- © 2002 by American Society of Clinical Oncology
Patient-Tailored Antiemetic Treatment With 5-Hydroxytryptamine Type 3 Receptor Antagonists According to Cytochrome P-450 2D6 Genotypes
- Rolf Kaiser,
- Orhan Sezer,
- Anja Papies,
- Steffen Bauer,
- Claudia Schelenz,
- Pierre-Benoit Tremblay,
- Kurt Possinger,
- Ivar Roots and
- Jürgen Brockmöller
- From the Institute of Clinical Pharmacology and Department of Hematology and Oncology, University Medical Center Charité, Humboldt University of Berlin, Berlin, and Department of Clinical Pharmacology, University Medical Center, Georg-August-University Göttingen, Göttingen, Germany.
- Address reprint requests to Rolf Kaiser, MD, Abteilung für Klinische Pharmakologie, Universitätsklinikum der Georg-August-Universität Göttingen, Robert Koch Str 40, D-37075 Göttingen, Germany: email: rolf.kaiser{at}med.uni-goettingen.de
Abstract
PURPOSE: The use of serotonin 5-hydroxytryptamine type 3 receptor antagonists has substantially reduced, but not eliminated, nausea and vomiting in cancer chemotherapy. This study sought to investigate whether efficacy of antiemetic treatment with ondansetron and tropisetron depends on cytochrome P-450 2D6 (CYP2D6) genotype, hypothesizing that the rapid and particularly the ultrarapid metabolizers of these drugs are at risk of being undertreated.
PATIENTS AND METHODS: Included in the study were 270 cancer patients receiving their first day of chemotherapy. Nausea and vomiting were documented using standardized interviews. The intensity of nausea was measured with visual analog scales before and twice during the chemotherapy. The relationship between the CYP2D6 genotypes and the tropisetron serum concentrations 3 and 6 hours after drug administration was analyzed in a subgroup of 42 patients. CYP2D6 genotyping was carried out by polymerase chain reaction–restriction fragment length polymorphism analysis.
RESULTS: Genetically defined poor metabolizers had higher serum concentrations of tropisetron than all other patients (P < .03). Approximately 30% of all patients receiving chemotherapy experienced nausea and vomiting. Genetically defined ultrarapid meta-bolizers of CYP2D6 substrates had higher frequency of vomiting within the first 4 hours (P < .001) and within the period 5 to 24 hours (P < .03) after treatment than all the other patients; the tendency for nausea was similar. This difference was more pronounced in patients treated with tropisetron than in those treated with ondansetron.
CONCLUSION: Antiemetic treatment with tropisetron or ondansetron could be improved by adjustment for the CYP2D6 genotype; approximately 50 subjects would have to be genotyped to protect one patient from severe emesis.
