliomas in neurofibromatosis patients, with a comprehensive review of the literature at that time. However, this series included patients that almost certainly would today be classified as having neurofibromatosis type 2, rather than type 1; histologic examination was not performed in all cases. Furthermore, the histologic studies were performed before the development of current histologic classification and grading schemes. Similar to the study of Ilgren et al (14), we found that gliomas in NF1 can arise in any location along the neuraxis, including within the spinal cord. One of their main findings was that cerebellar tumors had a worse prognosis. However, their cases included glioblastomas and ependymomas. Although there were only 5 cerebellar PAs in our study, they did not seem to behave more aggressively than those located in other brain regions. Most of the tumors that arise in the setting of NF1 are PAs and, unlike their sporadic counterparts, have a distinctive predilection to involve the optic nerve, chiasm, and hypothalamus. These optic gliomas are typically found in young children, afflicting 15% to 20% of patients with NF1 (15). Although these tumors seem to have a more indolent behavior than their sporadic counterparts and may even regress without treatment (6), they can cause significant morbidity and progress in some patients (7).

There is considerable uncertainty in predicting which optic pathway tumors will behave more aggressively. Cummings et al (16), in a study of 22 gliomas involving the optic nerve (19 PAs; 3 DAs), identified p53 and MIB-1 labeling index as potentially useful for identifying tumors with either a worse behavior or more infiltrative histology. Most of their cases were, however, sporadic, rather than associated with NF1. In the present study, MIB-1 was not a useful prognostic indicator in low-grade astrocytomas. Although strong diffuse p53 staining (3+) was only encountered in DAs, the numbers were too small to draw any firm conclusion. Twenty-four cases in the current series involved the optic pathways/ hypothalamic region, although the histology was more variable than in the study of Cummings (16), including 14 WHO Grade I PAs, 4 LGSI, 4 anaplastic astrocytomas, and 1 example each of pilomyxoid astrocytoma and ganglioglioma. Pilocytic astrocytomas with moderate to high mitotic activity (4-5/10 HPF) did not behave more aggressively than those with low proliferative rates. Furthermore, even the tumors that were classified as anaplastic astrocytomas of the optic pathways behaved in a clinically less aggressive fashion. These findings support the observations from previous studies that suggest that the optic pathway represents a permissive location for tumor growth in children with NF1. Recent reports using Nf1 genetically engineered mice suggest that specific growth cues from the tumor microenvironment may uniquely promote the growth of Nf1-deficient glial cells to facilitate and limit glioma proliferation within the optic pathway (17, 18).

In addition to the astrocytomas that fall neatly into either PA or DA categories, some NF1-associated astrocytomas display indeterminate or overlapping features that preclude definitive classification as one or the other histologic subtype. These gliomas are usually low grade (5), often with piloid cytologic features, but they lack the typical biphasic architecture, Rosenthal fibers, and eosinophilic granular bodies of classic PAs. Additionally, there is often an infiltrative growth pattern and/or increased proliferative activity providing additional sources of concern. In this study, we referred to such cases as LGSI. Our data suggest that there are no significant clinicopathologic or prognostic differences between this histologic subtype and classic PA when they arise in patients with NF1. In small biopsies, however, the LGSI diagnosis may simply reflect sampling error. For example, 1 of 4 LGSI cases with multiple pathologic specimens showed morphologic features of pilomyxoid astrocytoma at autopsy. A second case showed classic PA after a second resection. Pilomyxoid astrocytoma has been regarded as a morphologic subtype of PA with a predilection for the hypothalamic area and worse clinical behavior (e.g. leptomeningeal seeding) (19) and is therefore classified as a Grade II neoplasm in the most recent WHO classification (2). Two of our cases can be placed in this category, and at least 1 isolated case of pilomyxoid astrocytoma in NF1 has been previously reported (20).

Glioneuronal tumors may also arise in NF1 (11, 21, 22), and we found 1 conventional ganglioglioma in the current study. Desmoplastic infantile ganglioglioma represents a distinct clinicopathologic subtype initially identified in young children and characterized by large size, exuberant collagen deposition, and excellent overall prognosis (23). To our knowledge, this unusual subtype has not been previously described in a patient with NF1.

