In vitro versus in vivo correlations of chemosensitivity of human medulloblastoma

An in vitro clonogenic assay was used to test the chemosensitivity of the human medulloblastoma cell line TE-671. Dose-response relationships for reduction in colony formation were generated for cyclophosphamide, vincristine, Adriamycin, 1,3-bis(2-chloroethyl)-1-nitrosourea (NSC 409962), and 1,4-cyclohexadiene-1,4-dicarbamic acid, 2,5-bis(1-aziridinyl)-3,6-dioxo-diethylester (NSC 182986); and the in vitro drug dose at which there is a 75% reduction in the number of colonies in comparison to controls (ID75S) were derived from these data. Methotrexate produced no colony reduction at any dose tested up to 1000 micrograms/ml. The in vitro results were compared to growth delays in s.c. TE-671 xenografts in athymic mice treated with the same agents. Agents with an ID75 less than assumed in vivo plasma drug concentrations were all active in vivo, whereas two of the three agents with an ID75 greater than assumed in vivo plasma drug concentrations demonstrated no in vivo activity. These results suggest that for these agents, the relationship between the ID75 of the drug and its in vivo concentration allows in vitro clonogenic assay results to agree with in vivo growth delay responses.

Treatment of human glioma and medulloblastoma tumor lines in athymic mice with diaziquone and diaziquone-based drug combinations

We used diaziquone (NSC 182986) alone and in combination with other antineoplastic drugs to treat six human glioma and one human medulloblastoma tumor lines growing s.c. in athymic mice. Pharmacokinetic studies of diaziquone in the plasma of athymic mice indicated rapid clearance with a half-life of approximately 11.5 min. Diaziquone produced significant growth delays in at least one experiment using each of seven different tumor lines, and it produced consistent and significant delays in five of the seven. There was no obvious difference between a single dose and a dose administered once daily for 5 days, and tumor regressions to a volume smaller than that at treatment were uncommon in any of the single-drug experiments. Using our most extensively characterized human glioma line, D-54 MG, we found striking enhancement of the therapeutic effect by using nontoxic combinations of either diaziquone and carmustine (1,3-bis(2-chloroethyl)-1-nitrosourea, NSC 409962) or diaziquone and procarbazine (NSC 77213). These combinations produced significant increases in the median growth delay, significant increases in the number of tumor regressions, and some instances in which no palpable tumors were present 100 days after treatment. In contrast, in experiments using diaziquone -based chemotherapy combinations with either cyclophosphamide, cis-platinum, or vincristine, there was only slight enhancement of the therapeutic effect. These results, using human glioma and medulloblastoma tumor lines in athymic mice, suggest a broad range of activity of diaziquone against primary nervous system tumors and enhancement of its therapeutic effect with either 1,3-bis(2-chloroethyl)-1-nitrosourea or procarbazine. If Phase II and Phase III clinical trials corroborate these findings, the value of the nude mouse system for the evaluation of new therapeutic approaches to brain neoplasms would be further confirmed.

Characterization of three restricted specificity monoclonal antibodies raised against the human glioma cell line D-54 MG