NAUSEA AND VOMITING are still among the feared side effects of cancer chemotherapy, and the incidence of these adverse effects may influence the success of individual cancer therapy.1 Three different forms of vomiting or nausea induced by cancer chemotherapy can be distinguished: acute emesis within the first 24 hours; delayed emesis after the first 24 hours, up to 6 days later; and anticipatory emesis.2 Various, still not finally elucidated mechanisms contribute to these effects. Cisplatin, the archetypical emetogenic drug, increases the release of serotonin from enterochromaffin cells of the gut and, in parallel, increases the number of episodes of emesis.3 Correspondingly, less emesis and lower serotonin concentrations have been observed when patients are pretreated with an inhibitor of serotonin synthesis.4 Moreover, metoclopramide has serotonin antagonist properties at higher therapeutic doses, and the treatment with serotonin or 5-hydroxytryptamine receptor type 3 (5-HT3) antagonist reduces emesis significantly.5,6
Therefore, acute emesis seems to be provoked by a peripheral or central serotonin release with consecutive activation of 5-HT3 receptors on peripheral vagal fibers and central regions such as the area postrema and nucleus tractus solitarii.7-9 The correlation between serotonin release and nausea and vomiting has also been observed during treatment with nitrogen mustard, dacarbazine, and radiation.3,10,11 On the other hand, some anticancer drugs, such as cyclophosphamide, may not increase the serotonin release in humans (as opposed to animals), but the induced emesis is still sensitive to 5-HT3 receptor antagonists.10,12 Nevertheless, the prophylactic administration of 5-HT3 receptor antagonists such as ondansetron, tropisetron, granisetron, or dolasetron plays a major role in current antiemetic treatment13 and results in a significant improvement of quality of life in cancer patients.5,14
Female sex, younger age, no alcohol consumption, preexisting nausea, and the emetogenic level of the chemotherapeutic agents have been correlated with the individual risk of acute vomiting; however, approximately 20% to 30% of the patients do not respond satisfactorily to 5-HT3 receptor antagonists.9 One cause for such individual differences in drug response may be variation in drug biotransformation by genetically polymorphic enzymes, such as the hepatic cytochrome P-450 enzyme 2D6 (CYP2D6). All 5-HT3 receptor antagonists are metabolized by the cytochrome P-450 enzymes: tropisetron and dolasetron predominantly by CYP2D6, ondansetron partially by CYP2D6 but also by CYP3A4, CYP2E1, or CYP1A2, and granisetron mainly by CYP3A4.15-20 For the genetically polymorphic enzyme CYP2D6, several alleles have been detected that result in defective, qualitatively altered, diminished, or enhanced activity.21 Approximately 5% to 10% of whites, the so-called poor metabolizers (PM) of the model substrates debrisoquine and sparteine, completely lack CYP2D6 activity, and approximately 2% of whites are categorized as ultrarapid metabolizers (UM), with more than two active genes as a result of a duplication or even a several-fold amplification of the CYP2D6 gene.22 The proportion of PM and UM varies between different populations.23
We hypothesized that PM would have the highest concentration of 5-HT3 antagonists in blood and, consequently, the best protection from nausea and vomiting, whereas UM would have the worst protection from nausea and vomiting when given the standard dose. Therefore, we correlated the different CYP2D6 genotypes with the intensity of acute nausea and vomiting that occurred in patients treated with ondansetron or tropisetron within the first 24 hours after chemotherapy. As an additional confirmation, in a subgroup of patients, the association between the blood concentrations of tropisetron and the respective genotypes was investigated. If an association between CYP2D6 genotype and nausea and vomiting exists, then performing CYP2D6 genotyping before providing the patient with chemotherapy may permit the clinician to adjust the antiemetic therapy as needed to reduce these adverse effects.
PATIENTS AND METHODS
Patients
We conducted a prospective noninterventional cohort study to analyze the impact of functional polymorphisms of CYP2D6 on the antiemetic efficacy of tropisetron or ondansetron in cancer patients. From April 1998 to September 2000, consecutive adult patients scheduled to receive moderately to highly emetogenic chemotherapy either for the first time or as the first course of a chemotherapy regimen after relapse were enrolled onto the study. We included 270 patients (116 men, 154 women; 157 outpatients, 113 inpatients) at the Universitätsklinikum Charité and the community hospital Krankenhaus Moabit, Berlin, Germany. Mean age of the patients was 53.7 years (range, 18 to 83 years; SD, 13.3). From these patients, 32.5% had from breast cancer, 15.4% lung cancer, 14.2% non-Hodgkin’s lymphoma, 4.9% multiple myeloma, 4.9% Hodgkin’s disease, and 28.1% miscellaneous tumors.
