The impact of the CYP2D6 polymorphism
`on haloperidol pharmacokinetics and on the
`outcome of haloperidol treatment
`
`Objectives: The genetically polymorphic enzyme cytochrome P450 (CYP) 2D6 contributes to the biotrans-
`formation of the antipsychotic drug haloperidol. The impact of the polymorphism on haloperidol pharma-
`cokinetics, adverse events, and efficacy was prospectively evaluated under naturalistic conditions in 172
`unselected psychiatric inpatients with acute psychotic symptoms.
`Methods: Serum trough levels of haloperidol and reduced haloperidol of patients receiving clinically adjusted
`doses were analyzed on days 3, 14, and 28 after hospital admission. Adverse events such as extrapyramidal
`symptoms were assessed by standardized rating scales. Efficacy was documented by recording the change in
`positive and negative schizophrenic symptoms. These parameters were correlated with the CYP2D6 genotype
`determined by polymerase chain reaction analysis for alleles *1 to *15 and *17.
`Results: The serum concentrations showed wide interindividual variation. Reduced haloperidol trough levels
`and haloperidol total clearance correlated significantly with the number of active CYP2D6 genes. In addition,
`body weight and smoking had significant effects on haloperidol kinetics, whereas age, gender, and comedi-
`cation showed only slight effects. The ratings for pseudoparkinsonism were significantly higher in poor
`metabolizers of substrates of CYP2D6. On the other hand, there was a trend toward lower therapeutic
`efficacy with increasing number of active CYP2D6 genes.
`Conclusions: Treatment with haloperidol should be avoided in extremely slow and extremely rapid metabo-
`lizers of CYP2D6 substrates. Both genotyping and blood concentration measurement explained only a
`fraction of the adverse events; about 20 patients would have to be genotyped to achieve a significant benefit
`in 1 patient. It is interesting that genotyping was at least as good a predictor of adverse events as the measured
`drug concentrations. (Clin Pharmacol Ther 2002;72:438-52.)
`
`Ju¨rgen Brockmo¨ller, MD, Julia Kirchheiner, MD, Ju¨rgen Schmider, MD,
`Silke Walter, PhD, Christoph Sachse, PhD, Bruno Mu¨ller-Oerlinghausen, MD, and
`Ivar Roots, MD Berlin, Goettingen, and Dresden, Germany, and Groton, Conn
`
`The dopamine D2-receptor antagonist haloperidol is
`prescribed as a high-potency antipsychotic drug for the
`
`From the Institute of Clinical Pharmacology, University Medical
`Center Charite´, Humboldt University, Berlin; Department of Clin-
`ical Pharmacology, University Medical Center, Goettingen; Pfizer
`Inc, Groton; Cenix BioScience GmbH, Dresden; and Research
`Group of Clinical Psychopharmacology, University Medical Cen-
`ter Benjamin Franklin, Freie Universita¨t, Berlin.
`Supported by Humboldt University, Berlin, and German Ministry for
`Education and Research.
`Received for publication Jan 2, 2002; accepted June 12, 2002.
`Reprint requests: Ju¨rgen Brockmo¨ller, MD, Abteilung Klinische
`Pharmakologie, Universita¨tsklinikum, Robert-Koch-Strasse 40,
`37075 Goettingen, Germany.
`E-mail: jurgen.brockmoller@med.uni-goettingen.de
`Copyright © 2002 by the American Society for Clinical Pharmacol-
`ogy & Therapeutics.
`0009-9236/2002/$35.00 ⫹ 0 13/1/127494
`doi:10.1067/mcp.2002.127494
`
`438
`
`treatment of acute and chronic schizophrenia and other
`psychiatric disorders worldwide. Its high efficacy is
`compromised by serious extrapyramidal adverse reac-
`tions (acute dystonia, pseudoparkinsonism, akathisia,
`and tardive dyskinesia), which occur with a high
`incidence.1-3 The therapeutic serum concentration of
`haloperidol ranges from 5 to 17 ␮g/L according to
`several mostly small clinical trials,4-14 but the lower
`limit, in particular is not well defined.
`A linear correlation was observed between haloper-
`idol dose and its serum concentration.15 It was shown
`that dosages of haloperidol higher than 10 mg/d are
`usually not of additional benefit in the treatment of
`acute schizophrenia.16 However, interindividual varia-
`tions in haloperidol pharmacokinetics are consider-
`able.17 Haloperidol is extensively metabolized in the
`liver, and only 1% of the administered dose is excreted
`
`Vanda Exhibit 2026 - Page 1
`
`This article is protected by copyright and is provided by the University of Wisconsin-
`Madison under license from John Wiley & Sons. All rights reserved.
