Optimizing antimetabolite-based chemotherapy for the treatment of childhood acute lymphoblastic leukaemia.

Abstract

Therapy with antimetabolites forms an important part of the treatment of childhood acute lymphoblastic leukaemia (ALL). However, the use of these agents varies for dose, route and schedule of administration world-wide. Although the overall survival rates for children receiving predominantly antimetabolite-based therapy ( Veerman et al, 1996 ) and regimens containing anthracyclines and alkylating agents ( Reiter et al, 1994 ) are similar, the pharmacological guidance of antimetabolite-based therapy in childhood ALL has been shown to be associated with improved survival ( Evans et al, 1998 ). Moreover, the high lymphoid marrow relapse-free survival reported by Winick et al (1996) for children receiving a treatment regimen containing intensive oral methotrexate, oral 6-mercaptopurine and intravenous etoposide and cytosine arabinoside are among the highest reported to date. The optimization of antimetabolite-based therapy using schedules of administration which are based on the known clinical and cellular pharmacological properties of these agents may further improve the survival of childhood acute leukaemia. The purpose of this review will be to discuss the possible optimization of methotrexate, cytosine arabinoside, thiopurines and epipodophyllotoxins in the therapy of childhood ALL. For this, the discussion will focus on the known mechanisms of cytotoxicity of these agents in vitro, on their activity as single agents and on studies of the clinical and cellular pharmacology of these agents in the therapy of childhood ALL. Methotrexate (MTX) is a classic antifolate that can act at three cytotoxic loci within the cell, namely dihydrofolate reductase (DHFR), thymidylate synthase (TS) and aminoimidazole carboxymide ribonucleotide (AICAR) formyltransferase ( Fig 1). MTX requires specialized membrane transport via the reduced folate carrier (RFC) and is a substrate for intracellular polyglutamation by the enzyme folylpolyglutamate synthetase (FPGS), a process during which up to an additional six glutamate residues are added to form MTX polyglutamates (MTXPG). In contrast, MTXPG are subject to hydrolysis by the enzyme folylpolyglutamyl hydrolase (FPGH). Polyglutamation is known to enhance the intracellular retention of MTX, and the polyglutamation of MTX, especially to long-chain MTXPG with four to six glutamyl residues (MTXPG4−6), markedly enhances the intracellular retention of MTX and potency of the drug as an inhibitor of TS and AICAR (for a review, see Allegra, 1996). Effect of MTX on TS, DHFR and de novo purine synthesis. MTXPG, MTX polyglutamates; GARFT, glycinamide ribonucleotide formyltransferase; AICAR, aminoimidazole carboxymide ribonucleotide formyl transferase; FH4, tetrahydrofolate; 10-CH0-FH4, 10-formylterahydrofolate; 5,10-CH2-FH4, 5,10-methylenetetrahydrofolate; FH2PG, dihydrofolate polyglutamates; IMP, inosine monophosphate. The broken lines indicate loci for MTX or FH2 inhibition. For human leukaemia cell lines, MTX is cytotoxic at extracellular concentrations below 1 µm for both short- and long-term duration of exposure ( Takemura et al, 1996 ) . However, human leukaemia cell lines display lineage-specific differences with regard to total and MTXPG4−6 formation over a wide range of methotrexate concentrations ( Galpin et al, 1997 ). For NALM6 (B lineage) cells and CCRF-CEM (T lineage) cells, the formation of MTXPG is a saturable process and nears maximum rates (Emax) at extracellular MTX concentrations of 20 µm. However, the higher constitutive FPGS mRNA and protein expression for NALM6 cells allows these B-lineage cells to form sixfold higher levels of MTXPG4−6 than CCRF-CEM cells ( Galpin et al, 1997 ), a finding which is associated with greater sensitivity to MTX. Furthermore, the cytotoxicity of MTX in vitro has been shown to relate more to increases in the time of exposure than to increases in the extracellular concentration of the drug. For example, for L5178Y/Asn− murine lymphoblasts, a 10-fold increase in exposure time to MTX between 3 and 42 h resulted in a 100-fold increase