Effects of OCT1 Polymorphisms on the Cellular Uptake, Plasma Concentrations and Efficacy of the 5-HT3 Antagonists Tropisetron and Ondansetron

Effects of OCT1 polymorphisms on the cellular uptake, plasma concentrations and efficacy of the 5-HT3 antagonists tropisetron and ondansetron

MV Tzvetkov, AR Saadatmand, After uptake into liver cells, the antiemetic drugs tropisetron and ondansetron undergo metabolic inactivation by cytochrome P450 2D6 (CYP2D6). K Bokelmann, I Meineke, We investigated whether the hepatic organic cation transporter 1 (OCT1; R Kaiser and J Brockmo¨ller SLC22A1) mediates cellular uptake and whether common OCT1 loss-of- function polymorphisms affect pharmacokinetics and efficacy of both Department of Clinical Pharmacology, University þ Medical Center, Georg-August-University, drugs. Both tropisetron and ondansetron inhibited ASP uptake in OCT1- Go¨ttingen, Germany overexpressing HEK293 cells. Overexpression of wild-type, but not OCT1 loss-of-function variants, significantly increased tropisetron uptake. Corre- Correspondence: spondingly, patients with two loss-of-function OCT1 alleles had higher Dr M Tzvetkov, Department of Clinical tropisetron plasma concentrations (n ¼ 59, Po0.04) and higher clinical Pharmacology, University Hospital Go¨ttingen 4E1-323, Robert-Koch-Strasse 40, Go¨ttingen, efficacy (n ¼ 91, P ¼ 0.009) compared with carriers of fully active OCT1. D-37075, Germany. Overexpression of OCT1 did not increase ondansetron uptake. Nevertheless, E-mail: [email protected] OCT1 genotypes correlated with pharmacokinetics (n ¼ 45, Po0.05) and clinical efficacy (n ¼ 222, Po0.02) of ondansetron, the effect size of OCT1 genotypes on pharmacokinetics and efficacy was greater for tropisetron than for ondansetron. In conclusion, in addition to the known effects of CYP2D6, OCT1 deficiency may increase efficacy of tropisetron and potentially of ondansetron by limiting their hepatic uptake. The Pharmacogenomics Journal (2012) 12, 22–29; doi:10.1038/tpj.2010.75; published online 5 October 2010

Keywords: organic cation transporter OCT1; SLC22A1; 5-HT3 antagonists; tropisetron; ondansetron; antiemetic treatment

Introduction

Although significant progress has been achieved in supportive care of cancer patients, many patients still suffer from chemotherapy-induced nausea and vomiting. Nausea and vomiting in cancer patients receiving moderate and high emetogenic chemotherapy are treated by serotonin receptor type (5-hydroxy- 1 tryptamine 3 (5-HT3)) antagonists, such as tropisetron and ondansetron. But still, approximately 15–30% of the patients do not respond sufficiently to the current antiemetic therapy.1–3 One reason for the substantial differences in response may be genetic variability in drugs uptake and elimination. The major elimination route of tropisetron and ondansetron is via metabolic Received 11 March 2010; revised 17 August 2010; accepted 26 August 2010; inactivation by hepatic cytochrome P450 enzymes. Tropisetron is mainly, and published online 5 October 2010 ondansetron is partially, metabolized by cytochrome P450 2D6 (CYP2D6). Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 23

Genetic polymorphisms in CYP2D6 are known to affect the the periods 0–4 h and 5–24 h after the start of the pharmacokinetics and the efficacy of these two 5-HT3 chemotherapy. A vomiting episode was defined as a single antagonists, but these polymorphisms could explain only vomit or retch, or any number of continuous vomits or a small fraction of the interindividual variation in the retches. Vomiting or retching had to be absent for at least efficacy of antiemetic treatment.4 Compared with metabo- 1 min to be calculated as different episodes according to the lism, less is known about the cellular uptake of the two 5- definition of the Italian Group for Antiemetic Research. The

