The Association of Cytochrome P450 Genetic Polymorphisms With

Home , CYP2B6, CYP2C19, CYP2C9

ORIGINAL ARTICLE The association of cytochrome P450 genetic polymorphisms with sulfolane formation and the efﬁcacy of a busulfan-based conditioning regimen in pediatric patients undergoing hematopoietic stem cell transplantation

CRS Uppugunduri1,2, MA Rezgui3, PH Diaz1,2, AK Tyagi1,2, J Rousseau3, Y Daali2,4, M Duval3,5, H Bittencourt3,5, M Krajinovic3,5,6 and M Ansari1,2

Cytochrome P450 enzymes (CYPs) and flavin-containing monooxygenases (FMOs) likely have a role in the oxidation of intermediate metabolites of busulfan (Bu). In vitro studies to investigate the involvement of these enzymes are cumbersome because of the volatile nature of the intermediate metabolite tetrahydrothiophene (THT) and the lack of sensitive quantitation methods. This study explored the association between the CYP2C9, CYP2C19, CYP2B6 and FMO3 genotypes and sulfolane (Su, a water soluble metabolite of Bu) plasma levels in children undergoing hematopoietic stem cell transplantation (HSCT). The relationship between these genotypes and the effectiveness of myeloablative conditioning was also analyzed. Sixty-six children receiving an intravenous Bu-based myeloablative conditioning regimen were genotyped for common functional variant alleles in CYP2C9 (*2 and *3), CYP2C19 (*2 and *17), FMO3 (rs2266780, rs2266782 and rs1736557) and CYP2B6 (*5 and *9). The plasma levels of Bu and its metabolite Su were measured after the ninth Bu dose in a subset of 44 patients for whom plasma samples were available. The ratio of Bu to Su was considered the metabolic ratio (MR) and was compared across the genotype groups. Higher MRs were observed in CYP2C9*2 and *3 allele carriers (mean±s.d.: 7.8±3.6 in carriers vs 4.4±2.2 in non-carriers; P ¼ 0.003). An increased incidence of graft failure was observed among patients with an MR45 compared with those with MR values o5 (20% vs 0%; P ¼ 0.02). In contrast, a significantly higher incidence of relapse and graft failure (evaluated as event-free survival) was observed in patients with malignant disease who carried CYP2B6 alleles with reduced function on both chromosomes compared with carriers of at least one normal allele (100% vs 40%; P ¼ 0.0001). These results suggest that CYP2C9 has a role in the oxidation reactions of THT and indicate that it may be possible to predict the efficacy of Bu-based myeloablative conditioning before HSCT on the basis of CYP genotypes and Bu MRs.

The Pharmacogenomics Journal (2014) 14, 263–271; doi:10.1038/tpj.2013.38; published online 29 October 2013 Keywords: busulfan; CYP2C9; graft failure; HSCT; metabolic ratio; sulfolane

INTRODUCTION several studies have investigated the relationship between Busulfan (Bu) is a bi-functional alkylating agent and a widely genetic variants of the GSTA1 and GSTM1 genes and the 6–11 used component of myeloablative and non-myeloablative con- pharmacokinetics of Bu. ditioning regimens before hematopoietic stem cell transplantation Cytochrome P450 enzymes (CYPs) are likely involved in the 12 (HSCT) in children.1 Bu is metabolized via conjugation with oxidation reactions of THT to Su. However, data are not currently glutathione (GSH), which is predominantly catalyzed by available to specify which CYPs are involved. Limited data are glutathione-S-transferase alpha 1.2 The GSH conjugate available with regard to the inter-individual variability of CYP dissociates to tetrahydrothiophene (THT), which is oxidized to activity and expression levels in children. On the basis of the sulfolane (Su) and subsequently to 3-hydroxy Su.3,4 The metabolic observations of altered expression and the liver volumes of fetal fate of Bu is brieﬂy outlined in Figure 1. Inter-individual variability and adult livers, CYP activity might differ in children and adults or in Bu plasma levels occurs when orally or intravenously remain similar for all age groups (for example, CYP3A4/3A5, administered; this administration requires drug concentration 1A2).13–15 Nevertheless, the presence of a signiﬁcant proportion of monitoring to adjust the dosage and maintain the Bu levels within CYP2C isoforms in the livers of younger children suggests their the therapeutic range.5,6 Observed variability might be attributed importance with regard to the metabolism of certain drugs and to enzyme activity that alters the metabolism of Bu.7 Because GST the oxidation of Bu intermediate metabolites.15 CYPs, particularly isoforms are predominantly involved in the metabolism of Bu, CYP2C9 and CYP2C19, participate in the oxidation of sulfur in

1Department of Pediatrics, Onco-Hematology unit, University Hospital of Geneva, Geneva, Switzerland; 2CANSEARCH Research Laboratory, Geneva Medical University, Geneva, Switzerland; 3Charles-Bruneau Cancer Center, CHU Sainte-Justine Research Center, Montreal, Quebec, Canada; 4Department of Clinical Pharmacology and Toxicology, Geneva University Hospitals, Geneva, Switzerland; 5Department of Pediatrics, Charles-Bruneau Cancer Center, CHU Sainte-Justine Research Center, Montreal, Quebec, Canada and 6Department of Pharmacology, University of Montreal, Montreal, Quebec, Canada. Correspondence: Dr CRS Uppugunduri, CANSEARCH Research Laboratory, Geneva Medical University, 64 Avenue De La Roseraie, Geneva 1205, Switzerland or Dr M Ansari, Department of Pediatrics, Onco-Hematology Unit; Geneva University Hospital, Rue Willy Donze´ 6, Geneva CH-1211, Switzerland. E-mail: [email protected] or [email protected] Received 30 May 2013; revised 6 August 2013; accepted 9 September 2013; published online 29 October 2013 Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 264

Figure 1. Metabolic fate of busulfan. Cytochrome P450 enzymes likely have a role in the formation of sulfolane; however, the speciﬁc subfamily of enzymes involved is unknown. CYPs, cytochrome P450 isoform enzymes; CTH, cystathionine gamma-lyase; DPEP, dipeptidase/ cysteinylglycine; FMO, ﬂavin-containing monooxygenase; GGT, gamma-glutamyl transferase; GSTA1, GSTM1 and GSTP1, glutathione S transferase enzymes, subtypes alpha1, mu1 and pi1, respectively; NAT, N-acetyl transferase.