`
`

`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 72, NUMBER 4
`
`Brockmo¨ller et al 439
`
`unchanged in the urine.18 Hepatic biotransformation
`includes glucuronidation (about 50%-60% in vivo) and
`reduction and back-oxidation (23%), as well as
`N-dealkylation and pyridinium metabolite formation
`(about 20%-30%).19-22 The glucuronide metabolite is
`inactive, whereas reduced haloperidol has significant
`pharmacologic effects (eg, high-affinity binding to
`sigma-type opioid receptor– binding sites and to dopa-
`mine D2 and D3 receptors).23 The concentrations of
`haloperidol and reduced haloperidol appeared to corre-
`late with clinical response and adverse effects.24-27
`The formation of reduced haloperidol is mediated by
`a carbonyl reductase.28 Reduced haloperidol is partially
`back-oxidized to haloperidol. In vitro, this reaction is
`mediated by cytochrome P450 (CYP) 3A429-31; only
`one group of authors has reported a contribution of
`CYP2D6.32 The oxidation of haloperidol to a pyri-
`dinium metabolite and oxidative N-dealkylation also
`appeared to be mediated mainly by CYP3A4,31,33-35 but
`CYP2D6 appeared to be involved as well.31
`Several studies have investigated the contribution of
`CYP2D6 to haloperidol metabolism in healthy volun-
`teers or patients. Although in vitro data suggest that
`CYP2D6 might play only a minor role in haloperidol
`biotransformation in human beings, all studies except
`one (which included only one poor metabolizer of
`CYP2D6)36 have found an effect of CYP2D6 activity
`on haloperidol pharmacokinetics. Poor metabolizers ac-
`cording to CYP2D6 showed higher haloperidol serum
`concentrations, lower haloperidol clearance, and higher
`concentrations of reduced haloperidol than extensive
`metabolizers.37-42 These in vivo studies were either
`clinical trials that were conducted under standardized
`conditions in selected groups of inpatients36,38,40,43 or
`single-dose studies with healthy volunteers.37,42 One
`observational study examined the influence of CYP2D6
`genotype, smoking, and concomitant drug use in a
`naturalistic clinical setting,39 in which the total sample
`size was 92 participants. The reason for the discrepancy
`between the in vitro and in vivo data with regard to the
`role of the polymorphic CYP2D6 in haloperidol phar-
`macokinetics is not yet clear.
`Higher reduced haloperidol and haloperidol levels in
`blood are associated with a higher risk of adverse drug
`reactions. Reduced haloperidol levels were even more
`strongly correlated with extrapyramidal side effects and
`poorer clinical outcome than those of haloperidol.25,27
`In a retrospective analysis, the poor metabolizer phe-
`notype of CYP2D6 was found to be significantly over-
`represented in patients with adverse effects.44 This
`finding is corroborated by several reports on the higher
`incidence and severity of adverse drug effects in poor
`
`metabolizers of CYP2D6.44-46 We conducted a pro-
`spective naturalistic study,
`in which the impact of
`CYP2D6 on haloperidol pharmacokinetics, adverse
`events, and efficacy was tested in an unselected sample
`of 172 consecutively included inpatients. The aim was
`to determine the extent to which the CYP2D6 genotype
`influences the risk of adverse drug reactions, therapeu-
`tic outcome, and interindividual pharmacokinetic vari-
`ability under naturalistic clinical conditions. Further-
`more, these results might serve as an empiric basis for
`future genotype-based optimization of antipsychotic
`drug treatment.
`
`METHODS
`Study design
`This was a prospective, observational, multicenter
`study conducted at 5 psychiatric departments in Berlin,
`Germany. Patients were included within 3 days after
`hospital admission.
`
`Patients
`The study was approved by the ethics committees at
`the University Medical Center Benjamin Franklin and
`the University Medical Center Charite´, Humboldt Uni-
`versity, Berlin, Germany. One hundred seventy-five
`German psychiatric inpatients (34% women and 66%
`men), treated with haloperidol because of psychotic
`symptoms, were included. In accordance with the study
`protocol, patients were informed about the study and
`the genotyping for polymorphic drug-metabolizing en-
`zymes, and consent was obtained from each patient.
