Open Full Page

Research Article

Differential Effects of Targeted Disruption of Thiopurine Methyltransferase on Mercaptopurine and Thioguanine Pharmacodynamics

Christine Hartford,1,2 Erick Vasquez,1 Matthias Schwab,1,7 Mathew J. Edick,1,6 Jerold E. Rehg,3 Gerard Grosveld,4 Ching-Hon Pui,2,5,6 William E. Evans,1,5,6 and Mary V. Relling1,5,6

Departments of 1Pharmaceutical Sciences, 2Hematology-Oncology, 3Pathology, and 4Genetics, and 5Hematologic Malignancies Program, St. Jude Children’s Research Hospital, 6University of Tennessee, Memphis, Tennessee; and 7Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany, and University of Tuebingen, Tuebingen, Germany

Abstract TPMT polymorphism. Inherited as an autosomal codominant trait The recessive deficiency in thiopurine methyltransferase (6–8), TPMT activity in erythrocytes (reflective of other tissues as (TPMT), caused by germ-line polymorphisms in TPMT, can well; refs. 7, 9–11) exhibits a trimodal population frequency cause severe toxicity after mercaptopurine. However, the distribution (12–14). TPMT activity is inversely related to the significance of heterozygosity and the effect of the polymor- concentration of active thioguanine metabolites after administra- phism on thioguanine or in the absence of thiopurines is not tion of thiopurines (4). The rare homozygous deficient individuals known. To address these issues, we created a murine knockout develop extreme myelosuppression, which can be fatal when given of Tpmt. Pharmacokinetic and pharmacodynamic studies of usual doses of thiopurines (15), necessitating a 10-fold dosage À À mercaptopurine and thioguanine were done in Tpmt / , decrease to prevent such toxicity (16). Heterozygotes, constituting À Tpmt+/ , and Tpmt+/+ mice and variables were compared 10% of the population, often exhibit an intermediate degree of among genotypes. Methylated thiopurine and thioguanine toxicity following mercaptopurine (17), although the role of nucleotide metabolites differed among genotypes after treat- heterozygosity in dosing thiopurines is somewhat controversial. ment with mercaptopurine (P < 0.0001 and P = 0.044, Patients with homozygous wild-type (high) activity may be at respectively) and thioguanine (P = 0.011 and P = 0.002, increased risk of poor treatment response (e.g., persistence of respectively). Differences in toxicity among genotypes were leukemia; ref. 18). more pronounced following treatment with 10 daily doses Due to the inherent toxicity and possible carcinogenicity of of mercaptopurine at 100 mg/kg/d (0%, 68%, and 100% thiopurines (19–21), conducting randomized clinical trials to fully 50-day survival; P = 0.0003) than with thioguanine at 5 mg/ explore how best to use these agents in humans has ethical kg/d (0%, 33%, and 50% 15-day survival; P = 0.07) in the constraints. For example, there is considerable interest in studying À À À Tpmt / , Tpmt+/ , and Tpmt+/+ genotypes, respectively. Mye- the relative importance of the TPMT polymorphism for thioguanine losuppression and weight loss exhibited a haploinsufficient versus for mercaptopurine, particularly because recent clinical trials phenotype after mercaptopurine, whereas haploinsufficiency have suggested a lower relapse rate with thioguanine treatment in was less prominent with thioguanine. In the absence of childhood acute lymphoblastic leukemia (ALL). TPMT activates the drug challenge, there was no apparent phenotype. The mercaptopurine riboside intermediate metabolite of mercaptopu- murine model recapitulates many clinical features of the rine, but this methylated metabolite is not present after adminis- human polymorphism; indicates that mercaptopurine is more tration of thioguanine (22). Thus, it would be useful to directly affected by the TPMT polymorphism than thioguanine; and compare the effect of the TPMT polymorphism on the pharmaco- provides a preclinical system for establishing safer regimens of kinetics and pharmacodynamics of thioguanine versus mercapto- genetically influenced antileukemic drug therapy. [Cancer Res purine (3, 23). In addition, there has been much speculation about 2007;67(10):4965–72] whether the polymorphism in TPMT has any biological significance in the absence of thiopurine challenge (2, 24). Introduction In humans, the molecular basis of the TPMT polymorphism is largely related to three common nonsynonymous coding single- One of the clearest examples of a pharmacogenetic polymor- nucleotide polymorphisms (13), each of which renders the protein phism affecting drug efficacy and toxicity in humans is that caused unstable (25) and subject to enhanced ubiquitination and by thiopurine methyltransferase (TPMT) deficiency (1–5). Thio- degradation (26, 27). Thus, the homozygous deficiency in humans purines (mercaptopurine, thioguanine, and azathioprine) are is characterized by almost undetectable levels of TPMT protein; commonly used as antineoplastics and immunosuppressants, but heterozygotes have intermediate protein and activity levels; and their proper use remains uncertain and is complicated by the homozygous wild-type individuals have high levels of protein and activity (28, 29). We created a murine model to allow the controlled study of the

Note: C-H. Pui is an American Cancer Society Professor. TPMT polymorphism. This model has allowed us to compare the Requests for reprints: Mary V. Relling, Department of Pharmaceutical Sciences, pharmacokinetics and pharmacodynamics of thioguanine and St. Jude Children’s Research Hospital, 332 North Lauderdale, Memphis, TN 38105-2794. mercaptopurine in an in vivo isogenic system and to show that Phone: 901-495-2348; Fax: 901-525-6869; E-mail: [email protected]. I2007 American Association for Cancer Research. biological consequences of the TPMT polymorphism are evident doi:10.1158/0008-5472.CAN-06-3508 only in the presence of drug challenge. www.aacrjournals.org 4965 Cancer Res 2007; 67: (10). May 15, 2007

