Radon Prevention Program Hits High Gear

Surgical specimens from 15 medulloblastoma patients were used to establish early passage cultures. In vitro sensitivity to a battery of cytotoxic agents, including some in current medulloblastoma treatment protocols, was measured. Drug sensitivity was assessed at clinically relevant drug concentrations using the 3H-thymidine uptake method. Tumours were predicted to be sensitive if >37% were killed by exposure to drugs at clinically achievable levels. A poor response to vincristine (Vcr), cis-platin (CDDP), hydroxyurea (HU) or diaziquone (AZQ) (no responders), and cytosine arabinoside (AraC) (1/12), was seen. Nine of ten tumours tested were sensitive to mafosfamide (Mfs); seven out of 12 were sensitive to carmustine (BCNU), 12 of 13 to teniposide (VM-26) and seven of 13 to etoposide (VP16-213). VM-26 was the best of the agents tested with most tumours responding to very low concentrations of drug, suggesting that the role of epipodophyllotoxins in treatment of brain tumours be further investigated. Despite the marked sensitivity of the medulloblastomas to the epipodophyllotoxins, three early passage cultures were much more resistant to these drugs than the average for the group. The basis of this resistance

Medulloblastoma is an important paediatric brain tumour because of its high incidence and malignant behaviour. The overall disease free survival at 5 years, in patients receiving surgery and craniospinal axis radiotherapy, is approximately 50%. This is considered to be the upper limit of curability by these treatment modalities. Chemotherapy has been shown to be advantageous in selected patient groups either as adjuvant therapy or as treatment of recurrent disease. The value of chemotherapy in the treatment of medulloblastoma will be dependent on the availability and identification of drugs active against this tumour. At present, protocols for chemotherapy usually include the lipophilic compounds carmustine (BCNU) or lomustine and procarbazine. Vincristine (Vcr), whilst poorly absorbed by cerebrospinal fluid and brain, is also commonly used (Bloom, 1986;Workman, 1986). Currently '8 in 1F therapy is being evaluated (Pendergrass et al., 1987). Cytotoxicity has been assessed by clonal assays in agar or on plastic, incorporation of 3H-thymidine into DNA, uptake of vital dyes and other methods with much the same results. Predictions of resistance are close to 100% correct; prediction of sensitivity, whilst less reliable, still has a success rate of around 70% Moon et al., 1981;Rosenbloom et al., 1983;Bogdahn, 1983;Kornblith et al., 1981;Kimmel et al., 1987). We have examined drug sensitivity in human medulloblastoma by using 14 early passage cultures. Our study provides data on the degree of heterogeneity of response between medulloblastomas obtained from a number of patients. Furthermore, the cells were from early passage cultures, and hence should give a better indication of chemosensitivity of the tumours in vivo. The epipodophyllotoxins teniposide (VM-26) and etoposide (VP16-213) were identified as potentially useful agents in this disease. We investigated the basis of primary clinical resistance to these drugs by quantitating intracellular steady-state drug concentrations, drug-induced protein-linked DNA strand breaks, and topoisomerase II expression and activity.

Establishment of cell cultures
All studies were made on material collected at procedures performed for clinical reasons and remaining after sufficient tissue has been taken for clinical laboratory study. The histological diagnosis was confirmed by an independent neuropathologist (Professor B.W. Scheithauer). Over a period of 2 years, 16 medulloblastoma specimens were received from a total of 15 patients. The age of the patients ranged from 3 to 11 years (median 5.5 years) with a male to female ratio of 2:1. Most of the samples were from primary tumours however both a primary tumour (MD 6) and a metastasis removed at the same time (MD 6M) were obtained from one patient. A recurrent fourth ventricle tumour (MD 5) was removed 3 years after a patient had completed a full course of craniospinal irradiation and post irradiation chemotherapy and in another patient a frontal metastasis (MD 14) was removed 15 months following extirpation of the primary lesion and a full course of craniospinal irradiation.
Tumours samples were dissociated and cloned in agar as described by Ablett et al. (1984). Primary cultures, were established in tissue culture flasks from dissociated tumour cells in RPMI 1640 with 10% FCS, were incubated at 37°C in an environment of 5% 02, 5% C02, 90% N2 and 100% humidity. When the monolayers were almost confluent, the cultures were passed. Cultures were characterised around passage No. 4 and against around passage No. 10. Experiments were not performed on cells after 12 passages. They were routinely checked for mycoplasma.