`The mean age was 39 years. Most patients were diag-
`nosed as having an acute exacerbation of schizophrenia
`(n ⫽ 114) or schizoaffective psychosis (n ⫽ 40) ac-
`cording to the criteria of the International Statistical
`Classification of Diseases, 10th Revision. Brief psy-
`chotic disorder occurred in 10 patients, mania requiring
`antipsychotic medication in 7 patients, substance-
`related psychotic disorders in 2 patients, and borderline
`personality disorder, as well as delusional disorder, in 1
`patient. Patients with organic mental disorders, as di-
`agnosed by means of clinical history, computed tomog-
`raphy, or nuclear magnetic resonance imaging, were
`excluded.
`All medication was documented from hospital ad-
`mission until discharge or, in cases of prolonged hos-
`pitalization, for at least 28 days after admission. On
`average, patients took 4 medications during the study
`period. During the 3-day period before sample taking,
`32% of the haloperidol-treated patients received other
`antipsychotic comedication simultaneously. The anti-
`psychotic agents most frequently administered in addi-
`
`Vanda Exhibit 2026 - Page 2
`
`

`
`440 Brockmo¨ller et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`OCTOBER 2002
`
`tion to haloperidol were chlorprothixene (9%), perazine
`(6%), and promethazine (5%), all of which have low
`potency. In addition, 70% of the patients received ben-
`zodiazepines, 58% anticholinergics, and 34% other
`types of hypnotic drugs. Comedications were classified
`as CYP-interacting drugs and noninteracting drugs.
`CYP-interacting drugs for haloperidol were substances
`inhibiting CYP2D6.47 Because CYP3A4 plays a role in
`haloperidol metabolism, we also evaluated the effect of
`CYP3A4 inhibitory drugs (eg, ranitidine and levonorg-
`estrel) and CYP3A4 inducers (eg, carbamazepine).
`The course of illness was documented by the global
`clinical assessment scale and by the positive and neg-
`ative symptoms scale (PANSS)48 on days 3, 14, and 28.
`The instruments for the assessment of adverse effects
`included the extrapyramidal symptom (EPS) scale,49
`the abnormal involuntary movement scale,50 the Barnes
`akathisia rating scale,51 a questionnaire asking for all of
`the typical adverse events of antipsychotic agents, and
`the Hamilton depression rating scale for the assessment
`of depressive symptoms (which gives only the undif-
`ferentiated sum of depressive symptoms occurring as
`side effects of haloperidol
`treatment
`in addition to
`possible depressive symptoms resulting from the psy-
`chotic disease plus depressive symptoms resulting from
`possibly existing depressive comorbidity). For geno-
`typing, 9 mL of blood was taken from each patient, and
`haloperidol serum concentrations were determined at
`days 3, 14, and 28 or close to these dates. Psychiatric
`rating scales were assessed by psychiatrists who had no
`knowledge about the genotype data and serum concen-
`tration data when they performed the clinical examina-
`tion and data documentation (blinded assessment). To
`improve interrater comparability, regular monitoring
`sessions in 6-week intervals took place in which all
`clinical investigators of the study monitored 1 patient.
`These sessions were followed by discussions of the
`possible discrepancies in the scores obtained by differ-
`ent monitors. Furthermore, the persons who performed
`the genotyping and serum concentration analyses were
`blinded to the clinical data, and clinical monitors, at the
`time of their documentation, had no knowledge about
`the plasma concentrations or genotypes.
`
`HPLC quantification of haloperidol and reduced
`haloperidol in serum
`Standards of haloperidol and reduced haloperidol
`were obtained from Biotrend (Cologne, Germany). Se-
`rum was separated from whole blood and stored at
`⫺20°C. Six calibration standards with concentrations
`from 0 to 20␮g/L and 2 controls with 1.25 ␮g/L and
`7.5 ␮g/L were used. Haloperidol and reduced haloper-
`
`idol were quantified by HPLC with electrochemical
`detection as described by Walter et al.52 The lower limit
`of quantification was 0.31 ␮g/L for haloperidol and
`reduced haloperidol. The limit of detection was 40 pg
`for haloperidol and 50 pg for reduced haloperidol. The
`interassay coefficient of variation was 12.8% at 1.25
`␮g/L and 5.6% at 7.5 ␮g/L. Each patient sample was
`measured twice, and the mean was used for further
`calculations.