in cytotoxicity. In comparison, a 10-fold increase in drug concentration only resulted in a twofold increase in cytotoxicity, and this effect was lost for exposure times of greater than 6 h ( Keefe et al, 1982 ). Resistance to MTX may be mediated by several mechanisms that have been well characterized in vitro. These include reduced uptake by the RFC, alterations in DHFR structure which result in a reduced affinity of the enzyme for MTX, increased levels of DHFR, reduced FPGS activity and increased FPGH activity ( Allegra, 1996). During single-agent studies of MTX, which were performed in the 1950s under the auspices of the Acute Leukaemia Group B, MTX was administered as an age-dependent daily oral dose of 1·25–5 mg/d and, as a single agent, MTX could induce a clinical remission (sustained for a median duration of 4 months) in 29% of children with acute leukaemia ( Frei et al, 1961 ). A subsequent Acute leukaemia Group B study compared the survival of children with ALL who received either daily oral (3 mg/m2/d) or twice weekly parenteral (30 mg/m2 per dose) MTX as monotherapy until relapse after remission induction with vincristine and prednisolone ( Acute Leukaemia Group B, 1965). Whereas the median duration of complete remission (CR) for children receiving the parenteral schedule was 17 months, with an associated 2-year survival rate of 20%, the median remission duration for children receiving daily MTX was 3 months, and no children survived for 2 years. A further indication of the potential importance of MTX scheduling came from the study of Djerassi (1967), in which 80% survival at 30 months was found for 15 children receiving post-remission induction therapy with MTX. In this study, 180–525 mg/m2 MTX was administered as an intravenous infusion over 4 h, for two consecutive days at 3-weekly intervals. Therefore, these early clinical studies indicated that the antileukaemic efficacy of MTX might be improved by either increasing the dose of and time of exposure to the drug. In modern ALL treatment protocols, MTX is used in various protocols in remission induction, consolidation, maintenance and central nervous system (CNS)-directed therapy. Oral MTX is generally used during the maintenance period for treatment of childhood ALL, with oral doses ranging from 20 to 40 mg/m2 on a weekly basis ( Chessells et al, 1995 ; Evans et al, 1998 ). For intravenous therapy, intermediate dose MTX (500–1500 mg/m2 over 24 h) forms part of the consolidation therapy ( Evans et al, 1986 ; Camitta et al, 1994 ) and high-dose MTX (HDMTX) therapy (5–33 g/m2) is used in various protocols in an attempt to effect significant MTX exposures at disease sanctuary sites such as the CNS ( Borsi & Moe, 1987). Intrathecal methotrexate is generally administered in age-dependent doses ranging from 12·5 to 15 mg/m2 ( Chessells et al, 1995 ). The clinical pharmacology of MTX has been the subject of extensive investigation. Wide variation in the absorption of oral MTX has been observed, with median peak plasma concentrations of 0·7 µm (range 0·2–2·5 µm) being achieved with an oral dose of 20 mg/m2 ( Pearson et al, 1987 ). For HDMTX and intermediate dose MTX (IDMTX) administered intravenously, large interindividual variation in plasma steady-state MTX concentrations were found. Evans et al (1986) found median steady-state plasma MTX concentrations during IDMTX (1 g/m2) to vary between 9 and 25 µm. With HDMTX, steady-state plasma MTX concentrations of 50–100 µm can be achieved with 24-h intravenous infusions of 5 g/m2 MTX ( Donelli et al, 1995 ). After an intravenous infusion, MTX is eliminated biexponentially, with a t1/2α of 3 h and a t1/2β of 8–11 h ( Borsi & Moe, 1987). Although with intrathecal MTX concentrations of up to 100 µm can be achieved in the cerebrospinal fluid (CSF), MTX so administered tends to be poorly distributed throughout the CSF ( Shapiro et al, 1975 ) and is eliminated biexponetially with a t1/2α of 5·5 h and a t1/2β of 24 h ( Bleyer & Dedrick, 1977). The clinical pharmacology of IDMTX has been studied in relation to clinical outcome for childhood ALL. Evans et al (1984) demonstrated the importance of systemic exposure