Monoclonal antibodies ( MCAs ) have been derived from a fusion of P3-NS1/1-Ag 4-1 (NS1) myeloma cells and splenocytes immunized to human glioma cell line D-54 MG. MCAs 2F3 , 4C7 , and 5B7 were analyzed by cell surface radioimmunoassay (CS-RIA), quantitative absorption, indirect immunofluorescence, and peroxidase-anti-peroxidase (PAP) immunohistology of unfixed tissue samples. MCA 2F3 exhibits the most highly restricted pattern of reactivity we have observed, reacting only with 5/12 glioblastoma cell lines and 1/4 fetal skin lines by CS-RIA, and to 9/11 glioblastoma tissue samples by PAP and absorption analysis; this MCA is totally nonreactive with melanomas, neuroblastomas, meningiomas, and control non-central nervous system tumors, and to adult and fetal tissues including brain, thymus, spleen, liver, lung, heart, gut, skin, and muscle by PAP analysis. MCAs 4C7 and 5B7 demonstrate neuroectodermal tumor cross-reactivity profiles, reacting with either melanomas ( 5B7 ) or melanomas and neuroblastomas ( 4C7 ); both are reactive with fetal skin, brain, and thymus of less than or equal to 16 weeks of gestational age. Other than this latter fetal antigen reactivity, these MCAs share the same negative reactivity profile described above for MCA 2F3 . Data from experiments using control or 0.02% EDTA-treated confluent cell monolayers of D-54 MG as antibody absorbents showed that the antigens detected are present in the extracellular matrix material remaining following cell removal. The data presented here establish that these highly restrictive anti-human glioma cell line MCAs are expressed in primary human gliomas; that the markers defined are developmental in nature, in that they are expressed by human fetal tissue, but not by adult tissue; and that in conjunction with previously characterized specificities, these markers of antigenic heterogeneity will be valuable in model system studies of therapeutic response heterogeneity.

Monoclonal antibody localization in subcutaneous and intracranial human glioma xenografts: paired-label and imaging analysis

The murine antiglioma monoclonal antibody 81C6 has been shown to specifically localize in U-251 MG and D-54 MG human glioma subcutaneous and intracranial athymic mouse xenografts expressing the human gliomamesenchymal extracellular matrix glycoprotein GMEM. In paired-label studies 81C6 reached peak levels of localization in subcutaneous and intracranial xenografts in 24 to 48 hours and persisted there for an additional 5 to 7 days before declining. The percent localized 81C6 in U-251 MG subcutaneous xenografts was tumor size dependent ranging from 2% in 200-300 mg tumors to 15% in 1 gram tumors. Subcutaneous xenografts of U-251 MG were readily radioimaged from 1 up to 6 days following administration of 131-I-81C6. The specificity of 81C6 localization has also been demonstrated by tissue autoradiography, elution of radiolabeled 81C6 from tumor xenografts, and in vivo inhibition of radiolabeled 81C6 localization by unlabeled 81C6. D-54 MG intracranial xenografts were permeable to 81C6 and control immunoglobulin but specific localization occurred only with 81C6. Radiolabeled 81C6 monoclonal antibody and U-251 MG and D-54 MG human glioma xenografts provide a useful operationally tumor specific monoclonal antibody tumor localization model for the examination of monoclonal antibody pharmacokinetics, radioimaging, radioimmunotherapy and combined modality therapy in intracranial human glioma xenografts.

The biology of malignant gliomas--a comprehensive survey

Malignant gliomas constitute a morphologically diverse group of neoplasms that respond poorly to therapy. The median survival of patients with glioblastoma multiforme, the most common malignant glioma, is only 11 months. Several factors influence the therapeutic resistance of malignant gliomas, including tumor heterogeneity, local invasion, and non-uniform vascular permeability to therapeutic agents. Malignant gliomas vary in their expression of specific marker proteins such as glial fibrillary acidic protein and fibronectin, in their expression of specific cell surface antigens, in their in vitro growth properties and tumorigenicity in athymic nude mice, and in their chromosomal composition and DNA content. This biochemical, immunochemical, biologic, and chromosomal heterogeneity is observed between different gliomas and within individual gliomas both in vivo and in vitro. Marked vascular hyperplasia, a characteristic feature of malignant gliomas, is induced by growth factors that may stimulate not only vascular cells but also glioma cells. There is also heterogeneity in the ultrastructure and the permeability of glioma blood vessels. In animal brain tumor models, vascular permeability varies not only between different models but between different tumors and between different areas of a single tumor. An understanding of the cell biology of malignant gliomas may identify the factors that determine therapeutic resistance and provide strategies to overcome these factors.