`
`CYP2D6 genotyping
`Leukocytes were isolated after erythrocyte lysis, and
`deoxyribonucleic acid (DNA) was extracted in phenol-
`chloroform.53 DNA samples were processed by poly-
`merase chain reaction for detection of the CYP2D6
`alleles *1 to *15 and *17, as well as for gene duplica-
`tion of alleles *1 and *2. The methods of CYP2D6
`genotyping have been described previously.54 As
`shown in Table I, the CYP2D6 genotypes were classi-
`fied into groups with no active gene (poor metaboliz-
`ers), 1 active gene (intermediate metabolizers), 2 active
`genes (extensive metabolizers), or more than 2 active
`genes (ultrafast metabolizers), that is, individuals with
`gene duplications of allele *1 or *2. Genotyping was
`performed in 172 patients; the blood samples of 3
`subjects were lost before DNA extraction. After we had
`noted a surprisingly low frequency of genotypically
`identified poor metabolizers, 60% of all genotyping
`analyses were performed completely in duplicate by
`different investigators in a blinded fashion. No discrep-
`ancies with respect to the classification into the poor
`metabolizer or extensive metabolizer groups were
`found.
`
`Data evaluation
`Serum trough concentrations of haloperidol and the
`administered dosages of haloperidol were used to esti-
`mate the concentration-time curves by means of the
`Bayesian approach.55 In brief, WinNonlin, version 1.5
`(Pharsight Corporation, Mountain View, Calif), was
`used for the modeling of all individual concentration
`data on the basis of the complete individual dosing
`history. Concentrations after oral dosage and intramus-
`cular dosage of the depot form of the drug were mod-
`eled according to a first-order
`absorption and
`1-compartment disposition model. A population bio-
`availability of 60% for oral dosages, a bioavailability of
`100% for intramuscular and intravenous application, a
`total systemic clearance of 11.9 mL 䡠 min⫺1 䡠 kg⫺1
`(standard deviation [SD], 2.9 mL 䡠 min⫺1 䡠 kg⫺1), and
`a volume of distribution of 18 L/kg (SD, 7 L/kg) were
`applied,56 whereas absorption constants after oral dos-
`
`Vanda Exhibit 2026 - Page 3
`
`

`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 72, NUMBER 4
`
`Brockmo¨ller et al 441
`
`age of haloperidol and intramuscular dosage of halo-
`peridol decanoate were fixed at 1.5 h⫺1 and 0.006 h⫺1,
`respectively. By use of this Bayesian approach, the
`estimation of the individual pharmacokinetic parame-
`ters was stabilized through use of population data from
`the literature. It should be considered, however, that the
`variability resulting from the CYP2D6 genotype might
`be slightly underestimated because the Bayesian ap-
`proach as applied here may direct parameter estimation
`toward one common population mean not differentiated
`by
`genotype. Fig
`1
`shows
`the
`approximated
`concentration-time curves of 2 patients to illustrate how
`the individual dose schemes of each patient were con-
`sidered in the pharmacokinetic analysis. In Fig 1, A, the
`patient received haloperidol first orally and then intra-
`venously; in Fig 1, B, haloperidol was administered
`intramuscularly and orally. Fig 1, B, also illustrates
`how the combined oral and intramuscular medication
`results in high serum concentrations; this combined
`treatment was applied in 20% of the patients included
`in this study.
`
`Statistics
`Subgroups were formed according to the number of
`active CYP2D6 genes: 0 (poor metabolizers), 1 (inter-
`mediate metabolizers), 2 (extensive metabolizers), and
`more than 2 (ultrafast metabolizers). Nonparametric
`tests were used to evaluate differences between these
`groups (Kruskal-Wallis test or, as a test for trend, the
`Jonckheere-Terpstra test with a predefined trend ac-
`cording to the number of active CYP2D6 genes). All
`tests were performed with SPSS, version 10 (SPSS Inc,
`Chicago, Ill).
`Comedication was assessed as CYP2D6-inhibitory,
`CYP3A4-inhibitory, and CYP3A4-inducing. Inhibitory
`drugs were given a score of 1, inducing drugs were
`given a score of ⫺1, and substances without interaction
`were given a score of 0. If more than one comedication
`was administered, the scores were summed. Clearance
`was a dependent variable, and age, gender, weight,
`smoking habits, CYP2D6 genotype, and comedication
`factors were independent variables. The influence of
`factors such as age, weight, gender, smoking habits,
`and comedication on haloperidol clearance was tested
`with multiple linear regression analysis.