Patients who met one of the following criteria were excluded from participation: presence of nausea or vomiting before the chemotherapy; the use of antiemetics, benzodiazepines, neuroleptics, or radiation therapy in the 24 hours before administration of the chemotherapy; use of opioids within the last 2 weeks, and regular use of inducers of CYP2D6 (eg, rifampicin) or inhibitors of CYP2D6 (eg, quinidine, fluoxetine, haloperidol). We also excluded all patients with presence of concomitant diseases that might cause nausea or vomiting (eg, severe heart failure, ulcerations or obstructions of the upper gastrointestinal system, severe hepatic or renal dysfunction, brain metastases, patients with artificial stoma, pregnancy). From 286 patients primarily enrolled onto the study, 16 patients later had to be excluded for predefined reasons—for example, because of administration of antiemetics other than ondansetron or tropisetron or missing antiemetic drug treatment at day 1 of chemotherapy, or because patients did not complete all the questionnaires. Seven patients delivered incomplete data.
Emetogenic level at the day of the administered anticancer drugs was calculated according to the emetogenic classification scheme of Hesketh et al,24,25 and patients were grouped in five different emetogenic levels (level 1, n = 2; level 2, n = 55; level 3, n = 22; level 4, n = 95; level 5, n = 96). Cyclophosphamide was administered to 98 patients (mean dosage, 1,524 mg), either alone or in combination with various other cytostatic drugs. Cisplatin (mean dosage, 90 mg) and carboplatin (mean dosage, 448 mg) were given to 27 patients and 29 patients, respectively. All other patients (n = 116) received miscellaneous chemotherapeutic drugs. Glucocorticoids were administered to 151 patients, either as a part of the antineoplastic therapy or as additional antiemetic treatment.
Tropisetron (Navoban; Novartis Pharma, Basel, Switzerland) was given in a dosage of 5 mg once daily (n = 96), and ondansetron (Zofran; GlaxoSmithKline, Brentford, United Kingdom) was administered at a dosage of 8 mg twice daily (n = 174). Measurement of nausea and vomiting were performed immediately before the chemotherapy started, 4 hours after administration of chemotherapy (observation period 1), and then within the next 20 hours (5th to the 24th hour, observation period 2) at day 1 of the chemotherapy.
The timing within the first 24 hours and number of retching and vomiting episodes were recorded by the patients on diary cards. The intensity of nausea was assessed with the help of visual analog scales (which ranged from no nausea at 0 mm to most extensive nausea at 100 mm). An emetic episode was defined as a single vomit or retch, or any number of continuous vomits or retches. Vomiting or retching had to be absent for at least 1 minute to calculate different episodes of emesis according to the definition of the Italian Group for Antiemetic Research.26 Protection from nausea was regarded as incomplete when emetic episodes occurred or when nausea intensity was 20% above the baseline level. The study was approved by the ethics committee of the Universitätsklinikum Charité (Humboldt-Universität zu Berlin), and all patients gave written informed consent.
CYP2D6 Genotyping
High-molecular-weight genomic DNA was prepared from venous blood by use of the standard phenol chloroform extraction. All laboratory staff were blinded to the clinical observations. CYP2D6 genotyping was carried out according to Sachse et al.21 Polymerase chain reaction products were separated by agarose gel electrophoresis and stained with ethidium bromide for visualization.
Alleles *3, *4, *5, and *6 of CYP2D6 were considered to predict the deficient (PM) phenotype, whereas the allele *1 and the duplication (2x*1 or 2x*2) of the gene are coding for the active enzyme. By definition, PM are carriers of two of the alleles *3, *4, *5, and *6 of CYP2D6. Intermediate metabolizers (IM) have one active allele *1 (wild type). Extensive metabolizers (EM) have two active alleles *1 or one defective allele and one duplication allele. UM have one active allele *1 and one duplication allele or even two duplication alleles. Therefore, with respect to the genotype, we grouped the patients into four subgroups: PM (those with no active genes or one, two, or three active genes), IM, EM, and UM.
