Glial fibrillary acidic protein in stromal cells of some capillary hemangioblastomas: significance and possible implications of an immunoperoxidase study

Glial Fibrillary Acidic Protein (GFAP) in Human Cerebral Tumors. An Immunohistochemical Study

Glial fibrillary acidic protein (GFAP) was studied in 160 cerebral tumors, mostly of neuro-epithelial nature. It was positive in astroglial tumors with an intensity proportional to the degree of cell differentiation. It was sometimes positive also in non-astroglial tumors, such as oligodendrogliomas and ependymomas, and this finding is discussed in relation to genesis and diagnostic value. In medulloblastomas, there were also positive cells, which could be reactive glia cells included in the tumors or subependymal cells. The demonstration of GFAP is very useful in gliosarcomas for identifying the glial component. It was sometimes positive in hemangioblastomas, and it is discussed in view of the nature of the stromal cells of this tumor.

A case of cerebellar hemangioblastoma with rhabdoid features

We present an unusual case of cerebellar hemangioblastoma characterized by rhabdoid features. The patient was a 35-year-old Japanese man with occipital neuralgia and exacerbating blurred vision. Magnetic resonance imaging revealed a left posterior cranial fossa tumor, which was isointense on T1-weighted images and hyperintense on T2-weighted images with marked homogeneous enhancement. Histology of the surgically resected tumor showed cellular-type hemangioblastoma with extensive proliferation of rhabdoid cells Immunohistochemistry analysis showed tumor cells positive for inhibin A, CD56, vimentin, INI-1, and vascular endothelial growth factor; negative for PAX8, CD10, epithelial membrane antigen, cytokeratin, (AE1/3), alpha-smooth muscle actin and D2-40; and had focal positivity for glial fibrillary acidic protein and S100. The Ki-67 labeling index was <1 %. Ultrastructural analysis revealed large lipid droplets and abundant intracellular accumulation of intermediate filaments. Based on these findings, the diagnosis was hemangioblastoma with focal rhabdoid features. After a 14-month follow-up, there was no evidence of recurrence. This is the first report of hemangioblastoma with rhabdoid features in the central nervous system. In addition, we discuss the possible pathogenesis.

Supratentorial hemangioblastoma: clinical features, prognosis, and predictive value of location for von Hippel-Lindau disease

Supratentorial hemangioblastoma is a rare form of hemangioblastoma; little information is available regarding prognosis, treatment, and clinical characteristics, because the available literature is primarily composed of case reports and small case series. Therefore, we performed a systematic review of the literature to analyze clinical characteristics, disease progression, and surgical outcomes with respect to survival for supratentorial hemangioblastomas. The rate of progression-free survival (PFS) was determined using Kaplan-Meier analysis. Differences in categorical factors, including location of tumor and diagnosis of von Hippel-Lindau (VHL) disease, were analyzed using the Pearson χ(2) test. A total of 106 articles met the search criteria, which combined for a total of 132 patients. Of the patients with supratentorial tumors, 60% had VHL disease, and 31 (84%) of 37 patients with tumors in the sellar/suprasellar region had associated VHL (χ(2), P < .001). Five-year PFS for gross-total resection and subtotal resection were 100% and 53%, respectively (Log rank, P < .01). On the basis of our analysis of the literature on published cases of supratentorial hemangioblastoma, gross-total resection appears to be superior to other treatment modalities in extending PFS. Von Hippel-Lindau disease is positively correlated with supratentorial hemangioblastoma when compared with non-supratentorial CNS hemangioblastomas, particularly when present in the sellar/suprasellar region.

Central nervous system capillary haemangioblastoma: the pathologist's viewpoint

Haemangioblastomas are rare neoplasms of uncertain histogenesis. They represent 1.5-2.5% of intracranial tumours. While the cerebellum is by far the most frequent site, these lesions also tend to occur in the medulla and spinal cord. Most haemangioblastomas are sporadic but up to one quarter are associated with Von Hippel-Lindau disease (VHL). Although a fairly good number of haemangioblastomas were reported, a lack of side-by-side analysis of these reports has resulted in tentative conclusions that merely offer a first glimpse at their clinicopathologic diversity and histogenesis. To remedy this issue, this study presents a literature review concerning these lesions. Medline literature including both relevant monographs and clinicopathological case series. Haemangioblastomas occur either as a part of VHL disease (25-30%, inherited mutation of VHL gene on 3p25-26 chromosome) or as sporadic tumours (often with somatic mutation of VHL gene). They have diverse clinicopathologic presentations with cerebellar lesions having a better prognosis than their brainstem counterparts. Immunostaining is important for separation of haemangioblastomas from other tumours with similar histology. The rich vascularity of haemangioblastomas is due to overexpression of vascular endothelial growth factors. Moreover, 'stromal' cells represent the neoplastic cells of haemangioblastomas and are capable of forming blood islands with extramedullary haematopoiesis.

Cellular and reticular variants of haemangioblastoma revisited: a clinicopathologic study of 88 cases

The presence of histological variants of haemangioblastoma is well established, but data on the prognostic implications of histological subtyping are missing. We thus characterized clinical factors associated with histological subtypes, that is, of the cellular and reticular variant of haemangioblastoma, in a series of 88 consecutive primary haemangioblastomas of the central nervous system. Ten haemangioblastomas were classified as 'cellular' according to Cushing and Bailey. As compared to the more common 'reticular' variant (n = 78), the proportion of tumours containing glial fibrillary acidic protein-positive tumour cells (80% vs. 7%), as well as median Ki67 (MIB1) proliferation indices [4% (quartiles: 1-8%) vs. < 1% (<1-2%)], was significantly higher in cellular haemangioblastomas (P < 0.01). Recurrences were more frequent in the cellular variant [2/8 (25%) vs. 4/51 (8%)]. Kaplan-Meier analysis confirmed a significantly higher probability of recurrence in the cellular variant (Log-Rank test P < 0.01). Cox regression analysis not only confirmed the well established association of von Hippel-Lindau disease with tumour recurrence (P < 0.01), but also revealed an independent effect of histological subtype on the probability of recurrence (P < 0.05), whereas no significant influence of age, sex or tumour location was observed. To conclude, the results from this retrospective study suggest that histological subtyping of haemangioblastomas has prognostic implications and might contribute to identify patients at risk for recurrence.