THT; CYP2B6 and CYP3A5 are also involved to a certain extent.16,17 samples collected at 15, 30, 60, 120 and 180 min after the first dose.23 The In addition, flavin-containing monooxygenases (FMOs) likely pharmacokinetic parameters of the first dose were estimated using non- catalyze the oxidation reactions of Bu metabolites.18,19 FMO1 compartmental analysis (Winnolin, Pharsight Corporation, Version 3.1), and the dosage was adjusted at the fifth administration to achieve a target and FMO3 are predominantly expressed in fetal and adult liver, 1 respectively. Three non-synonymous variants at frequencies steady-state concentration (Css) of 600–900 ng ml . The hospital ethics committee approved the study protocol, and all patients/parents signed an higher than 5% were reported in the FMO3 of Caucasians from informed consent form (IRB number: 2450). This study was fully registered the HapMap project, whereas FMO1 does not have polymorphic 20 in a public trials registry as part of an ongoing prospective EBMT alleles in the coding gene region. In vitro studies of the multicenter study (registered at Clinical Trials.gov, NCT01257854). Blood functional role of specific CYP isoforms in the oxidation of Su and samples before conditioning were collected for genomic DNA extraction THT are difficult because of the volatile nature of THT and the lack and banking at CHU Sainte Justine’s. Plasma samples for MR analysis (after of highly sensitive analytical methods to quantify the metabolites. ninth dose) were available for 44 out of the 66 patients. Plasma samples Recently, we developed and validated a method to quantify Su in were collected between 10am and midday after infusion of the ninth dose the plasma of patients receiving Bu to facilitate investigations of of Bu, as a similar schedule was followed for Bu administration the role of Bu metabolites.21 This method was sensitive enough to (0800–1000 h). The time point for MR measurement was chosen on the basis of a pilot study that showed higher levels of Su in the plasma after detect in vivo Su levels from the plasma of patients receiving the infusion of the ninth dose.21 Table 1 displays the overall patient higher doses of Bu. Using this approach to estimate Su levels, we demographics and disease characteristics. Cyclosporine was administered explored the association between CYP2C9, CYP2C19, FMO3 and to all patients as prophylaxis for graft versus host disease (GVHD); steroids CYP2B6 genotypes and the metabolic ratio (MR) of Bu/Su after the were added to the cyclosporine regimen of 39 patients, and methotrexate ninth dose. We also explored the association between the Bu MRs was added to the cyclosporine regimen of 26 patients. Anti-thymocyte and CYP genotypes with regard to the effectiveness of globulin was administered to 78.8% (n ¼ 52) of the patients. Ursodeoxy- myeloablative conditioning. cholic acid was administered to all patients as a prophylaxis for VOD. Table 2 displays the demographics of the patients included in the MR analysis (n ¼ 44). Of the 44 patients included in the MR analysis, one MATERIALS AND METHODS received phenytoin, two received midazolam and the remainder received Patient samples and treatment regimen lorazepam for seizure prophylaxis. All patients received a Bu-based myeloablative conditioning regimen, with cyclophosphamide (CY) added Sixty-six pediatric patients at CHU Sainte Justine’s (Montreal, QC, Canada) for 41 patients, melphalan added for one patient and CY and etoposide who received a 2-h intravenous infusion of Bu (Busulfex, Otsuka added for one patient. The 44 patients included for MR analysis had Pharmaceuticals, Montreal, QC, Canada) four times per day for 4 days optimal liver and kidney function and had not received any medication before HSCT were included in this study. The details of the dosing schedule 6,22 that inhibited or induced CYP activity for at least 48 h before the and adjustments are detailed elsewhere. In brief, Bu was administered conditioning regimen, excluding one patient who had received phenytoin four times daily, with each dose administered as a 2-h intravenous infusion for seizure prophylaxis. every 6 h. The first Bu dose was age dependent as follows: 16 mg m 2 for infants p3 months old; 0.8 mg kg 1 for infants X3 months but o1 year of age; 1 mg kg 1 for children X1 year but o4 years of age; and Genotyping the CYP2C9, CYP2C19, FMO3 and CYP2B6 variants 0.8 mg kg 1 for children X4 years old. Bu levels were measured using a Genotyping of the common alleles of the CYP2C9, CYP2C19, FMO3 and modified high-performance liquid chromatography protocol from blood CYP2B6 genes was conducted using TaqMan Drug Metabolism Genotyping

The Pharmacogenomics Journal (2014), 263 – 271 & 2014 Macmillan Publishers Limited Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 265 50–100 ng of DNA, 1 assay reaction mixture and 1X TaqMan genotyping Table 1. Demographic characteristics of the patients genotyped for master mix. The loss-of-function or reduced-function single-nucleotide CYP2C9, CYP2C19, FMO3 and CYP2B6 (n ¼ 66) polymorphisms (SNPs) of CYP2C9 (*2,*3), CYP2C19 (*2), CYP2B6 (*5,*9) and CYP2C19 (*17) associated with a higher enzymatic activity were genotyped. Demographic characteristics Patients Three non-synonymous SNPs (with frequencies45% in Caucasians) in the FMO3 gene were also genotyped. Table 3 lists the SNPs, assay IDs and their N % effects on the function of the enzyme. The genotyping of these SNPs was also performed using genomic DNA extracted from the whole blood of 96 Gender unrelated healthy, Caucasian volunteers to verify the allele, genotype Male 33 50.0 frequencies and the Hardy–Weinberg equilibrium. Female 33 50.0