HT3 antagonists. first study has been approved by the ethics committee of the As 5-HT3 antagonists, like their natural counterpart Humboldt University Berlin and the second by the ethics serotonin, are moderately hydrophobic organic cations, committee of the Georg-August University in Go¨ttingen. All human organic cation transporters (OCTs) may be respon- patients have given written informed consent. sible for their cellular uptake. The OCT isoform OCT1 is by far the most highly expressed OCT in the human liver.5–7 Pharmacokinetic measurements The activity of OCT1 varies greatly between individuals Plasma concentrations of ondansetron and tropisetron were because of common genetic polymorphisms, including the measured by high-performance liquid chromatography amino acids substitution arginine61-to-cysteine (R61C), (HPLC) as described earlier.10 The limit of quantification of cysteine88-to-arginine (C88R), glycine401-to-serine (G410S), this method was 1.25 ng ml–1 for tropisetron and 0.62 ng ml–1 glycine465-to-arginine (G465R) and a deletion of methio- for ondansetron. Interassay coefficient of variation for the nine420 (M420del).7–9 In this study we analyzed whether concentration range between 1.9 and 15 ng ml–1 ranged OCT1 may mediate the cellular uptake of tropisetron and from 8.5 to 13.7% for tropisetron and from 4.8 to 9.4% for ondansetron and whether, and to what extent, genetic ondansetron. polymorphisms in OCT1 contribute to the variability in pharmacokinetics and therapeutic efficacy of tropisetron and OCT1 genotyping ondansetron in cancer patients. Genomic DNA was isolated from venous blood samples using an automated solid-phase extraction method according to the manufacturer’s instructions (EZ1 DNA blood Materials and methods 350 ml kit used with the Bio-robot EZ1; both from Qiagen, Hilden, Germany). The genotyping of OCT1 (alternative Clinical study gene name SLC22A1) was performed using a single-base The medical impact of OCT1 polymorphisms in the primer extension method as described before.7 The method interaction with CYP2D6 polymorphisms was analyzed in was modified to genotype only the amino acid substitutions a noninterventional clinical cohort study conducted at the R61C, C88R, G401S, M420del and G465R. Standard quality University Medical Center Charite´ and the community control procedures were performed as described earlier,11 hospital Krankenhaus Moabit, both in Berlin, Germany. showing 100% concordance in the sample calls. The study was designed to evaluate the pharmacogenetics of The PHASE 2.1 software12 was used to estimate the two cancer chemotherapy-induced nausea and vomiting in haplotypes of each individual. Ten independent runs with cancer patients receiving moderately to highly emetogenic different seed numbers were performed (used seeds 2, 1536, chemotherapy. More detailed results on the impact of 2936, 3123, 4957, 5283, 6757, 7992, 8633 and 9045) in order CYP2D6 polymorphisms in this context were published to avoid seed-biased allele assignments. elsewhere.4 This study included 270 patients with a mean age of 53.7 years (range 18–83 years; s.d. 13.3) who received Generation of HEK293 cell lines stably expressing OCT1 either ondansetron- or tropisetron-based antiemetic treat- Cells overexpressing human OCT1 were generated by ment. Of these patients, 32% had breast cancer, 15.4% lung targeted integration of OCT1 complementary DNA into cancer, 14.2% non-Hodgkin’s lymphoma, 4.9% multiple the genome of HEK293 cells using the Flp-In system myeloma, 4.9% Hodgkin’s disease and 28.1% miscellaneous (Invitrogen, Karlsruhe, Germany). Therefore, the OCT1 other tumors. A second study sample included 60 cancer open reading frame was cut out with HindIII and EcoRV patients who received ondansetron antiemetic treatment. from an OCT1 expression plasmid pcDNA3::OCT1 (kindly These patients were from the University Medical Center in supplied by Herman Koepsell and Valentin Gorboulev, Go¨ttingen with a mean age of 53.4 years (range 19–81 years; University of Wu¨rzburg, Germany) and ligated into s.d. 13). Of these patients, 66% had non-Hodgkin’s lym- pcDNA5.1 plasmid opened with the same restriction phoma, 11% Hodgkin’s disease, 3.1% multiple myeloma, enzymes. The resulting construct pcDNA5.1::OCT1 carries 3.1% lung cancer and 16.8% miscellaneous other tumors. the wild-type OCT1. In addition, point mutations that Tropisetron (Navoban; Novartis Pharma, Basel, Switzerland) represent the five lost-of-function amino acid substitutions was given in a dosage of 5 mg once daily (n ¼ 96), and in OCT1 were introduced into the pcDNA3::OCT1 construct ondansetron (Zofran; GlaxoSmithKline, Brentford, UK) was by targeted mutagenesis. The complete OCT1 open reading given in a dosage of 8 mg twice daily (n ¼ 174 for the first frame was sequenced to confirm the correctness of the and n ¼ 60 for the second study samples). Episodes of acute obtained OCT1 variants and then also re-cloned in the vomiting within the first 24 h after the start of chemothe- pcDNA5.1 plasmid. The different pcDNA5.1::OCT1 con- rapy were recorded by the patients in a diary, separately for structs were chromosomaly integrated in T-Rex HEK293 cells