Stromal cells in hemangioblastoma: neuroectodermal differentiation and morphological similarities to ependymoma

The histogenesis of stromal cells in hemangioblastoma is inconclusive despite a long-term controversy. An immunohistochemical and ultrastructural study was conducted for 17 cases of cerebellar hemangioblastoma. A wide range of immunohistological markers, targeting epithelial, mesenchymal, endothelial and neuroectodermal tissues, was used. In all cases, the microscopic hallmark characterizing hemangioblastomas, that is, lipid-containing stromal cells and a fine capillary network, known as a reticular variant, was noted. Stromal cells showed a variable immunoreactivity for neuroectodermal markers, such as S-100 protein, CD56, CD57, CD99, and neuron-specific enolase. This result, in conjunction with the absence of immunoreactivity for epithelial, mesenchymal, and endothelial markers, likely suggests neuroectodermal differentiation of stromal cells. In three cases, another component, known as a cellular variant, where epithelioid tumor cells were arranged in nests encircled by capillaries and/or in pseudorosette-like structures, was noted. Glial fibrillary acidic protein-immunoreactivity, which was totally absent in cases only showing the reticular pattern, was noted in two of them, suggesting a distinctive sign of glial differentiation in a proportion of hemangioblastomas. Ultrastructurally, microvilli-like projections in intracytoplasmic vacuoles were demonstrated in stromal cells. This result, taken together with the neuroectodermal hypothesis of stromal cells, suggests that hemangioblastomas may occasionally exhibit morphological similarities to ependymomas.

Lucien J. Rubinstein: enduring contributions to neuro-oncology

Dr. Lucien Rubinstein is best remembered for his significant contributions to the field of neuropathology, particularly in the classification of nervous system tumors. His accomplishments in basic neuro-oncology and in the formulation of diagnostic principles reflected a unique talent for synthesizing fundamental clinicopathological concepts based on skillful diagnostic investigation and a thorough understanding of neurobiology. Dr. Rubinstein was the leader in the establishment of cell cultures from central nervous system (CNS) tumors. He meticulously analyzed both light and electron microscopic features of CNS tumors, recorded his findings, and patiently drew sketches to be shared generously with his colleagues and students. As a pioneer in neuropathology, in his work Dr. Rubinstein set the foundation for many enduring concepts in neurosurgery, neuro-oncology, neurology, and basic tumor biology.

Transthyretin and transferrin in hemangioblastoma stromal cells

Hemangioblastoma, a rare benign tumor of the CNS, consists of two main components: capillaries and stromal cells. Despite many efforts, the histogenesis of stromal cells is still unknown. We took a new approach to clarify the origin of stromal cells using immunohistochemical methods. Paraffin-embedded tissue of 24 surgically removed hemangioblastomas of the CNS was examined with antibodies against transthyretin, transferrin, vimentin, NSE, protein S-100, CK 8, KL-1, EMA, CD34, factor VIII rAg, and collagen IV. Stromal cells showed a positive reaction with anti-transthyretin in 12 of 24 hemangioblastomas, a positive reaction with anti-transferrrin, to a different extent, in 13 of 24 cases, and many stromal cells expressed basal membrane collagen IV on the cell surface in 19 of 24 cases. The expression of transthyretin and transferrin in stromal cells of hemangioblastomas is reported for the first time, thus providing an antigenic profile of hemangioblastoma stromal cells that is very similar to that of immature choroid plexus epithelium. These findings support the notion that hemangioblastoma stromal cells may originate from the embryonal plexus epithelium. We discuss our results with special regard to stromal cell histogenesis, including a review of the literature.

Recurrent (Nonfamilial) Hemangioblastomas Involving Spinal Nerve Roots: Case Report

OBJECTIVE AND IMPORTANCE

Spinal nerve root hemangioblastomas are rare and are reported mainly in patients with von Hippel-Lindau (VHL) syndrome. The pathogenesis of so-called nonfamilial lesions is virtually unknown. We discuss, mainly from a molecular perspective, a unique patient with sporadic, recurrent hemangioblastomas restricted to spinal nerve roots.

CLINICAL PRESENTATION

A 53-year-old man who had had a surgically corrected lumbosacral meningomyelocele presented on at least three occasions during a 17-year period with multifocal capillary hemangioblastomas involving spinal nerve roots. On each occasion, tumors appeared on a different nerve root, with the majority located in the midcervical segments. The patient had no clinical features or family history of VHL syndrome.

TECHNIQUE

To obtain a clearer understanding of the pathogenesis of this unusual case and its relationship to VHL syndrome, molecular analysis of the VHL gene was performed by use of complete sequence analysis and loss of heterozygosity studies on deoxyribonucleic acid derived from the patient's blood leukocytes and three separately resected hemangioblastomas.

CONCLUSION

Germ-line molecular analysis performed on all three exons in the VHL gene coding region did not indicate that any mutations were present. Loss of heterozygosity analysis of deoxyribonucleic acid from the three hemangioblastoma resections showed normal heterozygosity in the 3p25–26 region. Complete VHL gene sequence analysis did not demonstrate a somatic mutation in the coding region of the VHL gene in any of the three tumors, thereby supporting the loss of heterozygosity data that a molecular event directly involving the VHL gene may not be the causative factor in their tumorigenesis.

