Immunohistochemical study of metastatic brain tumors with astroprotein (GFAP), a glia-specific protein

Transcription

1 J Neurosurg 62:414418, 1985 Immunohistochemical study of metastatic brain tumors with astroprotein (GFAP), a gliaspecific protein Tissue architecture and the origin of blood vessels TOSHIKI YOSHIMINE, M.D., YUKITAKA USHIO, M.D., TORU HAYAKAWA, M.D., HIROSHI HASEGAWA, M.D., NORIO ARITA, M.D., KAZt:O YAMADA, M.D., JAMSHID JAMSHIDI, M.D., AND HEITARO MOGAMI, M.D. Department of Neurosurgery, Osaka University Medical School, Osaka, Japan v Tissues from 12 metastatic tumors of the brain were studied immunohistochemically with an antiserum to a gliaspecific protein, astroprotein (glial fibrillary acidic protein, GFAP). Emphasis was laid on demonstrating the tissue architecture of metastatic lesions incorporating brainderived components (astrocytes and glial fibers). Of 12 samples, 11 manifested a number of irregular indentations at the tumor surface. These indentations, which contained astrocytic elements, extended into the tumor tissue in a tapering fashion. In seven cases, the deeper stromal portions of the tumor also contained astroprotein (GFAP)positive elements. The presence of this gilaspecific protein suggests that the stroma of the tumor tissue may in part be derived from preexisting brain tissue. This peculiar tissue architecture of the tumor supports the hypothesis that some of the blood vessels that are located in the stroma of the tumor tissue are also derived from the brain. These observations may be important in understanding the partial preservation of the bloodbrain barrier in metastatic brain tumors and the mode of growth of the metastatic lesion, and in selecting the type of chemotherapy that will be most effective in controlling this central nervous system complication of systemic malignancies. KEY WORDS 9 metastasis 9 brain tumor 9 astroprotein 9 glial fibrillary acidic protein 9 immunohistochemistry 9 bloodbrain barrier N EOPLASTIC cells derived from extraneural tissue may proliferate in the brain parenchyma and form metastatic nodules. Since the lesion is sharply demarcated and significant blood vessels feeding the tumor are not seen (at least on the macroscopic level), this type of tumor may be regarded incorrectly as an isolated tissue unit that grows independently within the brain. However, recent studies of capillary permeability have suggested that some of the blood, vessels in this type of tumor are of brain origin. 2'8 If this is in fact the case, and if some of the neoplastic development is dependent on the brain vessels within the tumor, j5'~8 those tumors can no longer be regarded as independent biological units. However, before this hypothesis can be accepted, further studies are needed to ascertain from an architectural basis how brain vessels can be included within the tumor. In the present study, we examined the relationship between brainderived and metastatic components within metastatic brain tumors to demonstrate basic structures of the tumor tissue where brain vessels may be intimately involved. A gliaspecific protein, astroprotein, L~2 or glial fibrillary acidic protein (GFAP) 5'7 was used as a marker for astrocytic elements that are apparently of brain origin. As astroprotein and GFAP have identical immunological characteristics, ~5 they will be referred to as "astroprotein (GFAP)" in this communication. Materials and Methods From among 3 consecutive patients with metastatic brain tumor, we selected 12 cases for the present study. Sites of the primary lesion and the metastasis are listed in Table 1. The secondary lesions were obtained and subjected to immunohistochemical study. These specimens contained a sufficient amount of surrounding brain tissue at the tumor margin to allow study of the interface between the tumor and the brain. One specimen (Case 3) was obtained at autopsy; all others were 414 J. Neurosurg. / Volume 62~March, 1985

2 GFAP study of metastatic brain tumors TABLE l Sites of prima v and metastatic lesions in this stud), Case Primary Metastasis No. Lesion 1 lung parietal 2 lung parietooccipital 3 lung frontal 4 lung parietooccipital 5 lung cerebeuar 6 lung parietal 7 lung cerebellar 8 lung parietal 9 breast cerebellar 1 breast occipital 11 breast parietal 12 esophagus cerebellar surgical specimens. Tissues were fixed in 1% buffered formalin and embedded in paraffin. Astroprotein (GFAP) was extracted from human glioma tissue and specific antiserum was raised in rabbits. These procedures and the immunological properties of the antiserum have been detailed elsewhere. 