Materials and Methods Chemicals. Mercaptopurine, thioguanine, 2-deoxyguanosine, 6-thioguanosine, 6-mercaptopurine riboside, and 6-methylmercaptopurine riboside were purchased from Sigma. Alkaline phosphatase was purchased from Promega. 6-Thio-2¶deoxyguanosine was purchased from R.I. Chemical, Inc. All other chemicals and reagents were of high-performance liquid chromatography grade. A folic acid–deficient purified diet for mice was purchased from TestDiet. Methotrexate and trimethoprim/sulfamethoxazole (Sulfatrim pediatric suspension) were purchased from Alpharma. Targeting construct design and generation. A clone containing the murine Tpmt gene was identified by screening a 129/Ola murine genomic DNA library, packaged in a P1 artificial chromosome vector (Genome Systems), with the assistance of Dr. Evgeny Krynetski and Dr. Mike Fessing (30, 31). The identified clone was verified to contain the Tpmt genomic locus by Southern blotting, sequencing, and fluorescent in situ hybridiza- tion. To generate the targeting construct, an 8.7-kb KpnI restriction endonuclease fragment encompassing part of exon III through a portion of intron 6 was subcloned into the pZERO-2 vector (Invitrogen). A herpes simplex virus-thymidine kinase gene cassette was cloned into the HindIII site of the pZERO vector at the 3¶ end of the Tpmt insert, and a 2.8-kb PacI/ XmaI fragment, internal to the 8.7-kb KpnI fragment and containing exons Figure 1. Erythrocyte TPMT activity differed by genotype. The mean (95% CI) À À À V and VI, was replaced with a neomycin phosphotransferase gene cassette. TPMT activity differed among +/+, +/ , and / genotypes for 129P1/ReJ [15.0 (13.40–16.10) versus 8.33 (7.60–8.80) versus 0.53 (0.10–0.90) units/mL Generation of targeted mice. E14 embryonic stem cells provided by the erythrocytes; P = 0.0273], NMRI-7C10 [15.03 (13.90–16.70) versus 8.03 laboratories of Dr. James Ihle and Dr. Gerard Grosveld (St. Jude Children’s (6.40–10.80) versus 1.47 (0.20–3.20) units/mL erythrocytes; P = 0.0273], Research Hospital) were electroporated in the presence of linearized Tpmt NMRI-5G3 [14.43 (13.40–15.70) versus 6.90 (6.50–7.30) versus 0.53 targeting vector DNA. After selection in G418 for 9 days, three successfully (0.40–0.60) units/mL erythrocytes; P = 0.0265], 129X1/SvJ [10.93 (9.70–12.90) versus 7.13 (5.30–8.96) versus 0.53 (0.15–0.91) units/mL erythrocytes; targeted clones (4B9, 7C10, and 5G3) were identified by PCR and confirmed P = 0.0273], C3H/HeJ [15.3 (14.40–16.80) versus 8.23 (7.60–9.00) units/mL by Southern blotting. C57BL/6 pseudopregnant mice were recipients of erythrocytes versus undetectable; P = 0.0241]. TPMT activity was similar in blastocyst injections and embryo transfers. Four germ-line transmitting patient samples from children with leukemia [18.88 (12.75–25.98) versus 8.97 (3.90–13.45), versus 0.56 (0.37–0.75) units/mL erythrocytes (P < 0.0001) in chimeric mice were produced (three from clone 5G3 and one from clone À À À +/À +/+, +/ , and / , respectively]. P values were determined by the 7C10). Heterozygous Tpmt (Tpmt ) chimeric mice produced offspring in Kruskal-Wallis test. the expected Mendelian distribution, indicating no in utero lethality due to loss of one or both functional Tpmt alleles. Because C57BL/6 have low To evaluate chronic toxicity that partially mimics the exposure conditions constitutive TPMT activity (32, 33), the mixed background 129/C57BL/6 of combination drug therapy given to children with leukemia, six mice of each chimeric mice were backcrossed into constitutively high TPMT activity genotype (at 2 months of age) were placed on a folate-deficient (undetectable strains (129P1/ReJ, NMRI, 129X1/SvJ, and C3H/HeJ) for five generations to folate) diet (Purina) because lymphoblastic leukemia therapy consists of produce f97% strain purity. All strains were generated from 5G3 clones. In chronic antifolates. To mimic the fact that thiopurines are given nearly addition, a strain of NMRI was also generated from clone 7C10. All continuously (for many months), thiopurines were administered via the experiments involving the production and use of the Tpmt knockout mouse drinking water. For chronic mercaptopurine experiments, the drinking water described herein were approved by the Institutional Animal Care and Use contained mercaptopurine (0.01 mg/mL) and sulfamethoxazole/trimetho- Committee of St. Jude Children’s Research Hospital. prim for the duration of the experiments to mimic the fact that children with Histologic evaluation. Histologic specimens were prepared from leukemia usually receive this antibiotic for the duration of their therapy. À 6-month-oldmiceof bothgendersandthree genotypes(Tpmt+/+, Tpmt+/ , and Given the average oral water intake of these mice, the estimated average oral À À and Tpmt / ) for screening of any pathologic changes associated with the intake of mercaptopurine was 2.5 mg/kg/d. Methotrexate was administered Tpmt-deficient genotype. All organs were dissected and fixed in 10% neutral i.p. at 1 mg/kg/wk beginning at 8 weeks of age, again to mimic the exposure of buffer formalin. Paraffin-embedded sections (5 Am) were visualized by H&E children with leukemia that accompanies chronic thiopurine therapy. staining. Moribund mice were sacrificed. After 9 weeks of this regimen, the oral dose TPMT activity. Blood (1 mL) was collected via heart puncture into tubes of mercaptopurine in all of the surviving mice was doubled to 5 mg/kg/d. containing sodium heparin. Erythrocyte lysates were prepared and analyzed Experiments for chronic toxicity after thioguanine were identical to those for TPMT activity by the nonchelated radiochemical assay of Szumlanski described above, except that mercaptopurine was replaced by thioguanine, et al. (34). The assay is based on the conversion of mercaptopurine to at an estimated oral dose of 1 mg/kg/d. After 5 weeks of chemotherapy, the radioactively labeled methylmercaptopurine with S-[methyl-14C]-adenosyl- oral dose of thioguanine was doubled to 2 mg/kg/d in the surviving mice. L-methionine as the methyl donor. To investigate the effect of a non-thiopurine drug in different Tpmt De novo purine synthesis assay. Bone marrow cells were flushed from genotypes, mice were treated with etoposide (100 mg/kg) by i.p. injection. femurs using PBS (pH 7.4), and the rate of de novo purine synthesis was Complete blood counts were measured before and every 3 days after determined by a 2-h ex vivo incubation of 5 Â 106 bone marrow cells with treatment. [14C]formate (35). Thiopurine bases and nucleosides were measured by high-pressure liquid Pharmacology studies. Mice from both genders, ages 8 to 13 weeks, chromatography in erythrocytes, plasma, and bone marrow cells before and were used in the pharmacology studies. For pharmacokinetic studies, after conversion with acid phosphatase, as described (36, 37). Methyl- mercaptopurine or thioguanine (50 mg/kg) was administered i.p., blood was thioguanosine nucleotides were measured in erythrocytes using a collected by heart puncture into heparinized tubes, and bone marrow cells modification of a prior method (38). were obtained by flushing with PBS (pH 7.4). Thiopurine pharmacology in children. To assess the extent to which To evaluate acute toxicity of thiopurines, mercaptopurine (100 mg/kg/d) murine data mirrored clinical data, we measured erythrocyte TPMT activity, alone or thioguanine (5 mg/kg/d) alone was administered i.p. After 10 days thiopurine metabolites, and toxicity (assessed as the time-dependent of treatment, complete blood counts and serum aspartate aminotransferase requirement for a thiopurine dose decrease in children treated for ALL, levels were measured. as we reported; refs. 17, 39). All clinical studies were approved by our