to MTX, with higher relapse rates occurring in those children with higher rates of clearance after a 1 g/m2 infusion. A further study of the relationship between systemic exposure to MTX and effect in which a significantly poorer event-free survival (EFS) for children who maintained median steady-state MTX concentrations of less than 16 µm during IDMTX-based therapy was demonstrated for children with standard risk ALL ( Evans et al, 1986 ). Further evidence to suggest a relationship between methotrexate exposure and outcome in childhood ALL has been found for children with higher-risk B-precursor ALL who received consolidation therapy based on intensive cycles of IDMTX. In this study, a significantly higher EFS was found for children who achieved median end-infusion MTX concentrations of > 11 µm than those with median end-infusion MTX concentrations below this value ( Camitta et al, 1994 ). Finally, a recently completed study from St Jude's Children's Research Hospital (Memphis, USA) demonstrated a significant survival advantage for children with ALL who were randomized to receive pharmacologically guided post-remission induction therapy with MTX, cytosine arabinoside (Ara-C) and teniposide compared with conventional dosing based on body surface area ( Evans et al, 1998 ). For children receiving individualized MTX, in which steady-state plasma MTX concentrations of at least 20 µm were maintained, a significant improvement in the 5-year continuous complete remission rate of 76% compared with 66% was found for children with B-precursor ALL. The cellular factors that may influence the sensitivity of leukaemic blasts to MTX have been investigated in leukaemic blasts. Reduced expression of the RFC ( Gorlick et al, 1997 ) and increased DHFR levels ( Matherly et al, 1997 ) are more common in relapsed ALL than at presentation. Moreover, heterogeneity of DHFR content between the lymphoblasts is more common in T-cell ALL than in B-lineage ALL and is associated with a poorer prognosis for children receiving chemotherapy containing oral MTX ( Matherly et al, 1995 ). In contrast, no lineage-specific differences in either RFC expression ( Belkov et al, 1999 ) or FPGH activity ( Rots et al, 1999 ) have been found between B-precursor and T-cell ALL blasts. The influence of FPGS on the cellular pharmacology of MTX in childhood ALL has been extensively studied over the past decade. Lineage- and age-specific differences in FPGS activity have been reported, with higher FPGS activity and hence MTXPG4−6 formation found for B-lineage vs. T-lineage lymphoblasts, B-lineage lymphoblasts vs. myeloblasts and children vs. adults with B-lineage ALL ( Goker et al, 1993 ). Moreover, investigation of the relationship between systemic exposure to MTX and the formation of intracellular MTXPG4−6 by leukaemic blasts in vivo has helped to define lineage-specific differences for the relationship between clinical and cellular pharmacological parameters of MTX and pharmacodynamic measures such as target enzyme inhibition and cytoreduction. For children randomized to receive either low-dose oral MTX (30 mg/m2 every 6 h for six doses), or intravenous IDMTX (1 g/m2 over 24 h) as a single agent before conventional remission induction therapy at presentation with ALL, the formation of both total MTXPG and MTXPG4−6in vivo has been found to be higher in B-lineage blasts than in T-lineage blasts and was related to the dose of MTX received at the initial randomization and the maximum plasma concentration of MTX achieved ( Synold et al, 1994 ) . In addition, whereas 81% of patients receiving IDMTX achieved 44-h MTXPG4−6 concentrations in excess of the IC95 for de novo purine synthesis, 46% of patients receiving the oral MTX schedule achieved this level of intracellular MTXPG4−6 ( Masson et al, 1996 ). In addition, the disappearance of lymphoblasts from the peripheral blood at 96 h was also found to relate to total and MTXPG4−6 levels, MTX steady-state concentration and area under the concentration-versus-time curve (AUC), and was more likely to occur in children receiving IDMTX than oral MTX. As the absence of circulating blasts by day 8 of remission