Retroperitoneal peripheral hemangioblastoma: a case report and review of the literature

Central nervous system hemangioblastomas are uncommon tumors of controversial etiology that are usually found in the posterior fossa of the cranial cavity, retina, and spinal cord. Peripheral involvement is rare; only isolated case reports have been identified. We report an unusual case of hemangioblastoma involving the retroperitoneum. A 47-year-old African-American man presented with polycythemia on routine laboratory testing. Computed tomography revealed a large retroperitoneal mass near the pancreas, in a left suprarenal location, without adrenal involvement and without attachment to a nerve. Although hemangioblastoma may be associated with the von Hippel-Lindau syndrome, this patient did not have any of the stigmata of this disease. The histologic features included a highly vascular tumor with cellular areas composed of plump, pleomorphic spindled and epithelioid (stromal) cells with variable cytoplasmic lipid vacuoles and hypocellular areas with inflammatory cells and collagenous fibrils. Immunohistochemical staining showed that the tumor (stromal) cells were positive for vimentin, calponin, S-100 protein, neuron-specific enolase, and CD57 and negative for glial fibrillary acidic protein, cytokeratins, epithelial membrane antigen, CD34, HMB-45, desmin, and the actins. These morphologic and immunohistochemical findings are consistent with hemangioblastoma. To our knowledge this is the first reported case of a hemangioblastoma in this location. Based on this case we conclude that hemangioblastoma may occur in the retroperitoneum and outside of the central nervous system in a patient without von Hippel-Lindau syndrome. The immunoprofile of this case suggests that hemangioblastomas are mesenchymal neoplasms exhibiting both neural and myofibroblastic differentiation.

Hemangioma of spinal nerve root

OBJECT

Hemangiomas of the spinal nerve roots are rare. Only 12 cases have been reported in the literature, all since 1965. These lesions occurred in adults, primarily in males, and were located in the cauda equina. Of the 12 lesions, eight were cavernous hemangiomas. The authors report here the clinicopathological features of 10 cases of hemangioma of the spinal nerve root, nine of which were of the capillary type.

METHODS

The group included six men and four women who ranged in age from 40 to 62 years. The majority of patients presented with pain and weakness. One patient also experienced worsening of pain during menses. In all cases, gadolinium-enhanced magnetic resonance imaging or computerized tomography myelography demonstrated a discrete intraspinal extramedullary mass. Administration of contrast agent resulted in uniform, intense enhancement. A gross-total resection was achieved in all but one case; in most cases, it was necessary to sacrifice the parent nerve root. Preoperative symptoms and signs improved in nine cases. On histological examination, all but one tumor, a spindle-cell hemangioendothelioma, were shown to be ordinary capillary hemangiomas. Two lesions exhibited a cavernous component, and five showed a partly solid growth pattern resembling juvenile capillary hemangioma. In all cases, the relationship with the nerve root was histologically confirmed. The tumor-nerve relation varied. The intact nerve was displaced by the epineurial mass in three cases. The fascicles appeared separated in six cases, and individual nerve fibers were seen throughout one lesion with endoneurial involvement.

CONCLUSION

. Hemangiomas of spinal nerve roots pose a challenging diagnostic problem. Knowledge of their existence is relevant in that they may mimic tumors and are amenable to surgical treatment.

Cerebellar clear cell ependymoma mimicking hemangioblastoma: its clinical and pathological features

BACKGROUND

As the clinical and pathological significance of cerebellar clear cell ependymoma (CCE) has not been recognized in the past, eight cases of cerebellar CCEs were studied.

METHODS

Subjects were six men and two women, and their ages ranged from 23 to 64 years old. Their neuroradiological, operative, and pathological features were reviewed.

RESULTS

The tumors frequently showed high vascularity on angiography and marked enhancement on computerized tomography. Grossly, six tumors exhibited a "cyst with mural nodule" and two were mostly solid containing small cysts. At operation, the tumor nodules were dark red in color and the cysts contained xanthochromic fluid. Microscopically, all tumors displayed mostly clear round cells and abundant blood vessels. Although they commonly lacked the characteristic features of ependymomas, immunostaining and electron microscopy confirmed their ependymal origin.

CONCLUSIONS

The results indicate that cerebellar CCEs closely mimic hemangioblastoma both clinically and pathologically, which suggests CCE should be considered in the differential diagnosis of vascular tumors of the cerebellum. Some diagnostic clues to differentiate CCE from hemangioblastoma are discussed.

Cavernous angioma associated with oligo-astrocytoma-like proliferation. Report of two cases and review of the literature with a reappraisal of the term "angioglioma"

Reporting two cases of cavernous angioma closely associated with oligo-astrocytoma we were stimulated to revise the proper use of the term "angioglioma", introduced by Councilman 80 years ago7. In the past this term was often used in a merely descriptive sense i.e. either to give a name to an exceptional coincidence of AVM and glioma or simply to describe a hypervascularized glioma. Today, according to the majority of authors, such a use of this term has to be rejected as inappropriate. Occasionally however the term angioglioma has been used to designate a true mixed composite neoplasm developed after the transformation of both glial and angiogenic elements determined by simultaneously or consecutively acting (even possibly intermingled) oncogenic factors. Among reported cases claimed to be examples of "true" angioglioma, the association cavernomaoligodendroglioma/astrocytoma seems the most intriguing besides being also the most frequent one. Opinions about this topic however did continue to be unequivocal. While some authors regarded the oligodendroglial proliferation associated with AVM as merely reactive or malformative, others considered it as truly neoplastic putting forward the hypothesis of a common viral or genetic aetiology for both AVM and glioma.

Hemangioblastoma of the radial nerve: case report

A case of hemangioblastoma of the right radial nerve is presented. Hemangioblastomas are rare vascular neoplasms characteristically associated with the Von Hippel-Lindau syndrome, in which they are found in the retina, the posterior fossa, and, less often, the spinal cord. Thought of as primary central nervous system neoplasms, hemangioblastomas are rarely found adjacent to the spinal cord involving proximal nerve roots, which represent border zones between the central and peripheral nervous systems. We could find no other report of a pure hemangioblastoma situated this far distally in the peripheral nervous system. The histological findings, immunohistochemistry, and electron microscopic findings of this lesion are discussed. This case supports the hypothesis that hemangioblastoma is not derived from astrocytes, because of the location of this tumor in the peripheral nervous system and glial fibrillary acidic protein negativity.