Cancer Res 2007; 67: (10). May 15, 2007 4966 www.aacrjournals.org

À Institutional Review Board. The clinical treatment protocol was one in heterozygous Tpmt (Tpmt+/ ) littermates by appearance; life span; which 91% of the weeks of therapy consisted solely of mercaptopurine and reproductive capacity; and histology of liver, lung, kidney, stomach, methotrexate. TPMT genotype was assessed directly by sequencing germ- duodenum, small and large intestine, spleen, thymus, lymph nodes, line DNA or by inference based on erythrocyte TPMT activity (10, 40). heart, adrenals, reproductive organs, bone marrow, and brain. De novo purine synthesis was measured in bone marrow at diagnosis of TPMT activity. Mean erythrocyte TPMT activity differed among leukemia (35). +/+ +/À À/À Statistical analysis. Pharmacokinetic and toxicity data are reported as Tpmt , Tpmt , and Tpmt genotypes in five different strains means with 95% confidence intervals (95% CI). Differences in pharmaco- of Tpmt knockout mice (Fig. 1). The NMRI-5G3 strain was used for kinetic and toxicity variables among all three genotypes were compared all further experiments. The magnitude of differences in TPMT using the Kruskal-Wallis analysis of ranks. Where only two groups were activity among Tpmt genotypes in mice was similar to that compared, the Mann-Whitney U test was used. The Kaplan-Meier method measured in 152 children with leukemia (Fig. 1; ref. 17). was used to calculate survival, and differences in survival among genotypes Endogenous purines. To assess possible effects of TPMT in the were analyzed using the log-rank test. P < 0.05 was considered statistically absence of exogenous thiopurine exposure, we documented that significant for all analyses. All analyses were two sided. the mean rate of de novo purine synthesis in bone marrow cells did À À À not differ among Tpmt+/+, Tpmt+/ , and Tpmt / genotypes Results [3.00 (95% CI, 2.89–3.10), 3.44 (95% CI, 2.11–4.76), and 3.43 (95% À À Homozygous deficient Tpmt (Tpmt / ) mice were viable, fertile, CI, 2.59–4.27) pmol newly synthesized purines/nmol unlabeled and indistinguishable from their wild-type Tpmt (Tpmt+/+)or purines/h, respectively; P = 0.35]. Likewise, the median rates of

Figure 2. A, TGN levels in bone marrow cytosol differed by genotype. TGN levels were measured in Tpmt mice4haftermercaptopurine (50 mg/kg; top) and thioguanine (50 mg/kg; bottom) treatments. Inset, clinical data for steady-state erythrocyte TGN levels from patients with leukemia. B, erythrocyte MeTIMP levels differed by genotype. MeTIMP levels were measured in Tpmt mice after mercaptopurine treatment (50 mg/kg; top) and MeTGN levels were measured after thioguanine treatment (50 mg/kg; bottom). Inset, clinical data for MeTIMP levels in patients (www.pharmgkb.org PS206669; ref. 17). Boxes, 25th, 50th, and 75th percentiles; whiskers, 5th and 95th percentiles. P values were determined by the Kruskal-Wallis test. www.aacrjournals.org 4967 Cancer Res 2007; 67: (10). May 15, 2007