induction therapy may be associated with a more favourable outcome for childhood ALL ( Gajjar et al, 1995 ), the more effective cytoreduction conferred by IDMTX than with oral MTX may be important in terms of antileukaemic efficacy. Furthermore, as with human leukaemia cell lines in vitro, B-lineage lymphoblasts formed threefold higher total MTXPG and MTXPG4−6 than T-lineage blasts ( Galpin et al, 1997 ). Moreover, when intracellular MTXPG4−6 was related to the steady-state plasma concentration of MTX achieved with either oral or IDMTX, for both B-lineage and T-lineage blasts a steep dose–response curve was found for steady-state MTX concentrations of less than 10 µm. As with human leukaemia cell lines in vitro, near maximum levels of intracellular MTXPG4−6 were found at MTX plasma concentrations of ≈ 20 µm ( Galpin et al, 1997 ). In summary, the clinical and cellular pharmacological studies presented above suggest that MTX infusions associated with steady-state plasma concentrations of 20–25 µm would be more likely to achieve maximal intracellular levels of MTXPG4−6, and hence maximal target enzyme inhibition, in both B-lineage and T-cell lymphoblasts. For lower-dose oral schedules, even when these are administered intensively, MTXPG4−6 formation may be limited by uptake via the RFC. This may, in part, explain the concentration-dependent effect for IDMTX reported by Evans et al (1986) . Indeed, in a recent Pediatric Oncology Group (POG) phase III study of children with lower risk ALL receiving consolidation therapy based on MTX and intravenous 6-MP, a superior EFS was demonstrated for children randomized to receive intravenous IDMTX than repetitive low-dose oral MTX ( Mahoney et al, 1998 ). To maximize the antileukaemic efficacy of PK-guided MTX, it may be necessary to administer a more prolonged pharmacologically guided IDMTX infusion, so that individual patients experience non-dose-limiting toxicity. Indeed, the administration of prolonged (42 h) HDMTX infusions are tolerated by children ( Bode et al, 1987 ), and a longer duration of second haematological remission has been reported for children receiving prolonged IDMTX than for children receiving shorter HDMTX infusions ( Wolfram et al, 1993 ). Ara-C, an analogue of the nucleoside deoxycytidine, enters cells by the nucleoside transporter and requires activation via three sequential kinases to form the cytotoxic metabolite Ara-CTP (for a review, see Chabner, 1996). At extracellular Ara-C concentrations of 10 µm or less, cellular uptake, and hence formation of Ara-CTP, is limited by the nucleoside transporter, the expression of which is relatively low in B-lineage lymphoblasts when compared with T-cell ALL and acute myelogenous leukaemia (AML) ( Wiley et al, 1985 ). This factor may play an important role in determining cellular sensitivity to Ara-C as a direct relationship between the level of incorporation of Ara-CTP into DNA and cytotoxicity exists for myeloid leukaemia cell lines ( Kufe & Spriggs, 1985). At concentrations > 10 µm, Ara-C enters cells by passive diffusion and the rate of Ara-CTP formation is limited by deoxycytidine kinase ( Fig 2). This kinase, which has an affinity constant for Ara-C of ≈ 5 µm, catalyses the first step in the metabolic activation of Ara-C to Ara-CTP ( Chabner, 1996). The major cytotoxic mechanism of Ara-C has been attributed to the incorporation of Ara-CTP into DNA, with cell death resulting from apoptosis ( Chabner, 1996). Metabolism of Ara-C. Ara-U, uracil arabinoside; Ara-C, cytosine arabinoside; Ara-CMP, cytosine arabinoside monophosphate; Ara-CDP, cytosine arabinoside diphosphate; Ara-CTP, cytosine arabinoside triphosphate; NDP, nucleoside diphosphate. For human leukaemia cell lines, once the cytotoxic threshold of 0·4 µm has been exceeded, the cytotoxicity of Ara-C has been shown to be related to the concentration × time product of exposure to the drug ( Chabner, 1996). For HL60 cells and human myeloblasts, maximal cytotoxicity is found when the duration of exposure to a 10 µm concentration is increased from 12 to 24 h ( Kufe & Spriggs, 1985). Although for cell lines with acquired resistance to Ara-C the most frequent