Fibrohistiocytic differentiation in capillary hemangioblastoma

Seven cases of capillary hemangioblastoma from the cerebellum and spinal cord were studied by immunohistochemical methods to determine the origin of the stromal cells. A subpopulation of factor XIIIa-positive tumor cells was a constant feature in hemangioblastomas. These stellate or spindle-shaped cells transformed into typical vacuolated stromal cells. Factor VIII-related antigen was limited to the vascular endothelium. Glial fibrillary acidic protein was present only in entrapped astrocytes. Staining for alpha-1-antitrypsin (alpha 1AT) and alpha-1-antichymotrypsin (alpha 1 ACT) was occasionally observed in stromal cells. It was concluded that the factor XIIIa-positive stromal cells in capillary hemangioblastoma indicate fibrohistiocytic differentiation, which is part of the differentiation spectrum of hemangiopericytomas.

The growth of two murine hemangioendotheliomas intracranially, subcutaneously, and in culture, and their comparison with human cerebellar hemangioblastomas: morphological and immunohistochemical studies

Two thorium dioxide-induced murine hemangioendotheliomas, 42021 TCT and 44347 TST, were grown subcutaneously (for up to 22 and 15 passages respectively) or intracranially (single passage) and were adapted to culture as a monolayer and, in a limited fashion, in an organ culture system or in rotary suspension. They remained viable and malignant following 20-21 years of storage in liquid nitrogen, and had ultrastructural similarities to human hemangioblastomas. The murine tumors were positive for Griffonia (Bandeiraea) simplicifolia isolectin B4 binding, establishing their endothelial nature; however, unlike human hemangioblastic tumors, they did not cross-react with antisera to human factor VIII or fibronectin and they did not demonstrate Ulex europaeus type I lectin (UEA I) binding (as is also the case for non-neoplastic murine vascular endothelial cells). A variety of morphological cell types in cultures derived from the tumors were also positive for Griffonia (Bandeiraea) simplicifolia isolectin B4 binding. Both murine hemangioendotheliomas, when implanted in the cerebrum, were potent inducers of reactive gliosis, but there was no evidence of uptake of glial fibrillary acidic protein. Unlike the human cerebellar hemangioblastomas, murine tumors were malignant and invasive and did not contain stromal cells, nor did they demonstrate Weibel-Palade bodies or extensive pinocytotic activity. Thus, the murine tumors appear to more closely resemble angiosarcomas or epitheloid hemangioblastomas than the cerebellar hemangioblastomas.

Retinal hemangioblastoma. A histologic, immunohistochemical, and ultrastructural evaluation

The authors evaluated the histologic, immunohistochemical, and ultrastructural characteristics of two eyes with retinal hemangioblastoma from patients with von Hippel-Lindau and von Hippel disease. Results of histologic evaluation showed the eyes to have degenerative changes and residual retinal hemangioblastoma. Immunohistochemical stains performed for MAC-387, factor XIIIa, lysozyme, alpha 1 anti-chymotrypsin (histiocyte markers), factor VIII-associated antigen, ulex europeaus (endothelial markers), neuron-specific enolase, chromogranin, neurofilament (neuroectodermal/neural/neuroendocrine markers) and glial fibrillary acid protein (glial marker) showed normal retinal vascular endothelium, neurons, and glial cells to stain where expected. Vascular endothelium in the retinal hemangioblastomas stained for factor VIII and ulex europeaus. Interstitial cells in the stroma of the tumors failed to stain for the histiocyte markers, chromogranin, and neurofilament. The stromal cells stained for glial fibrillary acid protein and neuron specific enolase. Ultrastructural findings in both eyes included endothelial/pericyte-lined vascular channels, elongated stromal cells, and plump, vacuolated stromal cells with ultrastructural features consistent with glial cells. This study supports the concept that retinal hemangioblastoma is composed of a proliferation of capillaries and glial cells.

Granular cell brain tumors of the laboratory rat: an immunohistochemical approach

We have studied paraffin-embedded specimens of 17 rat granular cell brain tumors (GCBT) from four long-term drug safety carcinogenicity studies by peroxidase-antiperoxidase (PAP) immunohistochemistry with either polyvalent or monoclonal antibodies against glial fibrillary acidic protein (GFAP), S-100 protein (S-100), Leu-7 epitopes, vimentin (VIM), keratin, desmin, and myelin basic protein. We have found that 9 of the 17 GCBT contained GFAP-positive, S-100-positive, and VIM-positive astrocytes, while GFAP-positive and VIM-positive granular cells were observed in 5 of these 9 tumors. Our findings indicate that astroglial cells are involved in rat GCBT and suggest that an astrocytic origin should be considered for these neoplasms.

Desmoplastic cerebral astrocytoma of infancy. Light microscopy, immunocytochemistry, and ultrastructure

The first detailed ultrastructural study of a superficial desmoplastic cerebral astrocytoma of infancy is reported. This is a neoplasm which mimics a mesenchymal tumor, originally described in 1984 under the name of "superficial cerebral astrocytoma attached to dura." This tumor, which is believed to have a good prognosis, was resected from the frontoparietal region of a 6.5-month-old girl, in whom it had presented as a large densely enhancing vascular and cystic mass. The ultrastructure of the tumor appeared to be distinctive, characterized by the absence of neuronal elements, and the presence of large amounts of redundant and sometimes extensively duplicated basal laminal material and collagen between nonpleomorphic and nonlipidized astrocytes, corresponding to the reticulin fibers seen by light microscopic analysis between the S-100 protein and glial fibrillary acidic protein (GFAP)-positive cells. The features of the tumor, its differential diagnosis, and its relationship to other pediatric supratentorial tumors are discussed.