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 24

(HEK293 cells genetically engineered to be used as hosts for phase (see below) and 50 ml was injected into the HPLC for targeted chromosomal integration; Invitrogen) using the quantification. Flp-In system according to the manufacturer’s instructions. Tropisetron and ondansetron quantifications were per- Successful genomic integration was confirmed by integra- formed using a modification of a previously described HPLC tion-specific PCR, and the overexpression of OCT1 was method.10 Briefly, tropisetron and ondansetron were quan- validated by real-time reverse transcriptase-PCR. Clones tified using a LaChrom HPLC system (Merck Hitachi, carrying the wild-type or the loss-of-function OCT1 variants Darmstadt, Germany) consisting of an interface (D-7000, showed on average 15-fold higher OCT1 mRNA expression Merck Hitachi), a pump (L-7100, Merck Hitachi), an compared with the control clone carrying genomic integra- automatic sampler (L-7200, Merck Hitachi), an ultraviolet tion of the pcDNA5.1 vector alone. No significant differ- detector (L-7400, Merck Hitachi) and a degasser (L-7614, ences were detected in the OCT1 mRNA expression between Merck Hitachi). The compounds were separated at room the wild-type and the lost-of-function variants. temperature on a LiChrospher 100 reverse phase-18e (5 mm, 4 Â 150 mm) column with a LiChrospher 100CN guard column (5 mm, both from Merck) and quantified by ultra- Inhibition of ASP þ uptake violet detection at 284 nm for tropisetron and 305 nm for A total of 800 000 cells were plated per single well of 12-well ondansetron. The mobile phase consisted of 20% acetoni- plates (Nunc, Langenselbold, Germany), which had trile and 80% 0.05 M sodium acetate buffer (pH 5.0) and was been precoated with poly-D-lysine (Sigma, Taufkirchen, delivered at a flow rate of 1.5 ml min–1. Tropisetron and Germany). The cells were grown for 48 h to reach ondansetron were detected as peaks with retention time of confluence. Prior to the uptake measurements, the cells 5.7 and 4.3 min, respectively, and quantified using peak area were washed twice with 1 ml warm (37 1C) Hank’s buffered with external standardization. The amount of intracellular salt solution (HBSS, Invitrogen). The uptake was initiated tropisetron and ondansetron was normalized to the total by adding 0.5 ml warm (37 1C) HBSS that contained 1 mM ASP number of cells. (4-(4-(dimethylamino)styryl)-N-methylpyridinium iodide; Invitrogen) and varying concentrations of the respective inhibitory substance. The uptake was stopped after 3 min by Results adding 2 ml ice-cold HBSS. The HBSS medium was discarded and the cells were lysed in 0.5 ml RIPA buffer (50 mM In vitro analyses of the OCT1-mediated uptake of tropisetron and Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% (v/v) NP-40 and ondansetron 0.25% (v/v) Na-deoxycholate; all purchased from Sigma). In order to analyze whether OCT1 mediates the cellular The amount of intracellular ASP þ was quantified by uptake of tropisetron and ondansetron, wild-type OCT1 was measuring the fluorescence at 485 nm excitation and overexpressed in HEK293, a human embryonic kidney cell 612 nm emission wavelengths using TECAN Ultra plate line, using stable genome integration. Both tropisetron and photometer (TECAN, Crailsheim, Germany). ondansetron effectively competed the uptake of the well- þ established OCT1 substrate ASP . Tropisetron IC50 was Direct quantification of cellular uptake of tropisetron and 8.5±1.4 mM and ondansetron IC50 was 63±15.6 mM ondansetron (Figure 1a). The inhibition constants were comparable A total of 9 million cells were plated on 100 mm plates (BD (ondansetron) or even tenfold lower (tropisetron) to the Falcon, Heidelberg, Germany), which had been precoated inhibition constant of quinine, a well-known OCT1 sub- with poly-D-lysine, and grown for 48 h to reach confluence. strate.13 The inhibition results corresponded to previously The number of used cells was based on the quantification reported data,13,14 but still do not clarify whether both drugs limit of 1.25 ng ml–1 (4.5 pmol ml–1 for tropisetron) of the are substrates that are actively transported by OCT1, or act HPLC detection method,10 assumed intracellular concentra- only as inhibitors of OCT1. tion of at least 1 mM and intracellular volume of 1 ml for To confirm tropisetron and ondansetron as OCT1 sub- 1 Â 106 cells. Prior to the uptake measurements, the cells strates, we measured the intracellular uptake of both drugs were washed with 10 ml warm (37 1C) HBSS. The uptake was using HPLC. Compared with cells transfected with blank initiated by the addition of 5 ml warm (37 1C) HBSS plasmid, the OCT1-overexpressing cells showed a 2.3-fold containing 1 mM tropisetron or ondansetron. The uptake increase in the intracellular accumulation of tropisetron was stopped after 3 min by the addition of 20 ml ice-cold (Po0.001, Student’s t-test; Figure 1b). This increase HBSS. The cells were washed twice with ice-cold HBSS and was completely reversible by MPP þ (1-methyl-4-phenylpy- an aliquot was used for determination of total protein ridinium), a specific model substrate and competitive amount in the samples. The remaining cells were lysed and inhibitor of OCT1.13 In the same transport assay, the the proteins were precipitated by adding 1 ml of a mixture of HEK293 cells showed high intracellular levels of ondanse- 80% acetonitrile and 20% 0.05 M sodium acetate (pH 5.0). tron, and OCT1 overexpression did not result in additional The samples were then centrifuged (10 min at 16 000 g)to increase of ondansetron uptake. pellet the cell debris, the supernatant was transferred into a Second, we analyzed how common polymorphisms in the 10 ml glass tube and evaporated to dryness at 40 1C under OCT1 gene may influence tropisetron uptake. Therefore, we nitrogen. The residue was reconstituted in 200 ml of mobile overexpressed OCT1 variants carrying the five common

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 25

substitutions were analyzed that are natively found as polymorphisms in Caucasian populations.7 In all variants, the uptake of tropisetron was reduced to the levels of the OCT1 non-expressing cells (Po0.001, one-way ANOVA; Figure 1b). Thus, the in vitro data showed that OCT1 transported tropisetron and that the common loss-of- function polymorphisms in OCT1 abolished tropisetron transport. Although ondansetron inhibited the OCT1 transport, a direct OCT1-mediated uptake of ondansetron could not be measured in our in vitro system.

