Eradication of Human Medulloblastoma Tumor Xenografts with a Combination of O⁶-Benzyl-2′-deoxyguanosine and 1,3-Bis(2-chloroethyl)-1-nitrosourea

INTRODUCTION

Inhibitors of MGMT3 potentiate the cytotoxic effect of chloroethylating and methylating antitumor drugs that produce O⁶-substituted guanine adducts. Such potentiation has been shown against a variety of tumors both in culture (1, 2, 3, 4, 5, 6, 7) and in animal models (8, 9, 10) using the prototype MGMT inhibitor BG, which is currently undergoing clinical trials (11) . BG is one of the most potent compounds in suppressing MGMT activity in vitro and has an ED₅₀ of <0.2 μm(12) . However, BG is only marginally soluble in aqueous solvents and has a short half-life in rodents and humans (13 , 14) . Rapid oxidation to O⁶-benzyl-8-oxoguanine, an equally potent MGMT inhibitor, accounts for part of its metabolic clearance and also for its persistent inhibitory capacity in vivo (13) . Decomposition involving the loss of the benzyl group and inhibitory activity is believed to be a prevailing pathway in the metabolism of BG, accounting for as much as 60% of metabolic clearance in humans (14) . These observations have stimulated the development of additional inhibitors with ED₅₀s comparable to that of BG but with greater solubilities in aqueous solvents to ensure rapid systemic distribution. Persistence of the parent compound or active metabolites to ensure prolonged suppression of MGMT, lack of toxicity due to the parent compound or its metabolites, and possibly expression of activity against mutant MGMT proteins that are resistant to BG (15, 16, 17) are also desirable features of a new generation of MGMT inactivators. In this context, we have tested several 9-substituted derivatives of BG, and we have obtained most promising results with dBG (18) . This compound, which has an ED₅₀ of 2 μm in cell-free systems and 0.5 μm in HeLa cells (12) , was expected to be less effective than BG in suppressing the growth of tumor xenografts in athymic mice. However, dBG was as effective as BG in delaying the growth of medulloblastoma xenografts (Daoy) in mice when used at equimolar doses to BG (8) . Because the dBG dose can be escalated more easily than BG, at least in the mouse model, we tested various doses of dBG to optimize the potentiation of BCNU. In previous experiments, we have shown that dBG is relatively resistant to metabolism as compared with BG (19 , 20) . dBG persists in the circulation for about 4 h after administration to rodents and reduces the MGMT activity in xenografted tumors to about 1% of its base value for at least 16 h after its administration (18) . Conversion of dBG to BG and to O⁶-benzyl-8-oxoguanine may account for the enhanced and prolonged suppression of MGMT even after the clearance of the parent nucleoside from circulation (8 , 18) that is required to effectively cross-link DNA after treatment with bifunctional nitrosoureas (21 , 22) . In this study, we demonstrate that optimal doses of dBG and BCNU in combination can actually eradicate human CNS tumor xenografts in the athymic mouse model.

MATERIALS AND METHODS

Chemicals.

dBG was synthesized and purified according to previously published methods (23 , 24) . ³H-labeled methylated DNA was prepared as described previously (20 , 25) with [³H-CH₃]MNU (specific activity, 17.5 Ci/mmol).

Animals.

Four-week-old BALB/c nu/nu athymic mice were purchased from Simonson (Gilroy, CA). Mice were maintained under barrier conditions and given sterilized food (Harlan Teklad laboratory diet) and water.

Tumor Lines.

Daoy, a hyperdiploid human medulloblastoma line, grows s.c. in athymic mice, with a doubling time of 4.9 days. Tumors grow upon injection of shredded tumor fragments. When growing as a xenograft in athymic mice, its untreated MGMT level is 375 ± 25 fmol/mg protein.

Drug Treatment.

All treatments were administered i.p. at a volume of 20–25 ml/m² (approximately 0.2 ml/mouse). dBG was dissolved in 40% PEG 400 and 60% PBS. The pH of the solvent was corrected to pH 7.0 with sodium bicarbonate before the addition of the drug. BCNU was administered in 5% ethanol in water. Drug doses were calculated in mg/m² using the formula meters (m) = weight (g)^2/3 × K × 10⁻⁴, where K is 10.5 for mice (26) . In animals of 20 ± 2 g used in this study, the area surface is approximately 2.6 times the weight of the animal.

