Thiarabine, 1-(4-Thio-ß-D-arabinofuranosyl)cytosine. A Deoxycytidine Analog With Excellent Anticancer Activity.

Thiarabine has demonstrated exceptional antitumor activity against numerous human tumor xenografts in mice, being superior to gemcitabine, clofarabine, or cytarabine. Unlike cytarabine, thiarabine demonstrated excellent activity against solid tumor xenografts, suggesting that this agent has the kind of robust activity in animal models that leads to clinical utility. Thiarabine is effective orally (bioavailability of approximately 16%) and with once per day dosing: Two characteristics that distinguish it from cytarabine. Although both the structure and basic mechanism of action of thiarabine are similar to that of cytarabine, there are many quantitative differences in the biochemical pharmacology of these two agents that can explain the superior antitumor activity of thiarabine. Two important attributes are the long retention time of the 5'-triphosphate of thiarabine in tumor cells and its potent inhibition of DNA synthesis. The biochemical pharmacology of thiarabine is also different from that of gemcitabine. Thiarabine has been evaluated in three phase I clinical trials, where it has demonstrated some activity in heavily pretreated patients with hematologic malignancies and solid tumors. Because of its impressive activity against numerous human tumor xenografts in mice, its unique biochemical activity, and encouraging clinical results in phase I clinical trials, we believe thiarabine should continue to be evaluated in the clinic for treatment of hematologic and/or solid tumors. The preclinical results to date (superior in vivo antitumor activity, oral bioavailability, and once per day dosing), suggest that thiarabine could replace cytarabine in the treatment of acute myelogenous leukemia.

Effect of expression of adenine phosphoribosyltransferase on the in vivo anti-tumor activity of prodrugs activated by E. coli purine nucleoside phosphorylase.

The use of E. coli purine nucleoside phosphorylase (PNP) to activate prodrugs has demonstrated excellent activity in the treatment of various human tumor xenografts in mice. E. coli PNP cleaves purine nucleoside analogs to generate toxic adenine analogs, which are activated by adenine phosphoribosyl transferase (APRT) to metabolites that inhibit RNA and protein synthesis. We created tumor cell lines that encode both E. coli PNP and excess levels of human APRT, and have used these new cell models to test the hypothesis that treatment of otherwise refractory human tumors could be enhanced by overexpression of APRT. In vivo studies with 6-methylpurine-2'-deoxyriboside (MeP-dR), 2-F-2'-deoxyadenosine (F-dAdo) or 9-ß-D-arabinofuranosyl-2-fluoroadenine 5'-monophosphate (F-araAMP) indicated that increased APRT in human tumor cells coexpressing E. coli PNP did not enhance either the activation or the anti-tumor activity of any of the three prodrugs. Interestingly, expression of excess APRT in bystander cells improved the activity of MeP-dR, but diminished the activity of F-araAMP. In vitro studies indicated that increasing the expression of APRT in the cells did not significantly increase the activation of MeP. These results provide insight into the mechanism of bystander killing of the E. coli PNP strategy, and suggest ways to enhance the approach that are independent of APRT.

In vivo antitumor activity of F 11782, a non-intercalating dual catalytic inhibitor of topoisomerases I and II, against a panel of human tumor xenografts.

The marked in vivo antitumor activity of F 11782 against murine experimental tumors (Kruczynski et al., Br J Cancer 83: 1516-24, 2000) has now been confirmed in a panel of human tumor xenografts. Using an intermittent schedule of six administrations over 2 weeks, F 11782 showed major activity in four of eight xenograft models. Excellent activity was noted versus the CAKI-1 (renal) model, with regressions at the two highest doses, and marked activity against DLD-1 (colon) xenografts, also resulting in regressions at the MTD. Marked antitumor activity was also observed against DU 145 (prostate) and GLC4 (small-cell lung) tumors. At optimal doses, significant T/C values ranged from 3 to 29%, with significant growth delays of 1.5-5.6, without major body weight loss. This tumor growth inhibition induced by F 11782 was sustained with time for > or = 6 weeks post implant. In contrast, no real activity was recorded against NCI-H460 (non small-cell lung) tumors and only minor responses, with optimal T/C values of < 42%, noted in the rapidly proliferating SF-295 (CNS) and LOX IMVI (melanoma) xenografts or the chemo-refractory LoVo (colon) model. Overall, this study showing a 50% response rate with definite antitumor activity across a broad spectrum, coupled with its unique mechanistic profile, has prompted the further development of F 11782.

