Effect of pharmacologic doses of zinc on the therapeutic index of brain tumor chemotherapy with carmustine.

To evaluate the potential differential effect of pretreatment with pharmacologic doses of the trace element zinc on the chemosensitivity of glioma cells and bone marrow cells for carmustine (BCNU), we performed in vitro and in vivo studies of zinc toxicity as well as of the combined treatment with zinc and the anticancer drug. We studied the in vitro effects on established human and rat glioma cell lines using a microcolorimetric growth assay and on murine bone marrow using a clonogenic assay for committed progenitor cells of the granulocyte-monocyte lineage. Zinc exposures of up to 100 microM for 120 h did not influence the growth of six of seven human glioma cell lines. Only U87MG demonstrated statistically significant toxicity during high zinc exposure (100 microM over 120 h). Dose-response growth curves generated for BCNU did not show protection against the anticancer agents by a 48-h pretreatment with different zinc concentrations. The clonogenic capacity of bone marrow cells was slightly reduced by in vitro culture for 24 and 48 h. Although this effect appeared to be more prominent in the presence of zinc supplementation, overall a statistically significant inhibition was seen only after exposure to a concentration of 100 microM zinc over 48 h. As compared with chemotherapy alone, in vitro pretreatment with 50 microM zinc over 48 h followed by chemotherapy resulted in an increased number of colony-forming unit-granulocyte monocyte (CFU-GM): CFU-GM increased by a factor of 2 for BCNU (60 microM x 2 h). This statistically significant in vitro chemoprotection would translate into a dose-protection factor of 1.5, i.e., for the same level of myelosuppression, zinc pretreatment would allow administration of a 50% increased dose of BCNU. The in vivo studies were performed in an s.c. xenograft model of the human glioma cell line U87MG in athymic mice. The maximal tolerable pretreatment with zinc was determined to be a 10-day course of daily i.p. injections of 10 mg/kg ZnCl2. The subsequent i.p. administration of the dose lethal to 10% of the mice (LD10) and of a 1.5 x LD10 dose of BCNU resulted in less bone marrow toxicity in pretreated animals than in non-zinc-pretreated mice as determined in a CFU-GM assay. Glioma colony-forming efficiency (CFE) assays, on the other hand, did not show any zinc-related difference in the BCNU sensitivity of U87MG.(ABSTRACT TRUNCATED AT 400 WORDS)

Experimental basis for increasing the therapeutic index of cis-diamminedicarboxylatocyclobutaneplatinum(II) in brain tumor therapy by a high-zinc diet.

Metallothionein (MT), a ubiquitous intracellular protein, confers resistance to the toxic effects of platinum compounds. Since a high-zinc diet has been shown to induce MT synthesis in extracerebral tissues but not in brain, we investigated whether it could provide an experimental basis for decreasing the hematotoxicity of carboplatin without impairing its activity against brain tumors. After 2 weeks on either a high-zinc diet or a control diet (zinc content, 180 vs 10 ppm), mice and rats received various doses of carboplatin or Hanks' balanced salt solution by i.p. injection. The hematotoxicity of carboplatin was evaluated with an assay of colony-forming units of granulocytes and mononuclear cells in mice. The high-zinc diet enabled a 50% increase in the carboplatin dose without increasing hematotoxicity. The antitumor activity was evaluated with an assay of the colony-forming efficiency of gliosarcoma cells from 9L brain tumors in rats. The high-zinc diet did not alter the efficacy of carboplatin against this brain tumor. Northern blot analysis confirmed that the high-zinc diet induced MT mRNA in the kidney but not in the brain of mice and rats; it also showed MT mRNA induction in bone marrow cells of mice but not in rat 9L brain tumors. These results suggest that increasing the dietary intake of zinc might increase the therapeutic index of carboplatin in the treatment of brain tumors.

Comparison of the cytotoxic activities of cisplatin and carboplatin against glioma cell lines at pharmacologically relevant drug exposures.