resistance mechanism described is reduced deoxycytidine kinase activity, the clinical relevance of this is not yet known ( Chabner, 1996). Evaluation of Ara-C as a single agent in childhood ALL was performed for children with acute leukaemia in first relapse. For children receiving Ara-C either as daily intravenous bolus injections of 3–5 mg/kg ( Howard et al, 1968 ) or prolonged intravenous infusions of 150–200 mg/m2/d for 5 consecutive days ( Wang et al, 1970 ) , complete remission rates (7% for ALL, 21% for AML) were low despite the occurrence of significant myelosuppression. Moreover, following the observation in the early 1980s that high-dose Ara-C could induce complete remissions in adults with relapsed–refractory AML ( Early et al, 1982 ), a single-agent study of high-dose Ara-C was performed for children with refractory leukaemia ( Ochs et al, 1984 ). Children receiving Ara-C as a continuous infusion at a dose of 5 g/m2/d for 96 h experienced severe toxicity, and, although this schedule was better tolerated for patients receiving 3·5 g/m2/d for 96 h, dose-limiting haematological toxicity of 3 weeks duration was observed. Of the 10 children studied, two patients achieved brief partial responses. Ara-C is currently used in many different dose and schedule combinations in the treatment of childhood ALL in modern protocols. For example, conventional doses of 200–300 mg/m2/d may be administered either as an intravenous bolus dose every 12 h for 5 d ( Chessells et al, 1995 ), or as an intravenous infusion of 300 mg/m2 over 4 h as part of consolidation therapy ( Evans et al, 1998 ). In addition, schedules utilizing intermediate dose Ara-C, i.e. 1 g/m2 as a continuous infusion over 24 h, have been described previously ( Harris, M.B., et al, 1998 ). High-dose Ara-C therapy, utilizing doses of up to 3 g/m2, has also been reported in the treatment of refractory ALL ( Harris, R.E., et al, 1998 ) and as part of multiagent consolidation therapy ( Reiter et al, 1994 ). Intrathecal Ara-C is administered at a dose of 30 mg/m2 (maximum 30 mg) in many protocols as part of triple intrathecal chemotherapy regimens ( Pullen et al, 1993 ). The pharmacokinetics of Ara-C are characterized by the rapid elimination of the drug from the plasma as a result of the deamination of Ara-C to the metabolite uridine arabinoside (Ara-U) by the enzyme cytidine deaminase. In adults, peak plasma concentrations of Ara-C reach 10 µm after a bolus dose of 100 mg/m2 and the drug is eliminated with a half-life of 7–20 min ( Slevin et al, 1983 ). The pharmacokinetics of Ara-C administered by similar conventional bolus doses have not been determined for children. For children receiving conventional dose Ara-C (300 mg/m2) as an intravenous infusion over 4 h, a greater than fourfold range in steady-state plasma concentrations (5–20 µm) and hence systemic exposure has been found for children with ALL ( Evans et al, 1991 ). For children receiving high-dose, i.e. 3000 mg/m2, Ara-C as a short infusion over 1 h, peak plasma concentrations in the range of 57–199 µm have been reported ( Avramis et al, 1989 ), followed by a biexponential decay with an average t1/2α of 17 min and an average terminal elimination phase (t1/2β) of 3·8 h. Approximately 12–14% of the simultaneous peak plasma concentrations are achieved in the CSF after intravenous bolus administration ( Chabner, 1996), and proportionately higher CSF levels are achieved with intravenous high-dose Ara-C regimens with CSF concentrations of 4–6 µm ( DeAngelis et al, 1992 ). However, the lower cytidine deaminase activity of the CSF results in less rapid clearance of Ara-C, with t1/2 values of 2–6 h demonstrated after conventional dose and high-dose intravenous Ara-C ( DeAngelis et al, 1992 ). The CSF pharmacokinetics of Ara-C after intravenous administration of the drug have not been reported in children. As with methotrexate, intrathecal administration of Ara-C results in high CSF concentrations. In adults, administration of 50 mg/m2 Ara-C yields peak CSF concentrations of 1 µm and concentrations above 0·4 µm are maintained for 24 h ( Ho & Frei, 1971) . The majority of the studies which have sought