Characterisation of early passage cultures The early passage cultures were characterised by immunohistochemical methods for neuron specific enolase (NSE) (Dako PAP kit, Dako Corp., Santa Barbara, USA) and by a panel of monoclonal antibodies (UJ13A and A2B5 which bind to normal and neoplastic neuroectodermal tissue, UJ18 1.4 which shows specificity to oncofoetal antigen expressed by foetal brain and neuroblastic tumours and UJ127.11 which recognises antigens of normal neuroectodermal tissue, neuronal tumours and schwannomas) all of which were kindly provided by Dr J.T. Kemshead (Brain Tumour Research Laboratory, Frenchay Hospital, Bristol, England) (Coakham et al., 1984). As frozen material was not available, immunohistochemical staining was not performed on the surgical specimen. Cytogenetic analysis was performed on early passage cultures as described by Sandburg (1980). Metaphases were Gbanded (G-bands with Trypsin and Giemsa). At least five metaphases were examined per culture. DNA flow cytometry was performed as described by Pemble et al. (1987). Normal HeLa cells were used as control.
Drug testing protocol A reduction in survival, at clinically achievable drug levels, to 37% was adopted for classifying tumours into sensitive and resistant groups Moon et al., 1981;Lihou & Smith, 1983& 1985Kornblith et al., 1981;Rosenbloom et al., 1983;Kimmel et al., 1987). The pharmacokinetic parameter C x T (concentration time.product) (Mellett, 1974;Alberts et al., 1980) was used as the measure of clinically achievable drug exposure. The C x T achieved in patients receiving standard drug doses was obtained from, or calculated from published data (Tables I, II). The in vitro C x T required to reduce survival to 37% was calculated simply by multiplying the D37, the drug concentration required to reduce survival to 37% of control values, by the number of hours of incubation i.e. 24 h (Tables I, II). This was compared with the C x T achievable in patients (Tables  I, II): tumours were classified as sensitive if the former figure was less than the latter.
were exposed to the phase-specific agents i.e. cytosine arabinoside (AraC), hydroxyurea (HU), VM-26 and VP16-213 for 24 h, since the cells grew slowly. The doubling time up to the 10th passage was around 3 days for all early passage cultures except MD 12 (1.2 days).
Cell survival Cell survival was determined by 3H-thymidine incorporation. Early passage cultures were seeded in 0.1 ml of medium, in 96-well microtitre plates 24 h before treatment with drug (six replicates for each drug concentration) at a cell concentration of 4 x I04ml-'. The cells were exposed to drugs whilst attached to the bottom of wells in microtitre plates. Control untreated cells were examined daily until almost confluent. Five to 7 days after treatment was initiated, the medium was replaced with medium containing [3H-methyl]-thymidine (5 ytCi ml-', 40 Ci mmol -; RadioChemical Centre, Amersham, UK).
After 24  Detection ofprotein-linked DNA strand breaks The SDS-K+ precipitation method used was a modification of that developed by Trask et al. (1984) and used to detect drug damage by Rowe et al. (1986). All early passage cultures used were harvested in log-phase growth. Cells were labelled overnight with 0.8 iLM ml-I of 3H-thymidine (specific activity 41 Ci mmol-') and cells (2 x 104 cells/vial) rested for 1 h at 37°C in a gassed (5% 02, 5% C02, 90% N2), humidified incubator. Drug was then added and cells incubated for the required period. At different time points, the suspension was diluted with 5 volumes Dulbecco's medium and spun at 150 x g for 10 min. Liquid scintillation counting was performed following transfer of the pellet resuspended in 200 of water at 65°C into scintillation tubes containing 3.5 ml of Instagel (Packard, Ill., USA). Additionally, cross-link repair was determined by further incubation in drug free media.
Preparation of nuclear lysate Nuclear lysates were prepared according to the method of Miller et al. (1981) but as limited numbers of cells were available the method was scaled down 100-fold. Between 106 and 107 cells were used to make the lysates which were stored at -20'C following dialysis against 30 mM potassium phosphate pH 7.0, 50% glycerol, 0.5 mM dithiothreitol and 0.1 mM EDTA.