Although most unusual tumors in our series were low-grade gliomas, caution is warranted because tumors with "indeterminate" or "unusual" histologic diagnoses in NF1 can also be high-grade tumors, as demonstrated by 2 of our cases. One was an anaplastic infiltrative astrocytoma with a peculiar perivascular arrangement reminiscent of pilomyxoid astrocytoma, and the second was an unusual tumor arising after radiation for a conventional PA. The latter tumor showed a solid architecture and eosinophilic granular bodies but had remarkable proliferative activity and extensive necrosis.

The main prognostic indicator in our current series was histologic subtype. Patients with DA had a reduced overall and recurrence free survival. Although these results may be expected from previous data based on sporadic gliomas, it is important to highlight the importance of accurate histologic classification, when possible, for patients with NF1. In this regard, earlier reports have focused on tumor location and concluded that patients with nonoptic pathway gliomas have a worse outcome (10). In the current study in which all tumors were classified histologically, we did not find significant prognostic differences when PAs or PAs and LGSIs were analyzed by tumor location. However, there was a trend for increased recurrence-free survival in optic pathway tumors. After adjusting for histology, patients older than 10 years had a reduced survival rate compared with younger patients.

Diffusely infiltrating astrocytoma as a group encompassed greater than one fourth of the patients, a significant proportion of the tumors in this series. This is in contrast to previous clinical and epidemiologic studies that report or estimate a lower frequency (4). These results might be related to our current series being comprised mostly of surgically obtained specimens at 2 large tertiary care centers. Diffusely infiltrating astrocytomas are more likely to be progressive and demonstrate alarming radiographic features and therefore prompt surgery. Furthermore, PA of the optic pathways, the predominant clinicopathologic type, is not usually biopsied or resected at the present time. Rather, it can be followed without intervention in most instances (6). Most DAs in our series seem to have arisen without a history of prior irradiation. This is consistent with prior publications (4).

In summary, we report the findings of a study with detailed histopathologic examination in a large cohort of NF1 patients. The spectrum of morphologic subtypes is wide in this population. A significant subset display unusual histologic features, but most tumors in this group are low grade and behave similarly to PA. The diagnosis of a DA subtype is the strongest prognostic indicator especially when arising outside of the optic pathways. For this reason, careful histologic classification and grading should be attempted whenever possible to provide an accurate diagnosis, prognostic information, and to guide appropriate therapy for individuals with NF1-associated gliomas.


1. Gutmann DH, Aylsworth A, Carey JC, et al. The diagnostic evaluation and multidisciplinary management of neurofibromatosis 1 and neurofibromatosis 2. JAMA 1997;278:51-57

2. Louis D, Ohgaki H, Wiestler O, Cavenee W. WHO Classification of Tumours of the Central Nervous System. Lyon, France: IARC Press, 2007

3. Stern J, DiGiacinto GV, Housepian EM. Neurofibromatosis and optic glioma: Clinical and morphological correlations. Neurosurgery 1979;4: 524-28

4. Gutmann DH, Rasmussen SA, Wolkenstein P, et al. Gliomas presenting after age 10 in individuals with neurofibromatosis type 1 (NF1). Neurology 2002;59:759-61

5. Leonard JR, Perry A, Rubin JB, et al. The role of surgical biopsy in the diagnosis of glioma in individuals with neurofibromatosis-1. Neurology 2006;67:1509-12

6. Parsa CF, Hoyt CS, Lesser RL, et al. Spontaneous regression of optic gliomas: Thirteen cases documented by serial neuroimaging. Arch Ophthalmol 2001;119:516-29

7. Thiagalingam S, Flaherty M, Billson F, et al. Neurofibromatosis type 1 and optic pathway gliomas: Follow-up of 54 patients. Ophthalmology 2004;111:568-77

8. Zeid JL, Charrow J, Sandu M, et al. Orbital optic nerve gliomas in children with neurofibromatosis type 1. J AAPOS 2006;10:534-39

9. King A, Listemick R, Charrow J, et al. Optic pathway gliomas in neurofibromatosis type 1: The effect of presenting symptoms on outcome. Am J Med Genet A 2003;122:95-99