Blood-brain barrier disruption in immature Fischer 344 rats

Methods for transiently disrupting the blood-brain barrier (BBB) which are consistent with survival are described for immature Fischer 344 rats weighing 40 to 99 gm. A catheter was retrogradely inserted into the external carotid artery to the level of the bifurcation. Perfusion of 1.4 M mannitol or 1.6 M arabinose, at a rate of 0.01 to 0.1 ml/sec for 30 seconds, resulted in transient BBB disruption as measured by Evans blue dye (EBD) staining. Higher flow rates or perfusion with 10% to 30% dimethyl sulfoxide were associated with a mortality rate ranging from 0% to 44%. Perfusion with 0.9% sodium chloride or intrafemoral artery perfusion with 1.4 M mannitol did not disrupt the BBB. Optimum BBB disruption as measured by EBD staining was achieved with 1.6 M arabinose at 0.026 ml/sec for 30 seconds, at which time all of the 42 experimental animals had BBB disruption; all of the animals so treated survived 2 weeks following perfusion. This technique will allow the efficacy of delivering chemotherapeutic agents following BBB disruption to be tested in several of the more commonly used small-animal models for brain-tumor research.

Carotid artery injections in 40- to 99-g Fischer rats: technical note and evaluation of blood flow by various injection techniques

Despite extensive clinical use of brain tumor chemotherapy via the internal carotid artery (ICA), demonstration of efficacy and toxicity screening of ICA chemotherapy in brain tumor models has been limited. A method for performing ICA injections in 40- to 99-g Fischer rats is described, with documentation of its effect upon cerebral blood flow. A 33 gauge cannula was secured into a PE-10 catheter and, after ligation of the external carotid (ECA) and the pterygopalatine arteries, injections were made into the common carotid artery (CCA). The reliability of this method compared to CCA injection with and without ligation of the ECA was evaluated utilizing 15-micrometers 103Ru microspheres. With this technique, 68.3 +/- 11.19% of the microspheres lodged in the brain, compared to 28.9 +/- 21.3% to 32.7 +/- 19.8% for the other techniques (P less than 0.01). With this technique, regional perfusion of brain tumors can be done in the avian sarcoma virus rat glioma model, the D-54 MG human glioma-immunosuppressed rat transplantation model, and any of the large rat brain tumor models. The relevance of this method of experimental regional perfusion for the preclinical assessment of the efficacy and of the toxicity of chemotherapy and immunotherapy via the ICA is discussed.

Intracerebral transplantation of a human glioma line in immunosuppressed rats

A model was developed for in vivo study of the human glioma-derived D-54 MG cell line in the brains of immunosuppressed Fischer 344 rats. The rats were injected with horse anti-rat thymocyte serum before and after intracerebral inoculation with 5 or 10 microliters of a D-54 MG tumor cell suspension. Reproducible mortality distributions were obtained, with deaths occurring 18 to 34 days after intracerebral inoculation. Tumors grew as well circumscribed intracerebral masses with sheets of anaplastic cells, areas of necrosis bordered by concentrated nuclei, and minimal lymphocytic infiltration. Cytogenetic analysis revealed the same general chromosome distribution and markers in the heterotransplanted glioma cells as in the cultured line. Blood-brain barrier disruption was demonstrated by intracerebral tumor staining after intravenous injection of Evans blue dye. The in vivo growth of D-54 MG in immunosuppressed rats provides a reliable experimental model for the study of chemotherapy, immunodiagnosis, and immunotherapy of a human glioma-derived tumor in an animal sufficiently large to evaluate intracarotid or intratumoral injection of therapeutic agents.

In vivo measurement of regional brain tissue pH using positron emission tomography

Carbon-11-labeled dimethyloxazolidinedione ([11C]DMO) was injected intravenously into human subjects, and serial positron emission tomographic (PET) scans were obtained until brain-blood equilibration was achieved or could be accurately predicted from dynamic PET and 11C blood data. Knowledge of regional brain-blood partition coefficients for DMO, together with measurements of arterial blood hematocrit and pH, permitted the calculation of regional brain tissue and tumor pH (rpH). [11C]DMO PET rpH values were similar to rpH values derived from quantitative autoradiographic measurements of [14C]DMO concentrations in rat brain slices.