Tropisetron Serum Concentration Analysis
Three and 6 hours after the administration of the 5-HT3 antagonist tropisetron, blood samples were drawn from the arm that had not been used for drug administration. Tropisetron hydrochloride was provided by Novartis Pharma (Basel, Switzerland). Tropisetron was extracted with dichloromethane under alkaline conditions, separated at room temperature on a Phenomenex Luna C18 HPLC-column (5 μm, 250 × 4.6 mm inner diameter; Phenomenex, Aschaffenburg, Germany) and quantified by ultraviolet detection at 284 nm. The mobile phase consisted of 20% of acetonitrile and 80% 0.05 M sodium hydrogen phosphate buffer, pH 5.0; the flow rate was 1.5 mL/min. Intra- and interassay coefficients of variation ranged from 1.5% to 7.5% and from 5.3% to 13.7%, respectively. The lower limit of quantification was 1.25 ng tropisetron/mL.
Statistical Methods
The mean number of vomiting episodes and the mean degree of nausea were compared with the Kruskal-Wallis test or with the Mann-Whitney U test. The pairwise comparison between groups was performed with the Wilcoxon rank sum test (SPSS version 8.01; SPSS, Inc, Chicago, IL). The limit of significance was set to .05.
RESULTS
Within the first 24 hours after administration of the chemotherapy, vomiting was observed in 58 (22.1%) of 270 patients and nausea in 94 (35.9%) of 270 patients. The mean number of vomiting episodes of all 270 patients was one (range, zero to 22 episodes), and the mean percentage of the visual analog scale for nausea was 15.6% (range, 0.0% to 98%). Figure 1 illustrates the data on acute nausea and vomiting stratified for the different emetogenic levels of chemotherapy as classified according to Hesketh et al.24,25 The percentage of patients with incomplete protection from nausea or vomiting seemed to be independent of the respective emetogenic level of the chemotherapy: vomiting (one or more episodes) was observed in 19.1% of the patients treated with anticancer drugs of the high-emetogenic level 5 and in 18.9% of the patients treated with the low emetogenic level 2.24,25 Nausea occurred in 40.4% of the patients treated with high-emetogenic drugs (level 4) and in 37.7% of the patients receiving a low-emetogenic therapy (level 2). The results were similar in inpatients and outpatients.
Fig 1. Percentage of patients with vomiting and nausea as function of the emetogenic level of the applied chemotherapy in the whole study group.
Patients treated with glucocorticoids, either as part of their therapeutic regimen or as additional antiemetic therapy, received better protection from nausea than patients without glucocorticoids (73.6% v 51.8%, P < .001, χ2 test). A similar trend was observed for vomiting and for the combined event of nausea and vomiting (detailed data not shown). When the patients were stratified by emetogenic levels (Table 1), patients treated with high-emetogenic level 4 chemotherapy without glucocorticoids experienced a two-fold higher intensity of nausea in observation period 1 (mean, 12.8% v 6.8%, P < .02, Mann-Whitney U test) and in observation period 2 (mean, 23.1%, v 11.9%, P = .01, Mann-Whitney U test) than patients who received glucocorticoid treatment. A similar trend was observed for vomiting.
Vomiting and Nausea Among Patients Treated With or Without Glucocorticoids
As illustrated in Fig 2, patients deficient for CYP2D6 activity had significantly higher serum concentrations of tropisetron 6 hours after administration, compared with patients with one or more active alleles (median concentration, 15.3 ng/mL v 4.9 ng/mL, P < .03, Mann-Whitney U test). The result was similar 3 hours after the administration, although statistically not significant (median concentration, 13.5 ng/mL v 8.0 ng/mL).
Fig 2. Box plots of serum concentrations (ng/mL) of tropisetron as function of the number of active CYP2D6 genes (3 and 6 hours after administration). The difference between concentration of poor metabolizers and all others was statistically significant (P < .03, Mann-Whitney U test) in the 6-hour measurement.