1~4 The modified peroxidaseantiperoxidase method of Sternberger 18 was applied to sections 6 u thick. Briefly, after treatment in methanol containing.3% hydrogen peroxide for 15 minutes, the sections were covered for 3 minutes with normal goat serum diluted 1:1 in.5 M Tris buffer (ph 7.6) containing.8% sodium chloride (Tris buffer). The sections were then exposed for 3 minutes to an antiserum reagent to astroprotein (GFAP) at 1:25 dilution. The tissues were rinsed in Tris buffer and placed for 3 minutes in goat antirabbit immunoglobulin G (IgG) at 1:2 dilution, then rinsed again and immersed for 3 minutes in a rabbit peroxidaseantiperoxidase complex at 1 : 1 dilution. The tissues were rinsed again and incubated for 5 to 8 minutes in Tris buffer containing.1% hydrogen peroxide and.5% 3,3'diaminobenzidine tetrahyclrochloride. The tissues were then washed and counterstained with Harris' hematoxylin. Control sections were treated with normal rabbit serum instead of specific antiserum at the same dilution. Three adjacent sections were processed for routine histological examination with hematoxylin and eosin (H & E), phosphotungstic acidhematoxylin (PTAH), and Laidraw's connective tissue (LCT) stains. Results Histological examinations confirmed secondary metastasis with various features. On immunohistochemical study, frank tumor cells were invariably negative for astroprotein (GFAP). However, there were a few positive cells and/or fibers in the tumor tissue; these will be described in detail. The specificity of the immunohistochemical method was confirmed by the absence of reaction products in the corresponding control sections. TABLE 2 lmmunohistochemical findings in 12 metastatic brain tumors* Case Surface Deep Tumor No. Indentations Stroma Cells~ * The presence and absence of astroprotein (GFAP)positive cells and/or fibers is indicated by or. respectively. The absence of the structure studied is indicated by. Islets of tumor cells in necrotic areas. Our immunohistochemical findings are summarized in Table 2. Examination of the tumor margins disclosed irregular surfaces with protrusions and indentations of various sizes and shapes. Protrusions of tumor tissue into the surrounding brain were noted in six cases. Indentations of the tumor surface were present in all but one case. Rather large indentations, or "recesses," were readily recognized on H & Estained sections (Fig. 1A). Immunoperoxidase staining disclosed that these recesses were filled with astroprotein (GFAP)positive cells and fibers (Fig. 1B). In those areas, the PTAH technique disclosed the presence of glial fibers; LCT stain was negative for connective tissue except for the walls of the blood vessels. Further examination of the tumor surface demonstrated still smaller recesses or "fissures" containing astroprotein (GFAP)positive elements (Fig. 1C). These fissures were not always recognized on H & Estained sections. Some PTAHstained sections did and others did not demonstrate glial fibers in these fissures. On immunoperoxidasetreated sections, the recesses or fissures could be shown to extend 1 to 2 mm into the tumor. In seven cases, astroprotein (GFAP)positive fibers were present in the deep stroreal structures within the tumor (Fig. 1D). Cell bodies positive for this gliaspecific protein (Fig. I E) were demonstrated in six cases. In three cases of papillary adenocarcinoma, astroprotein (GFAP)positive elements were seen in the surface indentations which formed the stem of the arborizing structure (Fig. 2 left), but not in the stroma of the papillary branches (Fig. 2 right). In eight cases, islands of tumor cells were present in necrotic areas. In three of these, astroprotein (GFAP) positive elements were seen in the central stroma of the islands (Fig. 3). Sections with LCT staining usually showed positive fibers; PTAH stain rarely disclosed glial fibers in those deeper stromal portions. The number of J. Neurosurg. / Volume 62/March,

3 T. Yoshimine, et al FiG. 1. Case 9. Photomicrographs of a cerebellar metastasis from an undifferentiated carcinoma of the breast. A: Section from the margin of the lesion showing irregular protrusions and indentations at the tumor surface. H & E, x 28. B: Immunoperoxidase preparation of a section adjacent to that shown in A. Gliosis of the surrounding brain tissue is seen. Astroprotein (GFAP)positive fibers and cells extend into numerous surface indentations (recesses). Antiastroprotein (GFAP), x 28. C: Highpower magnification of the same section as shown in B, disclosing still smaller indentations (fissures) containing astroprotein (GFAP)positive elements. Antiastroprotein (GFAP), x 7. D: Deeper portion of the tumor on the same section as shown in B. Stromal structures, including astroprotein (GFAP)positive fibers, are seen. Antiastroprotein (GFAP), x 14. E: The portion adjacent to that seen in D showing a stroma containing astroprotein (GFAP)positive fibers and cell bodies. Antiastroprotein (GFAP), x 14. astroprotein (GFAP)positive cells and fibers was somewhat higher in the t