Chromosomal composition of four permanent cultured cell lines derived from human gliomas

Four permanent cell lines derived from malignant human gliomas were karyotyped using Giemsa-trypsin banding. D-65 MG had a stemline with 44 chromosomes, including 11 markers: 1p+, 2q-, 3p-, 3q+, 4p-, 9q-, 11q+, 15q-, 17p+, 21p+, 22q-. The net effect after accounting for fragments in markers was: +8, -10, -16 -X. D-32 MG had chromosome counts 90-91 without a distinct stem karyotype. Modal cells contained from 3 to 5 copies of the normal autosomes and 5 markers: 1q-, 3q-, 7q-, 13q-, 18q-. D-32 MGCl2 had a complex karyotype containing 78-82 chromosomes. There was no stemline, and modal cells varied from one another primarily in their set of marker chromosomes. A total of 23 markers were seen in this line, 17 of which were present in most modal cells. They were partially characterized as: 1q+, 1q+, 2q-, 5q+, 7p-, 7q-, 8p+, 8p+, 9p+, 12p+, 14q-, 16q+, 19q+, 19q+, a small submetacentric chromosome of undetermined origin and two small isochromosomes, i(Dp or Gp) and i(17p or 18p). A-172 MG had a modal peak of 77 chromosomes within which no two cells were exactly alike. Ten markers seen in modal cells were: 1p-, 4p+, 6p+, 6p+, 6q-, 7p-, 9p-q+, 13q+ 14p+, 22q+. There were no normal copies of chromosomes #1, #6, #9, #14. These four glioma-derived cell lines possess unique karyotypes, but each displays some combination of the numerical and structural deviations generally associated with established glioma lines.

A model for human medulloblastoma. Growth, morphology, and chromosomal analysis in vitro and in athymic mice

A model was developed for the in vitro and in vivo study of the continuous cell line, TE-671, derived from a human medulloblastoma. TE-671 grew in vitro as sheets of uniform triangular- to spindle-shaped cells with round to oval nuclei and sparse cytoplasm. The in vitro population doubling time was 31.48 hours (h). Mean colony forming efficiency in an agarose medium was 9.0 +/- 3.0%. TE-671 grew subcutaneously in athymic mice as lobulated masses composed of sheets of small uniform cells with round hyperchromatic nuclei and scanty cytoplasm. Perivascular pseudorosettes were commonly seen in early in vivo passage. The in vivo doubling time of subcutaneous tumors was 3.17 +/- 1.02 days, with a latency to 500 mm3 of 20.66 +/- 1.48 days. The cell cycle time (Tc) was 24 h, S phase was 14 h, G1 phase was seven h, and G2 phase was three h. Intracerebral tumors grew as discrete masses composed of diffuse sheets of small anaplastic cells with frequent mitoses and perivascular pseudorosettes. Both in vitro and in vivo chromosome studies revealed modal chromosome counts in the near tetraploid range with the same marker chromosomes as described in the original report of this cell line. The in vitro and in vivo growth of TE-671 provides the best available experimental model for the analysis of the biological properties and chemosensitivity of human medulloblastoma.

Treatment of five subcutaneous human glioma tumor lines in athymic mice with carmustine, procarbazine, and mithramycin

Five human glioma tumor lines were transplanted subcutaneously in athymic nude mice. Three of the lines (D-54 MG, U-118 MG, and U-251 MG) were derived from permanent human glioma cell lines, while the other two lines (N-456 and N-519) were established as direct xenografts in mice. Growth rates varied among the lines, but all were treated at a common average tumor volume of 200-300 mm3. Three drugs of known clinical activity against human anaplastic glial tumors were tested in this system. Procarbazine (PCB) produced significant responses in all five tumor lines, carmustine produced significant growth delays in two, and mithramycin produced a slight growth delay in only one. A carmustine-PCB combination produced a greater therapeutic effect than either agent alone against D-54 MG, but the combination was no more effective than PCB alone against U-118 MG. These results parallel clinical experience and indicate that this is a useful model for testing agents of potential activity in brain tumor therapy.