Genotyping for CYP2D6 revealed that 7.8% of the 270 patients were deficient for the CYP2D6 gene (PM), 32.6% had one active allele, 58.1% had two active alleles (EM), and 1.5% had three active genes (UM). As illustrated in Fig 3, patients with three active CYP2D6 genes (UM) had a significantly higher mean number of vomiting episodes than all other patients in observation period 1 (mean ± SD episodes of vomiting, 2.3 ± 2.5 v 0.2 ± 1.0, P < .001, Mann-Whitney U test) and in observation period 2 (mean ± SD episodes of vomiting, 3.3 ± 3.5 v 0.8 ± 2.4, P < .03). A similar but statistically not significant trend was observed for nausea: UM had more severe nausea in the two study periods (mean ± SD, 22.3% ± 25.9% v 9.6% ± 16.4% and 46.8% ± 44.9% v 15.1% ± 22.2%, respectively) than all other patients. These results were similar for vomiting and nausea in the group of patients receiving glucocorticoids at both observation periods and in the group of patients who did not receive glucocorticoids (detailed data not shown).
Fig 3. Mean values (± SD) of vomiting as function of number of active CYP2D6 genes. Patients with three active genes had significantly more vomiting at both observation periods than all other patients (P < .001, P < .03, Mann-Whitney U test). A similar trend was observed for nausea. VAS, visual analog scale.
The effects of the CYP2D6 polymorphisms seen in the whole group of patients were similar with tropisetron treatment and after treatment with ondansetron (Fig 4): EM had the highest intensity of vomiting or nausea in both groups and observation periods, and PM patients demonstrated the lowest intensity of vomiting and nausea during the first observation period. None of the PM patients in the tropisetron group demonstrated vomiting (detailed data not shown).
Fig 4. Intensity of vomiting or nausea as a function of the CYP2D6 genotype for patients treated with tropisetron or ondansetron between the 5 to 24 hours after administration of the chemotherapy (mean ± SD).
DISCUSSION
Serotonin antagonists have considerably improved the antiemetic treatment in cancer patients. Nevertheless, approximately 20% to 30% of patients treated with highly emetogenic chemotherapies still suffer from vomiting or nausea within the first day after the administration of chemotherapy.9 The emetogenic level of the therapeutic regimen was determined according to the proposal of Hesketh et al,24,25 whose classification system is based on literature and clinical experience. Paradoxically, in the present study, the incidence of vomiting or nausea was independent from the respective emetogenic level of the chemotherapy (Fig 1). Patients receiving chemotherapy protocols with high emetogenic levels (levels 4 and 5) were more often treated with glucocorticoids and therefore had a better antiemetic protection. This may be the reason why a correlation between the emetogenic level of the administered chemotherapy and the number of nausea and vomiting episodes experienced by the patients is lacking in our study. Moreover, patients treated with glucocorticoids demonstrated an improved antiemetic protection principally and independent of the emetogenic level of the chemotherapy. For intermediate emetic risk, published guidelines recommend pretreatment with glucocorticoids, and for higher emetogenic risk, combined therapy with a 5-HT3 receptor antagonist is recommended.13,26-30
Our results clearly demonstrate that antiemetic treatment could be improved considerably by the identification of non-, low, and high responders on a pharmacogenetic basis, thus allowing the clinician to predict who will require a specific and individual antiemetic dose adaption. In the present study, UM for CYP2D6 had the highest score of vomiting and nausea, whereas PM demonstrated the lowest score and significantly higher tropisetron blood concentrations.
These effects are not only specific to tropisetron but were also observed with ondansetron, although to a lesser extent, because ondansetron is partly metabolized also by other cytochrome P-450 enzymes, as has been demonstrated in vitro (eg, CYP3A4).15,18 This may account for the high rate of vomiting and nausea that occurred in the PM group in observation period 2. Nevertheless, in carriers of the CYP2D6 gene duplication, a large fraction of ondansetron is probably rapidly eliminated via CYP2D6.