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Cancer Research de novo purine synthesis did not differ (P = 0.68) in bone marrow Heterozygotes had greater mean weight loss [À9.22 (95% CI, À15.0 leukemia cells obtained at diagnosis of B-lineage ALL among to À3.45) versus À1.26 (95% CI, À5.02 to 2.51) g; P = 0.025] and 105 children who were homozygous for wild-type TPMT (0.91; more anemia [5.67 (95% CI, 4.48–6.85) versus 9.91 (95% CI, 7.73– 95% CI, 5.10–5.82 pmol/nmol/h) compared with the 11 patients 12.10) g/dL; P = 0.01] than Tpmt+/+ littermates. Heterozygotes also with heterozygous TPMT genotype (0.88; 95% CI, 4.12–8.58 pmol/ had higher mean platelet counts [264 (95% CI, 174–299) versus 101 nmol/h). Thus, targeted disruption of Tpmt did not have obvious (95% CI, 70.7–131) Â103/AL; P = 0.006], greater mean aspartate effects on purine homeostasis. aminotransferase levels [82.2 (95% CI, 66.5–97.8) versus 51.8 (95% Pharmacokinetics. Thioguanine nucleotide (TGN) concentra- CI, 34.8–68.8) units/L; P = 0.023], and longer survival (P = 0.005) À À À tions in bone marrow cytosol differed among Tpmt+/+, Tpmt+/ , and than Tpmt / littermates. Thioguanine was more potent than À À and Tpmt / genotypes after treatment with both mercaptopurine mercaptopurine; preliminary experiments with 50, 25, and 10 mg/ [3.24 (95% CI, 2.59–3.88), 3.58 (95% CI, 2.40–4.76), and 5.13 (95% CI, kg/d were all quite toxic to all three genotypes. At 5 mg/kg/d of 3.84–6.42) pmol/5 Â 106 cells, respectively; P =0.044]and thioguanine i.p. as a single agent, all mice survived for the 10 days À À thioguanine [99.7 (95% CI, 76.0–123), 143 (95% CI, 109–177), and of treatment with the exception of one Tpmt / mouse, but all 216 (95% CI, 158–274) pmol/5 Â 106 cells, respectively; P = 0.002; three genotypes experienced profound myelosuppression and there À Fig. 2A]. We compared the Tpmt+/ genotype with both Tpmt+/+ were no differences in any of the hematologic variables measured À À and Tpmt / genotypes to evaluate whether there was an effect of among the genotypes or in overall survival (Figs. 4 and 5). When À À À haploinsufficiency. After treatment with mercaptopurine, there comparing Tpmt+/ with either Tpmt+/+ or Tpmt / genotype at was no significant difference in TGN concentration in bone this dose, only platelet count [115 (95% CI, 88.8–140) versus À marrow cytosol between Tpmt+/ and either Tpmt+/+ (P = 0.67) or 81.5 (95% CI, 62.9–100) Â103/AL; P = 0.048] and percent weight À À Tpmt / (P = 0.09) genotypes. After thioguanine treatment, the loss [À3.46 (95% CI, À6.90 to À1.60) versus À22.5 (95% CI, À26.2 À À difference in TGN concentration between Tpmt+/ and Tpmt+/+ did to À19.5) g; P = 0.014] differed significantly between Tpmt+/ and À À not reach statistical significance (P = 0.07), but the difference Tpmt / genotypes, respectively. We tested an even lower dose À À À between Tpmt +/ and Tpmt / genotypes was significant of thioguanine (2.5 mg/kg/d) and similarly found evidence of (P = 0.041). Consistent with clinical data (41, 42), the concentration myelosuppression in all genotypes. There was a significant of TGNs was much higher after thioguanine than after mercaptopurine treatment [mean, 99.7 (95% CI, 76.0–123) versus 3.24 (95% CI, 2.59–3.88) pmol/5 Â 106 cells for Tpmt+/+ and 216 (95% CI, 158–274) versus 5.13 (95% CI, 3.84–6.42) pmol/5 Â 106 cells for À À Tpmt / mice, respectively]. After treatment with mercaptopurine (50 mg/kg), there was a significant difference (P < 0.0001) in the erythrocyte concentration of methylthioinosine monophosphate (MeTIMP) among genotypes, with Tpmt+/+ mice having the highest concentration À (mean, 1,104; 95% CI, 902–1,306 pmol/8 Â 108 cells), Tpmt+/ mice having intermediate concentration (mean, 755; 95% CI, 568– À À 942 pmol/8 Â 108 cells; P = 0.015), and Tpmt / mice having no measurable MeTIMP (Fig. 2B). The methylated metabolites after thioguanine (50 mg/kg) revealed a significant difference in the erythrocyte methylthioguanine nucleotide (MeTGN) concentrations (P = 0.011) among the three genotypes, although the À difference between Tpmt+/+ and Tpmt+/ mice was not significant [135 (95% CI, 66.0–204) versus 166 (95% CI, 72–259) pmol/8 Â 108 cells, respectively; P = 0.34; Fig. 2B]. As has been observed in patients (41), the concentration of the relevant methyl metabolites was much higher after mercaptopurine than after thioguanine treatment [mean, 1,104 (95% CI, 902–1,306) versus 135 (95% CI, 66–204) pmol/8 Â 108 cells among Tpmt+/+ mice, respectively]. Clinical pharmacokinetic data from patients undergoing treatment for ALL were similar to those obtained from the mice. The mean levels of MeTIMP were 17,363 (95% CI, 14,543–20,184) versus 8,688 (95% CI, 1,244–18,619) versus 0.05 pmol/8 Â 108 cells in À À À TPMT +/+, TPMT +/ , and TPMT / patients, respectively (P = 0.007; Fig. 2B). Similarly, there was a significant difference among the genotypes in the levels of TGNs: 2.61 (95% CI, 2.43–2.78) versus 6.02 (95% CI, 3.60–8.43) versus 22.28 (95% CI, 49.7–94.3) pmol/5 Â 106 À À À cells in TPMT +/+, TPMT +/ , and TPMT / patients, respectively (P < 0.001; Fig. 2A). Toxicity. After 10 days of treatment with mercaptopurine Figure 3. Blood counts, serum aspartate aminotransferase levels, and (100 mg/kg/d i.p.) as a single agent, significant differences were percentage of body weight loss of Tpmt mice after 10 d of treatment with mercaptopurine as a single agent (100 mg/kg/d i.p.). Boxes, 25th, 50th, and 75th observed among the genotypes in survival, weight, complete blood percentiles; whiskers, 5th and 95th percentiles. P values were determined by counts, and serum aspartate aminotransferase levels (Figs. 3 and 4). the Kruskal-Wallis test.