Treatment of s.c. Tumors.

Fifty μl of Daoy tumor were implanted in the left flank of 5-week-old athymic mice weighing between 18 and 22 g. Visible tumors appeared in most of the animals within 3 weeks after implantation. The tumors were subsequently measured in two perpendicular dimensions, and their volumes were estimated using the formula (α² + β)/2, where α is the shorter of the two dimensions, and β is the longer of the two dimensions. Treatment was administered to animals with tumors ranging between 200 and 300 mm³. Tumors were measured every other day until their volumes exceeded five times the volume of the tumor at treatment. The data were analyzed using the Wilcoxon rank-sum test, comparing the time from treatment to five times treatment volume in individual animals in each of the groups. Growth delay was the difference between the median time to five times treatment volume in the treatment group minus the median time to five times treatment volume in the control group. The number of tumor regressions (the number of tumors with a volume less than that on the treatment day) occurring in each group was also determined. Groups were compared using the two-tailed Fisher’s exact test. Two control groups received 40% PEG and 60% PBS followed by 5% ethanol/water for 1 h or 40% PEG and 60% PBS and BCNU in 5% ethanol. Between 10 and 12 animals were used in each experimental group. All treatments were repeated, but the results of duplicate treatments are not reported unless significant differences in the tumor delay or the number of animal deaths were observed.

Determination of Acute Toxicity.

The effect of BCNU on the intestinal epithelium was determined in athymic mice bearing Daoy tumor xenografts 48 h after receiving each of the following treatments: (a) PEG/ethanol, (b) dBG (133 mg/m²)/ethanol; (c) dBG (133 mg/m²)/BCNU (17 or 25 mg/m²); (d) dBG (200 mg/m²)/BCNU(17 or 23 mg/m²); and (e) dBG (300 mg/m²)/BCNU (11 or 23 mg/m²) . Forty-eight h after the treatment, animals were killed, and their jejunums were removed, cut longitudinally, and fixed in formalin. Sections were stained with H&E and Oncor ApoTag (Oncor, Gaithersburg, MD) for apoptosis and examined microscopically. Degrees of toxicity were determined by scoring for inflammation, apoptotic bodies, mitosis, and hemorrhage.

Tissue MGMT Assay.

MGMT was determined in tumors and intestinal epithelium 2 h after the administration of various doses of dBG. Animals bearing Daoy xenografts were injected i.p. with various doses of dBG in 40%PEG and 60% PBS. Animals were sacrificed after 2 h, and tumors and the intestinal epithelium were removed. Tumors were also removed from animals 24 h after treatment with dBG at 133, 200, and 300 mg/m². Three animals were used for each of the doses selected. Tissue preparation for MGMT determinations and the MGMT assay were performed as described previously (25) .

RESULTS

Treatment Efficacy of s.c. Injections and Effect of dBG Dose.

The median time from treatment to five times the tumor volume at treatment for the Daoy medulloblastoma was 9 days when tumor-bearing animals were injected with the vehicle (40% PEG and 60% PBS and 5% ethanol) and 8.8 days when animals were treated with the same vehicle plus BCNU (23 mg/m²). On the other hand, when animals with Daoy tumors were treated with various combinations of dBG and BCNU, the median time to five times the tumor volume at treatment was markedly greater (Table 1)<$REFLINK> , and growth delay (as compared to treatment with BCNU only) was highly dependent on the dBG and BCNU doses. A dose of 133 mg/m² dBG in combination with 25 mg/m² BCNU resulted in a tumor growth delay of 30.2 days, with one death and an average weight loss of 5.3%. A dose of 200 mg/m² dBG in combination with doses of BCNU at 17 and 23 mg/m² resulted in tumor growth delays of 38.4 days and more than 81 days, respectively (Fig. 1)<$REFLINK> , with no deaths and only a modest average weight loss of 4.2% and 9.0%, respectively. A 300 mg/m² dose of dBG in combination with BCNU (17 mg/m²) produced a growth delay of only 14.4 days, with treatment-related mortality of 4 out of 12 animals and an average weight loss of 12.9%. Reduction of the BCNU dose to 11 mg/m² produced similar results. The combination of dBG (200 mg/m²) and BCNU (17 mg/m²) was curative for 3 of 11 animals with no measurable tumor for more than 90 days after treatment. Eight of 12 animals were tumor-free survivors for at least 90 days after treatment when the BCNU dose was escalated to 23 mg/m² in combination with 200 mg/m² dBG . There were no 90-day survivors after treatment with any of the other dosage combinations.