Changes in glial cell markers in recent and old demyelinated lesions in central pontine myelinolysis

An immunohistochemical study was performed to compare glial reactions in recent and old lesions of central pontine myelinolysis (CPM). Regions of demyelination and destruction of oligodendrocytes, showed reduced immunoreactivity of myelin basic protein (MBP), myelin-associated glycoprotein (MAG), transferrin, and carbonic anhydrase C (CA C). In addition, labeling of glial fibrillary acidic protein (GFAP) and S-100 protein revealed distinct dystrophic alterations of the astroglia. Remarkably, immunolabeling of GFAP was drastically reduced in astrocytic cytoplasm within freshly demyelinated lesions. Immunostaining of vimentin revealed a differential intracytoplasmic decoration of hypertrophic and dystrophic astrocytes in recent and old CPM lesions. Immunolabeling of desmin failed to stain glial cells. Monoclonal antibodies against HNK-1 exhibited greatly increased immunoreactivity both of persisting oligodendrocytes and of reactive fibrillary astrocytes in old CPM foci. In freshly demyelinated lesions, enhanced immunoreactivity of the X-hapten (3-fucosyl-N-acetyllactosamine) was prominent in astroglia and oligodendrocytes. Simultaneously, reactive astrocytes revealed intracytoplasmic labeling of laminin. Quantitation of GFAP+ astroglia in fresh CPM and control cases revealed an increase in the number of astrocytes within the demyelinated foci and in the surrounding non-demyelinated pontine tissue of CPM cases. The occurrence of astroglial alterations in the demyelinated foci of CPM could be interpreted as "astroglial dystrophy" which may represent a pathogenic factor in CPM. Furthermore, it is possible that changes of the glial microenvironment may influence the astroglia to revert transiently back to an immature phenotype as indicated by the enhanced expression of the X-hapten and HNK-1, and the de novo synthesis of vimentin and laminin.

Histopathological features of recurrent pleomorphic xanthoastrocytomas: further corroboration of the glial nature of this neoplasm. A study of 3 cases

Pleomorphic xanthoastrocytoma (PXA), a tumor most often presenting superficially over the cerebral hemisphere of young subjects, has certain morphological similarities to fibrous histiocytoma (or fibrous xanthoma) of the meninges and brain, namely the occurrence of lipid-laden neoplastic cells and, frequently, a dense reticulin fiber network. The detection of glial fibrillary acidic (GFA) protein in the tumor cells helped to establish its astrocytic derivation, but it has been advanced that, in spite of this agreed observation, the tumor should still be regarded as a fibrous xanthoma of meningeal origin. Although many patients have a long symptom-free postoperative survival, local recurrences at varying intervals after surgery have been noted in some instances. Weldon-Linne et al. first reported that such a recurrence had the morphology of a small-cell glioblastoma. We are reporting three further examples of locally recurrent neoplasms in patients whose original meningocerebral tumors had the typical features of PXA; the recurrences (developing 7 months, 7 years and 15 years, respectively, after surgery) were small-cell glioblastomas. The rich reticulin network present in the initial tumor was mostly lost in the recurrences. This anaplastic evolution further confirms the astrocytic nature of the PXA.

Immunohistochemical determination of protein kinase C expression and proliferative activity in human brain tumors

Protein kinase C (PKC), the major receptor for phorbol ester tumor promotors, is a phospholipid- and calcium-dependent phosphorylating enzyme which plays an important role in the intracellular signal transduction necessary for a variety of basic cellular functions including the control of cell proliferation. To determine the expression of PKC in human neurogenic tumors we investigated 121 tumors of the human nervous system by means of immunohistochemistry using the monoclonal antibody C5. The results were compared with immunohistochemical staining for intermediate filament proteins, desmoplakins, and the proliferation-associated nuclear antigen Ki-67. Besides strong staining of normal and reactive astrocytes, C5 immunoreactivity was consistently observed in tumor cells of all types of gliomas. However, the fraction of C5 positive tumor cells varied between the different tumor types with astrocytomas and subependymomas demonstrating the strongest immunoreactivity. In the other gliomas, especially those of higher malignancy, a considerable heterogeneity in C5 expression could be observed. There was a tendency for the percentage of C5 immunostained tumor cells being lower in high-grade gliomas compared to low-grade ones and comparison with Ki-67 staining frequently revealed an inverse relationship between proliferative activity and C5 immunoreactivity. Besides the gliomas we found 3 of 7 neurinomas and 6 of 18 meningiomas which were partially C5 positive. All other tumors investigated including medulloblastomas and metastatic carcinomas were C5 negative. Our results thus indicate that immunohistochemistry for PKC using the monoclonal antibody C5 could be an useful aid for histopathological tumor classification in neuro-oncology.

Use of immunohistochemical methods in the differential diagnosis between primary cerebellar haemangioblastoma and metastatic renal carcinoma

The potential role of immunohistochemistry in making the distinction between primary cerebellar haemangioblastoma and metastatic renal carcinoma was investigated by examining the reaction pattern of 10 cerebellar haemangioblastomas (seven women, three men, aged 20-40 years) and 10 primary renal carcinomas (six men, four women, aged 49-82 years) to a panel of epithelial, glial, and neural/neuroendocrine antisera. The tumour cell membranes of the renal carcinomas stained strongly with epithelial membrane antigen (EMA); membrane staining was totally absent in the haemangioblastomas. Strong neurone specific enolase (NSE) and S100 staining were also seen in haemangioblastomas but were more variable than EMA staining in renal carcinomas. It is concluded that a panel of antisera is required to distinguish between histologically similar areas in primary haemangioblastomas and metastatic renal carcinomas, and that while complementing conventional histological techniques, new problems of interpretation result which must be taken into account.

Glial fibrillary acidic protein in chondrocytes of elastic cartilage in the human epiglottis: an immunohistochemical study with polyvalent and monoclonal antibodies

Glial fibrillary acidic protein (GFAP), an intermediate filament, was first discovered in the cytoplasm of reactive astroglia and to this date is regarded as a rather characteristic component of both non-neoplastic and neoplastic astrocytes. It has, however, been found in other glial elements also, as well as in a few types of tissues outside the central nervous system. Chondrocytes in hyaline and fibrocartilage do not express GFAP, but in the elastic cartilage of the human epiglottis we found GFAP to be regularly present when tested with either polyclonal or monoclonal antibodies. While most types of cartilage in humans are believed to be of mesenchymal origin, embryologic relationship between epiglottal cartilage and the neural crest may play a role in the curious phenomenon of the regular presence of GFAP in the chondrocytes of the epiglottis.