Cancer Res 2007; 67: (10). May 15, 2007 4968 www.aacrjournals.org

Myelosuppression (measured by the absolute neutrophil count, hemoglobin, and platelet count) was significantly different among À À À Tpmt+/+, Tpmt+/ , and Tpmt / genotypes after the first 4 weeks of chronic oral mercaptopurine combined with methotrexate (data not shown). The poor survival rates precluded similar analyses with chronic oral thioguanine/methotrexate treatment. Hepatotoxicity is a known adverse effect of thiopurine therapy (43–45). In contrast to myelosuppression after mercaptopurine, it seems to occur more commonly among patients with high rather than low TPMT activity (46, 47). In mice, after acute exposure to mercaptopurine alone (100 mg/kg/d), serum aspartate aminotransferase levels were related to the number of wild-type copies of Tpmt (P = 0.014; Fig. 3), with heterozygotes having higher À À aminotransferase levels than Tpmt / mice [80.2 (95% CI, 56.2– 104) versus 51.8 (95% CI, 34.8–68.8) units/L, respectively; P = 0.022]. A similar relationship to hepatotoxicity was observed after chronic exposure to 8 weeks of oral mercaptopurine combined with methotrexate, with aspartate aminotransferase levels higher in À Tpmt+/+ than in Tpmt+/ littermates [126 (95% CI, 103–148) versus 91.17 (95% CI, 60.94–121.39) units/L, respectively; P = 0.045]; no À À Tpmt / mice survived to this time point. Interestingly, mean aspartate aminotransferase levels did not differ among genotypes after 10 days of thioguanine [67.00 (95% CI, 42.57–91.43), 60.6 (95% CI, 30.52–90.68), and 62.25 (95% CI, 25.17–99.33) units/L in

Figure 4. Cumulative proportion of Tpmt mice surviving after exposure to mercaptopurine (100 mg/kg/d; top) or thioguanine (5 mg/kg/d; bottom). P values were determined by the log-rank test. difference, however, in all complete blood count variables measured as well as in percent weight loss when comparing À À À À À Tpmt+/ and Tpmt / genotypes, with Tpmt / mice being more profoundly affected (data not shown). Thiopurines are often administered chronically as daily oral doses and combined with weekly methotrexate to treat ALL. We assessed the effects of chronic exposure by treating mice with daily oral mercaptopurine (2.5 mg/kg/d) or thioguanine (1 mg/kg/d) combined with weekly i.p. methotrexate (1 mg/kg). For both thiopurines, all of the À/À mice were sacrificed due to toxicity after 4 to 8 weeks, with little toxicity observed in the +/À and +/+ mice. Therefore, to hasten the onset and to increase the probability of toxicity in the surviving mice, the mercaptopurine dose was increased from 2.5 to 5 mg/kg/d beginning at 9 weeks, and the thioguanine dose was increased from 1 to 2 mg/kg/d at 5 weeks, leaving the methotrexate and trimethoprim/sulfamethoxazole À À doses unchanged. In this setting, survival of Tpmt / mice was 0% at 8 weeks following mercaptopurine and 0% at 4 weeks following thioguanine treatment, both being significantly lower À than their respective Tpmt+/ or Tpmt+/+ littermates (Fig. 6). The doses of thiopurines were increased at 9 and 5 weeks for Figure 5. Blood counts, serum aspartate aminotransferase levels, and mercaptopurine and thioguanine, respectively, and there was no percentage of body weight loss of Tpmt mice after 10 d of treatment with +/À +/+ thioguanine as a single agent (5 mg/kg/d i.p.). Boxes, 25th, 50th, and 75th difference in survival in Tpmt compared with Tpmt mice for percentiles; whiskers, 5th and 95th percentiles. P values were determined by either the former (P = 0.66) or the latter (P = 0.44) drug. the Kruskal-Wallis test. www.aacrjournals.org 4969 Cancer Res 2007; 67: (10). May 15, 2007

Figure 6. Cumulative proportion of Tpmt mice surviving after chronic exposure to mercaptopurine (2.5 mg/kg/d for 9 wk followed by 5 mg/kg/d; top)or thioguanine (1 mg/kg/d for 5 wk followed by 2 mg/kg/d; bottom), both combined with weekly methotrexate. Inset (top right), for comparison, the cumulative proportion of patients with leukemia tolerating daily mercaptopurine therapy (17) is shown; patients were censored at the time of toxicity necessitating a dosage decrease or for any events taking the patient off the study. P values were determined by the log-rank test.

À À À Tpmt+/+, Tpmt+/ , and Tpmt / , respectively; P = 0.53; Fig. 5]; poor and-error approach in the clinic. Therefore, the diagnosis of a defect survival precluded a similar comparison after chronic exposure to in TPMT can serve as a clinically useful basis on which to target oral thioguanine combined with methotrexate. thiopurines as the likely cause of myelosuppression in the setting of Etoposide as a negative control did not result in any differences multiagent chemotherapy. The clinical data supporting an in complete blood counts or in survival (P = 0.49) among genotypes increased risk of thiopurine-induced myelosuppression associated (data not shown). with the relatively rare homozygous deficiency are strong, with a >10-fold dosage reduction required in such patients (44). Our murine data are consistent with a dramatic effect of homozygous Discussion defects of TPMT. The magnitude of the effect of heterozygosity and We created a murine model that recapitulates all the known its implications for dose reductions clinically are less clear (17). elements of the human genetic polymorphism in TPMT. This Herein, with acute administration of thiopurines alone, we showed model therefore allowed us to address fundamental questions that heterozygosity affects mercaptopurine-induced myelosuppres- about the effects of TPMT on thiopurine dosing. sion (Fig. 3), but haploinsufficiency was more difficult to show after When thiopurines are used to treat cancer, they are almost always thioguanine treatment (Fig. 5). With chronic thiopurine exposures, used in a multidrug context. Because of overlapping toxicities and in the presence of methotrexate (mimicking the clinical among anticancer drugs, it is often not possible to discern which regimen used for treatment of ALL), toxicity after mercaptopurine À agent is the primary culprit causing myelosuppression using a trial- was only slightly greater in Tpmt+/ mice than in Tpmt+/+ mice