Acyl derivatives of demethylpenclomedine, an antitumor-active, non-neurotoxic metabolites of penclomedine.

PURPOSE: The purpose of this investigation was to compare the antitumor activities of a series of acyl derivatives of 4-demethylpenclomedine (DM-PEN), the major plasma metabolite of penclomedine (PEN) observed to be an active antitumor agent in vivo and non-neurotoxic in a rat model with that of DM-PEN. METHODS: Acyl derivatives were prepared from DM-PEN and evaluated in vivo against human MX-1 breast tumor xenografts implanted subcutaneously (s.c.) or intracerebrally (i.c.). Several derivatives were also evaluated against other human tumor xenografts and murine P388 leukemia cell lines. RESULTS: Several of the acyl derivatives were found to be superior to DM-PEN against MX-1, human ZR-75-1 breast tumor, human U251 CNS tumor and the P388 leukemia parent cell line and lines resistant to cyclophosphamide and carmustine. 4-Demethyl-4-methoxyacetylpenclomedine showed inferior activity to current clinical brain tumor drugs against a glioma cell line, superior activity to temozolomide and procarbazine against the derived mismatch repair-deficient cell line, and superior activity to cyclophosphamide and carmustine but inferior activity to temozolomide against two ependymoma cell lines, all of which were implanted s.c. CONCLUSION: Proposed mechanisms of activation and action of DM-PEN and the acyl derivatives support the potential clinical superiority of the acyl derivatives.

Metabolism of 4'-thio-beta-D-arabinofuranosylcytosine in CEM cells.

Because of the excellent in vivo activity of 4'-thio-beta-D-arabinofuranosylcytosine (T-araC) against a variety of human solid tumors, we have studied its metabolism in CEM cells to determine how the biochemical pharmacology of this compound differs from that of beta-D-arabinofuranosylcytosine (araC). Although there were many quantitative differences in the metabolism of T-araC and araC, the basic mechanism of action of T-araC was similar to that of araC: it was phosphorylated to T-araC-5'-triphosphate (T-araCTP) and inhibited DNA synthesis. The major differences between these two compounds were: (i) T-araC was phosphorylated to active metabolites at 1% the rate of araC; (ii) T-araCTP was 10- to 20-fold more potent as an inhibitor of DNA synthesis than was the 5'-triphosphate of araC (araCTP); (iii) the half-life of T-araCTP was twice that of araCTP; (iv) the catalytic efficiency of T-araC with cytidine deaminase was 10% that of araC; and (v) the 5'-monophosphate of araC was a better substrate for deoxycytidine 5'-monophosphate deaminase than was the 5'-monophosphate of T-araC. Of these differences in the metabolism of these two compounds, we propose that the prolonged retention of T-araCTP is a major factor contributing to the activity of T-araC against solid tumors. The data in this study represent another example of how relatively small structural changes in nucleoside analogs can profoundly affect the biochemical activity.

Synthesis of 4'-thio-beta-D-arabinofuranosylcytosine (4'-thio-ara-C) and comparison of its anticancer activity with that of ara-C.

4'-thio-beta-D-arabinofuranosylcytosine was synthesized by a facile route in high yields. It was evaluated for antitumor activity against a panel of human tumors, both in vitro and in vivo.

Preclinical antitumor activity of 2-chloro-9-(2-deoxy-2-fluoro-beta-D- arabinofuranosyl)adenine (Cl-F-ara-A).

Cl-F-ara-A, an analog of fludarabine, was evaluated against a spectrum of tumor systems in culture and in mice. The compound exhibited significant cytotoxicity against a variety of human tumor cell lines. More importantly, the compound showed selectivity in vivo, with excellent activity being demonstrated against human colon and renal tumors. Human nonsmall cell lung and prostate tumors were also sensitive in vivo to the compound, albeit at a reduced level.

Synthesis and evaluation of several new (2-chloroethyl)nitrosocarbamates as potential anticancer agents.