Does the pleomorphic xanthoastrocytoma exist? Problems in the application of immunological techniques to the classification of brain tumors

A case of a fibrous xanthomatous tumor of the meninges is reported. This is a rare tumor of childhood in which the characteristic pleomorphic histology contrasts with the good clinical prognosis. These tumors were reclassified as pleomorphic xanthoastrocytomas (PXA) due to their glial fibrillary acidic protein (GFAP) positivity. In the present tumor, GFAP was absent from nearly all cell bodies in most of the leptomeningeal regions of the tumor but could be detected with greater frequency at the cortical-leptomeningeal border zones and in the areas in which the tumor had infiltrated the cortex. All the tumor cells expressed vimentin and in, addition, most expressed alpha-1-antitrypsin, alpha-1-antichymotrypsin, tartrate-resistant acid phosphatase, common leukocyte antigen, and OKM1. This spectrum of marker staining corresponded not only to the pattern observed in two cutaneous fibrous histiocytomas and one malignant fibrous histiocytoma, but also to the results previously published in the literature with regard to fibrohistiocytic tumors. By contrast, this spectrum of monocytic-histiocytic marker staining was not seen in gliomas. We, therefore, regard the PXA as a mesenchymal tumor of the meninges, identical to benign fibrous histiocytomas elsewhere in the body. The possible reasons why this mesenchymal tumor can show GFAP positivity in the leptomeningeal border zone are discussed.

Stromal cells in cerebellar haemangioblastomas: an immunocytochemical study

The nature of the stromal cells in formalin-fixed paraffin-embedded material from 23 cerebellar haemangioblastomas was investigated using antisera to intermediate filaments (glial fibrillary acidic protein, vimentin and desmin), histiocytic markers (alpha 1-antitrypsin, alpha 1-antichymotrypsin and lysozyme), glycolytic enzymes (alpha and gamma enolase and aldolase C4) and the endothelial markers, factor VIII related antigen and Ulex europaeus I lectin. Most stromal cells stained positively for vimentin and the glycolytic enzymes. Occasional process-bearing cells within the stroma stained strongly for glial fibrillary acidic protein, alpha 1-antitrypsin and alpha 1-antichymotrypsin. No stromal cell staining for desmin, lysozyme or the endothelial markers was observed, although the latter stained the vascular endothelium within all neoplasms. The findings do not support previous suggestions of an endothelial or histiocytic origin for the stromal cells. They appear to be a heterogeneous population including entrapped reactive astrocytes and locally-derived non-angiogenic cells of neuroectodermal (pial) origin.

Gamma-enolase and glial fibrillary acidic protein in nervous system tumors. An immunohistochemical study using specific monoclonal antibodies

A large series of central and peripheral nervous system tumors was studied for the presence of glial fibrillary acidic protein (GFAP) and gamma-enolase (neuron-specific enolase, NSE), using specific monoclonal antibodies (mAbs). Occurrence in and specificity of GFAP to glial and mixed tumors was confirmed and depended on the malignancy grade and features such as meningeal invasion. Using a well-characterized mAb, gamma-enolase was demonstrated in neuronal, as well as in a whole range of non-neuronal tumors. This lack of specificity of gamma-enolase prohibits its use as an exclusive neuronal marker. Nevertheless quantization or comparison with other types of enolases could still prove to be useful in well-defined situations. The advantages inherent to mAbs and a highly sensitive detection system turn GFAP stainings into a specific and readily reproducible technique.

Haemangioblastoma. An immunohistochemical study of ten cases

Ten cases of cerebellar haemangioblastoma were studied using the immunoperoxidase technique for glial fibrillary acidic protein (GFAP), Factor VIII-related antigen (F8RA), Ulex europeus agglutinin 1 (UEA-1), S-100 protein, neurone-specific enolase (NSE), leucocyte common antigen, synaptophysin, chromogranin and eight polypeptide hormones (bombesin, pancreatic polypeptide, somatostatin, thyroglobulin, calcitonin, glucagon, insulin and gastrin). GFAP and S-100 were demonstrated at the periphery of all tumours and in small groups of cells in the centre of four cases. Most of these cells had the morphology of reactive astrocytes but some had the appearance of stromal cells. In general stromal cells gave negative results. F8RA and UEA-1 stained the endothelial cells in each case but there was no stromal cell reactivity. NSE was present in the stromal cell component of all tumours. There was no staining for synaptophysin, for chromogranin, or any of the polypeptide hormones. It therefore appears that some haemangioblastomas contain an admixed non-neoplastic astrocytic element. NSE, F8RA and UEA-1 staining demonstrates that the endothelial and stromal cell parts of the tumour are antigenically distinct. Recent reports of polypeptide hormone expression cannot be confirmed and it is therefore unlikely that stromal cells originate from primitive peptidergic neurones.

Capillary hemangioblastoma: histogenesis of stromal cells

The histogenesis of stromal cells in capillary hemangioblastoma has been the subject of debate. The light and electron microscopic studies of hemangioblastomas presented here showed pericytic and leiomyoblastic features in stromal cells. Cells cultured by the monolayer method showed similar features to those of the original tumors. Immunohistochemical studies for glial fibrillary acidic protein and factor VIII/von Willebrand factor indicated that stromal cells were antigenically distinct from astrocytes and endothelial cells. These findings suggest that stromal cells are closely related to pericytes and smooth muscle cells, and support Rhodin's speculation that pericytes serve as a precursor to smooth muscle cells.