Cancer Res 2007; 67: (10). May 15, 2007 4970 www.aacrjournals.org

(Fig. 6), and after thioguanine was indistinguishable between difficulty in discerning acute toxicity differences among TPMT À Tpmt+/ and Tpmt+/+ mice (Fig. 6). genotypic groups with thioguanine alone may be partly because The Tpmt activity differences among the genotypic groups were it was difficult to find a dose low enough that was not toxic in all identical to those observed in patients (Fig. 1) and clear evidence three genotypes; however, it is consistent with the extensive TGN for haploinsufficiency was conferred by heterozygosity, as is true formation after thioguanine compared with mercaptopurine. in humans (40). Thus, the model seems to be suitable for Thiopurines are associated with hepatotoxicity; hepatic trans- addressing thiopurine questions that cannot be feasibly tested in aminase elevations have been related to MeTIMP concentrations humans. Our data indicate that toxicity differences among after mercaptopurine (48) whereas hepatic veno-occlusive disease is patients of differing TPMT genotypes are likely due to differences more common after thioguanine than mercaptopurine and seems in metabolism of thiopurines and not to any underlying to be dose related (49). Our data are consistent with the hypothesis susceptibility caused by TPMT variation. There was no evident that TPMT may contribute to hepatocellular toxicity via methyla- phenotype produced by abrogation of the Tpmt gene in the tion of mercaptopurine metabolites (43, 45, 50) but does not relate absence of drug challenge. In this regard, we tested for more to thioguanine hepatotoxicity: we did not observe transaminasemia subtle biochemical effects of Tpmt deficiency by measuring after thioguanine regardless of Tpmt genotype. The clinical hepatic constitutive levels of de novo purine synthesis and found no toxicity due to thioguanine is likely due to an independent À À À difference among the Tpmt+/+, Tpmt+/ ,orTpmt / genotypes. mechanism. At necropsy, no hepatic pathology was observed after Because TPMT uses S-adenosylmethionine as a cofactor, which thioguanine treatment with the regimens used herein. plays a crucial role in purine and pyrimidine homeostasis, these In summary, the targeted disruption of the Tpmt gene findings indicate that TPMT is unlikely to play an important role recapitulates the essential elements of the human pharmacoge- in endogenous purine synthesis. Moreover, as expected, there netic polymorphism, both pharmacokinetically and pharmacody- were no differences in toxicity among genotypes after treatment namically, but the model also further differentiates specific adverse with etoposide, an agent that is not metabolized by TPMT. and pharmacologic effects of thioguanine versus mercaptopurine. Methylated active metabolites of mercaptopurine (MeTIMP) Such murine models of human germ-line genetic variability provide differed among Tpmt genotypes, mirroring the differences observed a laboratory basis for the systematic optimization of dosages of among patients (Fig. 2B). MeTIMP concentrations are much higher, anticancer medications. relative to dose or to active TGN metabolites, following mercaptopurine than are the analogous methylated metabolites (MeTGN) Acknowledgments after thioguanine (Fig. 2B). Interestingly, haploinsufficiency was not evident for production of MeTGN after thioguanine as it was for Received 9/27/2006; revised 1/12/2007; accepted 3/7/2007. Grant support: National Cancer Institute grants T32-CA070089, CA 51001, CA production of MeTIMP after mercaptopurine (Fig. 2B), which is 36401, and CA21765; NIH/National Institute of General Medical Sciences Pharmaco- consistent with haploinsufficiency being more evident for toxicity genetics Research Network and Database, grants U01 GM61393 and U01 GM61374 after mercaptopurine than thioguanine. Active TGN metabolites (http://pharmgkb.org/); a Center of Excellence grant from the State of Tennessee; American Lebanese Syrian Associated Charities; and the Robert-Bosch Foundation, differed by Tpmt genotype after mercaptopurine and after Stuttgart, Germany. thioguanine, as was anticipated from clinical data (41). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance In humans, thioguanine and mercaptopurine show substantially with 18 U.S.C. Section 1734 solely to indicate this fact. different pharmacologic profiles, despite sharing common TGN We thank our clinical staff, research nurses, and patients and their parents for metabolites. Mercaptopurine is more subject to methylation to participation in this study; Paxton Baker, Yaqin Chu, May Chung, Nancy Duran, Natalya Lenchik, Margaret Needham, and Emily Melton for outstanding technical both the inactive base and the active MeTIMP than is thioguanine, assistance; and Nancy Kornegay and Mark Wilkinson for computer and database for which anabolism to TGNs plays a greater quantitative role. The expertise.