Carboplatin has lower nephro- and neurotoxicities and better penetration into brain tissue than cisplatin. If carboplatin has comparable cytotoxicity against glioma cells, it might have a therapeutic advantage in the treatment of malignant gliomas. Using an assay of colony-forming efficiency, we compared the cytotoxicity of these two drugs in human glioma cell lines SF-126, SF-188, U87-MG, and U251-MG. The experiments were designed so that the product of in vitro drug concentration (C) and time (T) would encompass the same range of values as the C x T of the ultrafilterable platinum plasma fraction as determined by pharmacokinetic studies in man. The in vitro stability of the drugs was evaluated by measuring the cytotoxicity of aged drugs with a microculture tetrazolium assay. Cisplatin and carboplatin were both stable during the 2-h treatment. The cytotoxic activities of these drugs at clinically achievable levels of drug exposure were of the same order of magnitude. These results, in conjunction with the lower nephro- and neurotoxicities of carboplatin, the higher platinum levels in brain tissue after treatment with carboplatin, and the encouraging results of carboplatin in the clinical treatment of brain tumors that have been demonstrated in other studies, suggest that carboplatin might be preferable to cisplatin in the treatment of patients with malignant glioma.

Experimental basis for increasing the therapeutic index of carboplatin in brain tumor therapy by pretreatment with WR compounds.

We tested an experimental strategy to decrease the dose-limiting hematotoxicity of carboplatin without compromising its activity against brain tumors. The effect of pretreatment with WR-1065, a chemomodifier that penetrates brain poorly, on carboplatin's cytotoxicity was evaluated in human hematopoietic granulocyte-monocyte progenitor cells and in three human glioblastoma cell lines. WR-1065 reduced bone marrow toxicity without decreasing carboplatin's activity against glioblastoma cells. These results suggest that the therapeutic index of carboplatin might be increased in the treatment of malignant brain tumors.

Laboratory studies of berberine used alone and in combination with 1,3-bis(2-chloroethyl)-1-nitrosourea to treat malignant brain tumors.

Berberine was evaluated for antitumor activity against malignant brain tumors. In addition, studies on combination of berberine with 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) were done. Several experimental approaches were used. In vitro studies were performed on a series of 6 human malignant brain tumor cell lines and rat 9L brain tumor cells (gliosarcoma) by using a colony forming efficiency (CFE) assay. 9L was also evaluated by a sister chromatid exchange (SCE) assay. In addition, in vivo treatment of intracerebral 9L solid brain tumors was analyzed by CFE assay. Berberine used alone at a dose of 150 micrograms/ml showed an average of 91% cell kill (1.08 log kill) for the 6 malignant brain tumor cell lines. On the average, BCNU alone at a dose of 23 microM gave a 43% cell kill (0.24 log kill). Treatment with a combination of berberine and BCNU at 23 microM showed additive effects with an average of 97% cell kill (1.55 log kill). The relative number of SCEs for 9L cells was increased 2.7 times over background following in vitro treatment with 150 micrograms/ml berberine. Following in vivo treatment of animals harboring solid 9L brain tumors with 10 mg/kg of berberine, an 80.9% cell kill (0.69 log kill) was noted. This activity is equivalent to treatment with 1/3 LD10 dose of BCNU (4.44 mg/kg). In vivo combination treatment with berberine and BCNU showed additive cytotoxicity. Using a BCNU-resistant 9L subline (9L-2), treatment with berberine in combination with BCNU also demonstrated additive cytotoxicity. In conclusion, our results indicate that berberine has potent antitumor activity against human and rat malignant brain tumors.(ABSTRACT TRUNCATED AT 250 WORDS)

Partial characterization of a soluble mitogenic factor from medulloblastoma.

To determine how medulloblastoma cells might influence the proliferation and phenotype of normal stromal cells, normal human leptomeningeal cells were treated in culture with medulloblastoma-conditioned medium; their ability to incorporate tritiated thymidine and synthesize collagen was measured. The treated leptomeningeal cells had a significantly greater uptake of tritiated thymidine and grew faster than control leptomeningeal cells. Immunofluorescence studies demonstrated a greater intensity of staining for procollagen type III in the cell layer of the treated cultures than in control cultures; diethylaminoethyl (DEAE)-cellulose chromatography of the medium showed that the treated cells synthesized predominantly type III collagen, whereas control cells synthesized type I collagen. Analysis of the medulloblastoma-conditioned medium revealed that the soluble factor responsible for these effects in an acid- and heat-stable protein. The increased proliferation and altered collagen synthesis induced in leptomeningeal cell cultures by a soluble factor from a medulloblastoma are examples of how tumor and stromal elements interact, and may be related to the process of desmoplasia often observed in medulloblastomas in vivo.

Comparison of in vitro cloning assays for drug sensitivity testing of human brain tumours.