Effect of dBG Dose on MGMT Levels in Tumor and Normal Tissue and Associated BCNU-related Toxicity.

Inhibition of MGMT activity in Daoy tumors implanted in athymic mice to levels as low as 8% of the base value was accomplished within 2 h after administration of 100 mg/m² dBG (Fig. 2)<$REFLINK> . Doses higher than 150 mg/m² dBG reduced the activity to less than 10 fmol/mg protein, a state that is considered comparable to the mer-negative phenotype (nonexpressive ). MGMT activity remained suppressed for at least 16 h after dBG administration. At doses of 134, 200, and 300 mg/m² dBG, the levels of MGMT in the tumor were 3.5 ± 2%, 1.6 ± 1%, and less than 1% of the baseline level at 16 h after treatment. However, at 24 h, the MGMT levels were elevated to 20 ± 3%, 15 ± 2%, and 14 ± 4% of the baseline, respectively. The MGMT activity of the intestinal epithelium of the host was more difficult to suppress by dBG and remained nearly unaffected at doses less than 100 mg/m². Suppression of MGMT activity in the intestinal epithelium was observed at dBG doses greater than 150 mg/m², but residual activity (12–18 fmol/mg protein) was still detectable at 300 mg/m². Acute toxicity of BCNU administered 1 h after dBG was usually manifested as early as 24 h by weight loss. Pathological changes such as inflammation and expansion of lumina propria of the villi were seen in the intestinal epithelium of animals treated with the highest dose of dBG (300 mg/m²) and a dose of BCNU as low as 11 mg/m². No significant toxicity or animal acute deaths were observed at dBG doses of ≤200 mg/m², despite the escalation of BCNU from 17 to 23 mg/m² (Table 2)<$REFLINK> . Hemorrhage and severe villi disorganization were evident only in animals treated with 300 mg/m² dBG and 23 mg/m² BCNU 48 h after treatment. No other gross lesions were identified in moribund animals after treatments with toxic combinations of dBG and BCNU.

DISCUSSION

Nitrosoureas were originally chosen for the treatment of CNS tumors because of their physicochemical properties, which allow them to readily penetrate the blood-brain barrier (27) . Although chemotherapy with chloroethylating agents (such as BCNU) and, most recently, with methylating agents (such as procarbazine or temozolomide) can be effective in individual cases, it has been difficult to demonstrate a consistent and significant benefit from chemotherapy in patients with primary CNS tumors with these drugs (28, 29, 30) .

Failure of nitrosoureas against CNS tumors results from tumor cell resistance to DNA damage impacted by these drugs (31) . Chloroethylnitrosoureas lead to DNA interstrand cross-links by an initial rapid reaction with the O⁶ position of guanine, followed by a rearrangement and a subsequent slow reaction with cytosine on the complementary strand (32 , 33) . Repair of the initial O⁶-chloroethylguanine adduct by MGMT prevents cross-link formation (34 , 35) . Thus, MGMT expression has been shown to be important in the resistance of human CNS tumors to nitrosoureas. In a retrospective study of patients with anaplastic gliomas who had been treated on various protocols with radiation therapy and BCNU, Belanisch et al. (36) found that patients whose tumors were “low” in MGMT activity (by immunohistochemistry) had a significantly better outcome than patients with “high” MGMT activity. This relationship remained statistically significant when other known prognostic factors were considered. Jaeckle et al. (37) has recently analyzed the largest subset of patients from the Belanisch series (the Southwest Oncology Group patients), all of whom were treated on the same protocol. The statistical relationship between MGMT expression and outcome was even stronger in this subset, despite the fact that the total number of patients was smaller.