Seven new (2-chloroethyl)nitrosocarbamates have been synthesized as potential anticancer alkylating agents. These compounds were designed with carrier moieties that would either act as prodrugs or confer water solubility. All compounds were screened in an in vitro panel of five human tumor cell lines: CAKI-1 (renal), DLD-1 (colon), NCI-H23 (lung), SK-MEL-28 (melanoma), and SNB-7 (CNS). Several agents showed good activity with IC(50) values in the range of 1-10 microg/mL against at least two of the cell lines. One compound, carbamic acid, (2-chloroethyl)nitroso-4-acetoxybenzyl ester (3), was selected for further study in vivo against intraperitoneally implanted P388 murine leukemia. In addition to the aforementioned compound, both carbamic acid, (2-chloroethyl)nitroso-4-nitrobenzyl ester (9) and carbamic acid, (2-chloroethyl)nitroso-2,3,4, 6-tetra-O-acetyl-1-alpha,beta-D-glucopyranose ester (24) were evaluated against subcutaneously implanted M5076 murine sarcoma in mice. None of these compounds were active in vivo.

Synthesis and structure activity relationships of 5-substituted-4'-thio-beta-D-arabinofuranosylcytosines.

Four 5-substituted (chloro, fluoro, bromo, methyl) 1-(4-thio-beta-D-arabinofuranosyl)cytosines and their alpha anomers were synthesized by a facile route in high yields. All of these nucleosides were evaluated for cytotoxicity against a panel of human tumor cell lines in vitro. Only 5-fluoro-1-(4-thio-beta-D-arabinofuranosyl)cytosine was found to be highly cytotoxic in all the cell lines and was further evaluated in vivo.

Superior in vivo experimental antitumour activity of vinflunine, relative to vinorelbine, in a panel of human tumour xenografts.

The antitumour activity of vinflunine, 20',20'-dichloro-3',4'-dihydrovinorelbine, a fluorinated Vinca alkaloid obtained by reaction in superacid media, was evaluated in comparison with vinorelbine against a series of subcutaneously-implanted human tumour xenografts. The tumours studied were established from bladder (BXF1299), pancreas (PAXF546), kidney (RXF944LX), colon (DLD-1, HT-29, TC37), central nervous system (SF-295), small cell lung (NCI-H69) and prostate (PC-3). Vinflunine or vinorelbine was administered as four weekly intraperitoneal treatments, within dose ranges of 5-80 or 0.63-10 mg/kg/injection, respectively. The overall antitumour activity of vinflunine was superior to that of vinorelbine. Vinflunine showed high activity against RXF944LX and NCI-H69 xenografts and moderate activity against PAXF546, PC-3 and TC37 tumours, achieving an overall response of 64%. This contrasts with a 27% response with vinorelbine, which proved only moderately active against RXF944LX and TC37 xenografts. These results confirm and extend our previous report of the broad spectrum of in vivo antitumour activity of vinflunine and reinforce its potential as a valuable addition to current chemotherapeutic agents.

Gene therapy of cancer: activation of nucleoside prodrugs with E. coli purine nucleoside phosphorylase.

During the last few years, many gene therapy strategies have been developed for various disease targets. The development of anticancer gene therapy strategies to selectively generate cytotoxic nucleoside or nucleotide analogs is an attractive goal. One such approach involves the delivery of herpes simplex virus thymidine kinase followed by the acyclic nucleoside analog ganciclovir. We have developed another gene therapy methodology for the treatment of cancer that has several significant attributes. Specifically, our approach involves the delivery of E. coli purine nucleoside phosphorylase, followed by treatment with a relatively non-toxic nucleoside prodrug that is cleaved by the enzyme to a toxic compound. This presentation describes the concept, details our search for suitable prodrugs, and summarizes the current biological data.

Potent antitumor activity of a novel nucleoside analogue, BCH-4556 (beta-L-dioxolane-cytidine), in human renal cell carcinoma xenograft tumor models.

Beta-L-Dioxolane-cytidine (BCH-4556) is a novel anticancer nucleoside analogue with a stereochemically unnatural beta-L configuration. This compound was previously shown to have a potent antitumor activity in human prostate and hepatocellular xenograft tumor models (K. L. Grove et al., Cancer Res., 55: 3008-3011, 1995). Herein, we extended the efficacy validation of BCH-4556 to renal cell carcinoma (RCC) cell lines both in vitro and in vivo. In vitro cytotoxicity and proliferation inhibition determinations in human RCC cell lines CAKI-1, CAKI-2, 786-0, and A498 produced IC50 concentrations ranging from 15-35 nM. In vivo antitumor activity was consistent with the in vitro sensitivity. BCH-4556 was very effective in human RCC tumor xenograft models, including CAKI-1, A498, RXF-393, and SN12C carcinomas. Very good responses were observed in animals bearing CAKI-1, A498, and RXF-393 RCC tumors given i.p. doses of 10, 25, and 50 mg/kg twice a day for 5 days, with complete regression recorded in most of the animals tested. Curative activity was also observed, with 40-60% of animals remaining tumor free in all three RCC models at the day of study termination. Significant tumor shrinkage was also evident in the SN12C model. BCH-4556 efficacy evaluation in the orthotopic subrenal capsule tumor models demonstrated a potent tumor growth inhibition against human CAKI-1 xenografts and tumor stasis against mouse Renca tumors. BCH-4556 was also effective in inhibiting the growth of rebound CAKI-1 tumors after the administration of a second treatment cycle. The observed antitumor activity of BCH-4556 in several RCC human solid tumor xenografts, including the lethal RXF-393 model, warrants further investigation of this novel nucleoside analogue in clinical trials of RCC.