Three in vitro clonogenic assays were used to determine the sensitivity of an established human glioblastoma cell line (U251-MG) to five chemotherapeutic agents. The colony-forming efficiency of untreated culture was 0.695 +/- 0.170 in a monolayer assay with irradiated feeder cells, 0.018 +/- 0.006 in a low-O2 agar assay, and 0.049 +/- 0.021 in a two-layer agar system with nutrient-enriched medium (p less than 0.001). Comparison of the slope of the regression line for the dose-response curve and the interpolated ID90 for each drug showed that U251-MG was equally sensitive to aziridinylbenzoquinone and dianhydrogalactitol in all three assays. The sensitivity of this cell line to 1,3-bis-(2-chloroethyl)-1-nitrosourea (BCNU), cis-dichlorodiammineplatinum (II) (CDDP) and 9-hydroxy-2-N-methylellipticine (HME), however, varied depending on the assay used. In no instance did U251-MG show greater sensitivity (lower ID90 or steeper slope) in the low-O2 agar assay than in the other assays. BCNU and CDDP were least active in the monolayer assay, whereas HME showed both the lowest ID90 and steepest slope using this technique. We conclude that different in vitro tumour clonogenic assays show different colony-forming efficiencies for the same cell line and may show different responses to certain drugs. Identification of accurate predictive models of drug sensitivity will require correlative in vivo and in vitro studies.

Growth of a medulloblastoma on normal leptomeningeal cells in culture: interaction of tumor cells and normal cells.

Well-characterized medulloblastoma cells growing in suspension were placed on top of a confluent monolayer of leptomeningeal cells. In contrast to cells placed on plastic alone, which did not grow or attach, the medulloblastoma cells attached readily to the leptomeningeal cells and grew to form enlarging spheroids. The growth of these spheroids was supported with minimal essential medium containing 10% fetal calf serum or with human cerebrospinal fluid. Medulloblastoma cells grown on plastic remained viable for 7 to 10 days, whereas those grown on a monolayer of leptomeningeal cells remained viable for 40 days. Electron microscopy demonstrated increased interdigitation of the plasma membrane at the sites of contact between leptomeningeal and medulloblastoma cells, the deposition of fine, basement membrane-like material between the two cell types, and an increased number of cytoskeletal filaments in the leptomeningeal cells. We conclude that medulloblastoma cells can be grown on a leptomeningeal monolayer. This in vitro system may be useful in studying the mechanisms by which medulloblastoma cells attach to leptomeningeal elements and grow after dissemination in cerebrospinal fluid.

Establishment and characterization of a cell line from a human gliosarcoma.

A human gliosarcoma culture was characterized from the time of inception to the time of establishment of the cell line (SF-539 BT). Immunohistochemical analysis of the original tumor showed 2 distinct regions of cells. The gliomatous regions were identified by immunostains for glial fibrillary acidic protein and the sarcomatous regions by immunostains for laminin, collagen type IV, procollagen type III, and fibronectin. In early-passage culture, both types of cells maintained their characteristic immunohistochemical profiles; however, after the fourth subcultivation in monolayer culture, no cells expressing glial fibrillary acidic protein could be identified. All cells had become morphologically uniform and expressed laminin, collagen type IV, procollagen type III, and fibronectin only. The immunostaining profile of clones grown in soft agar was similar to that of cells in monolayer culture. At establishment, SF-539 BT has a saturation density of 1.3 X 10(6) cells/25 sq cm, a doubling time of 32 h, and a plating efficiency of 22% in monolayer culture. The tumor cell line is resistant to 1,3-bis(2-chloroethyl)-1-nitrosourea, has an abnormal karyotype, grows anchorage independently, and forms a tumor that most closely resembles a spindle cell sarcoma in athymic mice. Its ultrastructure in monolayer culture consists of large cells with an expanded rough endoplasmic reticulum and abundant multivesicular bodies; in athymic mice, extracellular collagen fiber formation is prominent. DEAE-cellulose chromatography and sodium dodecyl sulfate-polyacrylamide gel electrophoresis of cultures labeled with [3H] proline demonstrated interstitial collagen formation. We conclude that the cell line at establishment is a collagen-producing spindle cell sarcoma that resembles the sarcomatous regions of the original mixed tumor. Further cell separation and characterization studies are needed to determine the pathogenesis of mixed tumors such as gliosarcoma.