Selective localization of gamma-enolase in stromal cells of cerebellar hemangioblastomas

Three cases of cerebellar hemangioblastoma were studied using the immunoperoxidase technique to localize gamma-enolase, also known as neuron-specific enolase. The stromal cells demonstrated positive staining for gamma-enolase, while endothelial cells and pericytes showed no reactivity. Two vascular lesions, an angiosarcoma and a cutaneous angioma, were studied and found to be nonreactive for gamma-enolase. All tumors were also tested for factor VIII/von Willebrand factor, glial fibrillary acidic protein, and the S-100 protein. The lack of expression of gamma-enolase in endothelial cells of hemangioblastomas demonstrates a clear antigenic distinction from neighboring gamma-enolase-positive stromal cells. The significance of this finding and its implications for stromal cell histogenesis are discussed.

Expression of glial filament protein (GFP) in nerve sheaths and non-neural cells re-examined using monoclonal antibodies, with special emphasis on the co-expression of GFP and cytokeratins in epithelial cells of human salivary gland and pleomorphic adenomas

We describe two novel monoclonal antibodies specific for glial filament protein (GFP), i.e., GF12.23 and GF12.24 (both IgG2a]. These cross-react over a broad range of species with epitopes located in the alpha-helical rod domain typical of all intermediate filament (IF) proteins. These monoclonal antibodies were used, in conjunction with other monoclonal GFP antibodies, rabbit antiserum to GFP, and various antibodies to other cytoskeletal proteins, to examine the occurrence of GFP in cells outside of the central nervous system of rodents, cows, and humans. We detected some scattered GFP-containing cells in the neural sheaths in some species but not in others, and we obtained different results when comparing the rabbit antisera with the monoclonal GFP antibodies. In the enteric glia of rats, we observed GFP-positive cells with all of the antibodies used, whereas in human intestine, the various monoclonal antibodies showed no reaction with any intestinal cells. Similarly, no GFP was detected in surface cells of the lens of cows and rats using any of the GFP antibodies, whereas some reaction was seen in murine lens tissue. We were also unable to detect GFP-positive cells in human, bovine, or rat liver with any of the monoclonal antibodies, which is in contrast to the reactivity of the rabbit GFP antisera with some stellate perisinusoidal cells of rat but not bovine or human liver. The possible reasons for the discrepancies between the different species and the different antibody preparations used are discussed. In addition, using double-label immunofluorescence microscopy, we showed that normal human parotid glands contain a certain type of epithelial cell that co-expresses cytokeratins and desmosomal proteins with GFP. The histological distribution of these GFP-positive cells suggests that they represent a subset of the myoepithelial cells present in this tissue. Cells co-expressing cytokeratins and GFP - in some cases, apparently together with vimentin as the third IF protein present - were also identified in tumors derived from this salivary-gland epithelium, i.e., pleomorphic adenomas, in which GFP-positive cells were relatively frequent in the myxoid and chondroid components, thus confirming the work of other investigators. Possible implications for the concept of histogenesis of these tumor cells are discussed, as are possible mechanisms resulting in the co-expression of IF proteins.

Immunohistochemical demonstration of vimentin in human cerebral tumors

The distribution of vimentin (VIM) has been histochemically investigated in 53 cerebral tumors and compared in gliomas to that of glial fibrillary acidic protein (GFAP). In gliomas VIM is less positive than GFAP, but shows the same distribution. It cannot be considered as indicating immaturity of glial tumor cells. VIM is also positive in glial processes of cerebellar pilocytic astrocytomas, in Schwann cells of neurinomas and in endothelial cells of all oncotypes. In medulloblastomas, VIM decorates reactive glia cells. A diffuse positive reaction has been observed in meningiomas. In hemangioblastomas, besides intervascular and endothelial cells, groups of polygonal cells are intensely positive for both VIM and GFAP. The interpretation of VIM in cerebral tumors is largely based on the distribution patterns of this intermediate filament in the developing CNS of rodents.

Non-glial specificities of immunocytochemistry for the glial fibrillary acidic protein (GFAP). Triple expression of GFAP, vimentin and cytokeratins in papillary meningioma and metastasizing renal carcinoma

In an extensive immunocytochemistry study for glial fibrillary acidic protein (GFAP) of human neuropathological biopsy or autopsy tissue specimens examined for diagnostic or research purposes, rare non-glial specificities of the GFAP immunostain were observed: Schwann cells of some small nerves in salivary gland, renal capsule, and in epidural fat adjacent to a metastatic carcinoma, Schwann and satellite cells in a spinal ganglion invaded by tumor, chondrocytes of epiglottic cartilage, few cells of a malignant pleomorphic adenoma of salivary gland, most cells of a recurrent papillary meningioma with areas similar to the hemangiopericytic variant, and many cells of a renal carcinoma metastatic to brain; the primary renal tumor had been operated 4 years earlier and focally contained some GFAP-positive cells. To ascertain the specificity of such unexpected immunoreactivities for GFAP and to exclude possible crossreactivities with other intermediate filament (IF) proteins, a panel of different antibodies was used for immunocytochemistry with the peroxidase-antiperoxidase (polyclonal antisera) or labeled biotin-avidin (monoclonal antibodies) techniques: two monoclonal and four polyclonal anti-GFAP, three monoclonal and one polyclonal anti-cytokeratins (CK), and two monoclonal anti-vimentin (VIM) antibodies. Triple expression of GFAP, VIM and CK was found in the papillary meningioma (in patterns suggesting frequent co-localization), in the metastatic carcinoma (in patterns suggesting little co-localization), and in the pleomorphic adenoma (only few GFAP-positive cells). Co-expression of GFAP and VIM was seen in epiglottic chondrocytes and reactive astroglia; another metastatic carcinoma was labeled only for CKs. In the light of previous reports on non-glial specificities of the GFAP immunostain, and of the consistency of our immunostaining results obtained by all monospecific anti-GFAP antibodies as well as the lack of immunocytochemically evident crossreactivity with other IF proteins, authentic GFAP production by some rare non-glial tissues and tumors is suggested.