Effects of OCT1 polymorphisms on the pharmacokinetics and efficacy of tropisetron and ondansetron We analyzed whether the OCT1 genetic polymorphisms modified the pharmacokinetics and clinical effects of the

5-HT3 antagonists. For that purpose, we utilized the data and DNA of patients from a previous study on the antiemetic treatment efficacy of tropisetron and ondansetron against chemotherapy-induced nausea and vomiting.4 All 270 patients were Caucasians and had received standard doses of either tropisetron (n ¼ 96) or ondansetron (n ¼ 174) as prophylactic antiemetic treatment. DNA was not available from nine patients, and clinical data were incomplete for additional eight patients. Thus, data of 253 patients were available for the efficacy analyses. Thereof, 104 patients agreed to participate in a pharmacokinetic substudy. The 253 patients were genotyped for the five common amino acid substitutions in OCT1, that is, R61C, C88R, G401S, M420del or G465R. The genotypes obtained were in Hardy– Weinberg equilibrium and the allele frequencies corresponded to those previously reported in Caucasians.7,9 Corresponding to our results on tropisetron with the cell line experiments, we regarded as fully active those OCT1 alleles that lacked any of the five amino acid substitutions (Figure 2a). In all, 31 patients (12%) carried two deficient OCT1 alleles, 96 (38%) carried only one deficient and one Figure 1 Organic cation transporter 1 (OCT1)-mediated cellular uptake fully active OCT1 allele and 126 (50%) carried two fully of tropisetron and ondansetron. (a) Both tropisetron and ondansetron active OCT1 alleles. inhibit the cellular uptake of the typical OCT1 substrate ASP þ (4-(4- (dimethylamino)styryl)-N-methylpyridinium iodide (4-Di-1-ASP). OCT1- First, we analyzed the plasma concentrations of tropise- overexpressing HEK293 cells were incubated with 1 mM ASP þ for 3 min tron and ondansetron in relation to the number of fully in the presence of increasing concentrations of tropisetron, ondansetron, active OCT1 alleles. The plasma concentrations of tropise- the weak OCT1 inhibitor tetraethylammonium (TEA þ ) and the strong tron were measured at 3 h (n ¼ 59) and at 6 h (n ¼ 39) after OCT1 inhibitor quinine. The graph shows mean and s.e. of three administration. The plasma concentrations of ondansetron independent experiments. (b) Direct measurements of cellular uptake of were measured at 3 h after administration (n ¼ 45). The tropisetron in HEK293 cells. Control cells (pcDNA5.1, gray bars) and cells plasma concentrations of tropisetron at 3 and 6 h after overexpressing the wild-type (WT, black bar) or polymorphic OCT1 forms administration and of ondansetron at 3 h after administra- (black bars) were incubated with 1 mM tropisetron for 3 min. The intracellular accumulation of tropisetron was measured by high-perfor- tion were highest in the subgroups of patients lacking any mance liquid chromatography (HPLC). The cellular uptake of tropisetron fully active OCT1 allele and decreased with the increasing was blocked in the cells expressing wild-type OCT1 using 100 mM MPP þ number of fully active OCT1 alleles (Po0.05; Figures 2b–d). (gray hatched bars). The analyzed polymorphic OCT1 forms carried the Second, we analyzed the efficacy of antiemetic therapy in single amino acid substitutions R61C, G401S, M420del or the combina- relation to the number of fully active OCT1 alleles. tions of M420del with C88R and of M420del with G465R that represent all Antiemetic therapy efficacy was measured by the number commonly existing OCT1 alleles in Caucasians. The graph shows mean and of vomiting episodes in the first 24 h of the chemotherapy. s.e. of three independent experiments. Among the 253 patients, from whom the efficacy and genetic data were available, vomiting was observed in 59 amino acid substitutions, R61C, C88R, G401S, M420del or patients (23.2%). Correspondingly to the increased plasma G465R,7,9 and compared their activity with the activity of concentrations of tropisetron and ondansetron, the patients the wild-type OCT1. Only those combinations of amino acid lacking any fully active OCT1 allele vomited more than