Topoisomerase II activity Topoisomerase II activity of nuclear lysate from medulloblastoma early passage cultures was assessed using the P4 DNA unknotting assay (Liu & Davis, 1981). Incubation was performed for 1 h. The protein content of nuclear lysates was measured by the Coomassie blue method of Bradford (1976).
Topoisomerase II expression The topoisomerase II content was assessed by immunoblotting of cell extracts using anti-topoisomerase II antisera (Halligan et al., 1985). The topoisomerase II antibody was kindly donated by Dr L.F. Liu (Johns Hopkins University, School of Medicine, Baltimore, USA). Murine erythroleukaemia cells (MELC), clone 745 of C. Friend, also donated by Dr L.F. Liu, were used as controls.

Results
In vitro growth All tumours plated in agar formed colonies (40 cells or more) after 2 to 4 weeks culture. The cloning efficiency in agar, of the four primary cultures tested, was very low (MD 4, 0.010 ± 0.001; MD 5, 0.0011 ± 0.0005; MD 7, 0.0023 ± 0.0009; MD 9, 0.0004 ± 0.0003; MD 10, 0.0015 ± 0.0002; mean ± standard deviation). Most of the specimens provided were not of sufficient size for plating in agar and therefore primary cultures from these tumours were established in liquid culture on plastic. Fourteen of 16 samples, including a primary tumour and spinal metastasis, were established in liquid culture.
Characterisation of early passage cultures All early passage cultures were positive for the following antibodies: UJ127.11 which recognises antigens on normal ectodermal tissue, neuronal tumours and schwannomas; UJ13A and A2B5 which recognise normal and neoplastic neuroectodermal tissue; and UJ181.4 which recognises an oncofoetal antigen expressed by foetal brain and neuroblastic tumours. Staining for NSE was negative in all early passage cultures. Karyotyping was successfully performed in eight out of ten early passage cultures of which four were abnormal (Table III). Of the four specimens submitted for flow cytometry, two were aneuploid (Table III). Proliferative activity of early passage cultures Although the experiments were performed at the earliest possible passages, the variation in control 3H-Thymidine incorporation between experiments was still quite high, decreasing as the cultures became older. Despite this variation, a reliable estimate of drug sensitivity was obtained. The standard error of the mean of survival at a given drug concentration obtained for three to ten experiments was usually less than 10%. The average control values for incorporated d.p.m. are given in Figure 1. This average was taken from all cytotoxicity experiments (approximately 30 to 100 independent experiments per culture, 12 control wells per experiment).
Sensitivity to phase-specific agents All three phase-specific agents in the '8 in 1' protocol (Pendergrass et al., 1987) were tested i.c. AraC, HU and Vcr, and the calculated D37 values are given in Table I. There was a general trend for faster growing cells to be more sensitive to the phase-specific agents. However, Spearman rank correlation analysis was performed for each set of data and no significant association was found between drug sensitivity and d.p.m. in control cultures of the same experiments, indicating that factors other than proliferative status also influenced sensitivity to these agents.
There was a heterogeneous response to the phase-specific agents, with the highest D37 being 40 times (Vcr), 43 times 100,000 r 3 4 5 6 7 9 10 *eduflblastoma ulture (AraC) or 88 times (HU) higher than the lowest measured. The fastest growing early passage culture, MD 12, was by far the most sensitive to all three agents, but no culture was consistently more resistant than others. Spearman rank correlation analysis revealed no significant cross-sensitivity between any combination of these agents. This again indicates that factors other than those affecting cell growth determined in vitro sensitivity.
Only one tumour was considered sensitive to clinically achievable levels of AraC and none to either HU or Vcr (Table I). A very high concentration of HU was required to kill cells when compared with plasma drug levels.
Sensitivity to epipodophyllotoxins Similar experiments to those described above were carried out using the epipodophyllotoxins VM-26 and VP16-213, nonintercalative topoisomerase II poisons (Liu, 1989). The response was heterogenous (Table I); the range of response to VM-26 being greater than that for any other drug tested. The in vitro response was not related to d.p.m. in control cultures, but sensitivity to VM-26 was highly, significantly correlated with response to VP16-213 (r = 0.852, P<0.002, n = 13) ( Table IV).