to relate the clinical and cellular pharmacology of Ara-C to measures of patient outcome have been performed in relation to adult AML (for a review, see Chabner, 1996). Although during therapy with high-dose Ara-C a large interindividual variation in intracellular Ara-CTP formation does not relate to differences in pharmacokinetic parameters ( Hiddeman et al, 1992 ), Plunkett et al (1987) demonstrated that Ara-CTP formation in blasts from adult patients with relapsed acute leukaemia to be saturable. In the last study, no differences were found for Ara-CTP accumulation and intracellular AUC for patients receiving Ara-C at doses of 0·5, 1 or 3 g/m2 as a 2-h infusion. However, when the infusion rate was decreased to 0·4 or 0·3 g/m2 over 2 h, with resulting steady-state Ara-C concentrations of < 7 µm, Ara-CTP accumulation and intracellular AUC values were substantially reduced. Furthermore, an increase in intracellular peak Ara-CTP levels and AUC was found when the infusion time for a dose of 3 g/m2 Ara-C was increased between 1 and 4 h. The relationship between the clinical and cellular pharmacology of Ara-C and response from children with acute leukaemia has only been studied in a small number of children with multiply relapsed acute leukaemia. Avramis et al (1987) found a wide interpatient variability in the peak Ara-CTP levels for children with ALL, and no clear relationship between plasma Ara-C AUC and the cellular AUC of Ara-CTP in leukaemic blasts was found. This may have reflected interindividual differences in the activity of deoxcytidine kinase. Similarly, for children receiving conventional dose Ara-C (200 mg/m2/48 h as continuous infusion) as part of induction therapy for AML, a similar wide interindividual variation in intracellular Ara-CTP levels has been described ( Boos et al, 1996 ), despite Ara-CTP levels being ≈ 10-fold lower than those associated with high-dose infusions. Lineage-specific differences in Ara-CTP retention have been found ex vivo for leukaemic blasts obtained from children with acute leukaemia ( Boos et al, 1996 ). Whereas B-precursor ALL blasts retained an average 67% of Ara-CTP, as measured at 3 h after a 1-h exposure to 4 µm Ara-C, T-cell ALL and AML blasts retained ≈ 37% of the end exposure Ara-CTP levels. However, the relationship among Ara-C pharmacokinetic parameters, Ara-CTP formation and retention, Ara-CTP incorporation into DNA and patient outcome is yet to be determined for childhood ALL and AML. A measure of the sensitivity of lymphoblasts ex vivo to Ara-C has been assessed with the MTT assay, and median IC50 values of 2 µm have been reported for children presenting with ALL ( Kaspers et al, 1991 ). Therefore, from the above description of the cellular pharmacology of Ara-C, more prolonged infusion schedules of Ara-C might be expected to increase the likelihood of incorporation of Ara-CTP into the DNA of leukaemic cells during the S-phase of the cell cycle. An example of such a regimen has been described by Avramis et al (1989) , in which patients received an individualized loading dose of Ara-C followed by a continuous 72-h infusion at an average 130 mg/m2/h, designed to achieve steady-state plasma concentrations in the range of 20–35 µm (four times the Km for deoxycytidine kinase purified from the leukaemic cells of individual patients). In summary, the efficacy of Ara-C as an antileukaemic agent may be optimized by achieving extracellular concentrations, i.e. steady-state plasma concentrations of 20–25 µm, which may maximize the formation of intracellular Ara-CTP in lymphoblasts. As with MTX, optimization of Ara-C may require the administration of a continuous infusion of pharmacologically guided Ara-C for a duration which is associated with non-dose-limiting toxicity. This strategy may confer added benefit in view of the relatively high nucleoside transporter expression of T-cell ALL compared with B-lineage ALL ( Chabner, 1996) and the longer Ara-CTP retention times described for B-lineage ALL. Indeed, in the study of conventional vs. individualized chemotherapy for childhood ALL reported by Evans et al (1998) , a trend towards improved survival was found