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 26

three times less frequently than patients with one or two episodes in the carriers of fully active OCT1, P ¼ 0.009; fully active OCT1 alleles (0.35 compared with 1.17 mean Figure 3b), and a similar strong trend in the same direction episodes of vomiting in the first 24 h, P ¼ 0.007; Figure 3a). was present in the group of patients who had received The effect was present both in the first 4 h and in the ondansetron (P ¼ 0.062, data not shown). We confirmed the following 20 h after initiation of the chemotherapy effect of OCT1 genotypes on ondansetron efficacy in an (P ¼ 0.003 and P ¼ 0.037, respectively; data not shown). independent second sample of 60 patients. The patients The effect was significant in the subgroup of patients who lacking fully active OCT1 (n ¼ 8) did not vomit at all, had received tropisetron (a mean of 0.08 episodes of whereas the patients with one or two fully active OCT1 vomiting in the non-carriers compared with a mean of 0.8 alleles experienced 0.75 mean episodes of vomiting in the first 24 h (n ¼ 52). Combined analyses of the two groups who had received ondansetron showed a significant decrease in vomiting incidence in the patients lacking any fully active OCT1. A mean of 0.37 episodes of vomiting was observed in the group lacking fully active OCT1 compared with a mean of 1.27 episodes of vomiting observed in the carriers of one or two fully active OCT1 gene copies (P ¼ 0.018; Figure 3c). Finally, we analyzed the combined effects of OCT1 and CYP2D6 polymorphisms on the plasma concentrations and the efficacy of tropisetron and ondansetron. As shown earlier in the same sample of patients, high activity of CYP2D6 is associated with rapid elimination of the antiemetic drugs and poor antiemetic efficacy.4 Therefore, to exclude confounding effects of CYP2D6 activity, we adjusted the effects of the OCT1 genotypes for the effects of the CYP2D6 genotypes using multiple linear regression. The patients were stratified into four groups according to their number of active CYP2D6 alleles: 0 active alleles (poor metabolizers), 1 (intermediate metabolizers), 2 (extensive metabolizers) and 3 and more active alleles (ultra-rapid metabolizers). After the adjustment for the number of active CYP2D6 alleles, the plasma concentrations of tropisetron remained significantly dependent on the OCT1 genotypes at both 3 and 6 h after the administration (P ¼ 0.02 and P ¼ 0.04, respectively). OCT1 genotype explained 8.1% of the variation in tropisetron plasma concentrations at 3 h and 11.3% at 6 h after administration (calculated as r2 from linear regression analysis). CYP2D6 genotype explained 9.4% of the variation in tropisetron plasma concentrations at 3 h and 12% at 6 h after administration. The effect size of the OCT1 polymorphisms on tropisetron concentrations was lower than this of the CYP2D6 polymorphisms. The increases in the mean tropisetron concentrations between carriers of two deficient

Figure 2 Plasma concentrations of tropisetron and ondansetron depend on organic cation transporter 1 (OCT1) genotype. The haplotype combination of the five common functional amino acid substitutions in the OCT1 gene and their frequencies are shown for the 253 cancer patients from the first study sample (a). One fully active allele and five deficient OCT1 alleles, the latter with a combined allele frequency of 29.7%, were observed. Minor alleles are shown in gray. Plasma concentrations of tropisetron measured at 3 h (b)and6h(c)afterdrug administration and plasma concentrations of ondansetron measured at 3 h after administration (d)areshownindependenceonthenumberoffully active OCT1 alleles. For organizational reasons (mostly outpatients), in the ondansetron pharmacokinetic substudy, only 3-h samples could be taken. Bars represent mean and s.e. of drug concentrations 3 h after administration. The number of analyzed patients with different OCT1 alleles is indicated (n). P-values are significances from linear regression analysis.

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 27

Also, the plasma concentrations of ondansetron remained significantly dependent on the OCT1 genotypes after adjustment for the effect of the CYP2D6 genotypes (P ¼ 0.04). OCT1 genotype explained 9% of the variation in ondansetron plasma concentrations (calculated as r2 from linear regression analysis). The effect of the OCT1 genotypes on antiemetic therapy efficacy also remained significant after adjustment for the CYP2D6 genotypes (Po0.03 in multiple linear regression analyses). OCT1 genotype explained 1.8% of the variation in frequencies of vomiting and CYP2D6 genotype explained only 1.2%. Also, the effect size of the OCT1 polymorphisms on vomiting was stronger than the effect size because of the CYP2D6 polymorphism. The difference in mean number of episodes of vomiting between carriers of two deficient versus two fully active OCT1 alleles was 76%, whereas the corresponding difference between carriers of two deficient versus two fully active CYP2D6 alleles was only 57%. The CYP2D6 gene dose-like effect was present only at the background of highly active OCT1 (two fully active OCT1 alleles, data not shown), which appeared mechanistically plausible: if the transport is limiting, the metabolism is less dependent on the intrinsic intracellular metabolic capacity. However, the sample size was too small for further in-depth analysis of the OCT1–CYP2D6 gene–gene interactions on tropisetron and ondansetron efficacy.

Discussion

In the past 30 years, convincing data have emerged about the impact of polymorphisms in cytochrome P450 enzymes on the pharmacokinetics and clinical effects of many drugs.15,16 However, before a drug can be metabolized, it must enter the cell, and the data presented here showed how polymorphisms in a transporter mediating the cellular uptake of a drug may modify pharmacokinetics and clinical effects. Specifically, our data suggest that patients lacking fully active OCT1 have impaired cellular uptake, reduced bio-inactivation and consequently a prolonged residence of the drug in the circulation, resulting in better efficacy. This was consistently shown for tropisetron, and several data indicated that this may also be the case for ondansetron. Effects of the loss-of-function polymorphisms in the liver Figure 3 The therapy efficacy of tropisetron and ondansetron depends uptake transporter OCT1 were comparable in strength with on organic cation transporter 1 (OCT1) genotype. Frequency of vomiting the effects of loss-of-function polymorphisms in the drug- under treatment is shown in dependence on the number of fully active metabolizing enzyme CYP2D6.4 Thus, our data illustrate OCT1 alleles for all 253 patients of the first study sample (a) and that the carrier-mediated hepatic uptake may be a rate- separately for the patients treated with tropisetron (b) or ondansetron limiting factor for the metabolism of organic cationic drugs. (c). For ondansetron, the data from the combined first and second study Thus, the rate-limiting effect of genetic polymorphisms in sample are shown. Bars represent mean and s.e. of vomiting episodes in OCT1 on the uptake and metabolism of organic cationic the first 24 h after the chemotherapy began. The number of patients drugs may correspond to the well-studied effects of genetic with different OCT1 genotypes is indicated (n). polymorphisms in OATP1B1 on the metabolism of some organic anionic drugs.17–20 compared with two fully active OCT1 alleles were 4.9-fold at Our in vitro and in vivo data were consistent regarding the 3 h and twofold at 6 h after administration. The correspond- involvement of OCT1 in the cellular uptake, consequently ing increases for two deficient versus two fully active in the pharmacokinetics and finally in the efficacy of CYP2D6 alleles were 11-fold and 2.8-fold. tropisetron. First, tropisetron inhibited OCT1-mediated