10. Guillamo JS, Creange A, Kalifa C, et al. Prognostic factors of CNS tumours in Neurofibromatosis 1 (NF1): A retrospective study of 104 patients. Brain 2003;126:152-60

11. Vinchon M, Soto-Ares G, Ruchoux MM, et al. Cerebellar gliomas in children with NF1: Pathology and surgery. Childs Nerv Syst 2000;16: 417-20

12. Riffaud L, Vinchon M, Ragragui O, et al. Hemispheric cerebral gliomas in children with NF1: Arguments for a long-term follow-up. Childs Nerv Syst 2002;18:43-47

13. Rodriguez HA, Berthrong M. Multiple primary intracranial tumors in von Recklinghausen's neurofibromatosis. Arch Neurol 1966;14: 467-75

14. Ilgren E, Kinnier-Wilson L, Stiller C. Gliomas in neurofibromatosis: A series of 89 cases with evidence for enhanced malignancy in associated cerebellar astrocytomas. Pathol Annu 1985;20:331-58

15. Listernick R Ferner RE, Liu GT, et al. Optic pathway gliomas in neurofibromatosis-1: Controversies and recommendations. Ann Neurol 2007;61:189-98

16. Cummings TJ, Provenzale JM, Hunter SB, et al. Gliomas of the optic nerve: Histological, immunohistochemical (MIB-1 and p53), and MRI analysis. Acta Neuropathol (Berl) 2000;99:563-70 17. Daginakatte GC, Gutmann DH. Neurofibromatosis-1 (Nf1) heterozygous brain microglia elaborate paracrine factors that promote Nf1-deficient astrocyte and glioma growth. Hum Mol Genet 2007;16:1098-1112

18. Warrington NM, Woerner BM, Daginakatte GC, et al. Spatiotemporal differences in CXCL12 expression and cyclic AMP underlie the unique pattern of optic glioma growth in neurofibromatosis type 1. Cancer Res 2007;67:8588-95

19. Tihan T, Fisher PG, Kepner JL, et al. Pediatric astrocytomas with monomorphous pilomyxoid features and a less favorable outcome. J Neuropathol Exp Neurol 1999;58:1061-68

20. Khanani MF, Hawkins C, Shroff M, et al. Pilomyxoid astrocytoma in a patient with neurofibromatosis. Pediatr Blood Cancer 2006;46: 377-80

21. Parizel PM, Martin JJ, Van Vyve M, et al. Cerebral ganglioglioma and neurofibromatosis type I. Case report and review of the literature. Neuroradiology 1991;33:357-59

22. Fedi M, Anne Mitchell L, Kalnins RM, et al. Glioneuronal tumours in neurofibromatosis type 1: MRI-pathological study. J Clin Neurosci 2004;11:745-47

23. VandenBerg SR, May EE, Rubinstein LJ, et al. Desmoplastic supratentorial neuroepithelial tumors of infancy with divergent differentiation potential ("desmoplastic infantile gangliogliomas"). Report on 11 cases of a distinctive embryonal tumor with favorable prognosis. J Neurosurg 1987;66:58-71

Fausto J. Rodriguez, MD, Arie Perry, MD, David H. Gutmann, MD, PhD, Brian Patrick O'Neill, MD, Jeffrey Leonard, MD, Sandra Bryant, MS, and Caterina Giannini, MD, PhD

From the Departments of Laboratory Medicine and Pathology (FJR, CG), Neurology (BPON), and Biostatistics (SB), Mayo Clinic College of Medicine, Rochester, Minnesota; and Division of Neuropathology (AP), Departments of Neurology (DHG) and Neurosurgery (JL), Washington University School of Medicine, St. Louis, Missouri.

Send correspondence and reprint requests to: Fausto J. Rodriguez, MD, Mayo Clinic College of Medicine, 200 First Street SW, Mayo Clinic, Rochester, MN 55905; E-mail: [email protected]

Supported in part by training Grant No. T32 NS07494-04 from the National Institutes of Health (to F.J.R.).

Copyright Lippincott Williams & Wilkins Mar 2008

(c) 2008 Journal of Neuropathology and Experimental Neurology. Provided by ProQuest Information and Learning. All rights Reserved.