u m o r periphery than in the central tumor portion; however, it was not unusual for considerably deep portions to demonstrate positive elements. As the plane of section may vary from case to case, the depth of the relevant portion could not be determined. Discussion As astroprotein (GFAP) can be detected exclusively in astrocytes and some of the ependymal cells by immunoperoxidase methods, 2~22this protein may serve as a reliable marker for glial cells. Tumors originating in tissues outside the nervous system are thought not to 416 contain astroprotein (GFAP). Teratomas arising from extraneural organs represent a rare exceptionj 3'~9 The association of cells positive for this protein with the metastatic brain lesion has been reported by several authors who simply thought that these cells were reactive astrocytes of the brain trapped in the tumor. 4'6'~6 The distribution of those astrocytes and the implication of their presence in the composition of the tumor tissue were not studied in detail. Our study disclosed a number of "recesses" and "fissures" at the tumor surface; they projected into the tumor tissue and contained astrocytic elements. In the deeper portions of the tumor, the astrocytes were 1 J. Neurosurg. / Volume 62/March, 1985

4 GFAP study of metastatic brain tumors FIG. 2. Case 4. Photomicrographs of an occipitoparietal metastasis from a lung adenocarcinoma. Left: Papillary growth of the tumor and the surrounding brain tissue (upper left) is seen. H & E, 28. Right: Immunoperoxidase preparation of the section adjacent to that shown left. Astroprotein (GFAP)positive fibers are demonstrated in the stroma of the stem of the papillary structure but not in distal branches. Antiastroprotein (GFAP), 7. FIG. 3. Case 1. Immunoperoxidase section of the occipital metastasis from a breast adenocarcinoma. An island of tumor cells is noted in the central necrotic portion. Astroprorein (GFAP)positive cells and fibers are present in lhe central stroma and the stromal tissue entering it. Antiastroprotein (GFAP), 27. cated exclusively in the stroma. The tissue architecture of the tumor suggested that some of the stromal tissue was an extension of brain tissue in the surface indentations. The capillary endothelium of blood vessels in normal brain exhibits a distinctive ultrastructure and the vessels have limited permeability to various substances, z~ In metastatic brain tumors, the capillary ultrastructure exhibits abnormalities which may account for defects in the barrier function of the capillary endothelium, especially in the central portions of the tumors. ~~ Long ~2 reported that these capillaries resembled those of the original tissue rather than brain capillaries. However, autoradiographic studies with radioactive tracers in our and other laboratories demonstrated that experimental metastatic brain tumors less than 1 m m in J. Neurosurg. / Volume 62/March, 1985 diameter have an intact bloodbrain barrier, and in some of the larger tumors the barrier function is partially preserved at the periphery of the lesion. 2'8 During the early phase of tumor growth, a small nidus of metastatic cells may derive its blood supply from preexisting brain vessels with an intact bloodbrain barrier. 17 The situation may be similar at the periphery of larger tumors. We observed the distribution of glial elements in the stromal portion of the tumor; this suggests that a part of the stromal tissue may be of brain origin. As blood vessels are also located in the stroma of tumor tissue, our findings support the hypothesis that some of these blood vessels may be derived from the brain. Similar architectural characteristics were suggested in cases of medulloblastoma by Cox, 3 who studied the inclusion of astrocytes in primary brain tumors. Based on findings obtained with classical histological methods, he suggested that the brainderived astrocytes associated with the vascular system may form a stroma within the tumor. As astroprotein (GFAP) is a specific and sensitive marker for glial elements, and is absent in metastatic extraneural cells, the architectural characteristics were better appreciated in the metastatic tumors we subjected to immunohistochemical study. The incorporation of astrocytes and glial fibers was most c o m m o n at the periphery of the tumor. The presence of brainderived vessels may be restricted to areas within a few millimeters of the t u m o r surface. In an electron microscopic study, Long 12 discovered a few blood vessels in the peripheral portion of the metastatic brain tumor that exhibited evidence of barrier structures, while vessels in the more central portion showed no ultrastructural evidence for barrier function. Thus, the latter may be products of the t u m o r tissue. Three of our cases with metastatic papillary adenocarcinoma showed astrocytic inclusions only in the surface inden 417