Race The quantification of Bu and Su levels Caucasian 54 81.8 The plasma samples were collected at the end of the ninth dose infusion Native American 2 3.1 and immediately frozen at 80 1C. Bu levels were estimated using dried African-American 9 13.6 plasma spot sampling (5 ml) followed by an LC-MS/MS analysis.24 In brief, Asian 1 1.5 5 ml of plasma was spotted on Whatman 903 protein saver cards (Whatman Switzerland GmbH, Wuhrmattstrasse, Bottmingen, Switzerland) and dried Diagnosis at room temperature for 15 min. These dried plasma spots were then cut ALL 6 9.1 and placed in an injector vial followed by brief vortexing with 100 mlof AML 20 30.3 methanol containing the internal standard D8-Bu (100 ng ml 1). Five MDS 16 24.2 microliters of the extracted solvent was injected into an API 4000 triple Hemoglobinopathies 8 12.1 quadrupole mass spectrometer (AB Sciex, Concord, ON, Canada) controlled Immunodeficiencies 8 12.1 by Analyst 1.5.1 software. The mass spectrometer was operated in the Metabolic disease 4 6.1 multiple reaction-monitoring mode with positive-ion electrospray Hemophagocytic syndrome 3 4.5 ionization, and the multiple reaction-monitoring transitions were 264.1- 151.1 for Bu and 272.1-159.1 for D8-Bu, with a dwell time of 150 ms for Disease status (leukemia) Bu and internal standard Bu-D8. The assay displayed good inter- and intra- CR1 14 53.8 day trueness and precisions less than ±15%, with a lower quantification CR2/CR3/relapse 7/3/2 26.9/11.5/7.7 limit of 50 ng ml 1. External quality control samples were included in each analysis. HLA compatibility Su levels were measured using liquid–liquid extraction with 100 mlof MUD 11 31.8 plasma followed by a GC MS/MS analysis.21 In brief, 100 ml of plasma was MRD 21 16.7 spiked with D8-Su (500 ng ml 1), followed by the addition of 50 mlof1M MM related donor 3 4.5 NaOH and 500 ml of ethyl acetate. The mixture was then vortexed MM unrelated donor 31 47.0 horizontally for 5 min, and 400 ml of ethyl acetate phase was separated. Fifty microliters of isopropanol was added to the acetate phase, followed Stem cell source by evaporation under N2 until only the isopropanol phase remained. One BM 26 39.4 microliter of this isopropanol phase was injected into a Varian CP 3800 gas Cord blood 40 60.6 chromatograph (Walnut Creek, CA, USA) equipped with a Varian CP-8400 autosampler and a Varian 300-MS triple quadrupole mass spectrometer. Conditioning The assay had good reliability and precision (less than ±15%) between 20 Bu/Cy 59 89.4 and 400 ng ml 1. Appropriate external quality controls were included in Bu/Cy/VP16 5 7.6 the analysis.21 The MRs were calculated using the Bu:Su ratio measured Bu/Mel 1 1.5 after the ninth dose. Because the dose adjustments were performed at the Bu/Cy/Mel 1 1.5 fifth dose level, the MRs measured after the ninth dose represent dose- adjusted MRs. The initial dose pharmacokinetic parameters estimated for Seizure prophylaxis these patients were published previously.6,11 Pierre Fabre Laboratories Lorazepam 52 78.8 conducted a cross-calibration study that included St. Justine’s Center, Phenytoin 3 4.5 Geneva Center, and others to validate the analytical methods used. Both Midazolam 11 16.7 the Geneva and Canadian centers passed the cross-validation because they generated similar results using different analytical methods (data Median Range available upon request). Age (years) 6.9 0.1–19.9 Weight (kg) 24.6 4.3–95.6 BMI (kg m 2) 17.5 13.2–34.2 Clinical outcomes Busulfan first dose (mg) 20.0 4.0–65.0 The treatment regimen, the definition of clinical outcomes and other Busulfan Css (ng ml 1) 702.4 418.9–1480.5 cohort details are described elsewhere.6,11 The Seattle criteria for veno- Busulfan ninth dose (mg) 27.0 4.5–72.0 occlusive disease or sinusoidal obstruction syndrome as well as the 1994 Follow-up in days after transplant 1000.5 28–3390 Consensus Conference guidelines on acute GVHD were used.25,26 Because Nucleated cells (108 kg 1) 2.7 0.5–21.0 our cohort was composed of patients with malignant and non-malignant CD34 þ cells (108 kg 1) 0.006 0.0007–1.3 disease, the effectiveness of myeloablative conditioning was evaluated as the incidence of graft failure in the entire cohort and event-free survival Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid (EFS) in patients with either malignant or pre-malignant disease. EFS was leukemia; BM, bone marrow; BMI, body mass index; Bu, busulfan; CR1 to defined as the time from the day of transplant to the occurrence of any CR3, complete remission first to third; Css, steady-state concentration event (relapse or graft failure, whichever occurs first). All analyses were 1 adjusted with the first dose administered in mg kg ; Cy, cyclopho- performed with either exclusion of dead patients or inclusion as censored sphamide; MDS, myelodysplastic syndrome; MM, mismatch; MRD, matched for the event as discussed under the respective results. Graft failure was related donor; MUD, matched unrelated donor; VP16, etoposide. defined as either persistent pancytopenia without the hematological recovery of donor cells after 28 days (primary) or a rapid decline in the neutrophil count requiring cell infusions after successful engraftment (secondary). The occurrence of disease relapse in patients with malignant Assays (Life Technologies Europe B.V., Zug, Switzerland) on a StepOnePlus disease within 1 month of graft failure was not considered to be graft Real-Time PCR System (Life Technologies Europe B.V.). Appropriate positive failure; disease relapse was defined as either the appearance of blasts in controls were included in each experiment, and all samples were peripheral blood or the infiltration of bone marrow by more than 5% of the genotyped in duplicate. Each genotyping reaction (25 ml) included blasts. Hemorrhagic cystitis was defined as the presence of macroscopic

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 263 – 271 Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 266 Table 2. The demographic characteristics and metabolic ratios of busulfan of the patients included in the metabolic ratio comparison study

Characteristic mean±s.d. (median) CYP2C9*1*1 (n ¼ 28) CYP2C9*2 or *3 carriers CYP2C19*17 (CC; CYP2C19*17 (CT or TT; (n ¼ 16) n ¼ 30) n ¼ 14)

Age 7.7±6.2 (6.9) 6.5±4.8 (5.9) 5.9±5.5 (4.8) 9.9±5.3 (10.7) Body weight (kg) 27.9±20.5 (24.1) 26.9±20.1 (23.3) 24.3±20.5 (20.0) 34.3±18.2 (32.1) Body mass index (kg m 2) 18.2±3.8 (17.2) 17.9±3.1 (17.0) 17.9±3.6 (17.0) 18.3±3.4 (17.8)

Gender, N (%) Males 13 (46.4) 9 (56.3) 13 (43.3) 9 (64.3) Females 15 (53.6) 7 (43.7) 17 (56.6) 5 (35.7) Malignant disease or pre-malignancies, N (%) 14 (50.0) 13 (81.3) 18 (60.0) 9 (64.3) Non-malignant disease, N (%) 14 (50.0) 3 (18.7) 12 (40.0) 5 (35.7) Busulfan first dose (mg) 22.4±15.9 (19.0) 21.9±15.9 (19.0) 19.9±16.2 (15.5) 27.1±14.2 (25.5) Busulfan ninth dose in (mg) 27.3±16.9 (25.0) 26.9±18.6 (21.5) 23.7±16.8 (17.0) 34.6±16.7 (34.5) Busulfan dose ratio (adjusted vs first dose) 1.3±0.4 (1.3) 1.2±0.2 (1.2) 1.2±0.3 (1.2) 1.4±0.4 (1.4) Busulfan after ninth dose (ng ml 1) 1106.8±413.6 1561.8±653.4 1351.4±580.9 1102±459.9 (1163.4) (1120.1) (1328.3) (1222.4) Sulfolane after ninth dose (ng ml 1) 286.1±116.9 (264.6) 227.9±109.7 (205.3) 243.3±118.8 (234.4) 311.1±100.5 (291.5) Metabolic ratio (busulfan/sulfolane) 4.4±2.2 (4.3) 7.8±3.6 (6.7) 6.5±3.3 (5.5) 3.9±2.3 (3.6) Busulfan first dose AUC (mM min 1) 812.4±280.1 (756.1) 841.3±149.2 (814.6) 822.3±212.7 (778.1) 824.3±297.2 (717.8) Busulfan Css (ng ml 1, estimated from first 729.6±269.4 (661.2) 733±126.2 (720.5) 710±178.2 (694.6) 774.6±308.7 (648.4) dose) Busulfan first dose clearance 4.4±1.3 (4.4) 4.1±0.7 (4.1) 4.4±1.1 (4.2) 4.2±1.3 (4.5) (ml min 1 kg 1) Abbreviations: AUC, area under the curve; Css, steady-state concentration. Malignancies and pre-malignancies include acute lymphoblastic leukemia, acute myeloid leukemia and myelodysplastic syndrome. The group CYP2C9*2 or *3 carriers includes patients with genotypes of *1*2 or *2*2 or *1*3 or *2*3. Both AUC and Css were adjusted with dose administered in mg kg 1. The MRs observed among genotype groups are marked in bold font.