In vivo gene therapy of cancer with E. coli purine nucleoside phosphorylase.

We have developed a new strategy for the gene therapy of cancer based on the activation of purine nucleoside analogs by transduced E. coli purine nucleoside phosphorylase (PNP, E.C. 2.4.2.1). The approach is designed to generate antimetabolites intracellularly that would be too toxic for systemic administration. To determine whether this strategy could be used to kill tumor cells without host toxicity, nude mice bearing human malignant D54MG glioma tumors expressing E. coli PNP (D54-PNP) were treated with either 6-methylpurine-2'-deoxyriboside (MeP-dR) or arabinofuranosyl-2-fluoroadenine monophosphate (F-araAMP, fludarabine, a precursor of F-araA). Both prodrugs exhibited significant antitumor activity against established D54-PNP tumors at doses that produced no discernible systemic toxicity. Significantly, MeP-dR was curative against this slow growing solid tumor after only 3 doses. The antitumor effects showed a dose dependence on both the amount of prodrug given and the level of E. coli PNP expression within tumor xenografts. These results indicated that a strategy using E. coli PNP to create highly toxic, membrane permeant compounds that kill both replicating and nonreplicating cells is feasible in vivo, further supporting development of this cancer gene therapy approach.

Stabilized analogs of thymopentin. 1. 4,5-Ketomethylene pseudopeptides.

The pentapeptide, thymopentin (Arg1-Lys2-Asp3-Val4-Tyr5) is known for its activity as an immunomodulating drug, but with limited half-life in plasma. In this first paper of a series of three studies, the synthesis of analogs stabilized at the peptide bond between the C-terminal amino acids via insertion of a ketomethylene moiety is described. N-Blocked pseudopeptides containing Val(k)Phe, Ala(k)Phe, and Val(k)Val units were prepared and attached to chloromethyl Merrifield resin via the carboxy terminal. Removal of the N-BOC group by trifluoroacetic acid was followed by sequential coupling with N-BOC dipeptides of aspartic acid to yield resin-bound N-BOC pseudotetrapeptides. Removal of N-BOC and coupling with N-BOC-r-N-tosylarginine followed by total cleavage of blocking groups and resin by HF afforded the target pseudopentapeptides. The analogs were found to compete favorably with thymopentin for binding to CEM cells, but binding was reduced by about 20-30% on average. All analogs showed significant enhancement of half-life versus thymopentin in mouse serum, but most showed only modest improvement in human serum. Insertion of proline or norleucine at position 2 in the chain caused a substantial increase in half-life (3-4-fold), while N-methylnorleucine conferred complete stability in the analogs.

Stabilized analogs of thymopentin. 2. 1,2- and 3,4-ketomethylene pseudopeptides.

In this second paper in a series of three studies of stable analogs of thymopentin (Arg1-Lys2-Asp3-Val4-Tyr5), the synthesis of analogs stabilized at peptide bonds 1,2 and 3,4 via insertion of ketomethylene units is described. A tris(carbobenzyloxy)arginyl(k)norleucine pseudopeptide was synthesized and coupled to Asp-Val-Phe-resin units followed by HF cleavage to prepare Arg(k)Nle-Asp-Val-Phe analogs. Preparation of N-BOC Asp(k)Val and N-BOC Asp(k)Ala units followed by coupling to Phe- or Tyr-resin units provided resin-bound pseudotripeptide substrates for attachment of various arginyl dipeptides. Cleavage from the resin afforded 3,4-ketomethylene-stabilized pseudopeptide analogs of thymopentin. The Arg-Lys-Asp(k)Val-Phe and Arg-Lys-Asp(k)Val-Tyr analogs were more strongly bound to CEM cells than thymopentin itself. There was significant enhancement of stability in serum for the analogs, especially those containing Arg(k)Nle or Arg-NMeLys moieties at the 1,2-peptide bond.