Twelve of 13 (92%) tumours were considered sensitive to clinically achievable levels of VM-26 as were seven out of 13 (54%) to VP16-213 (Table I). This difference in responsiveness was due to two factors; on a molar basis, VM-26 was far more cytotoxic than VP16-213, and the plasma pharmacokinetics reported for VP16-213 are worse than for VM-26, because lower doses of VP16-213 must be administered and for the same dose the C x T and the peak plasma concentration are lower (Table I). Most medulloblastoma cells were very sensitive to VM-26. Ten of the sensitive early passage cultures responded at levels which were less than 1/10 of that achievable in patients and for one culture (MD 12) this figure was 1/600. In this regard, VM-26 was the best agent tested in this study.
Three early passage cultures however, were markedly more resistant to VM-26 than average. The D37 values for these cultures were 15.17IAM for MD5; 49.30 ,sm; for MD 7 and 11.60JM for MD 11. These cells were approximately 400-1600 times as resistant as the most sensitive culture (MD 12, D37 0.03 JAM) and 20 to 80 times more resistant than average for the remaining cultures (D37 0.62 JAM ± 0.35 JAM s.d). The same three early passage cultures were also resistant to VP16-213. The difference between sensitive and resistant cultures was not as marked as for VM-26, for D37 for the VP16-213 resistant cultures being around 300 times that of MD 12 and 13 times the average for the other nine early passage cultures.
Sensitivity to alkylating and cross-linking agents Medulloblastoma early passage cultures were also tested for sensitivity to alkylating and cross-linking agents i.e. BCNU, Mfs, CDDP and AZQ (Table II). A spectrum of responses  Tables I and II were analysed for cross-sensitivity relationships between different agents. All possible combinations were tested, and in addition the relationship between response to a drug and the d.p.m. counted in control, untreated plates (ie. an indication of the proliferative status of the cells) was also tested. ar, the probability of there being a positive relationship between sensitivity to two different agents, or between an agent and the proliferative status of control cells.
bNumber of early passage cultures tested. was observed but the differences between the most sensitive and most resistant cultures (D37 resistant cultures/D37 sensitive culture) were less for this group of agents than for the phase-specific agents. These ratios were five for BCNU, 23 for Mfs, 9 for CDDP and 16 for AZQ.
Early passage cultures resistant to the epipodophyllotoxins (MD 5, MD 6, MD 7 and MD 11) were generally more resistant than average to alkylating agents as well. MD 6 however appeared to be relatively sensitive to AZQ. Again, MD 12 was the most sensitive to all these agents, but as mentioned above, the differences between cultures were not as great for this class of agents.
None of the tumours were sensitive to CDDP or AZQ but seven of 12 were sensitive to BCNU and nine out of ten were sensitive to Mfs (Table II). In all cases, the level of in vitro drug exposure had to be of the same order of magnitude as the maximum clinically achievable level to reduce survival to 37% of control, untreated cells. In other words, none of the tumours tested were very sensitive to any agent from this group.
Sensitivity to amsacrine (m-AMSA) Sensitivity to a DNA intercalating topoisomerase II poison (M-AMSA) (Liu, 1989) was tested ( Figure 2). The range of responses to this agent was large, the highest D37 calculated being 399 times the lowest. MD 5 and MD7 were very resistant to this agent, whilst MD 11 was only marginally more resistant than average. Rank correlation analysis revealed significant relationships between sensitivity to this agent and epidodophyllotoxins, and also with Mfs (Table  IV).
Patterns of cross-resistance and collateral sensitivity As stated previously, the data in Tables I and II was analysed for cross-resistance and collateral sensitivity relationships by the Spearman rank correlation test (Table IV). Cross resistance between VM-26 and VP16-213 was confirmed. There was also a significant relationship between resistance to these agents and to the other topoisomerase II poison tested, m-AMSA.