hematuria unrelated to other causes of bleeding and requiring medical patients are shown in Table 3. All of the studied SNPs were in the intervention. Hardy–Weinberg equilibrium in patients except for CYP2C9*2. The analysis between the MRs and the CYP2C9, CYP2C19, Statistical analyses CYP2B6 and FMO3 genotypes revealed a significant association The genotype frequencies from healthy volunteers and patients were between the MRs and CYP2C9 and CYP2C19 (Table 2 and checked for the Hardy–Weinberg equilibrium by comparing their Supplementary Table 1). Higher MRs were observed among observed and expected genotype frequencies using the w2-test. The patients carrying the *2 allele of CYP2C9 and a combination of the MRs between the different genotype groups were compared using non- *2 and *3 alleles (Figures 2a and b; P ¼ 0.01) than those carrying parametric tests. The between-group distribution of age and gender was the CYP2C9*1*1 genotype. In contrast, a significant decrease was compared using a Student’s t-test and a w2-test, respectively. Log- observed in the MRs of CYP2C19*17 carriers compared with transformed MR values were used for a stepwise multiple linear patients without this allele (Figures 2c, P ¼ 0.01). Table 2 displays regression analysis against CYP genotypes, FMO3 genotypes, diagnosis, the details of the CYP2C9 and CYP2C19 genotypes with regard to age and gender. The cumulative incidences of EFS and graft failure were estimated using Kaplan–Meier curves, and the log-rank test was used to MRs, as well as patient demographic and PK data. Between-group compare the between-group differences in a univariate analysis. Only differences were not found for the distributions of age, weight, variables with Po0.1 in the univariate analysis were included in the diagnosis, Bu-dose ratios (that is the ratio of the adjusted dose to following multivariate analysis. Risk factors for EFS, graft failure and the first dose administered) or the first dose PK data with regard to relapse were evaluated using the forward conditional Cox-regression the CYP2C9 genotype (Table 2); however, the distribution of age model in the multivariate analysis. The variables analyzed in the differed between patients with different CYP2C19*17 genotypes. univariate and subsequent multivariate analyses included gender, age A significant negative correlation between patient age and the Bu (above vs below 4 years), diagnosis (malignant vs non-malignant), MRs MR was also noted (r ¼0.34; P ¼ 0.024; Supplementary Table 2). (below and above median), CYP genotypes (normal vs reduced function), The patients were stratified into two groups of older and younger the number of infused nucleated cells (above vs below the median), the number of CD34 þ cells infused (above vs below the median), graft than 4 years of age following an age-based dosing schedule. source (bone marrow vs cord blood), GVHD prophylaxis (cyclosporine and Differences in the MRs were observed between older and younger cyclosporine þ methotrexate vs cyclosporine þ steroids or mycopheno- children (Figure 3a, P ¼ 0.01). We also analyzed whether the late mofetil), Css (above vs below the median), malignant disease status relationship between the MRs and CYP2C9/CYP2C19 was age (first complete remission vs second or third remission) and HLA matching dependent. Significant between-group differences in the MRs (complete match vs mismatch). Two-sided tests were used, and a P-value based on the CYP2C9 and CYP2C19 genotypes were observed only o0.05 was considered significant. All data are represented as the among patients older than 4 years (Figures 3b and c; P ¼ 0.001 mean±s.d. (or median and percentages) unless otherwise specified. Data and P ¼ 0.04, respectively). When all patients were analyzed using analyses were performed using (IBM SPSS Statistics for Windows, Version a multivariate linear regression, the inclusion of age along with 19.0,IBM,Armonk,NY,USA). CYP2C9 genotype improved the model and explained B35% of the variability in the MRs, up from the 27% explained by the RESULTS CYP2C9 genotype alone (P ¼ 0.0001). The relationship between genotype and Bu MR All of the studied polymorphisms (SNPs) in the CYP2C9, CYP2C19, MR, CYP genotype and the efficacy of the Bu-based conditioning FMO3 and CYP2B6 genes were in the Hardy Weinberg equili- regimen brium among the 96 unrelated, healthy volunteers (data not Supplementary Tables 3 and 4 outline the clinical outcomes shown). The frequencies of the minor allele and genotypes from 66 observed among patients included for the genotyping and MR

The Pharmacogenomics Journal (2014), 263 – 271 & 2014 Macmillan Publishers Limited Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 267

Table 3. Allele and genotype frequencies of the SNPs in the study population (n ¼ 66)

Genetic Reference Alleles Amino acid Functional impact on TaqMan DME Minor allele Heterozygosity variation SNP ID (major/ change enzyme activity Assay ID frequency (%) n (%) minor)

CYP2C9*2 rs1799853 C/T Arg144Cys Decreased C__25625805_10 12.9 13.6 CYP2C9*3 rs1057910 G/A Ile359Leu Decreased C__27104892_10 7.6 15.2 CYP2C19*2 rs4244285 G/A Splicing No enzyme activity C__25986767_70 12.9 22.7 defect CYP2C19*17 rs12248560 C/T Promoter Increased C__469857_10 19.7 30.3 SNP CYP2B6*5 rs3211371 C/T Arg487Cys Decreased C__30634242_40 14.4 22.7 CYP2B6*9 rs3745274 G/T Gln172His Decreased C__7817765_60 25.8 27.3 FMO3 rs2266780 A/G Glu308Gly Missense variationa C__2220257_30 18.0 25.8 FMO3 rs2266782 G/A Glu158Lys Missense variationa C__2461179_30 38.5 36.4 FMO3 rs1736557 G/A Val257Met Missense variationa C__8698544_30 8.1 16.0 aSingle-nucleotide polymorphisms (SNPs) in FMO3 gene were predicted to be tolerant using prediction tools such as sorting intolerant from tolerant (SIFT) and protein analysis through evolutionary relationships (PANTHER), except rs2266780 that was predicted to be damaging to gene function by SIFT. The total number of subjects genotyped for cytochrome P450 genetic variants is 66, out of which the MRs of 44 patients were available for comparison. The allele frequency is derived from genotype data obtained in 66 individuals.