Immunobiology of primary intracranial tumors. Part 7: Active immunization of patients with anaplastic human glioma cells: a pilot study

Twenty patients with malignant gliomas were selected for active immunization within 4 weeks following surgery. Each patient had a Karnofsky Functional Rating equal to or greater than 70, a peripheral blood lymphocyte count equal to or greater than 1000 cells/cu mm, skin test responses to one or more of four recall antigens, peripheral blood T-cells equal to or greater than half that of control, and was not receiving steroid therapy at the time of entry into the study. Each patient received subcutaneous inoculations with one of two human glioma tissue culture cell lines (D-54MG or U-251MG) monthly, with 500 micrograms of bacillus Calmette-Guérin cell wall (BCG-CW) being included with the first inoculation. Each patient also received levamisole, 2.5 mg/kg 3 days per week every other week. Radiotherapy and chemotherapy with BCNU were begun after the first month of immunization. Follow-up evaluations included computerized tomography brain scans, neurological examinations, Karnofsky Functional Ratings, and studies of general immune competence. No evidence of allergic encephalomyelitis was noted clinically, nor was any gross or microscopic evidence of such pathology obtained upon autopsy of three of these patients. Serial studies of general immune competence showed no alterations from those previously described with non-immunized patients. Patients who were inoculated with the U-251MG cell line have had a longer survival time compared to those inoculated with the D-54MG cell line (p less than 0.0590) or compared to 58 historical cases of glioma patients treated with levamisole, radiation therapy, and chemotherapy alone (p less than 0.02).

Immunobiology of primary intracranial tumors. Part 8: Serological responses to active immunization of patients with anaplastic gliomas

Serial serological studies were carried out on 19 of 20 patients with malignant gliomas who were actively immunized with one of two human glioma tissue culture cell lines (D-54MG or U-251MG). Most patients mounted a significant serum reaction to histocompatibility antigens (HLA's), as well as an antibody response to fetal bovine serum (FBS) which was added to the glioma-cell inoculum. These two sources of antibody accounted for greater than 90% of the antibody induced by these inoculations. Two patients continued to have significant amounts of binding antibody to the original immunizing cell line following exhaustive absorptions of FBS and these two had all remaining significant antibody removed by further absorption of the serum against the 2-T osteogenic sarcoma tissue culture cell line known to possess antigens cross-reactive with human gliomas. One single patient continued to show significant antibody binding to the original glioma cell line following absorption against FBS, human platelets, and the 2-T cell line, and therefore seems to have produced glioma-distinctive antibodies in response to immunization. The antibody preparation from this patient was also cytotoxic against the original glioma cell line, as well as another recently cultured human glioblastoma cell line. The significance of these serological studies is discussed as it relates to immunological responses patients with gliomas may make to active immunization.

Regional measurements of blood-to-tissue transport in experimental RG-2 rat gliomas

Regional measurements of blood-to-tissue transport were performed in transplanted RG-2 rat gliomas using [alpha- 14C]aminoisobutyric acid (AIB), quantitative autoradiography, and equations to express a unidirectional transfer constant. Thirty-eight intracranial tumors in ten rats were analyzed according to location; 23 intraparenchymal tumors, eight meningeal tumors, six fourth-ventricular tumors, and one third-ventricular tumor were studied. Except for the small third-ventricular tumor, the transfer constant (K) for AIB was similar in all groups and ranged from 0.031 to 0.038 ml/g/min. Within individual tumors, regional variation of K was also small, although some local variation could be correlated with histological features. The K for AIB decreased in brain around tumor and, at a distance of 300 microns from tumor edge, had returned to values similar to those of normal cortex (0.002 ml/g/min). An average extraction fraction (E) of 0.09 was calculated for AIB in the RG-2 tumors. The low E suggests that delivery of water-soluble chemotherapeutic drugs to RG-2 tumors should be limited more by capillary permeability or surface area than by blood flow. RG-2 is an ideal experimental tumor with which to test drug delivery and the methods that attempt to increase drug delivery in brain tumors.