Characterization of normal human brain cultures. Evidence for the outgrowth of leptomeningeal cells.

To establish the histogenetic identity of the predominant cell type in monolayer cultures of normal human adult brain, eight brain specimens were placed into culture and characterized according to cell kinetics, karyotype, antigenic expression, and ultrastructural features. The protein profiles of both the cell layer and the medium were analyzed in selected cultures using sodium dodecyl sulfate polyacrylamide gel electrophoresis and diethylaminoethyl cellulose chromatography. All cultures displayed a limited life span in vitro; marked contact inhibition at confluence; a normal karyotype; an intracytoplasmic and extracellular glycoprotein profile consisting of fibronectin, procollagen type III, laminin, and collagen type IV; specialized intercellular junctions; and interstitial collagen chain synthesis. All of these features were identified in our previous study of human leptomeningeal cultures. The results of immunocytochemical staining for glial fibrillary acidic protein were negative in all cultures of normal human brain, except in early passages in two cultures, which lost the glial cell marker during subsequent passages; immunostains for vimentin were positive in all cells in all cultures. These results support the hypothesis that, in this study, cultures derived from normal human brain are not of glial origin. Our findings also suggest that glial cells are less well-suited to monolayer growth under our culture conditions than are other cell types in enzyme-dissociated brain tissue placed in culture, especially leptomeningeal cells. The identification of leptomeningeal cells as the predominant cell type in normal human brain cultures may prove useful in attempts to foster the growth of human glial cells by culturing brain samples under conditions that prohibit the growth of leptomeningeal cells. Under such conditions, astrocytes, oligodendroglia, and ependymal cells could be isolated with greater ease and cultured separately. These purified cultures of different glial cell types would then provide a more relevant in vitro model for studying human neurological diseases.

An ultrastructural and immunocytochemical analysis of leptomeningeal and meningioma cultures.

Using immunocytochemical methods, we localized several glycoproteins of the extracellular matrix to leptomeningeal cells and meningiomas in vitro. Three cell lines derived from normal human leptomeninges and seven from meningiomas were studied by indirect immunofluorescence to evaluate the cellular production of fibronectin, laminin, collagen type IV, and procollagen type III. All leptomeningeal cell lines stained intensely and uniformly for all matrix proteins; all meningioma cell cultures stained uniformly, but the intensity of staining varied considerably. After removal of the cells in culture adherent to glass with 25 mM ammonium hydroxide, indirect immunofluorescence demonstrated an exuberant residual extracellular residue enriched with fibronectin, laminin, collagen type IV, and procollagen type III. Electron microscopic examination of all leptomeningeal and meningioma cultures revealed desmosomes and dense tonofilament formation; in addition, granular, filamentous basement membrane-like material was abundant in the extracellular spaces of all cultures. Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis of the cell layer of two leptomeningeal and four meningioma cultures showed production of interstitial collagen types I and III; diethylaminoethyl (DEAE)-cellulose chromatography of the medium demonstrated preferential production of procollagen type I. Our findings show conclusively that normal arachnoid cells in vitro synthesize several of the collagen subtypes and may be responsible for the "fibrous response" of the leptomeninges to trauma, infection, or infiltration by tumor. The similarities between leptomeningeal cells and meningiomas demonstrated by electron microscopy and by indirect immunofluorescence support the notion that meningiomas are derived from arachnoid cells. The localization of various mesenchymal glycoproteins within the intra- and extracellular spaces and the ubiquity of specialized intercellular junctions suggest that leptomeningeal cells in culture have the potential to behave like both stromal and epithelial cells.

Variable response to 1,3-bis(2-chloroethyl)-1-nitrosourea of human glioma cells sorted according to DNA content.

Malignant gliomas, especially glioblastoma multiforme, are composed of a considerable number of cells undergoing DNA synthesis and cells that contain various amounts of DNA. We have characterized cells dissociated from biopsy specimens of such tumors by sorting vital cell populations on the basis of DNA content and have determined the sensitivity of sorted cells to BCNU. Cultured human glioma cells from seven malignant gliomas were stained with Hoechst 33342 dye; the colony-forming efficiency (CFE) of stained tumor cells ranged from 2-50%. Stained cells were sorted on a fluorescence-activated cell sorter (FACS-III). The surviving fraction of each population was determined after exposure to graded doses (2.5-10 micrograms/ml X 2 h) of BCNU. Sorted populations of cells from 5 tumors had dose-response curves that were similar, although differences in cell kill of up to a half-log were commonly found between cells from different DNA peaks treated with the same BCNU dose. For two tumors, cells from the first peak (smallest DNA content) had distinct BCNU sensitivity compared to cells from the second and third peaks (largest DNA content); compared to other tumors in the series, cell kill differences were significant and greater than 1 log in magnitude for one anaplastic astrocytoma. This lack of uniformity in the response of cells within a tumor demonstrates the problem imposed by heterogeneity with regard to the interpretation of chemosensitivity testing of all cells within a single tumor.