analyses. We analyzed the efficacy of the myeloablative con- (P ¼ 0.4), GVHD prophylaxis (P ¼ 0.4) or HLA match (0.9). However, ditioning regimen as graft failure and disease relapse (in Css was significantly associated with EFS, with improved EFS malignancies) with regard to the MR and CYP genotypes. Of the among patients with a Css value below the median (P ¼ 0.001). 44 patients included in the MR analysis, four patients died, four The multivariate analysis revealed that MR was independently patients experienced graft failure and nine patients (out of 27 associated with EFS (hazard ratio (HR) for MR45 ¼ 4.8, 95% patients with malignant or pre-malignant disease) relapsed. The CIs ¼ 1.2–19.9; P ¼ 0.02). These events occurred only among four patients with graft failure received Bu followed by CY, and patients who received the Bu-CY regimen, except for one three of them had cord blood as a graft source. In a univariate individual who received Bu-CY and etoposide. Significant associa- analysis, only the MR and Css (P ¼ 0.02) were associated with the tions were not observed between CYP genotypes and other incidence of graft failure (Figure 4a; P ¼ 0.02). All four patients with outcomes, including VOD, hemorrhagic cystitis or acute GVHD; graft failure had an MR45 and a Bu Css above the median however, higher incidences of VOD and hemorrhagic cystitis were (702.4 ng ml 1). Of these four patients, two were in first complete observed among patients carrying normal CYP2C alleles. remission (acute myeloid leukemia) before transplantation, and When compared with the global cohort of 66 patients, graft the other two were diagnosed with myelodysplastic syndrome. Four failure was not associated with CYP2C9 (P ¼ 0.1) or CYP2B6 patients died; two patients in each group were excluded from the genotypes or with other co-variants according to a univariate analysis because the primary cause of their deaths was infection. analysis. Five patients who died before a graft failure were However, similar results were observed when the deceased patients excluded from the graft failure and EFS analysis. EFS was were censored and included in the analysis. A significant association significantly associated with the CYP2B6 genotype (Figure 4c, of graft failure was not found with regard to the graft source P ¼ 0.0001) and Css (P ¼ 0.03) among patients with malignancies. (P ¼ 0.4), diagnosis (Po0.1), disease status (P ¼ 0.6), GVHD prophy- A higher incidence of events occurred among patients carrying a laxis (P ¼ 0.6), HLA match (P ¼ 1), age (P ¼ 0.1), cell dose (P ¼ 0.6) or pair of non-functional CYP2B6 alleles (100%) and Css values above CYP2C9 genotypes (Po0.1) in this subset of patients. The multi- the median (60%) compared with those carrying at least one variate analysis did not reveal a significant relationship with regard functional allele (40%) and Css values below the median (31.6%). to MRs (P ¼ 0.2). Another factor influencing efficacy of Bu is the total Three patients who died before the event were excluded from the amount of Bu administered. Patients with an MR 45oro5 received analysis. Similar results were observed after including the similar cumulative Bu doses. Patients with an MRo5 received a deceased patients as censored for the event. Finally, we merely Bu dose (mean±s.d.) of 16.3±2.9 mg kg 1, whereas those with an observed declining trends of EFS with regard to low cell number MR45 received 15.7±2.8 mg kg 1.Ofthe20patientswithan (P ¼ 0.1), cord blood source (P ¼ 0.3), mismatched HLA (P ¼ 0.1), MR45, three had no change in their dose, two required a decrease conditioning other than Bu-CY (P ¼ 0.3), female gender (P ¼ 0.2), and 15 required an increase in their dose. Of the 24 patients with an disease status (P ¼ 0.3) and carriers of non-functional CYP2C alleles MRo5, two had no change in their dose, three required a decrease (P ¼ 0.3). The multivariate analysis revealed that the CYP2B6 and 19 required an increase in their dose. Significant between-group genotype was independently associated with EFS (HR for non- differences were not observed with regard to the dose ratios functional genotype carriers ¼ 7.3, 95% CIs ¼ 1.7–31.0; P ¼ 0.007). (adjusted to initial dose prescribed), Css or AUCs of first dose adjusted to the dose. Similarly, an improved EFS (that is the lower incidences of DISCUSSION relapse and rejection) was observed among patients with CYP2C9 genotype and Bu/Su MR malignant disease with an MRo5 compared with those with an We observed higher MRs among carriers of the CYP2C9 *2 and *3 MR45 (Figure 4b; P ¼ 0.01). Because infections resulted in death alleles, which corresponded to reduced enzymatic function, for the majority of patients, death was not considered an event- whereas lower MRs were observed among carriers of the high- free survival advantage or disadvantage. Of the three patients activity, CYP2C19*17, allele. Interestingly, the CYP2C9*1 allele is in who died, two were in the group with an MRo5, and the other strong linkage disequilibrium with the CYP2C19*17 allele.27 In was in the group with an MR45. No significant associations were addition, the increased frequency of lower MRs in patients older observed with regard to disease status (P ¼ 0.2), cell dose (P ¼ 0.7), than 4 years was noted in the group with the CYP2C19*17 allele. cell source (P ¼ 0.4), CYP2C9 genotype (P ¼ 0.2), age (0.2), gender Thus, the lower MRs observed in CYP2C19*17 carriers might be

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 263 – 271 Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 268

Figure 2. The inﬂuence of CYP2C9 and CYP2C19 genotypes on the busulfan/sulfolane MR. The MRs were calculated using the Bu and Su levels measured after infusion of the ninth dose. The P-values obtained using the Mann–Whitney U tests are shown on the plots. (a) Patients were divided into three groups on the basis of their CYP2C9*2 genotype: CC (n ¼ 35), CT (n ¼ 6) and TT (n ¼ 3). The MR means±s.d. (medians) for the above groups were 5.1±2.9 (4.6), 6.2±1.5 (6.2) and 11.5±2.8 (13.1), respectively. (b) Patients were grouped as having either the CYP2C9*1*1 genotype (n ¼ 28) or at least one copy of either *2 or *3 (n ¼ 16). The MR means±s.d. (medians) for variant allele carriers and individuals with the *1*1 genotype were 7.9±3.6 (6.7) and 4.4±2.2 (4.3), respectively. (c) Patients were grouped on the basis of the CYP2C19*17 genotype (CT and TT: n ¼ 14, 3.9±2.3 (3.6); CC: n ¼ 30.6±3.3 (5.5)).