Stabilized analogs of thymopentin. 3. Evaluation of ketomethylene pseudopeptides for antiarthritic properties.

This study analyzed the role of ketomethylene pseudopeptides of thymopentin as potential agents for the treatment of arthritis. The analogs were tested in vivo using assessment of inflammation and antibody production in the mouse type II collagen arthritis model and the rat adjuvant arthritis model. The compounds were also tested for immune-potentiating activity in vitro using induction of the lymphocyte marker, Thy-1.2, in mouse spleen cells and stimulation of T-cell proliferation. The results show that certain of the compounds exhibit disease-remitting properties for arthritis as evidenced by reduction of paw swelling in the mouse and rat models and decreased incidence of disease in the mouse model. The active compounds were dose specific and represented a range in efficacy. In spite of effects on arthritis, type II collagen antibody levels were not altered in the mouse model. Selected compounds also exhibited immune potentiating properties as evidenced by induction of Thy-1.2 expression and stimulation of T-cell proliferation. The absence of effect of the compounds on type II collagen antibody production suggests that the antiarthritic activity of the effective compounds results from alteration of cell-mediated immunity.

4-Demethylpenclomedine, an antitumor-active, potentially nonneurotoxic metabolite of penclomedine.

Penclomedine [3,5-dichloro-4,6-dimethoxy-2-(trichloromethyl)pyridine], an antitumor agent, is currently in Phase I clinical trials and is believed to be a prodrug. In these studies, cerebellar effects have been dose limiting. Previous studies identified 4-demethylpenclomedine (4-DM-PEN) as the major plasma metabolite in rodents and humans. 4-DM-PEN was demonstrated to be an antitumor-active metabolite of penclomedine in vivo when evaluated against the penclomedine-sensitive MX-1 human breast tumor xenograft implanted either s.c. or intracerebrally and is believed to be on the metabolic activation pathway of penclomedine. Because earlier studies revealed an absence of neurotoxic cerebellar effects for 4-DM-PEN in contrast to penclomedine in a rat model, this metabolite may be a candidate for an alternative to penclomedine in the clinic for treatment of breast cancer or brain tumors, if the cerebellar effects of penclomedine preclude its further clinical development. Because neither penclomedine nor 4-DM-PEN were very active in vitro, the metabolism of penclomedine was also investigated using rat liver microsomes in an attempt to identify the ultimate active form of the drug. Metabolites and putative metabolites were prepared by chemical synthesis for antitumor evaluation in vitro and in vivo. A reductive metabolite, alpha,alpha-didechloro-PEN, was observed to be much more cytotoxic than penclomedine or 4-DM-PEN in vitro, but evaluation of this and the other metabolites and putative metabolites in vivo against the MX-1 tumor failed to identify any active metabolite among the structures evaluated other than 4-DM-PEN. The limited activity of 4-DM-PEN in vitro indicates that it, like penclomedine, is also a prodrug, demonstrating a need for additional studies on the metabolic activation of penclomedine to identify the ultimate active form of the drug.

Analogues of methotrexate in rheumatoid arthritis. 1. Effects of 10-deazaaminopterin analogues on type II collagen-induced arthritis in mice.

Carbonation of the dianions (LDA) of 5-methylthiophene-2-carboxylic, 2-methylpyridine-5-carboxylic, and 3-methylpyridine-6-carboxylic acids provided the respective carboxy heteroarylacetic acids. The crude diacids were directly esterified in MeOH-HCl to afford the diesters. Alkylation of the sodio anions with ethyl iodide yielded the appropriate alpha-ethyl diesters. The anions of the various diester substrates were then alkylated by 2,4-diamino-6-(bromomethyl)-pteridine followed by ester saponification at room temperature to afford the respective 2,4-diamino-4-deoxy-10-carboxy-10-deazapteroic acids. The 10-carboxyl group was readily decarboxylated by heating in DMSO at temperatures of 110-135 degrees C to give the diamino 10-deaza heteropteroic acid intermediates. Coupling with diethyl L-glutamate followed by ester hydrolysis afforded the target aminopterins. The analogues were evaluated for antiinflammatory effect in the mouse type II collagen model. The thiophene analogue of 10-ethyl-10-deazaaminopterin was found to be an effective inhibitor in terms of reduced visual evidence of inflammation and swelling as determined by caliper measurement.