Comedication may also have an influence on the genotype-phenotype correlation: an EM of substrates of CYP2D6 could be converted through inhibition to a PM; however, we excluded such cases of enzyme inhibition from the analysis. With respect to enzyme induction, there is a large amount of data demonstrating that CYP2D6 in the EM subgroup is not inducible to a relevant extent.31 For the PM subgroup, because of the function-disrupting nature of the genetic variations, a PM can never be converted to an EM. With respect to ondansetron, which is partially metabolized by CYP3A4, enzyme induction should be considered, but patients receiving any strong CYP3A4 inducer were excluded.32
As expected, we saw greater differences in the median of tropisetron concentration at 6 hours compared with 3 hours because the 3-hour concentrations should depend more on the volume of distribution, whereas at 6 hours, the individual elimination capacity should be the major factor. But UM demonstrated significant higher mean number of vomiting episodes in the first and the second observation period, although the differences of the plasma concentrations in the first observation period were not statistically significant. One reason for this might be that these patients may have reached lower nontherapeutic concentrations faster than all other patients. However, this discrepancy cannot be finally clarified because data for concentration analyses were only from a small subset of patients (n = 42), and only two sampling times for pharmacokinetic were chosen.
A disputed bell-shaped dose-response curve of the 5-HT3 antagonists would not allow the effect of UM to be counteracted simply by giving all patients higher doses of 5-HT3 antagonists, especially when treated with ondansetron.9,33 However, with current doses, our data provided no indication that those with highest blood concentrations were overtreated. The effect of this study is due to a few subjects who were UM and who all had a high extent of vomiting or nausea. Because of the low frequency (1.5% to 2%) of genetically defined UM in the German population, it would be necessary to genotype approximately 50 patients for CYP2D6 to prevent one case of severe vomiting or nausea. Analogous to the evaluation of drugs, where effectiveness is defined by the number needed to treat, effectiveness of genotyping might be defined by the number needed to genotype. Because of the low frequency of the UM among whites, this number would be relatively high. However, the frequency of UM is different in other ethnic groups: in Northern European countries, 2% to 4%; in the Mediterranean area, 7% to 12%; in Ethiopia, 29%; and in Saudi Arabia, 21%.23,34 Therefore, the higher proportion of UM in these regions may influence the efficacy of antiemetic treatment in cancer patients to a higher extent than in the Northern European countries. If UM were identified before chemotherapy starts, antiemetic treatment could be adapted to their specific needs, and severe nausea or vomiting episodes could be prevented—for example, by the use of 5-HT3 antagonists, which are not substrates of CYP2D6.
In conclusion, UM of CYP2D6 demonstrate the highest incidence and severity of emesis and nausea after cancer chemotherapy when ondansetron or tropisetron was given as antiemetic treatment. In these patients, a different antiemetic approach or significantly higher doses are required. Genotyping for CYP2D6 before the start of the chemotherapy or the use of antiemetic drugs that are not metabolized by CYP2D6 may further reduce nausea and vomiting. But cancer or noncancer patients with opioid-induced emesis or patients with postoperative nausea and vomiting could also benefit from this approach. Further studies including only patients receiving exactly one chemotherapeutic scheme and one antiemetic treatment should be performed to further confirm our results. In the future, cancer chemotherapy gene chips will test for a number of genes, thus allowing the clinician to predict the patient’s drug response and risk for adverse events. It is likely that CYP2D6 is among these genes.
Acknowledgments
Supported in part by grant nos. 01EC9408 and 01ZZ9511 from the German Ministry for Education and Research.
ACKNOWLEDGMENT
We thank Drs M. Schweigert, E. Späth-Schwalbe, D. Lüftner, and Th. Beinert and N. Rösler, A. Flögel, E. Arnold, Dr M. Kaiser, and Th. Fiedler for supporting the study from the clinical side and collecting the patients. We thank Dr G. Laschinski for careful revision of the manuscript. We thank Prof Dr H. Hellriegel for supporting this study in his department.
- Received September 14, 2001.
- Accepted March 18, 2002.