Figure 3. The inﬂuence of age and the age-related CYP2C9 and CYP2C19 genotypes on the busulfan/sulfolane MR. (a) The mean±s.d. (median) MRs for patients under 4 years of age (n ¼ 16) and those above 4 years of age (n ¼ 28) were 6.9±2.9 (6.0) and 4.9±3.2 (4.0), respectively. (b) The MRs for the CYP2C9 *2 and *3 genotypes in patients older than 4 years of age. The mean±s.d. (median) of CYP2C9*1*1 (n ¼ 18) carriers was 3.7±2.1 (3.5); the mean±s.d. (median) of patients with a variant allele was 7.3±3.8 (n ¼ 10). (c) MRs with regard to the CYP2C19*17 genotype in patients older than 4 years of age. The mean±s.d. (median) of CC carriers (n ¼ 16) was 6.02±3.5 (5.2); the mean±s.d. (median) of carriers of either CT or TT (n ¼ 12) was 3.6±2.3 (3.3).

due to age and the presence of the CYP2C9*1 allele (a reflection of formation of Su.13,15 Similarly, the association between MRs and normal CYP2C9 function and higher Su formation, independent of CYP2C9 genotype was apparent only in this age group; however, CYP2C19*17 allele status). Our finding suggests that CYP2C9 is the absence of this effect in younger children might also be due to involved in the oxidation of THT to Su, and individuals with a the small sample size. A significant difference between the CYP2C9 lower enzymatic activity have reduced Su formation. We verified genotypes was observed only with regard to MRs and not the first that this association was not due to differences in diagnosis, given dose PK parameters (Table 2), which suggests that CYP2C9 has a that reports suggest that non-malignant conditions (for example, role in the formation of Su from THT. thalassemia and possible liver damage) can affect metabolic enzyme function.28,29 A multivariate, stepwise linear regression analysis that included diagnosis, CYP, FMO3 genotypes and age MR, CYP genotype and the efficacy of a Bu-based conditioning revealed that only age and the CYP2C9*2,*3 alleles were regimen associated with lower and higher MRs, respectively. This result In the malignant group, patients with an MR above the median suggests that the disease status at the time of diagnosis might not had a higher incidence of graft failure and decreased EFS (20 and influence the results. 23.1%, respectively) than those with an MR below the median (0 Interestingly, significantly lower MRs were observed among and 72.7%, respectively). As previously observed,22 ahigherMR patients aged 4 years and older, which indicates increased CYP indicates a higher Bu level and lower Su level; the latter is most likely activity in this age group and the involvement of CYPs in the due to the reduced CYP2C9 activity needed to form Su. The MR

The Pharmacogenomics Journal (2014), 263 – 271 & 2014 Macmillan Publishers Limited Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 269

Figure 4. The influence of the busulfan MR and the CYP2B6 genotype on graft failure and EFS. The number of patients with graft failure or relapse versus the total number of patients in each group is shown on the plots, with the respective P-values obtained by log-rank comparisons. (a) The cumulative incidence of graft failure among patients with an MR below or above the median. (b) EFS (relapse and graft failure) among patients with malignant tumors and an MR below or above the median. (c) EFS based on CYP2B6 genotypes. values did not correlate with the cumulative Bu dose received, the size, however, we were unable to analyze the relationships among Css of Bu or the dose ratio (adjusted dose/first dose). However, drug transporter function, plasma Su levels and toxicity; these because more than 95% of the patients received CY following Bu, associations should be evaluated in the future. However, higher the increased EFS in patients with MRs below the median (which levels of Su might also indicate the increased depletion of reflects normal CYP2C9 activity) might be attributed to the increased glutathione as a sulfonium ion conjugate; thus, normal CYP2C9 formation of active CY metabolites and improved immuno- activity might predispose patients to the toxic events initiated by suppression.30,31 The roles of the CYP2C9 and 19 genotypes have CY active metabolites when the conjugation and the elimination of been reported in CY activation.30,32,33 However, a significant CY active metabolites are compromised. In these patients, association was not found between the trends observed for the changing the order from Bu-CY to CY-Bu might be advantageous CYP2C9 genotype and either graft failure or EFS in malignancies, to reduce toxicity, as has been reported in retrospective and which suggests that other enzymes have a role in either the prospective clinical studies.37,38 Because Su is heterocyclic, it might formation of Su or CY activation.34,35 Alternatively, these results induce some of the CYPs involved in the activation of CY; thus, might reflect a lack of power to detect the association because of higher levels of Su indicate increased activation of CY and its the small sample size and involvement of multiple factors. related toxicity. Unfortunately, we were unable to collect CY Recently, Hassan and Andersson36 proposed a model that kinetics data with regard to Su formation. An ongoing study at our illustrated the role of drug transporters with regard to the presence center is currently evaluating this perspective. of increased sulfonium ions to determine the response and toxicity Previous studies using animal models suggest that Su affects to Bu. These authors hypothesized that the sulfonium ion blast cells.39 In this context, increased Su formation might indicate conjugate of Bu might be a substrate for membrane transporters the increased conjugation of Bu and might suggest the extent of that, under normal circumstances, expel sulfonium ion conjugates the b-elimination reaction of sulfonium ions formed to yield THT from cells. However, the extent of sulfonium ion formation (a precursor of Su) and g-glutamyldehydroalanylglycine (an irrever- depends on the availability of glutathione for conjugation, as sible protein glutathionylating agent), which might sensitize cells to well as on the activity of the conjugating enzymes glutathione-S- Bu.40 Our laboratory is currently examining the possible interactions transferase alpha 1 and glutathione-S-transferase mu1.11 On the between Bu and Su at physiologically observed concentrations and basis of the observations of our cohort, we hypothesized that, durations with regard to lymphoblastoid cell lines. unlike the Bu Css (which is associated with toxicity), the extent of We observed a reduction in the EFS of patients with malignant Su formation might indicate increased conjugation, thereby disease who were homozygous for the CYP2B6*5 and *9 alleles, resulting in a greater efficacy of Bu. Owing to the small sample which have reduced enzymatic activity, compared with the carriers