for children with T-cell ALL. For these children, steady-state plasma Ara-C concentrations in the range of 5·5–15 µm were achieved for the duration of the 4-h infusion, and neither the Ara-C plasma concentrations achieved nor time of exposure may be optimal. Furthermore, a trend towards improved EFS has been demonstrated for children with B-precursor ALL, who received consolidation therapy containing a 1 g/m2 infusion of Ara-C administered over 24 h ( Harris, M.B., et al, 1998 ). Although in the last study the pharmacokinetics of Ara-C were not determined, the infusion rate is not likely to be ‘biochemically optimal’ ( Avramis et al, 1998 ). The thiopurine antimetabolites 6-mercaptopurine (6-MP) and 6-thioguanine (6-TG) are prodrugs, and after cellular uptake via the nucleoside transporter are subject to anabolic reactions to form thioguanine nucleotides (TGNs) and catabolic reactions to form metabolites that are predominantly inactive ( Fig 3). Several enzymes are involved in the intracellular anabolism of thiopurines to form TGNs ( Elion, 1989), with the first step being catalysed by hypoxanthine-guanine phosphoribosyl transferase (HGPRT). Alternatively, 6-MP and 6-TG are subject to S-methylation by the enzyme thiopurine S-methyltransferase (TPMT), yielding S-methylated nucleobases which are inactive ( Krynetski et al, 1995 ). However, TPMT also catalyses the S-methylation of the principal nucleotide metabolites of 6-MP and 6-TG to form metabolites ( Fig 3) which are inhibitors of de novo purine synthesis ( Shi et al, 1998 ). TPMT activity in humans exhibits genetic polymorphism, with about 1/300 people in Caucasian populations inheriting TPMT deficiency as an autosomal recessive trait. In contrast, whereas 89% of individuals are homozygous for non-mutated TPMT alleles and have high TPMT activity, 11% are heterozygotes and have intermediate levels of TPMT activity ( Lennard et al, 1997 ). 6-MP and 6-TG are also subject to inactivation by xanthine oxidase (XO) and guanase respectively ( Hande & Garrow, 1996). Moreover, intracellular nucleotides such as TIMP are also subject to intracellular degradation by cytoplasmic 5′-nucleotidase and non-specific (acid and alkaline) phosphatases ( Pieters et al, 1992a ). Thiopurine metabolism. PRPP, phosphoribosylpyrophosphate; PRA, 5-phosphoribosyl-1-amine; IMP, inosine monophosphate; TIMP, 6-thioinosine monophosphate; mTIMP, methyl-6-thioinosine monophosphate; TXMP, 6-thioxanthine monophosphate; TGMP, 6-thioguanine monophosphate; TGDP, 6-thioguanine diphosphate; TdGDP, 6-thiodeoxyguanine diphosphate; TdGTP, 6-thiodeoxyguanine triphosphate; HGPRT, hypoxanthine-guanine phosphoribosyl transferase. The broken line indicates inhibition of PRPP amidotransferase. The cytotoxicity of 6-MP and 6-TG is thought to depend mainly on the incorporation of TGNs into DNA, with resulting abnormal DNA–protein interactions which interfere with the function of DNA polymerases, ligases and endonucleases. Moreover, thiopurines may also exert cytotoxicity as inhibitors of de novo purine synthesis ( Hande & Garrow, 1996). In addition, the mismatch repair pathway may play a role in thiopurine-mediated cytotoxicity via the recognition of misincorporated TGNs ( Fink et al, 1998 ). For both 6-MP and 6-TG, the relationship between incorporation of TGNs into DNA and cytotoxicity have been investigated in the mouse lymphoma L5178Y cell line. Whereas delayed growth inhibitory effects after a 13-h exposure time were only found for 6-MP concentrations of 25–100 µm, 6-TG-mediated cytotoxicity was maximal at a concentration of 2 µm ( Tidd & Paterson, 1974). Similarly, whereas prolonged exposure to 6-MP concentrations > 10 µm are required for significant cytotoxicity against a variety of T-cell human leukaemia cell lines ( Adamson et al, 1994 ; da Silva et al, 1996 ), even short-term exposure to 0·5 µm 6-TG is associated with significant cell kill ( Adamson et al, 1994 ). Therefore, 6-TG appears to be more potent than 6-MP in terms of in vitro cytotoxicity. Resistance to 6-MP in vitro has been reported to result from decreased HGPRT activity ( Zimm et al, 1985 ). The activity of 6-MP as a single agent in ch