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 28

Table 1 Comparison of physicochemical properties that may these populations).9 Interestingly, the most common varia- affect the membrane permeability of tropisetron and tion in OCT1 coding region, the methionine420 deletion ondansetron (allele OCT1*2), seems to have highly variable effects on OCT1 activity, depending on the substrate used. In our Drug pKa % charged % uncharged Lipophilicity study, OCT1*2 showed no activity for tropisetron uptake at pH 7.4 at pH 7.4 logD at pH 7.4 (Figure 1), but showed close to 30% of the wild-type activity for metformin uptake21 and close to 100% activity for MMP þ Tropisetron 8.9 96.6 3.4 1.36 8,9 and ASP þ (Tzvetkov et al., unpublished data). Thus, the Ondansetron 7.4 50.0 50.0 2.28 frequency of individuals lacking active OCT1-mediated mem- The pKa value of ondansetron is according to Maschru et al.,29 and all other branetransportmaybedependentontherespectivesubstrate. physicochemical properties were calculated using ADMET Predictor Software 4.0 Clinical consequences for patients lacking any fully active (Simulations Plus, Lancaster, CA, USA). OCT1, and consequently having diminished bio-inactivation of tropisetron and ondansetron, may include not only transport. Second, OCT1 overexpression resulted in an increased incidence of adverse effects, but also improved increased uptake of tropisetron that was completely blocked efficacy of the antiemetic treatment. Constipation is a þ 23 by addition of the typical OCT1 substrate MMP acting as a relevant adverse effect of 5-HT3 antagonists, and one competitive inhibitor of tropisetron transport. Third, genetic might speculate that the risk for constipation may be polymorphisms known to cause reduction or loss of OCT1 increased in a case of OCT1 deficiency. However, generally function with other substrates21 also resulted in the loss of this class of drugs has only relatively few adverse effects.3 On in vitro uptake of tropisetron (all in Figure 1). Fourth, the drug the other hand, an increased antiemetic efficacy mostly plasma concentrations (Figure 2), and fifth, the frequency of without major adverse effects, which is expected according vomiting of cancer patients under tropisetron treatment to our data in patients lacking active OCT1 function, is a depended on the presence of loss-of-function OCT1 alleles therapeutic advantage that would not require any change in (Figure 3). Sixth, the in vivo effects of OCT1 could not be the dosing or scheduling of the antiemetic treatment. From explained by the potentially confounding CYP2D6 genotypes. that perspective, there would be no need for OCT1 Plasma concentrations and efficacy of ondansetron were genotyping of patients who would be treated with these also dependent on OCT1 genotypes, although the genotype drugs. However, as has been shown, carriers of three and effects were less strong compared with tropisetron. Ondan- more active CYP2D6 alleles, are at higher risk of antiemetic setron was also able to inhibit ASP uptake, but we were not therapy failure.4,24,25 These patients may indeed be particu- able to measure directly the cellular uptake of ondansetron larly prone to antiemetic treatment failure if they carry in in vitro. One explanation for the consistently weaker in vivo addition two fully active OCT1 alleles. This rare group of and in vitro effects of ondansetron may be the lower pKa patients may profit from both OCT1 and CYP2D6 genotyp- value of ondansetron (Table 1). Thus, a higher fraction of ing. In this rare group, a combined genotyping of OCT1 and ondansetron is not protonated in blood (or in the buffer CYP2D6 may have clinical consequences like dose escalation used in the in vitro uptake experiments). In addition, or selection of other 5-HT3 antagonists that are not ondansetron is more lypophilic (Table 1) and showed lower dependent on polymorphic CYP2D6. How pharmacokinetics affinity to OCT1 than tropisetron (IC50 of 63.9 compared and efficacy of tropisetron and ondansetron may be altered in with 8.6 mM; Figure 1a). Taken together, these data suggest CYP2D6 ultra-rapid metabolizers that lack any fully active that ondansetron may be more prone to enter cells by OCT1 allele is an interesting question. Unfortunately, our passive diffusion than tropisetron, and may be less depen- sample size was too small to answer that question. dent on OCT1. Still, similar to tropisetron, loss-of-function Drug–drug interactions on OCT1 may also account for OCT1 genotypes lead to increased plasma concentrations additional interindividual variability in the hepatic uptake and efficacy of ondansetron. The last finding was observed in of tropisetron and ondansetron. A number of drugs, two independent samples, but again the effect of OCT1 including the antidiabetic drug metformin, the anti- variants on ondansetron was weaker than on tropisetron. malarial drug quinine and the chemotherapeutic drug There may be many additional differences between our oxaliplatin, but not cis- or carboplatin, are substrates for in vitro cell culture system and the liver. An unknown OCT1.7,13,21,26,27 Although metformin and quinine are not alternative uptake transporter for ondansetron may be present typically co-administrated with tropisetron or ondansetron, in the HEK293 cells. Also, pH gradients or the OCT1 patients treated with oxaliplatin may be expected to have phosphorylation status, which may modulate substrate speci- reduced hepatocellular uptake and thus reduced metabolism ficity,22 may differ between the in vitro model and the human and increased efficacy of tropisetron and potentially of liver. However, this remains completely speculative at present. ondansetron. However, there were no patients treated with The lack of any fully active OCT1 is a frequent condition. oxaliplatin in our clinical sample to prove this assumption. In our study, the patients lacking any relevant OCT1 activity OCT1 may transport serotonin itself and, by its expression were close to 12%. The observed OCT1 allele frequencies in the intestine, may mediate reuptake of serotonin resembled those found in other Caucasians samples,7 but released from the enterochromaffine cells upon cytostatic may by lower in Africans and Asians (mostly because of the chemotherapy.28 However, in this case OCT1 loss-of-func- lower prevalence of the methionine420 deletion in most of tion variants would be expected to result in an increased