Applications of monoclonal antibodies in the diagnosis and treatment of primary brain tumors

The development of monoclonal antibodies has resulted in marked expansion in understanding the central nervous system (CNS). This has been especially true in the study of human neuroectodermal tumors where monoclonal antibodies have been used as physiological probes to define and characterize human neuroectodermal tumor-associated antigens. Utilizing monoclonal antibodies, neuroectodermal tumor-associated antigens have been described in four broad categories; biochemically defined markers, shared nervous system-lymphoid cell markers, shared neuroectodermal-oncofetal markers, and putative restricted tumor markers. Preliminary data have demonstrated the ability to localize animal and human tumors in vitro, ex vivo, and in vivo. Early application of monoclonal antibody technology to neuroimmunology and neuro-oncology has resulted in a new awareness of the complex relationships that exist within the CNS. Their specificity and reproducibility may provide the means to qualitatively and quantitatively define the phenotypic heterogeneity of human neuroectodermal tumors. Potentially, monoclonal antibodies, alone or as carriers of radionuclides, drugs, or toxins, may allow successful diagnosis and treatment of human neuroectodermal tumors.

Ultrastructure of cerebellar hemangioblastoma. Some new observations on the stromal cells

Five cases of hemangioblastoma were studied by electron microscopy with particular attention to the stromal cells and their surrounding structures. Most of the stromal cells of the tumor had abundant clear cytoplasm containing rough endoplasmic reticulum, mitochondria, scattered fibrils, and large lipid inclusions. They were usually aggregated without intervening cells. In the perivascular areas, their surfaces facing the perivascular collagen were surrounded by basal lamina. Their apposed cell membranes had occasional adhesive devices. Occasional, long, apparently cylindrical processes of the stromal cell cytoplasm were observed in some cases. These processes contained intermediate filaments of undetermined nature and microtubules. In the border zone between the tumor and the surrounding brain, the stromal cells were occasionally surrounded by narrow sheets of dark cell processes containing fibrils and glycogen granules, consistent with astrocytic processes. Altered neuronal elements were also observed in this area.

Immunohistochemistry of central nervous system tumors. Its contributions to neurosurgical diagnosis

Immunofluorescence and immunoperoxidase (peroxidase-antiperoxidase, PAP) techniques for the demonstration of neural and non-neural cell markers are contributing greatly to increase the diagnostic accuracy of difficult tumors of the central nervous system. Well characterized nervous system markers include glial fibrillary acidic (GFA) protein, the three protein subunits of neurofilaments, neuron-specific enolase (NSE), myelin basic protein, and S-100 protein. The most important and reliable of these is GFA protein, which is widely in use for the immunohistochemical diagnosis of tumors of the glioma group. Its many practical applications are reviewed and illustrated. Other neural markers, in particular the specificity of NSE and S-100 protein, need to be critically evaluated. Problems related to the immunohistochemical diagnosis of central neuroepithelial tumors of putative neuroblastic origin remain complex and still need to be resolved. Non-neural markers, such as vimentin, desmin, cytokeratins, Factor VIII, alpha-fetoprotein, human chorionic gonadotropin, and immunoglobulins have well defined, although more restricted, applications in surgical neuropathology.

Immunohistochemical study of fibronectin in hemangioblastomas and hemangiopericytomas

Eight hemangioblastomas and two hemangiopericytomas were studied using indirect immunoperoxidase stains for fibronectin (FN) and glial fibrillary acidic protein (GFAP) in formalin-fixed, paraffin-embedded surgical specimens. Stromal cells in hemangioblastomas were GFAP-negative and showed variable FN expression, while GFAP-positive cells were FN-negative, thus suggesting that the stromal cells are not derived from astrocytes. Hemangiopericytoma cells were poorly to intermediately FN-positive. The origin of stromal cells is discussed in the light of their fine structure and the immunohistochemical stains with other cell markers.

Distribution of endothelial and basement membrane markers in angiogenic tumors of the nervous system

The distribution of two endothelial cell markers Factor-VIII-related antigen and Ulex europaeus agglutinin was examined by immunoperoxidase and immunofluorescence techniques in paraffin-embedded specimens representing the three main types of angiogenic neoplasms of the nervous system, hemangioblastoma, hemangioendothelioma and hemangiopericytoma. In addition, the distribution of the basement membrane (BM) marker, laminin, was studied in the same tumors. It was found that Ulex europaeus agglutinin was a more sensitive marker of neoplastic endothelial cells than Factor-VIII-related antigen. Both markers only stained endothelial cells, while the tumor cells of hemangiopericytomas and the stromal cells of hemangioblastomas remained unstained. These findings do not support the view that the stromal cells of hemangioblastomas are derived from endothelial cells. With antiserum to laminin a typical staining pattern could be noticed in each tumor, showing the architectural relationships of the cells very clearly. In all three tumor types laminin was only found in the BM of the vessels, not in the interstices of the neoplastic cells outside vessel lumina. Therefore, the reticulin network previously found between the individual cells of hemangiopericytomas does not correspond to BM. It is concluded that both Ulex europaeus agglutinin and laminin antisera could be valuable new aids for the diagnosis of the three tumor types.

Glial and nonglial neoplasms evaluated on frozen section by double immunofluorescence for fibronectin and glial fibrillary acidic protein

Antisera against fibronectin (FN) and glial fibrillary acidic protein (GFAP) were used to combined on frozen sections of surgical biopsies to distinguish between astrocytoma and hemangioblastoma, meningioma, or sarcoma. The nonglial neoplasms contained FN in their parenchyma, whereas the gliomas contained GFAP. The immunofluorescent staining procedure takes 10 min. The immunoperoxidase staining for the permanent files takes 30 min. The stains have been used in this institute for more than 1 year, and 26 tumors have been evaluated.

Invited review. Nervous system antigens

Some of the better characterized proteins (markers) of the nervous system are described. The availability of specific antibodies to these markers has allowed the localization and assay of the proteins in tissue and biological fluids. There is some evidence that autosensitization may occur. Clinical application of these markers includes the evaluation and prognostic significance after stroke and head injury. The diagnostic histopathological use of the markers in the investigation of various tumours is summarized.