References 8. Weinshilboum RM. Human pharmacogenetics of Thiopurine methyltransferase activity in American methyl conjugation. Fed Proc 1984;43:2303–7. white subjects and black subjects. Clin Pharmacol Ther 1. McLeod H, Krynetski EY, Relling MV, Evans WE. 9. Weinshilboum RM. Methylation pharmacogenetics: 1994;55:15–20. Genetic polymorphism of thiopurine methyltransferase thiopurine methyltransferase as a model system. 15. Schutz E, Gummert J, Mohr F, Oellerich M. Azathi- and its clinical relevance for childhood acute lympho- Xenobiotica 1992;22:1055–71. oprine-induced myelosuppression in thiopurine meth- blastic leukemia. Leukemia 2000;14:567–72. 10. McLeod HL, Relling MV, Liu Q, Pui CH, Evans WE. yltransferase deficient heart transplant recipient. Lancet 2. Krynetski EY, Tai HL, Yates CR, et al. Genetic Polymorphic thiopurine methyltransferase in erythro- 1993;341:436. polymorphism of thiopurine S-methyltransferase: clini- cytes is indicative of activity in leukemic blasts from 16. Evans WE, Horner M, Chu YQ, Kalwinsky D, Roberts cal importance and molecular mechanisms. Pharmaco- children with acute lymphoblastic leukemia. Blood 1995; WM. Altered mercaptopurine metabolism, toxic effects, genetics 1996;6:279–90. 85:1897–902. and dosage requirement in a thiopurine methyltransfer- 3. Weinshilboum R. Inheritance and drug response. N 11. Schaeffeler E, Fischer C, Brockmeier D, et al. ase-deficient child with acute lymphocytic leukemia. Engl J Med 2003;348:529–37. Comprehensive analysis of thiopurine S-methyltrans- J Pediatr 1991;119:985–9. 4. Lennard L, Lilleyman JS, Van Loon J, Weinshilboum ferase phenotype-genotype correlation in a large 17. Relling MV, Hancock ML, Rivera GK, et al. Mercap- RM. Genetic variation in response to 6-mercaptopurine population of German-Caucasians and identification topurine therapy intolerance and heterozygosity at the for childhood acute lymphoblastic leukaemia. Lancet of novel TPMT variants. Pharmacogenetics 2004;14: thiopurine S-methyltransferase gene locus. J Natl 1990;336:225–9. 407–17. Cancer Inst 1999;91:2001–8. 5. Balis FM, Adamson PC. Application of pharmacoge- 12. Weinshilboum RM, Sladek SL. Mercaptopurine phar- 18. Stanulla M, Schaeffeler E, Flohr T, et al. Thiopurine netics to optimization of mercaptopurine dosing. J Natl macogenetics: monogenic inheritance of erythrocyte methyltransferase (TPMT) genotype and early treatment Cancer Inst 1999;91:1983–5. thiopurine methyltransferase activity. Am J Hum Genet response to mercaptopurine in childhood acute lym- 6. Otterness DM, Weinshilboum RM. Mouse thiopurine 1980;32:651–62. phoblastic leukemia. JAMA 2005;293:1485–9. methyltransferase pharmacogenetics: monogenic inhe- 13. Otterness D, Szumlanski C, Lennard L, et al. Human 19. Relling MV, Yanishevski Y, Nemec J, et al. Etoposide ritance. J Pharmacol Exp Ther 1987;240:817–24. thiopurine methyltransferase pharmacogenetics: gene and antimetabolite pharmacology in patients who 7. Vuchetich JP, Weinshilboum RM, Price RA. Segregation sequence polymorphisms. Clin Pharmacol Ther 1997;62: develop secondary acute myeloid leukemia. Leukemia analysis of human red blood cell thiopurine methyl- 60–73. 1998;12:346–52. transferase activity. Genet Epidemiol 1995;12:1–11. 14. McLeod HR, Lin JS, Scott EP, Pui CH, Evans WE. 20. Relling MV, Pui CH, Sandlund JT, et al. Adverse www.aacrjournals.org 4971 Cancer Res 2007; 67: (10). May 15, 2007