The Pharmacogenomics Journal Tropisetron and ondansetron transport via OCT1 MV Tzvetkov et al 29

incidence of vomiting, which was apparently not observed organic cation transporter hOCT1 and their functional consequences. (Figure 3). A possible explanation is that loss-of-function Pharmacogenetics 2002; 12: 591–595. OCT1 polymorphisms affect the metabolism of 5-HT 9 Shu Y, Leabman MK, Feng B, Mangravite LM, Huang CC, Stryke D et al. 3 Evolutionary conservation predicts function of variants of the human organic antagonists stronger than the intestinal serotonin reuptake. cation transporter, OCT1. Proc Natl Acad Sci USA 2003; 100: 5902–5907. In addition, quantitatively, the role of OCT1 in the reuptake 10 Bauer S, Sto¨mer E, Kaiser R, Tremblay PB, Brockmo¨ller J, Roots I. of serotonin in intestine may be limited, and other Simultaneous determination of ondansetron and tropisetron in human plasma transporters, such as the high-affinity serotonin reuptake using HPLC with UV detection. Biomed Chromatogr 2002; 16: 187–190. 11 Tzvetkov MV, Meineke I, Sehrt D, Vormfelde SV, Brockmo¨ller J. 28 transporter SERT, may be more important. Amelogenin-based sex identification as a strategy to control the identity Finally, we should acknowledge some limitations of our of DNA samples in genetic-association studies. Pharmacogenomics study. Because of the limited sample size, we were not able 2010; 11: 449–457. to analyze interesting combinations of genotypes such as 12 Stephens M, Donnelly P. A comparison of Bayesian methods for haplotype reconstruction from population genotype data. Am J Hum the CYP2D6 ultra-rapid metabolizers combined with the Genet 2003; 73: 1162–1169. OCT1 low transporters. However, both tropisetron and 13 Koepsell H, Lips K, Volk C. Polyspecific organic cation transporters: ondansetron showed a consistent pattern in both pharma- structure, function, physiological roles, and biopharmaceutical implica- cokinetics and antiemetic effects, which confirms the tions. Pharm Res 2007; 24: 1227–1251. 14 Ahlin G, Karlsson J, Pedersen JM, Gustavsson L, Larsson R, Matsson P et al. combined impact of OCT1 and CYP2D6 in antiemetic Structural requirements for drug inhibition of the liver specific human treatment with ondansetron and tropisetron. Our data also organic cation transport protein 1. J Med Chem 2008; 51: 5932–5942. suggest that the dependence on OCT1-mediated uptake into 15 Brockmo¨ller J, Kirchheiner J, Meisel C, Roots I. Pharmacogenetic the liver may increase with the decrease in the lipophilicity diagnostics of cytochrome P450 polymorphisms in clinical drug develop- ment and in drug treatment. Pharmacogenomics 2000; 1: 125–151. of the drug. We think that the data presented here may 16 Ingelman-Sundberg M. Pharmacogenetics of cytochrome P450 and its stimulate further systematic research on the interactions applications in drug therapy: the past, present and future. Trends between polymorphisms affecting cellular uptake and those Pharmacol Sci 2004; 25: 193–200. affecting metabolism of drugs. 17 Link E, Parish S, Armitage J, Bowman L, Heath S, Matsuda F et al. SLCO1B1 variants and statin-induced myopathy–a genomewide study. N Engl J Med 2008; 359: 789–799. Conflict of interest 18 Romaine SP, Bailey KM, Hall AS, Balmforth AJ. The influence of SLCO1B1 (OATP1B1) gene polymorphisms on response to statin therapy. Pharmacogenomics J 10:1–11. The authors declare no conflict of interest. 19 Vormfelde SV, Toliat MR, Schirmer M, Meineke I, Nu¨rnberg P, Brockmoller J. The polymorphisms Asn130Asp and Val174Ala in OATP1B1 Acknowledgments and the CYP2C9 allele *3 independently affect torsemide pharmacokinetics and pharmacodynamics. Clin Pharmacol Ther 2008; 83: 815–817. 