Regional measurements of blood flow in experimental RG-2 rat gliomas

Regional measurements of blood flow (F) were performed in transplanted intracerebral RG-2 rat gliomas using [14C]iodoantipyrine, Kety-Schmidt blood flow equations, and quantitative autoradiography. Twenty-nine intracranial tumors in ten rats were analyzed by location; 18 intraparenchymal, seven meningeal, two third-ventricular, and two fourth-ventricular tumors were studied. For all tumors, averaged mean F was 91 +/- 33 (S.D.) ml/hg/min. In all but one tumor, mean F was intermediate between normal cortex and corpus callosum values. There was moderate regional variation: averaged mean F was lower in tumor center (78 +/- 47 ml/hg/min) than in tumor periphery (93 +/- 30 ml/hg/min). Within individual tumors, F showed moderate variation which correlated to some extent with histological features; a regional F of less than 10 ml/hg/min was observed in only one tumor within an area of necrosis. F in regions of brain immediately surrounding the tumor was higher than in tumor periphery. Blood flow to RG-2 tumors seems unlikely to limit drug delivery any more than to normal brain, and the consistent levels from tumor to tumor and within individual tumors make the RG-2 model an excellent one with which to study drug delivery in experimental brain tumors.

Chemotherapy and radiation therapy of human medulloblastoma in athymic nude mice

The human medulloblastoma cell line TE-671 was grown s.c. and intracranially in athymic nude mice. Tumor-bearing animals treated with chemotherapeutic agents or radiation were compared to untreated tumor-bearing controls. Tumors growing s.c. were sensitive to cyclophosphamide and vincristine with growth delays in duplicate trials of 15.8/16.5 and 12.9/15.0 days, respectively. These tumors were minimally responsive to the 2,5-bis(1-aziridinyl-3,6-dioxodiethyl ester of 1,4-cyclohexadiene-1,4-dicarbamic acid (NSC 182986) and cis-diamminedichloroplatinum II and unresponsive to methotrexate, 2,4-diamino-6-(2,5-dimethoxybenzyl)-5-methylpyrido(2,3-d)pyrimidine (NSC 351521), 1,3-bis(2-chloroethyl)-1-nitrosourea (NSC 409962), and procarbazine. Radiation therapy with 2500 or 1500 rads as a single fraction produced a marked response, with growth delays of 39.5 and 21.1 days, respectively. Cyclophosphamide produced a significant (p less than 0.0005) increase in the median survival of mice with intracranial tumors. Vincristine produced a minimal increase in the median survival while no response was seen to the 2,5-bis(1-aziridinyl-3,6-dioxodiethyl ester of (1,4-cyclohexadiene-1,4-dicarbamic acid at the dose level and schedule tested. This model system will allow further analysis of the therapeutic sensitivity of human medulloblastoma to other agents or combined-modality regimens.

Demonstration of complex antigenic heterogeneity in a human glioma cell line and eight derived clones by specific monoclonal antibodies

We have investigated the antigenic heterogeneity of human glioma cells and its correlation with other parameters of tumor cell heterogeneity (karyotype, 2':3' cyclic nucleotide 3'-phosphohydrolase expression, in vitro morphology) using the established human glioma cell line D-54 MG and eight single-cell-derived clones. The panel of antibodies used included 3 previously described heterologous sera raised against human gliomas and lamb oligodendroglia and 10 monoclonal antibodies with demonstrated reactivity for tumors of neuroectodermal origin, human fetal tissue, or human Thy-1. Antigen expression was determined by cell surface radioimmunoassay and peroxidase-antiperoxidase immunohistology. The use of a monoclonal antibody panel composed of ten reagents of varied specificity resulted in the demonstration of highly variable and complex antigenic patterns on the cell surfaces of cloned subpopulations of the human glioma cell line D-54 MG. Only one antigen, human Thy-1, was present on the parent line and all clones; the remaining nine antigens exhibited a distribution unrelated to other predictive parameters of genotypic or phenotypic heterogeneity such as karyotype, 2':3' cyclic nucleotide 3'-phosphohydrolase expression, or in vitro morphology. With the exception of clones 3 and 4, which shared a common antigen profile but exhibited distinctly different in vitro morphological patterns, the detected antigenic profile of each clone was distinct, with the proportion of expressed antigens ranging from 2 of 10 (clone 2) to 10 of 10 (clone 1). The demonstration of distinct, selectively maintained cell subpopulations within a human glioma cell line has direct implications for immunotherapeutic designs.