Improved treatment of a brain-tumor model. Part 2: Sequential therapy with BCNU and 5-fluorouracil.

A combination chemotherapy regimen for brain tumors was developed, based on investigations of the survival of animals harboring the intracerebral 9L rat brain-tumor model and on analyses of their clonogenic tumor cells. Fischer 344 rats harboring 9L brain tumors were treated with 2-day courses of 5-fluorouracil (5-FU), in order to expose all cycling tumor cells to the drug during DNA synthesis and achieve maximum anti-tumor activity for this cell-cycle-specific anti-metabolite. Although a 74% cell kill was obtained for a total dose of 45 mg/kg or greater, animal life span was not increased over that of untreated tumor-bearing controls. However, when 5-FU (48 to 96 mg/kg total dose over 2 days) was administered after a single LD10 dose of BCNU (13.3 mg/kg), additive cell kill was suggested. In three large series, long-term animal survivors and occasional tumor cures were observed with this drug combination, a result never observed following BCNU alone. Schedule dependency was not apparent. A previously published protocol for treating recurrent malignant gliomas with sequential courses of BCNU and 5-FU was partially planned based upon these initial observations. Anti-tumor activity with the combination of drugs was superior to therapy with BCNU alone. Both animal and human studies confirm that, contrary to presently accepted oncological tenets, a chemotherapeutic agent that kills significant numbers of tumor cells but is clinically ineffective when given alone might, nevertheless, be useful in combination therapy regimens.

Stem cell studies of human malignant brain tumors. Part 1: Development of the stem cell assay and its potential.

A stem cell assay for human malignant gliomas has been developed. Cells obtained from tumor biopsies grew into colonies composed of malignant glial cells, as documented by histochemical, immunohistochemical, and immunobiological techniques. Studies suggest that the disaggregated cells are representative of the cells within the solid tumor. Clonogenic cells were obtained from 48 tumors and analyzed for their in vitro sensitivity to graded doses of 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU). The in vitro anti-tumor activity of BCNU at clinically achievable doses was compared to clinical response to the agent based on changes in computerized tomographic scan, radionuclide brain scan, and neurological examinations. Twenty-two patients received nitrosoureas before or after tumor specimen analysis, and were eligible for in vitro-in situ correlations. Clinical tumor sensitivity to nitrosoureas was predicted by culture results in 42% of all evaluable patients, and clinical resistance was predicted in 100%. The capability of the assay can be appreciated best for the 13 patients not treated with BCNU prior to culture; the in vitro prediction of clinical sensitivity and resistance was 71% and 100%, respectively. Preliminary findings show that clinical tumor resistance to BCNU may result from "intrinsic" cell resistance in some patients and from inadequate delivery of drug to tumor cells in other cases. The potential utility of this method to study the reason(s) for tumor cell resistance to drugs, to screen new chemotherapeutic agents, to individualize patient treatment, and to investigate tumor biology is discussed.

Improved treatment of a brain-tumor model. Part 1: Advantages of single- over multiple-dose BCNU schedules.

Clonogenic cell and animal survival studies were used to determine the most effective BCNU therapy schedule in the 9L rat brain-tumor model. Survival of tumor cells following a single LD10 dose of BCNU (13.3 mg/kg intraperitoneally) was compared to cell survival after one to four daily 0.5 X LD10 doses. The posttreatment kinetics of surviving clonogenic cells were investigated at various times after BCNU was given in single doses of 0.25 to 1 X LD10 and in two daily doses of 0.5 X LD10. The cell kill was greater, time to reinitiation of cell growth was later, posttreatment rate of clonogenic cell proliferation was slower, and the interval to total repopulation of the clonogenic cell pool was longer with a single LD10 dose as compared to the multiple-dose schedules. Animal survival studies confirmed that a single LD10 dose of BCNU was at least as effective as a cumulative level of up to 1 1/2 times that amount when treatment was administered in smaller doses, regardless of the fractionation schedule. Clinical experience with patients harboring malignant brain tumors has shown that a single BCNU dose of 185 to 200 mg/sq m is tolerated well. Results of these animal experiments suggest that this therapy should have anti-tumor activity at least equivalent to the more commonly employed schedule of 80 mg/sq m/day given for 3 days. Although direct comparison of treatment efficacy using the two schedules is not possible, no adverse clinical effects have been observed with the recently adopted single-dose schedule. Furthermore, the duration of patient hospitalization for chemotherapy has decreased.