5 T. Yoshimine, et al. tations and not in the deeper stroma. This type of tumor may be capable of developing its own stromal components and blood vessels. Alternatively, even brainderived vessels may undergo significant alterations in order to adapt to the new environment and thus lose their original endothelial ultrastructure. Malignancyrelated changes of the capillary endothelium have been well documented in cases with primary brain tumors. 9'~ We do not wish to underestimate observations of extensive destruction of the bloodbrain barrier either in experimental or clinical situations. 2"8"2~ Our study was intended to provide anatomical support to explain the partial preservation of the barrier function we observed in experimental metastatic tumors of the brain. We suggest the possibility that active growth at the periphery of this neoplasm may partially depend on the blood supply from preexisting cerebral vessels with intact or partially preserved capillary permeability characteristics. This kind of knowledge may be important in understanding the pathophysiology of these lesions and in devising chemotherapeutic control of this complication in the central nervous system. Acknowledgments We are indebted to Professor Takesada Mori and Dr. Kazayoshi Morimoto for their continued encouragement. References 1. Benda P, Mori T, Sweet WH: Demonstration of an astrocytespecific cerebroprotein by an immunofluorescence study of human brain tumors. J Neurosurg 33: , Blasberg RG, Groothuis D, Molnar P: Application of quantitative autoradiographic measurements in experimental brain tumor models. Semin Neurol 1:23221, Cox LB: The cytology of the glioma group; with special reference to the inclusion of cells derived from the invaded tissue. Am J Pathol 9:839898, Deck JHN, Eng LF, Bigbee J, et al: The role of glial fibrillary acidic protein in the diagnosis of central nervous system tumors. Acta Neuropathol 42:18319, Eng LF, Gerstl B, Vanderhaeghen J J: A study of proteins in old multiple sclerosis plaques. Trans Am Soc Neurochem 1:42, 197 (Abstract) 6. Eng LF, Rubinstein LJ: Contribution of immunohistochemistry to diagnostic problems of human cerebral tumors. J Histochem Cytochem 26:513522, Eng LF, Vanderhaeghen J J, Bignami A, et al: An acidic protein isolated from fibrous astrocytes. Brain Res 28: , Hasegawa H, Ushio Y, Hayakawa T, et al: Changes of the bloodbrain barrier in experimental metastatic brain tumors. J Neurosurg 59:3431, Hirano A, Matsui T: Vascular structures in brain tumors. Hum Pathoi 6:611621, Hirano A, Zimmerman HM: Fenestrated blood vessels in a metastatic renal carcinoma in the brain. Lab Invest 26:465468, Long DM: Capillary ultrastructure and the bloodbrain barrier in human malignant brain tumors. J Neurnsurg 32:127144, Long DM: Capillary ultrastructure in human metastatic brain tumors. J Neurosurg 51:5358, Ludwin SK, Eng LF, VandenBerg SR, et al: Glial differentiation in an experimental mouse teratoma. J Neuropathol Exp Neurol 35:12, 1976 (Abstract) 14. Mori T, Mogami H, Benda PF, et al: An astrocytespecific cerebroprotein in normal brain and in human gliomas. Nenrol Med Chit 1:1314, 1968 (Abstract) 15. Mori T, Morimoto K, Ushio Y, et al: [On the identity of astroprotein (Moil) and glial fibrillary acidic protein (Eng).] Igaku No Ayumi 92:1617, 1975 (Jpn) 16. Palfreyman JW, Thomas DGT, Ratcliffe JG, et al: Glial fibrillary acidic protein (GFAP). Purification from human fibrillary astrocytoma, development and validation of a radioimmunoassay for GFAPIike immunoactivity. J Neurol Sci 41:11113, Shapiro WR: Chemotherapy of metastatic central nervous system carcinoma, in Weiss L, Gilbert HA, Posner JB (eds): Brain Metastasis. Boston: GK Hall, 198, pp Sternberger LA: Immunoeytochemistry, ed 2. New York: John Wiley & Sons, 1979, pp VandenBerg SR, Ludwin SK, Herman MM, et al: In vitro astrocytic differentiation from embryoid bodies of an experimental mouse testicular teratoma. Am J Pathol 83:197212, Vick NA, Khandekar JD, Bigner DD: Chemotherapy of brain tumors. The "bloodbrain barrier" is not a factor. Arch Neuroi 34:523526, Yoshimine T, Moilmoto K, Homburger HA, el al: lmmunohistochemical localization of creatine kinase BBisoenzyme in human brain: comparison with tubulin and astroprotein. Brain Res 265:1118, Yoshimine T, Ushio Y, Hayakawa T, et al: Ependymal reaction to stab wounds in rat brains. Immunocytochemical study with antiserum to astroprotein. Neurol Med Chit 22:1923, 1982 Manuscript received July 13, Present address for Dr. Hasegawa: Department of Neurosurgery, Osaka Koseinenkin Hospital, Osaka, Japan. Address reprint requests to: Toshiki Yoshimine, M.D., Department of Neurosurgery, Osaka University Medical School, 115, Fukushima, Fukushima, Osaka 553, Japan. 418 J~ Neurosurg. / Volume 62/March, 1985