`
`RESULTS
`Among the 175 patients included, 5 (2.9%) were
`CYP2D6 ultrafast metabolizers (⬎2 active alleles), 106
`(60.6%) were extensive metabolizers with 2 active
`genes, 56 (32.0%) were intermediate metabolizers with
`1 active gene, and only 5 (2.9%) were identified by
`
`Table I. CYP2D6 genotypes of patients
`Metabolic
`Total
`CYP2D6
`phenotype*
`No. of patients
`genotypes†
`
`No. of
`patients
`
`Ultrafast
`
`5 (3%)
`
`Extensive
`
`106 (61%)
`
`Intermediate
`
`56 (32%)
`
`Poor
`
`5 (3%)
`
`2x2/1
`2x2/2
`2x2/4
`1/1
`1/2
`2/2
`1/10
`1/17
`1/9
`1x2/3
`2/10
`9/10
`9/9
`2/9
`2/3
`2/4
`2/5
`2/7
`1/3
`1/4
`1/5
`1/6
`3/9
`4/10
`4/9
`4/4
`4/5
`4/6
`
`3
`2
`1
`23
`49
`20
`1
`1
`2
`1
`5
`1
`1
`1
`4
`13
`3
`1
`7
`18
`3
`2
`1
`3
`1
`3
`1
`1
`
`*Prediction of metabolic activity of CYP2D6 according to genotype. Poor
`metabolizer, zero active genes; intermediate metabolizer, 1 active gene; exten-
`sive metabolizer, 2 active genes; ultrafast metabolizer, 3 or more active genes.
`†According to the international nomenclature.69
`
`genotype as poor metabolizers (ie, subjects without any
`CYP2D6 activity for genetic reasons). A list of the
`allele combinations found is presented in Table I (no
`blood samples for genotyping were available from 3
`patients).
`The mean haloperidol serum trough concentration
`was 8.4 ␮g/L (SD, 6.7 ␮g/L). The broad variability in
`serum concentrations is only partially explained by the
`different doses administered. The coefficient of deter-
`mination (r2) of dose was 0.31 (P ⬍ .001 for correla-
`tion) for haloperidol serum concentration and 0.10
`(P ⬍ .001) for reduced haloperidol serum concentra-
`tion, when correlated with the dose administered within
`22 hours before the blood sample was drawn. The
`frequency distributions and the correlations with the
`previously given doses are illustrated in Fig 2.
`The group mean of haloperidol concentrations for
`each CYP2D6 metabolizer group was calculated from
`the mean
`haloperidol
`concentrations
`for
`every
`
`Vanda Exhibit 2026 - Page 4
`
`

`
`442 Brockmo¨ller et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`OCTOBER 2002
`
`Fig 1. Illustration of typical concentration-time courses of haloperidol in 2 patients who received
`different application forms of haloperidol. A, Haloperidol serum concentration in a patient who
`received multiple intravenous and oral doses. B, Haloperidol concentration in a patient who received
`haloperidol intramuscularly and orally. The haloperidol concentrations measured are depicted by
`circles.
`
`haloperidol-treated patient. A trend toward higher hal-
`operidol serum concentrations was detected in individ-
`uals with low CYP2D6 activity or without CYP2D6
`activity (Table II). This trend was statistically signifi-
`cant (P ⫽ .05, Jonckheere-Terpstra trend test) when
`serum concentrations were normalized for the total
`dose administered within the 24-hour period before
`blood samples were taken. The total haloperidol clear-
`ance was calculated by taking each individual dose into
`account, and this parameter depended significantly on
`the number of active alleles of CYP2D6 (P ⫽ .034). As
`illustrated in Fig 3, individuals with more than 2 active
`alleles, classified as ultrafast metabolizers, had the
`highest total haloperidol clearance (mean, 57.3 L/h),
`which was 60% higher than total clearance in the poor
`metabolizer group (mean, 34.7 L/h).
`Genotypically identified poor metabolizers also
`showed significantly higher serum levels of reduced
`haloperidol (P ⫽ .004, Jonckheere-Terpstra trend test)
`
`(Table II). The CYP2D6 effect on reduced haloperidol
`levels is illustrated in Fig 4; the effect was even more
`pronounced for the ratio of reduced haloperidol
`to
`haloperidol (P ⫽ .002).
`Multivariate analysis revealed that body weight,
`smoking, and number of active CYP2D6 alleles were
`the most important determinants of haloperidol clear-
`ance, whereas the comedication exhibited only moder-
`ate effects (Table III). For CYP3A4 inhibitors and
`inducers, correlations were not statistically significant.