20 Werner D, Werner U, Meybaum A, Schmidt B, Umbreen S, Grosch A We are grateful to Herman Koepsell and Valentin Gorboulev for et al. Determinants of steady-state torasemide pharmacokinetics: kindly providing the initial OCT1-expressing construct and Rene impact of pharmacogenetic factors, gender and angiotensin II receptor Niehus for his help with the genotyping. In addition, we thank all blockers. Clin Pharmacokinet 2008; 47: 323–332. patients who participated in the clinical investigation. This project 21 Shu Y, Sheardown SA, Brown C, Owen RP, Zhang S, Castro RA et al. was financially supported by DFG grant GRK1034/2 to MVT and ARS. Effect of genetic variation in the organic cation transporter 1 (OCT1) on metformin action. J Clin Invest 2007; 117: 1422–1431. 22 Ciarimboli G, Koepsell H, Iordanova M, Gorboulev V, Durner B, Lang D References et al. Individual PKC-phosphorylation sites in organic cation transporter 1 determine substrate selectivity and transport regulation. JAmSoc 1 Roila F, Hesketh PJ, Herrstedt J. Prevention of chemotherapy- and Nephrol 2005; 16: 1562–1570. radiotherapy-induced emesis: results of the 2004 Perugia International 23 Horton R. Lotronex and the FDA: a fatal erosion of integrity. Lancet Antiemetic Consensus Conference. Ann Oncol 2006; 17: 20–28. 2001; 357: 1544–1545. 2 Aapro M. 5-HT(3)-receptor antagonists in the management of nausea 24 Candiotti KA, Birnbach DJ, Lubarsky DA, Nhuch F, Kamat A, Koch WH and vomiting in cancer and cancer treatment. Oncology 2005; 69: 97–109. et al. The impact of pharmacogenomics on postoperative nausea and 3 Hesketh PJ. Chemotherapy-induced nausea and vomiting. N Engl J Med vomiting: do CYP2D6 allele copy number and polymorphisms affect the 2008; 358: 2482–2494. success or failure of ondansetron prophylaxis? Anesthesiology 2005; 4 Kaiser R, Sezer O, Papies A, Bauer S, Schelenz C, Tremblay PB et al. 102: 543–549. Patient-tailored antiemetic treatment with 5-hydroxytryptamine type 3 25 Janicki PK, Schuler HG, Jarzembowski TM, Rossi M 2nd. Prevention of receptor antagonists according to cytochrome P-450 2D6 genotypes. postoperative nausea and vomiting with granisetron and dolasetron in J Clin Oncol 2002; 20: 2805–2811. relation to CYP 2D6 genotype. Anesth Analg 2006; 102: 1127–1133. 5 HilgendorfC,AhlinG,SeithelA,ArturssonP,UngellAL,KarlssonJ.Exp- 26 Yonezawa A, Masuda S, Yokoo S, Katsura T, Inui K. Cisplatin and ression of thirty-six drug transporter genes in human intestine, liver, kidney, oxaliplatin, but not carboplatin and nedaplatin, are substrates for and organotypic cell lines. Drug Metab Dispos 2007; 35: 1333–1340. human organic cation transporters (SLC22A1-3 and multidrug and 6 Nies AT, Herrmann E, Brom M, Keppler D. Vectorial transport of the toxin extrusion family). J Pharmacol Exp Ther 2006; 319: 879–886. plant alkaloid berberine by double-transfected cells expressing the 27 Zhang S, Lovejoy KS, Shima JE, Lagpacan LL, Shu Y, Lapuk A et al. human organic cation transporter 1 (OCT1, SLC22A1) and the efflux Organic cation transporters are determinants of oxaliplatin cytotoxicity. pump MDR1 P-glycoprotein (ABCB1). Naunyn Schmiedebergs Arch Cancer Res 2006; 66: 8847–8857. Pharmacol 2008; 376: 449–461. 28 Chen JJ, Li Z, Pan H, Murphy DL, Tamir H, Koepsell H et al. Maintenance 7 Tzvetkov MV, Vormfelde SV, Balen D, Meineke I, Schmidt T, Sehrt D of serotonin in the intestinal mucosa and ganglia of mice that lack the et al. The effects of genetic polymorphisms in the organic cation high-affinity serotonin transporter: abnormal intestinal motility and the transporters OCT1, OCT2, and OCT3 on the renal clearance of expression of cation transporters. J Neurosci 2001; 21: 6348–6361. metformin. Clin Pharmacol Ther 2009; 86: 299–306. 29 Mashru RC, Sutariya VB, Sankalia MG, Sankalia JM. Effect of pH on in 8 Kerb R, Brinkmann U, Chatskaia N, Gorbunov D, Gorboulev V, vitro permeation of ondansetron hydrochloride across porcine buccal Mornhinweg E et al. Identification of genetic variations of the human mucosa. Pharm Dev Technol 2005; 10: 241–247.

The Pharmacogenomics Journal