Monoclonal antibodies reactive with epitopes restricted to glial fibrillary acidic proteins of several species

The highly reproducible histochemical localization of glial fibrillary acidic protein (GFAP)+ qualifies it as an important marker of astrocytes in both research and clinical applications. The primary objective of this study was to produce monoclonal antibodies having the advantage of invariant specificity, affinity, and titer to GFAP-specific epitopes of wide species distribution. We report here the characterization of four monoclonal antibodies that recognize the same or spatially close epitopes specific to GFAP. The epitope(s) detected has been phylogenetically conserved; human, bovine, ovine, canine, porcine, rabbit, guinea pig, rat, murine, and chicken brain homogenates all specifically absorb monoclonal antibody activity. Of importance to the routine application of these new anti-GFAP monoclonal antibodies is the demonstration here of the stability of the antigen-antibody interaction in normal, reactive, and neoplastic astrocytes of both rat and human origin following various methods of fixation.

Intracerebral growth of a human glioma tumor line in athymic mice and treatment with procarbazine, 1,3-bis(2-chloroethyl)-1-nitrosourea, aziridinylbenzoquinone, and cis-platinum

Tumors derived from the established human glioma cell line D-54 MG were transplanted intracerebrally into athymic "nude" mice. All control mice developed tumors in the brain after the transplantation of 10(5) cells and died within 30 days. Tumor was purely intracerebral in the presymptomatic stages, including the time of drug treatment. The injection of serial cell dilutions reduced the tumor incidence and prolonged survival. Procarbazine (PCB), 1,3-bis(2-chloroethyl)-1-nitrosourea, diaziquone (aziridinylbenzoquinone), and cis-platinum each produced statistically significant increases in survival when administered on Day 10 after transplantation, and PCB alone cured most of the animals. These results extend the characterization of the human glioma line D-54 MG and confirm the value of the athymic mouse for the testing of chemotherapeutic agents of interest in brain tumor therapy.

Human glioma-mesenchymal extracellular matrix antigen defined by monoclonal antibody

The distribution and localization of a glioma-associated antigen defined by monoclonal antibody 81C6 has been examined using human cultured cell lines and tissues. Monoclonal antibody 81C6 was selected from a hybridoma fusion of spleen cells of mice immunized with the glial fibrillary acidic protein-positive human glioma cell line U-251 MG. Results of cell surface radioimmunoassay and absorption analysis demonstrated that 81C6 defined a glioma-mesenchymal extracellular matrix (GMEM) antigen expressed by 14 of 16 gliomas, 1 of 3 neuroblastomas, 1 of 7 melanomas, 2 of 6 sarcoma cell lines, and 8 of 9 cultured fibroblast lines. GMEM was not expressed by carcinoma or by the myeloid-lymphoid cell lines examined. Within the central nervous system, GMEM was expressed in 10 of 11 glioblastomas but was undetected in 5 of 6 astrocytomas and in normal adult and fetal brain by peroxidase-antiperoxidase immunohistology. In glioblastomas, the GMEM antigen was localized to basement membranes of the distinctive glomeruloid endothelial proliferations and hyperplastic blood vessels. The GMEM antigen was also expressed in 3 of 3 glioblastoma cell lines and 6 of 8 glioblastoma biopsy xenografts in athymic nude mice. Among non-central nervous system tissues and tumors, GMEM was found by peroxidase-antiperoxidase immunohistology in normal liver sinusoids, spleen red pulp sinusoids, kidney medullary tubule interstitium, and glomerular mesangium and in association with vascular and stromal elements of several undifferentiated tumors. The GMEM antigen is distinct from previously described forms of fibronectin, laminin, collagen types I to V, hyaluronic acid, chondroitin sulfate, and heparin, as determined by absorption analysis and immunohistological localization in tissues. The expression of GMEM in glioblastoma but not normal brain, association with glioblastoma-proliferative endothelium basement membranes, and expression in glioblastoma cell lines and nude mouse xenografts suggest that GMEM may be a useful marker of gliomas in vivo and in vitro.