Analogues of methotrexate in rheumatoid arthritis. 2. Effects of 5-deazaaminopterin, 5,10-dideazaaminopterin, and analogues on type II collagen-induced arthritis in mice.

Twenty-six compounds derived from the 5-deaza- and 5,10-dideazaaminopterin series of aminopterin analogues were evaluated for antiarthritic activity in the mouse type II collagen model. New compounds in the 5-deaza series were prepared by alkylation of an appropriate N-substituted (4-aminobenzoyl)-L-glutamic acid dialkyl ester or N-(5-amino-2-thenoyl)-L-glutamate diester with a 2,4-diamino-5-alkyl-6-(bromomethyl)-5-deazapteridine. The resultant 5-deazaaminopterin diesters were saponified to provide the target 5-deaza analogues. 5,10-Dideazaaminopterins were synthesized by similar alkylation of the carbanions of appropriate 4-carboxyphenylacetic, (5-carboxy-2-thienyl)acetic, or (5-carboxy-2-pyridyl)acetic acid dimethyl esters. The diesters of the 2,4-diamino-4-deoxy-10-carboxy-5,10-dideazapteroic acid types so obtained were saponified and then readily decarboxylated by heating in Me2SO solution to provide the 2,4-diamino-5,10-dideazapteroic acid-type intermediates. Peptide coupling with diethyl L-glutamate followed by ester hydrolysis at room temperature afforded the new 5,10-dideazaaminopterin analogues. 5-Deazaaminopterins bearing an alkyl substituent at the 5-position were generally quite effective as antiinflammatory agents. Thus 5-propyl-5-deazaaminopterin, 5-methyl-10-propargyl-5-deazaaminopterin, 5-methyl-10-allyl-5-deazaaminopterin, 5-ethyl-5-deazamethotrexate, and 2,5-disubstituted thiophene analogue of 5-methyl-5-deazaaminopterin showed potencies greater than methotrexate by intraperitoneal or oral administration and were active over a considerably broader dose range. Useful activity in the 5,10-dideaza series was only observed for 5,10-dideazaaminopterin and its 10-methyl analogue. Alkyl substitution at C-5 or C-10 was generally detrimental to antiinflammatory activity in this series.

Preclinical antitumor activity of bizelesin in mice.

Bizelesin (U-77779, NSC 615291), a synthetic analogue of the cytotoxic antibiotic CC-1065, is a bifunctional alkylating agent that produces DNA interstrand cross-links. Bizelesin was evaluated for antitumor activity against a broad spectrum of syngeneic murine tumors and human tumor xenografts in mice. Systemic drug administration produced >6.7 log10 cell kill against i.p. implanted P388 and L1210 leukemias and 80% tumor-free survivors against s.c. implanted L1210. Against i.p. implanted B16 melanoma, i.p. drug administration produced a 158%; increase in life span with 25% tumor-free survivors, whereas i.v. drug administration produced only a 67% increase in life span with no tumor-free survivors. More than 1.0 log10 cell kill was observed at low microgram/kg doses in several human tumor models representing diverse histiotypes (CAKI-1 renal, LX-1 lung, HT-29 colon, LOX IMVI and UACC-62 melanomas, and MX-1 mammary). Less than 1.0 log10 cell kill was exhibited in other tumor models (Lewis lung, colon 38, pancreatic 02, MCF7 mammary, and SK-MEL-3 melanoma). Bizelesin was optimally active when administered i.v. Although antitumor activity was independent of the schedule of administration, greater total doses were tolerated on the more prolonged schedules in any given experiment. Therapeutic doses of bizelesin did not produce delayed deaths, which had previously been observed for the parent compound CC-1065. However, recovery of lost weight was not attained until 16-30 days posttherapy. Bizelesin was as active against murine leukemia sublines resistant to cisplatin, melphalan, and 1,3-bis-(2-chloroethyl)-1-nitrosourea as against the parental line but was totally inactive against a doxorubicin-resistant subline. The complete cross-resistance of the doxorubicin-resistant subline to bizelesin suggests that bizelesin may be a substrate for the efflux pump that causes multidrug resistance. Due to its breadth of antitumor activity, potency, unique mechanism of action, and lack of cross-resistance with other alkylating agents, bizelesin was selected for development in clinical trials by the National Cancer Institute and the Upjohn Company. Toxicological studies and pharmaceutical development have been completed, and clinical trials are planned to start in the summer of 1996.