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 263 – 271 Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 270 of the *1*1 genotype. CYP2B6 contributes to CY activation.35 This organization of biological material and clinical data. We thank Professor Yves observed association might be due to an allele*1-related increase Theóret for the first dose pharmacokinetic-guided dosing in the cohort. We thank Ms in the availability of active CY that leads to effective Pauline Mathot for her assistance in genotyping the samples for some of the SNPs. immunosuppression and prevents graft failure. Melason et al.41 has also analyzed the CYP2B6 genotypes with regard to HSCT outcomes. In contrast to our observations, these authors observed REFERENCES that patients with ultra-rapid CYP2B6 metabolizers had worse 1 Schattenberg AV, Levenga TH. Differences between the different conditioning progression-free survival than those with poor metabolizers. regimens for allogeneic stem cell transplantation. Curr Opin Oncol 2006; 18: Several reasons might account for this discrepancy; Melason 667–670. et al.35,41 considered carriers of the CYP2B6*5 allele (*1/*5 and 2 Czerwinski M, Gibbs JP, Slattery JT. Busulfan conjugation by glutathione *5/*5) to be extensive metabolizers, which contradicts previous S-transferases alpha, mu, and pi. Drug Metab Dispos 1996; 24: 1015–1019. reports of the reduced function of this allele. These patients in this 3 Cooper AJ, Younis IR, Niatsetskaya ZV, Krasnikov BF, Pinto JT, Petros WP et al. Metabolism of the cysteine S-conjugate of busulfan involves a beta-lyase reaction. cohort also received the total body irradiation; thus, the Drug Metab Dispos 2008; 36: 1546–1552. occurrences of relapse and rejection might have been prevented 4 Hassan M, Oberg G, Ehrsson H, Ehrnebo M, Wallin I, Smedmyr B et al. Pharma- 41 to a greater extent. We did not investigate the presence of the cokinetic and metabolic studies of high-dose busulphan in adults. Eur J Clin ultra-rapid metabolizing variants of CYP2B6. Moreover, the Pharmacol 1989; 36: 525–530. genotypes with presumably normal enzymatic functions might 5 Malar R, Sjoo F, Rentsch K, Hassan M, Gungor T. Therapeutic drug monitoring is also include ultra-rapid metabolizers. Our observations support the essential for intravenous busulfan therapy in pediatric hematopoietic stem cell study that showed a trend of reduced relapse risk among normal recipients. Pediatr Transplant 2011; 15: 580–588. and ultra-rapid metabolizing carriers of CYP2B6 with leukemia who 6 Ansari M, Lauzon-Joset JF, Vachon MF, Duval M, Theoret Y, Champagne MA et al. received CY and total body irradiation before HSCT.42 Patients who Influence of GST gene polymorphisms on busulfan pharmacokinetics in children. carry normally functioning alleles of CYP2B6 also experienced a Bone Marrow Transplant 2010; 45: 261–267. 7 Poonkuzhali B, Chandy M, Srivastava A, Dennison D, Krishnamoorthy R. Glu- greater degree of toxicity in the first 100 days, suggesting tathione S-transferase activity influences busulfan pharmacokinetics in patients predictor function of the CYP2B6 genotypes for conditioning with beta thalassemia major undergoing bone marrow transplantation. Drug 42 regimens containing CY as one of the components. Future Metab Dispos 2001; 29: 264–267. studies examining the combined effect of the CYP2C19, 2C9 and 8 Zwaveling J, Press RR, Bredius RG, van Derstraaten TR, den Hartigh J, Bartelink IH 2B6 genotypes on CY PK and activation should investigate the role et al. Glutathione S-transferase polymorphisms are not associated with popula- they might have in CY-related HSCT outcomes. We did not observe tion pharmacokinetic parameters of busulfan in pediatric patients. Ther Drug an association between the CYP2B6 genotype and the occurrence Monit 2008; 30: 504–510. of either VOD or hemorrhagic cystitis, in contrast to the findings of 9 Ansari M, Krajinovic M. Can the pharmacogenetics of GST gene polymorphisms Rocha et al.43 but in support of those of Melason et al.41 This lack of predict the dose of busulfan in pediatric hematopoietic stem cell transplantation? association might be due to the weak statistical power of our Pharmacogenomics 2009; 10: 1729–1732. 10 ten Brink MH, Wessels JA, den Hartigh J, van der Straaten T, von dem Borne PA, study. The power of our study for the observed graft failure Guchelaar HJ et al. Effect of genetic polymorphisms in genes encoding GST incidence was 60% at an alpha level of 0.05. Given the limited isoenzymes on BU pharmacokinetics in adult patients undergoing hematopoietic number of patients available, we were only able to detect SCT. Bone Marrow Transplant 2012; 47: 190–195. nonsignificant trends. Therefore, these results must be replicated 11 Ansari M, Rezgui M, Theóret Y, Uppugunduri CRS, Mezziani S, Vachon MF et al. in a large prospective cohort. Finally, the role of other genetic Glutathione S transferase gene variations influence busulfan pharmacokinetics variations (for example, GSTA1 and GSTM1) should be investigated and outcome of hematopoietic stem cell transplantation in pediatric patients. with regard to CYP genetic variants and transporter variants using Bone Marrow Transplant 2013; 48: 939–946. the gene–gene and gene–environment interaction approaches to 12 Gulbis AM, Culotta KS, Jones RB, Andersson BS. Busulfan and metronidazole: an predict the outcomes of Bu-based conditioning before HSCT in often forgotten but significant drug interaction. Ann Pharmacother 2011; 45: e39. children. Several other risk factors, including immunosuppressive 13 Blanco JG, Harrison PL, Evans WE, Relling MV. Human cytochrome P450 maximal treatments for GVHD prophylaxis, cell source, cell number, disease activities in pediatric versus adult liver. Drug Metab Dispos 2000; 28: 379–382. 14 Hakkola J, Pasanen M, Purkunen R, Saarikoski S, Pelkonen O, Maenpaa J et al. status, conditioning regimen, co-medication, HLA matching and Expression of xenobiotic-metabolizing cytochrome P450 forms in human adult anti-HLA antibody status in cord-blood source, also modulate the and fetal liver. Biochem Pharmacol 1994; 48:59–64. genetic risk for EFS. A multicenter EBMT-labeled study to evaluate 15 Treluyer JM, Gueret G, Cheron G, Sonnier M, Cresteil T. Developmental expression and validate the observations made in this study is ongoing. of CYP2C and CYP2C-dependent activities in the human liver: in-vivo/in-vitro In conclusion, we found an association between the CYP2C9 correlation and inducibility. Pharmacogenetics 1997; 7: 441–452. genotype and the Bu MR. Decreased Su formation was observed 16 Orr ST, Ripp SL, Ballard TE, Henderson JL, Scott DO, Obach RS et al. Mechanism- among carriers of the CYP2C9 alleles associated with a decreased based inactivation (MBI) of cytochrome P450 enzymes: structure-activity enzyme activity, which suggests that CYP2C9 has a role in the relationships and discovery strategies to mitigate drug-drug interaction risks. oxidation of THT to Su. MRs also predicted graft failure and EFS. We J Med Chem 2012; 55: 4896–4933. observed an association between the CYP2B6 alleles encoding 17 Appendix: Drug Metabolizing Enzymes and Biotransformation Reactions. In: Zhang D, Surapaneni S (eds) ADME-Enabling Technologies in Drug Design and proteins with reduced enzymatic activity and reduced EFS. We Development. John Wiley & Sons, Inc.: Hoboken, New Jersey, 2012, pp 545–565. provided preliminary evidence regarding the roles of the MR and the 18 Krueger SK, Williams DE. Mammalian flavin-containing monooxygenases: struc- CYP2B6 genotype to predict the clinical outcome of HSCT; however, ture/function, genetic polymorphisms and role in drug metabolism. Pharmacol additional studies are needed to establish their clinical utility. Ther 2005; 106: 357–387. 19 Damani LA, Houdi AA. Cytochrome P-450 and FAD-monooxygenase mediated S- and N-oxygenations. Drug Metabol Drug Interact 1988; 6: 235–244. CONFLICT OF INTEREST 20 The International HapMap Project. Nature 2003; 426: 789–796. 21 Versace F, Uppugunduri CR, Krajinovic M, Theoret Y, Gumy-Pause F, Mangin P The authors declare no conflict of interest. et al. A novel method for quantification of sulfolane (a metabolite of busulfan) in plasma by gas chromatography-tandem mass spectrometry. Anal Bioanal Chem 2012; 404: 831–838. ACKNOWLEDGMENTS 22 Ansari M, Theóret Y, Rezgui M, Peters C, Mezziani S, Desjean C et al. Association We are grateful to all patients and their parents for participating in this study. This between busulfan exposure and outcome in children receiving intravenous investigation was supported by grants provided by CANSEARCH, the Geneva Cancer busulfan before hematopoietic stem cell transplantation, on behalf of the League and the Hans Wilsdorf and Charles Bruneau Foundations. We thank pediatric disease working parties of the european blood and marrow transplant MF Vachon, C Desjean and M Labuda for the collection, maintenance and group. Ther Drug Monit 2013 (in press).