effect of anticonvulsants on efficacy of chemotherapy ing and characterization of a common polymorphism. 40. Yates CR, Krynetski EY, Loennechen T, et al. for acute lymphoblastic leukaemia. Lancet 2000;356: DNA Cell Biol 1996;15:17–30. Molecular diagnosis of thiopurine S-methyltransferase 285–90. 30. Krynetski EY, Fessing MY, Yates CR, Sun D, Schuetz deficiency: genetic basis for azathioprine and mercap- 21. Thompsen J, Schroder H, Kristinsson J, et al. Possible JD, Evans WE. Promoter and intronic sequences of the topurine intolerance. Ann Intern Med 1997;126:608–14. carcinogenic effect of 6-mercaptopurine on bone human thiopurine S-methyltransferase (TPMT) gene 41. Erb N, Harms DO, Janka-Schaub G. Pharmacokinetics marrow stem cells: relation to thiopurine metabolism. isolated from a human PAC1 genomic library. Pharm and metabolism of thiopurines in children with acute Cancer 1999;86:1080–6. Res 1997;14:1672–8. lymphoblastic leukemia receiving 6-thioguanine versus 22. Dervieux T, Blanco JG, Krynetski EY, Vanin EF, 31. Fessing MY, Belkov VM, Krynetski EY, Evans WE. 6-mercaptopurine. Cancer Chemother Pharmacol 1998; Roussel MF, Relling MV. Differing contribution of Molecular cloning and functional characterization of 42:266–72. thiopurine methyltransferase to mercaptopurine versus the cDNA encoding the murine thiopurine S-methyl- 42. Lennard L, Davies HA, Lilleyman JS. Is 6-thioguanine thioguanine effects in human leukemic cells. Cancer Res transferase (TPMT). FEBS Lett 1998;424:143–5. more appropriate than 6-mercaptopurine for children 2001;61:5810–6. 32. Hernandez JS, Van Loon JA, Otterness DM, with acute lymphoblastic leukaemia? Br J Cancer 1993; 23. Evans WE. Pharmacogenetics of thiopurine S-meth- Weinshilboum RM. Mouse thiopurine methyltransferase 68:186–90. yltransferase and thiopurine therapy. Ther Drug Monit pharmacogenetics: correlation of immunoreactive pro- 43. Nygaard U, Toft N, Schmiegelow K. Methylated 2004;26:186–91. tein and enzymatic activity. J Pharmacol Exp Ther 1990; metabolites of 6-mercaptopurine are associated with 24. Scheuermann TH, Lolis E, Hodsdon ME. Tertiary 252:568–73. hepatotoxicity. Clin Pharmacol Ther 2004;75:274–81. structure of thiopurine methyltransferase from Pseudo- 33. Watters JW, Zhang W, Meucci MA, Hou W, Ma MK, 44. Evans WE, Hon YY, Bomgaars L, et al. Preponderance monas syringae, a bacterial orthologue of a poly- McLeod HL. Analysis of variation in mouse TPMT of thiopurine S-methyltransferase deficiency and het- morphic, drug-metabolizing enzyme. J Mol Biol 2003; genotype, expression and activity. Pharmacogenetics erozygosity among patients intolerant to mercaptopu- 333:573–85. 2004;14:247–54. rine or azathioprine. J Clin Oncol 2001;19:2293–301. 25. Tai HL, Krynetski EY, Schuetz EG, Yanishevski Y, 34. Szumlanski CL, Honchel R, Scott MC, Weinshilboum 45. Berkovitch M, Matsui D, Zipursky A, et al. Hepato- Evans WE. Enhanced proteolysis of thiopurine RM. Human liver thiopurine methyltransferase pharma- toxicity of 6-mercaptopurine in childhood acute lym- S-methyltransferase (TPMT) encoded by mutant alleles cogenetics: biochemical properties, liver-erythrocyte phocytic leukemia: pharmacokinetic characteristics. in humans (TPMT*3A, TPMT*2): mechanisms for the correlation and presence of isozymes. Pharmacoge- Med Pediatr Oncol 1996;26:85–9. genetic polymorphism of TPMT activity. Proc Natl Acad netics 1992;2:148–59. 46. Kontorinis N, Agarwal K, Gondolesi G, Fiel MI, Sci U S A 1997;94:6444–9. 35. Masson E, Synold TW, Relling MV, et al. Allopurinol O’Rourke M, Schiano TD. Diagnosis of 6 mercaptopu- 26. Tai HL, Fessing M, Bonten EJ, et al. Enhanced inhibits de novo purine synthesis in lymphoblasts of rine hepatotoxicity post liver transplantation utilizing proteasomal degradation of mutant human thiopurine children with acute lymphoblastic leukemia. Leukemia metabolite assays. Am J Transplant 2004;4:1539–42. S-methyltransferase (TPMT) in mammalian cells: mech- 1996;10:56–60. 47. Heckmann JM, Lambson EM, Little F, Owne EP. anism for TPMT protein deficiency inherited by 36. Dervieux T, Chu YQ, Su E, Pui CH, Evans WE, Relling Thiopurine methyltransferase (TPMT) heterozygosity TPMT*2, TPMT*3A, TPMT*3B or TPMT*3C. Pharmaco- MV. HPLC determination of thiopurine nucleosides and and enzyme activity as predictive tests for the develop- genetics 1999;9:641–50. nucleotides in vivo in lymphoblast following mercapto- ment of azathioprine-related adverse events. J Neurol Sci 27. Wang L, Sullivan W, Toft D, Weinshilboum R. purine. Clin Chem 2002;48:161–8. 2005;231:71–80. Thiopurine S-methyltransferase pharmacogenetics: 37. Su Y, Hon YY, Chu Y, Van de Poll ME, Relling MV. 48. Dubinsky MC, Yang H, Hassard PV, et al. 6-MP chaperone protein association and allozyme degrada- Assay of 6-mercaptopurine and its metabolites in metabolite profiles provide a biochemical explanation tion. Pharmacogenetics 2003;13:555–64. patient plasma by high-performance liquid chromatog- for 6-MP resistance in patients with inflammatory bowel 28. Loennechen T, Yates CR, Fessing MY, Relling MV, raphy with diode-array detection. J Chromatogr B disease. Gastroenterology 2002;122:904–15. Krynetski EY, Evans WE. Isolation of a human Biomed Sci Appl 1999;732:459–68. 49. Stoneham S, Lennard L, Coen P, Lilleyman J, Saha V. thiopurine S-methyltransferase (TPMT) complementary 38. Herrlinger KR, Fellermann K, Fischer C, et al. Veno-occlusive disease in patients receiving thiopurines DNA with a single nucleotide transition A719G Thioguanine-nucleotides do not predict efficacy of during maintenance therapy for childhood acute (TPMT*3C) and its association with loss of TPMT thioguanine in Crohn’s disease. Aliment Pharmacol lymphoblastic leukaemia. Br J Haematol 2003;123:100–2. protein and catalytic activity in humans. Clin Pharmacol Ther 2004;19:1269–76. 50. Dubinsky MC, Lamothe S, Yang HY, et al. Pharma- Ther 1998;64:46–51. 39. Relling MV, Pui CH, Cheng C, Evans WEE. Thiopurine cogenomics and metabolite measurement for 6-mer- 29. Szumlanski C, Otterness D, Her C, et al. Thiopurine methyltransferase in acute lymphoblastic leukemia. captopurine therapy in inflammatory bowel disease. methyltransferase pharmacogenetics: human gene clon- Blood 2006;107:843–4. Gastroenterology 2000;118:705–13.

Cancer Res 2007; 67: (10). May 15, 2007 4972 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Differential Effects of Targeted Disruption of Thiopurine Methyltransferase on Mercaptopurine and Thioguanine Pharmacodynamics

Christine Hartford, Erick Vasquez, Matthias Schwab, et al.

Cancer Res 2007;67:4965-4972.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/10/4965

Cited articles This article cites 50 articles, 8 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/10/4965.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/10/4965.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/10/4965. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.