There was no apparent relationship between sensitivity to either VM-26, VP16-213 or m-AMSA and Vcr suggesting that the phenomenon of 'pleiotropic drug resistance' may not be responsible for the patterns of cross-resistance seen. Doseresponse experiments with VM-26 were also performed in the presence of agents known to reverse this type of resistance: verapamil (1O JM) (Yalowich & Ross, 1985) and cyclosporin A (7 and 13 1ig ml-') (Slater et al., 1986). Neither of these agents significantly altered sensitivity to VM-26 in any of the early passage cultures tested; all cultures were tested except for MD 10 and MD 17. A surprising finding was the association between Mfs sensitivity and response to either VP16-213 or m-AMSA. The relationship with VM-26 was not significant. The only dgent for which sensitivity was significantly linked with the amount of 3H-thymidine incorporation in control cells was AZQ. There was no significant difference between results obtained from MD 6, the primary tumour and MD 6M, a metastasis obtained from the same patient. were calculated as outlined for Table I. -Intracellular accumulation of VP16-213 Cellular uptake in medulloblastoma early passage cultures measured at time points ranging from 5 to 120 min demonstrated that steady-state was achieved in all cultures by 60 min (Figure 3). Cellular uptake at 120 min in nine medulloblastoma cultures ranged from 4.88 to 20.26 pmol 10' cells (Table V). The most resistant cultures fell in the middle of this range (MD 5,9.04;MD 7,15.69;MD 11,10.30 pmoles 10-cells), whilst the hypersensitive culture, MD 12, had one of the lowest values. Estimation of cell diameters allowed calculation of the cellular volume and thus the intracellular drug concentration. Cell diameters ranged from 17.8 to 19.9ym. The standard deviation ranged from 5% to 8% of the mean, indicating no significant difference between cell diameters of different early passage cultures. It was therefore calculated that all of these cultures were capable of concentrating the drug. When the steady-state drug concentration was compared with cellular drug sensitivity, no correlation was found.
Protein-linked DNA strand break activity Results were expressed as the ratio of the d.p.m. precipitated in the presence of drug over d.p.m. precipitated without drug (control). The fact that some precipitable counts were measured in the absence of drug is an indication of the basal level of topoisomerase II activity present. A range of VM-26 concentrations from 1 lAM to 100 0tM were used initially  Protein-linked DNA strand breaks were compared in all available medulloblastoma early passage cultures following a 1 h incubation with 1OiM and 30 JAM VM-26. The latter concentration was chosen since this was likely to result in maximum DNA damage due to protein-linked DNA strand breaks, is a clinically achievable concentration and is highly cytotoxic in the 3H-thymidine assay. At the lower concentration of 10JM VM-26, the number of protein-linked DNA strand breaks was elevated by a factor of approximately ten times control. At 30 iJM VM-26, the number of protein-linked DNA strand breaks was elevated by a factor of around 20 times control. Some cultures exhibited less protein-linked DNA strand break activity. MD 5, in particular, was resistant to this form of drug-induced damage. No overall correlation between epipodophyllotoxin sensitivity and induction of protein-linked DNA strand breaks was apparent.
In medulloblastoma early passage cultures, VM-26 was much more cytotoxic to cells on an equimolar basis than VP16-213 (Table I). Experiments were performed to compare the number of protein-linked DNA strand breaks produced by  in four medulloblastoma cultures. Two of the four medulloblastoma early passage culture (MD 7, MD 11) were resistant to epipodophyllotoxins. The number of protein-linked DNA strand breaks produced by VM-26 was approximately 1.8 times greater than that produced by VP16-213.
Protein-linked DNA strand break time-course Production of protein-linked DNA strand breaks by continuous exposure to VM-26 (1O JM and 30 JM), was measured in medulloblastoma early passage cultures for time points 10 min to 24 h ( Figure 5). Generally, maximum proteinlinked DNA strand break frequency occurred at 1 h for both drug concentrations. The time-course of induction of proteinlinked DNA strand breaks at 30 JAM VM-26 in the VM-26 resistant cultures MD 7 and MD 11 was no different from that of the sensitive lines MD 12, MD 14. In the VM-26 resistant culture MD 5, however, the frequency of proteinlinked DNA strand breaks did not vary with exposure duration. In spite of the continuous presence of VM-26, at time points after 1 h, d.p.m. precipitated by SDS-K+ approached control values.