Tumor resistance due to MGMT expression can be overcome with agents that inactivate the MGMT protein and reduce the efficiency of repair of O⁶-chloroethylguanine adducts. The first compound to be tested in vivo as both an inactivator of MGMT and a potentiator of BCNU was BG. This compound is capable of rapidly inactivating high levels of MGMT for prolonged time periods at relatively low concentrations (7) . When BG is administered to animals bearing a MGMT-positive, BCNU-resistant human tumor, MGMT activity in the tumor is inhibited for several hours, and during that time, the tumor becomes highly sensitive to BCNU (8 , 9) . BG is not toxic as a single agent, and Phase I trials have indicated that 100–200 mg/m² BG results in nearly complete elimination of MGMT activity in brain tumors as early as 4 h after administration.4 In glioblastomas, such activity remains low for at least 18 h after administration ,(11) .

We have previously shown that dBG, a 9-substituted derivative of BG, can be as effective as BG in potentiating BCNU against human CNS tumor xenografts in athymic mice (8) , despite the difference in the ED₅₀s of the two compounds, which favors BG as a more effective inhibitor of MGMT and predicts that BG might be more effective than dBG in potentiating the antitumor activity of BCNU or other alkylating chemotherapeutic drugs. The reason for the discrepancy between the in vitro and in vivo results is not fully understood. Examination of the pharmacokinetics and metabolism of dBG in rodents has shown that the unexpectedly robust capacity of dBG to potentiate BCNU against tumors is most likely due to its rapid systemic distribution to tissues and its metabolic conversion to compounds that are more active as MGMT inhibitors than dBG itself. In rats, dBG is converted to BG, which is found in the circulation and in several tissues examined at concentrations peaking between 2 and 6 h after dBG administration (20) . Because MGMT levels are nearly depleted in both liver and tumor 1 h after the administration of dBG, it is speculated that dBG itself reacts with the MGMT, causing its initial decline. A subsequent decline in activity and prolonged suppression to about 1% of the baseline activity for at least 16 h after dBG treatment probably results from the generation of BG and its metabolite O⁶-benzyl-8-oxoguanine (13 , 19) . Such potent and extensive suppression is thought to lead to maximum effectiveness of cross-linking after BCNU treatment (21 , 22) . Thus, dBG in effect acts a pro-drug for the more active MGMT inhibitors BG and O⁶-benzyl-8′-oxoguanine, and the greater solubility and tissue distribution of dBG enhance its therapeutic effect in combination with BCNU.

In this study, we have demonstrated that dBG in combination with BCNU not only delays the growth of s.c. MGMT-positive tumors but can also cause their eradication in a large percentage of animals receiving treatment at optimal doses. A dose of 200 mg/m² dBG appears to reduce MGMT activity in the tumor to a degree and duration that allow maximum efficacy of BCNU without excessive toxicity. An equivalent therapeutic effect has not been found with BG and BCNU at any of the doses tested thus far in animals (8 , 9) . One of the recognized problems with MGMT inhibitors, including dBG, is the potentiation of BCNU toxicity in the host. BCNU is tolerated well as a single agent in athymic mice at doses as high as 75 mg/m². However, coadministration of BG at a dose of 240 mg/m² results in 100% morbidity at doses of BCNU above 50 mg/m² (9) . A dose of 300 mg/m² dBG in combination with 11–17 mg/m² BCNU results in the death of 33% of the animals. The toxicities of dBG plus BCNU are attributed to the depletion of normal tissue MGMT activity. Thus, optimal dosage combinations of dBG and BCNU are capable of producing a profound therapeutic effect without mortality. We believe this can be explained by postulating that a dBG dose can be found that maximizes the sensitivity of the tumor to BCNU but allows normal tissue to retain some MGMT activity and resistance to BCNU effects. Thus, the important observation is that the optimal dose of this class of agents is not necessarily the highest possible dose and that adjustments of the doses of both the MGMT inhibitor and the alkylating drug can have profound effects on both efficacy and toxicity. Both the toxicity and tumor suppression data presented here also demonstrate that although the ED₅₀ dose is a good prognostic factor for predicting the effectiveness of MGMT inactivators in potentiating BCNU against tumors, it is only a rough guide for predicting the outcome of in vivo experiments. Metabolism and clearance of the inhibitor are of paramount importance in determining the outcome of animal and human trials. Additional effort must be devoted to understanding the metabolism of other MGMT inhibitors and preparing additional drugs that are resistant to rapid metabolic degradation and are more specific for MGMT inactivation in tumors than in normal tissues.