The Pharmacogenomics Journal (2014), 263 – 271 & 2014 Macmillan Publishers Limited Busulfan metabolic ratio and CYP2C9 genotypes CRS Uppugunduri et al 271 23 Rifai N, Sakamoto M, Lafi M, Guinan E. Measurement of plasma busulfan con- hematopoietic stem cell transplantation. Eur J Clin Pharmacol 2013; 69: centration by high-performance liquid chromatography with ultraviolet detec- 1543–1551. tion. Ther Drug Monit 1997; 19: 169–174. 34 Begg EJ, Helsby NA, Jensen BP. Pharmacogenetics of drug-metabolizing enzymes: 24 Ansari M, Uppugunduri CR, Deglon J, Theoret Y, Versace F, Gumy-Pause F et al. the prodrug hypothesis. Pharmacogenomics 2012; 13: 83–89. A simplified method for busulfan monitoring using dried blood spot in combi- 35 Helsby NA, Hui CY, Goldthorpe MA, Coller JK, Soh MC, Gow PJ et al. The combined nation with liquid chromatography/tandem mass spectrometry. Rapid Commun impact of CYP2C19 and CYP2B6 pharmacogenetics on cyclophosphamide Mass Spectrom 2012; 26: 1437–1446. bioactivation. Br J Clin Pharmacol 2010; 70: 844–853. 25 McDonald GB, Hinds MS, Fisher LD, Schoch HG, Wolford JL, Banaji M et al. 36 Hassan M, Andersson BS. Role of pharmacogenetics in busulfan/cyclopho- Veno-occlusive disease of the liver and multiorgan failure after bone marrow sphamide conditioning therapy prior to hematopoietic stem cell transplantation. transplantation: a cohort study of 355 patients. Ann Intern Med 1993; 118: Pharmacogenomics 2013; 14: 75–87. 255–267. 37 Rezvani AR, McCune JS, Storer BE, Batchelder A, Kida A, Deeg HJ et al. Cyclo- 26 Przepiorka D, Weisdorf D, Martin P, Klingemann HG, Beatty P, Hows J et al. 1994 phosphamide followed by intravenous targeted busulfan for allogeneic hema- Consensus conference on acute GVHD grading. Bone marrow transplant 1995; 15: topoietic cell transplantation: pharmacokinetics and clinical outcomes. Biol Blood 825–828. Marrow Transplant 2013; 19: 1033–1039. 27 Pedersen RS, Brasch-Andersen C, Sim SC, Bergmann TK, Halling J, Petersen MS 38 Cantoni N, Gerull S, Heim D, Halter J, Bucher C, Buser A et al. Order of application et al. Linkage disequilibrium between the CYP2C19*17 allele and wildtype and liver toxicity in patients given BU and CY containing conditioning regimens CYP2C8 and CYP2C9 alleles: identification of CYP2C haplotypes in healthy Nordic for allogeneic hematopoietic SCT. Bone marrow transplant 2011; 46: 344–349. populations. Eur J Clin Pharmacol 2010; 66: 1199–1205. 39 Zhu ZH, Sun ML, Li ZS, Yang ZC, Zhang TB, Heng ZC et al. An investigation of the 28 McCune JS, Gibbs JP, Slattery JT. Plasma concentration monitoring of busulfan: maximum allowable concentration of sulfolane in surface water. Hua Xi Yi Ke Da does it improve clinical outcome? Clin Pharmacokinet 2000; 39: 155–165. Xue Xue Bao 1987; 18: 376–380. 29 Srivastava A, Poonkuzhali B, Shaji RV, George B, Mathews V, Chandy M et al. 40 Cooper AJ, Pinto JT, Callery PS. Reversible and irreversible protein glutathionylation: Glutathione S-transferase M1 polymorphism: a risk factor for hepatic venoocclu- biological and clinical aspects. Expert Opin Drug Metab Toxicol 2011; 7:891–910. sive disease in bone marrow transplantation. Blood 2004; 104: 1574–1577. 41 Melanson SE, Stevenson K, Kim H, Antin JH, Court MH, Ho VT et al. Allelic varia- 30 Balasubramanian P, Desire S, Panetta JC, Lakshmi KM, Mathews V, George B et al. tions in CYP2B6 and CYP2C19 and survival of patients receiving cyclopho- Population pharmacokinetics of cyclophosphamide in patients with thalassemia sphamide prior to myeloablative hematopoietic stem cell transplantation. Am J major undergoing HSCT. Bone Marrow Transplant 2012; 47: 1178–1185. Hematol 2010; 85: 967–971. 31 de Jonge ME, Huitema AD, Rodenhuis S, Beijnen JH. Clinical pharmacokinetics of 42 Black JL, Litzow MR, Hogan WJ, O’Kane DJ, Walker DL, Lesnick TG et al. Correlation cyclophosphamide. Clin Pharmacokinet 2005; 44: 1135–1164. of CYP2B6, CYP2C19, ABCC4 and SOD2 genotype with outcomes in allogeneic 32 Timm R, Kaiser R, Lotsch J, Heider U, Sezer O, Weisz K et al. Association of blood and marrow transplant patients. Leuk Res 2012; 36: 59–66. cyclophosphamide pharmacokinetics to polymorphic cytochrome P450 2C19. 43 Rocha V, Porcher R, Fernandes JF, Filion A, Bittencourt H, Silva Jr. W et al. Asso- Pharmacogenomics J 2005; 5: 365–373. ciation of drug metabolism gene polymorphisms with toxicities, graft-versus-host 33 Kim IW, Yun HY, Choi B, Han N, Kim MG, Park S et al. Population pharmaco- disease and survival after HLA-identical sibling hematopoietic stem cell trans- kinetics analysis of cyclophosphamide with genetic effects in patients undergoing plantation for patients with leukemia. Leukemia 2009; 23: 545–556.

Supplementary Information accompanies the paper on the The Pharmacogenomics Journal website (http://www.nature.com/tpj)

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 263 – 271