After long term exposure to VM-26 (24 h) the frequency of protein-linked DNA strand breaks at 10 JAM and 30 gM concentrations was not significantly different from control values for early passage cultures other than MD 7. In this culture, the frequency of protein-linked DNA strand breaks was 5.1 and 8.4 times control at 10 JM and 30 JLM respectively. r.n. mAMSA-induced protein-linked DNA strand breaks Protein-linked DNA strand break activity was examined following exposure to m-AMSA (3 jig ml-I for 1 h and 24 h).
This concentration was chosen because it was a clinically achievable level and resulted in significant cytotoxicity (being of the order of the D37 for the medulloblastoma group). At this concentration m-AMSA was readily soluble in RPMI 1640 10% FCS. At 1 h, significant protein-linked DNA strand break activity was observed which was roughly equivalent to that produced by 10OiM VM-26. Again, few links were observed in the MD 5 early passage culture. After 24 h, protein-linked DNA strand breaks were still apparent but were lower than at 1 h. Topoisomerase II activity Topoisomerase II activity in nuclear lysates from MD 5, 7, 12 and 14 cells was assayed using a P4 unknotting assay. Extracts from MD 12 had the greatest amount of activity, detectable with as little as 0.1 jig nuclear protein. Activity in the other early passage cultures was much lower than MD 12 with detectable activity at 1 jig protein for MD 11 and 3 jig for both MD 7 and MD 14. There was no detectable activity in nuclear lysates from MD 5 cells.
Topoisomerase II expression In three medulloblastoma early passage cultures, MD 7, MD 12, MD 14, a band of similar molecular weight to topoisomerase II (170 kDa) previously characterised in mouse friend leukaemia cells (Bodley et al., 1987; personal communication L.F. Liu, 1990), was identified. Consistent with the results of the activity assay, the amount of enzyme was high in MD 12 cells and much lower in the other two cultures, with more immunoreactive protein in MD 7 than MD 14. Insufficient cells were available to enable enzyme expression in MD 5 and MD 11. Two bands of high molecular weight greater than 180 kDa were recognised in the MD 7 cells but not the other lines. A number of bands of lower molecular weight were seen in all early passage cultures and presumably represent fragmented protein.

Discussion
The benefits of surgery and radiotherapy for treatment of malignant disease are generally considered to be approaching the maximal attainable limits. Further improvement in disease free survival may be achieved by chemotherapy but there is considerable heterogeneity of response to different agents; both between patients and between tumour types. The present study concentrated on medulloblastoma; a tumour with a poor prognosis but where chemotherapy has been shown to be advantageous for some patients (Evans et al., 1990;Tait et al., 1990;Packer et al., 1991;Bloom, 1986). We have examined the in vitro sensitivity of early passaged cultures from 13 medulloblastoma patients to a number of drugs. Since the cultures have not been extensively passage, the possibility of selection from an originally heterogenous sample of tumour cells is therefore diminished. The results should give an indication of the degree of heterogeneity of response between different tumours.
The clinical criteria for selection of tumour samples for use in this study were based on: (a) demographic characteristics and clinical presentation; (b) the radiologic appearance; (c) the histologic appearance. The identity of the cultured cells was determined on morphological (Rubinstein & Herman, 1986;Gerosa et al., 1987) immunohistochemical, cytogenetic and DNA flow cgtometric analysis. The immunohistochemical profile was determined using a panel of murine monoclonal antibodies (UJ13A, UJI81.4, UJ127.11 and A2B5). These antibodies have been successfully used for the characterisation of surgical material and established cell lines (Kemshead & Coakham, 1983;Bourne et al., 1986;He et al., 1989;Coakham & Bourne, 1989).
As with surgical specimens of gliomas (Rosenbloom et al., 1983), medulloblastoma primary cultures exhibited poor cloning efficiencies in semisolid media in contrast to the high cloning efficiency seen in an established medulloblastoma cell line Daoy (Jacobsen et al., 1985). However, medulloblastoma is unlike gliomas in that clonal growth on plastic is hard to assess due to cell mobility and poor growth at low cell density. In this study, we used an end-point of cellular 3Hthymidine uptake 5-7 days after drug exposure as an indicator of drug toxicity. This has been found to correlate with clonal growth in agar (Parsons & Brown, 1979).