Clonogenicity of multiple populations of human glioma cells in vitro sorted by DNA content.

Malignant human gliomas have a highly variable distribution of cell nuclei, consisting of diploid and/or other populations in terms of nuclear DNA content. In order to study in vitro clonogenicity of each population, dissociated or cultured human glioma cells were stained with 20 microM/ml of Hoechst 33342 dye (which stains viable DNA with minimal cell kill), and were sorted sterile into separate populations, based on specific nuclear DNA content, for clonogenicity assay. The colony-forming efficiency (CFE) of tumor cells plated immediately after disaggregation of the biopsy specimens ranged from 0.0044 to 0.149%, and the CFEs increased dramatically with successive passages (to 5 to 40%). The CFEs of the individual populations sorted according to DNA content were similar within individual tumors. These results suggest not only that malignant gliomas are composed of multiple populations in terms of DNA content, but also that each of these populations contain clonogenic cells. The morphologic structure of cells within and among colonies did not appear to relate to DNA content.

Age-related chemosensitivity of stem cells from human malignant brain tumours.

After radiation therapy and chemotherapy for malignant glioma, patients aged 50 or under survive longer than patients over 50. Data from Brain Tumor Study Group trials show that, without treatment, these age groups have similar survival; therefore unperturbed tumour growth does not account for the difference. Sixteen consecutive patients with malignant glioma were studied, half of whom were less than or equal to 50 years of age; none had been treated before initial surgery, and all were subsequently treated with radiation and chemotherapeutic agents (in all but two patients, with nitrosoureas). Median survival of those aged greater than 50 was less than or equal to 50 years was 54 + weeks whereas that of those aged greater than 50 was 37 weeks. The longer survival for younger patients could not be attributed to tumour type, size, or location, pretreatment Karnofsky status, or mode of treatment. In-vitro sensitivity testing of clonogenic cells obtained from biopsy specimens showed that tumour cells from seven of eight patients aged less than or equal to 50 years were sensitive to 1,3-bis (2-chloroethyl)-1-nitrosourea (greater than 40% cell kill at clinically achievable concentrations) compared with only one patient with sensitive cells out of eight older patients. Patient age was inversely correlated with in-vitro cell kill, and patients with sensitive cells were significantly younger than those with resistant cells. Therefore influence of age on survival after treatment of malignant gliomas is probably due to inherent differences in the sensitivity of clonogenic cells to radiation and/or chemotherapeutic agents.

Potentials and limitations of a clonogenic cell assay for human brain tumors.

Single cells from biopsies of ten malignant gliomas were treated with 1,3-bis (2-chloroethyl)-1-nitrosourea (BCNU) in vitro, and tumor cell survival was compared to patient response to nitrosourea therapy. The limitations of the methods and their interpretations and future prospects for clonogenic cell analyses are discussed.

Potentials and possible pitfalls of human stem cell analysis.

A clonogenic cell assay for malignant brain tumors that permits the evaluation of tumor cell sensitivity to BCNU and that correlates with patient response to BCNU has been developed. The potential for a stem cell analysis of human tumors has been demonstrated by studies of the reasons for clinical drug failure, tumor heterogeneity, and age-response relationships. The basic requirements of a stem cell assay include the ability to dissociate representative single cells from solid tumors, to optimize culture conditions, and to characterize the growth of colonies. Exposure of cells to a drug in vitro must be comparable to the in situ situation; possible significant differences between short-term and "continuous" treatment methods are emphasized. Also discussed are criteria for in vitro sensitivity of cells, problems inherent in the "early" evaluation of cultures (at the cell "cluster" stage), and the effects of system errors, which if overcome should lead to the development of analytic methods with a maximum sensitivity and predictive value.