Growth, morphology, and serial transplantation of anaplastic human gliomas in athymic mice

Sixteen of 17 anaplastic human gliomas (AHGs) transplanted into athymic mice produced progressively growing subcutaneous nodules at the site of implantation. Thirty-four of 68 animals (50%) receiving transplanted tissue developed 500 m3 tumors in 24 to 364 days. Fourteen AHG were passed to a second animal generation, and 11 showed continued growth. Eight of these were serially passed, with one reaching a sixth animal generation, four reaching a fifth, and a three third. Once growth occurred in a second animal generation, no AHGs were lost in subsequent passages because of failure to grow. Of 234 animals receiving tumors beyond the first animal generation, 189 (80.8%) developed tumors. Average doubling times of the exponentially growing tumors in serial passage ranged from 3.0 to 19.1 days. This growth rate tended to increase and stabilize in early animal passages. The tumors growing in animals contained cell types which were present in the original human tumors, including fibrillary and protoplasmic astrocytes, small anaplastic cells, gemistocytes, anaplastic spindle cells, and multinucleate giant cells. Glial fibrillary acidic protein (GFAP) was detected in 15 of 17 biopsy specimens and in 12 of 14 AHGs in animals. These data illustrate the value of the athymic mouse system for the investigation of human brain tumors by demonstrating a high rate of successful transplantation, quantitative growth data on serially passed tumors, and morphological and immunochemical resemblance of the tumors in mice to the original human tumors.

Spatial distribution of proliferating cells in avian sarcoma virus-induced gliomas

We studied the regional distribution of proliferating tumor cells in five avian sarcoma virus-induced gliomas. The labeling index and spatial distribution of [3H]thymidine (dThd)-labeled tumor cells were determined in serial sections of each tumor with a computer-assisted digitizing system. The density of [3H]dThd-labeled cells showed marked regional variation in each tumor, and the ratio of the density of [3H]dThd-labeled cells in tumor periphery to tumor center varied from 0.86 to 1.38. The labeling index generally, but not always, reflected [3H]dThd-labeled cell density. This study indicates that proliferating pools of glioma tumor cells exhibit regional variability in concentration and that the highest numbers of proliferating cells may be predominantly located in central regions of tumor and not in tumor periphery as assumed previously. In all tumors, large numbers of proliferating cells were present in all parts of the tumor.

Study of glial fibrillary acidic protein in a human glioma cell line grown in culture and as a solid tumor

A continuous human glioma cell line grown in culture and as a solid tumor was analyzed for glial fibrillary acidic (GFA) protein. This material provided a rich source for GFA protein that could also be manipulated and controlled. Immunoperoxidase staining at the light and electron microscopic levels revealed that the cell culture and tumor specimens were strongly positive for GFA protein. When aqueous soluble fractions of the cell culture and tumor were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, electroblotted onto nitrocellulose, and stained immunochemically, they contained exclusively low molecular weight (41--43K-dalton) GFA peptides. SDS (0.15%)-soluble fractions contained either low molecular weight only (culture) or a mixture of peptides ranging from 41 to 49K daltons. SDS (1%) extracts of either cell culture or tumor contained only 49K-dalton GFA protein. Two-dimensional gel separation revealed that the GFA protein extracted from either the culture or tumor with 1% SDS resolved to two or three spots at pH 5.8. Low molecular weight GFA peptides (less than 49K daltons) in aqueous and 0.15% SDS-soluble extracts became increasingly more acidic with decreasing molecular weight. The extremely rapid degradation seen suggests that this cell line may be a valuable system for further study of intermediate filament protein turnover.