The cut-off used for sensitivity in the present study was that found by others to predict clinical response (approxi-I mately 100% correct for resistance, 70% for sensitivity) Moon et al., 1981;Rosenbloom et al., 1983;Bogdahn, 1983;Kornblith et al., 1981;Kimmel et al., 1987); except that the D37 rather than D4, was used. The parameter used as a measure of maximum possible in vivo drug exposure (in the clinical situation) was the C x T (drug concentration x duration of exposure). This is considered to be a more relevant pharmacokinetic parameter than concentration or duration of exposure alone in determining response to drugs (Mellett, 1974;Alberts et al., 1980). The C x T of in vitro exposure was calculated to parallel reported in vivo C x T values. Exposures of 4 h duration were used for drugs which are not phase-specific. A 24 h incubation was used for drugs which demonstrate phase-specificity and have relatively long plasma half-lives. This approach has been successful in this laboratory in studies of acute myeloid leukaemia (Lihou & Smith, 1985).
We have thus found that two of the agents conventionally used for treatment of medulloblastoma are active against most of our early passage cultures in vitro i.e. Mfs, the cyclophosphamide analogue and BCNU. On the other hand, some agents did not exhibit activity, i.e. cis-platin, hydroxyurea, vincristine and cytosine arabinoside. The relatively new agent, diaziquone (AZQ) was also not effective. The most promising results were those for the epipodophyllotoxin VM-26. The epipodophyllotoxins are generally believed to be of equivalent clinical value, with no established evidence of superiority of one agent over another (Issell et al., 1982). Our study demonstrated in vitro cross-resistance between these agents, but we found that in all cases VM-26 was more cytotoxic to medulloblastoma than VP16-213. Issell et al. (1982) also found a significant difference between activity of these two compounds against a sample of brain tumours. In the present study, D37 values for VP16-213 were higher in all cases than for VM-26 (1.3 to 32.8 times VM-26 D37). When in vivo pharmacokinetics are considered, this difference in D37 meant that 12/13 medulloblastomas were predicted to be clinically sensitive to VM-26 but only 7/13 to VP16-213. When cytotoxic concentrations were compared with clinically achievable drug concentrations, VM-26 was the best of all the agents tested with the D37 for most early passage cultures being around one tenth of the plasma C x T. Friedman et al. (1988) found the medulloblastoma lines D283 Med and Daoy were sensitive to active cyclophosphamide analogues. The ID90's of D283 Med and Daoy to 4-hydroperoxycyclophosphamide (22.47 ylM and 29.90 jtM respectively for a 1 h incubation) are much lower than D37 values measured for our early passage medulloblastoma cells, with the exception of MD 12. This is not unexpected as it is common for established cell lines to be more sensitive to cytotoxic drugs than early passage tumour cells (Weisenthal, 1981;Tveit et al., 1981).
Additionally, Friedman et al. (1988) found BCNU to be ineffective in vivo, which the authors suggested was due to the presence in the xenografts of high levels of 06-alkylguanine-DNA alkyltransferase (Schold et al., 1989). Sensitivity to BCNU was demonstrated in 7/12 of the early passage cultures tested in this study. Our results are similar to those obtained by Rosenbloom et al. (1983) and Kornblith et al. (1981). Medulloblastomas as a group are more sensitive to BCNU than the gliomas but exhibit the same type of doseresponse curves to BCNU. These curves are characterised by a small response at low doses of drug, with a greater degree of sensitivity exhibited above a limiting concentration. This point of inflection on the dose-response curve unfortunately often occurs above C x Tm.. Reports of the possible benefits obtained by intracarotid administration of this short-lived compound (Feun et al., 1985;Watne et al., 1990) are consistent with this high dose requirement. Because of the identification of the epipodophyllotoxins as agents active against medulloblastoma and the availability of three early passage cultures with marked clinically acquired resistance to these agents, the biochemical basis of sensitivity/ resistance was investigated.
A number of authors have characterised the basis of epipo-dophyllotoxin resistance in non-medulloblastoma cell lines and found it to be due to alterations in drug transport and associated with 'pleiotropic drug resistance' (Yalowich & Ross, 1985). Although the medulloblastoma cells exhibited cross-resistance between epipodophyllotoxins, m-AMSA and Mfs, there was no cross-resistance with Vcr, a drug usually associated with multiple drug resistance. Further experiments showed that the resistance could not be reversed by verapamil, nor was there any meaningful difference between steady-state intracellular drug concentrations in the different early passage cultures. Thus the phenomenon of 'pleiotropic drug resistance,' or a more specific drug transport mechanism, was excluded as the mechanism of resistance in our early passage cultures.
Cross-resistance between VM-26 and m-AMSA suggested that modifications in amount or activity of topoisomerase II may be responsible for the difference in drug sensitivity. Both epipodophyllotoxins and m-AMSA, together with anthracyclics, ellipticine and actinomycin D have been shown to inhibit topoisomerase II activity and induce the formation of enzyme-linked DNA strand breaks (Zwelling, 1985). The level of protein-linked DNA breaks is most closely associated with cytotoxicity (Rowe et al., 1986), whilst inhibition of enzyme actually appears to be less important. The amount of protein-linked DNA strand breaks found and the subsequent cytotoxicity appear to be related to the amount of enzyme present; cells with a higher enzyme content as measured by either immune serum or activity assays are more sensitive to the topoisomerase II active drugs (Pommier et al., 1986;Per et al., 1987). This, however, is not always the case. It has been proposed that an alteration of activity (Glisson et al., 1986) or differential activity of two different forms of the enzyme explains decreased protein-linked DNA strand breaks in resistant cells . More recently, a second form of topoisomerase II, the p180 form, or topoisomerase II P has been characterised and this may explain non-MDR mediated resistance to topoisomerase II a inhibitors (Drake et al., 1989). We have identified heavier forms of topoisomerase II in MD 7. However, the topoisomerase II P form described is not identified by the monoclonal antibody used in this study. This suggests that the heavier form identified in MD 7 may be transformed topoisomerase II a (Dr L.F. Liu, personal communication). Further characterisation of these forms is beyond the scope of this study.
The level of protein-linked DNA strand breaks induced by epipodophyllotoxins in medulloblastoma early passage cultures was not indicative of cytotoxicity. The profile of resistant early passage culture MD 5 was typical of reported resistant non-medulloblastoma cell lines: very low crosslinking, undetectable enzyme activity. It was also the slowest growing culture. The other epipodophyllotoxin resistant early passage cultures (MD 7, MD 11) however, did not differ from sensitive cultures or the hypersensitive culture (MD 12) in protein-linked DNA strand break activity or enzyme activity. Drug resistance could not be explained by differences in repair of cross-linking.
Protein-linked DNA strand breaks in the presence of VM-26 increased up to about 1 h, and then decreased. At 4 h there were fewer K+-SDS precipitable counts than at earlier time points although VM-26 was still present and levels in controls did not decrease. It seems unlikely that this is due to repair in the presence of drug. It may be an indication that programmed cell death has been initiated and the relevant endonuclease has begun the process of DNA fragmentation, documented as an early stage of this phenomenon (Wyllie, 1987). After 24 h continuous exposure to VM-26 the precipitable counts approach control values indicating either that the cells are refractory to further damage and complete repair of lesions has been carried out, or extensive degradation of the DNA attached to protein has occurred. This will be further investigated by gel electrophoresis of DNA from these 24 h treated cells.
In conclusion, our study demonstrates a heterogeneity of response of a number of early passage medulloblastoma cultures to the chemotherapeutic agents tested. The epipodo-phyllotoxins, in particular VM-26, were the best class of chemotherapy agents tested, while some drugs in clinical use, were found to be ineffective. In vitro chemosensitivity testing of early passage medulloblastoma cultures, more representative of the tumour than established cell lines, provides a method for the selection of agents for use in chemotherapy regimens. With the current trend for pre-irradiation chemotherapy and use of chemotherapy alone in babies and very young children, this should be a valuable method, but the time interval necessary to test a battery of chemotherapeutic agents and the limited amount of material often available for multiple drug testing detract from its clinical applicability. Manipulation of early passage cultures, with cell growth factors, providing that this does not induce changes in genotype, may provide a rapid in vitro method which would allow selection of a patient specific chemotherapy protocol.