`Of the patients, 63% were smokers and 37% were
`nonsmokers. The mean haloperidol clearance was only
`marginally higher in smokers than in nonsmokers (49.7
`L/h versus 44.3 L/h). The correlation between the num-
`ber of cigarettes smoked per day and haloperidol serum
`concentrations was not statistically significant, but the
`corresponding correlation with reduced haloperidol
`was statistically significant (P ⫽ .02, Spearman rank
`correlation analysis).
`
`Vanda Exhibit 2026 - Page 5
`
`

`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 72, NUMBER 4
`
`Brockmo¨ller et al 443
`
`Fig 2. Frequency distribution of the measured haloperidol (A) and reduced haloperidol (B)
`concentrations and correlations between dose and concentrations of haloperidol (C) and reduced
`haloperidol (D).
`
`Table II. Dependence of haloperidol trough serum concentration, haloperidol clearance, reduced haloperidol serum
`concentration, and reduced haloperidol/haloperidol ratio on the number of CYP2D6 alleles
`
`Daily dose
`(mean ⫾ SD)
`(mg)
`
`Haloperidol
`Concentration
`Clearance
`(mean ⫾ SD)
`(mean ⫾ SD)
`(␮g/L)
`(L/h)
`
`Reduced haloperidol
`Concentration
`(mean ⫾ SD)
`(␮g/L)
`
`RH/H
`(mean ⫾ SD)
`
`No.
`of samples
`analyzed
`
`CYP2D6 activity*
`Ultrafast
`Extensive
`Intermediate
`Poor
`Jonckheere-Terpstra test
`
`14 ⫾ 10
`12 ⫾ 12
`12 ⫾ 10
`13 ⫾ 9
`
`7.0 ⫾ 5.7
`7.3 ⫾ 5.5
`8.6 ⫾ 5.8
`6.9 ⫾ 4.8
`NS
`
`57.3 ⫾ 31.7
`48.7 ⫾ 20.9
`44.1 ⫾ 25.4
`34.7 ⫾ 13.7
`P ⫽ .034
`
`7.2 ⫾ 8.6
`2.0 ⫾ 2.0
`4.2 ⫾ 5.6
`9.5 ⫾ 7.0
`P ⫽ .004
`
`1.4 ⫾ 1.7
`0.3 ⫾ 0.3
`0.4 ⫾ 0.3
`1.6 ⫾ 0.9
`P ⫽ .002
`
`5
`106
`56
`5
`172
`
`Mean daily doses of haloperidol were calculated as the mean of the haloperidol doses taken over a 24-hour period before haloperidol concentration measurement. Note
`that for estimation of total haloperidol clearances each individual dosage history was precisely considered. If more than one concentration was measured per patient, the
`mean of the serum concentrations was taken. RH/H, Reduced haloperidol/haloperidol ratio; NS, not significant.
`*Enzyme activity predicted from CYP2D6 genotype.
`
`Pseudoparkinsonism (EPS) was evaluated with the
`Simpson and Angus score49 at 3 predefined times (Ta-
`
`ble IV). Poor metabolizers of substrates of CYP2D6
`had statistically significant higher EPS sum scores than
`
`Vanda Exhibit 2026 - Page 6
`
`

`
`444 Brockmo¨ller et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`OCTOBER 2002
`
`Fig 3. Dependence of haloperidol clearance on CYP2D6
`genotype. The different CYP2D6 genotypes are coded accord-
`ing to their number of active genes as follows: 0, complete
`(homozygous) deficiency; 1, heterozygous carriers of 1 active
`and 1 deficient allele; 2, carriers of 2 active genes; 3, carriers
`of 1 active and 1 duplication allele or 2 duplication alleles.
`The trend was statistically significant (P ⫽ .034). Box plots
`generated with SPSS, version 10.0 software show medians
`(black lines), interquartile ranges (gray boxes), and ranges of
`measured data (error bars).
`
`Table III. Dependence of haloperidol clearance on
`age, gender, comedication, and CYP2D6 genotype as
`analyzed by multiple linear regression analysis (n ⫽
`172; r2 ⫽ 0.82)
`
`Parameter
`
`Age
`Gender
`Weight
`Smoking habits
`CYP2D6 genotype
`CYP2D6 inhibitors
`CYP3A4 interactors
`
`Regression
`coefficient
`
`Level of
`significance
`
`0.2
`3.8
`0.4
`7.4
`0.4
`⫺0.9
`3.1
`
`.3
`.2
`<.001
`.04
`.002
`.5
`.3
`
`The parameters gender, CYP2D6 genotype, CYP2D6 inhibitors, and
`CYP3A4 interactors were coded in the regression analysis as follows: gender
`(1, male; 2, female); CYP2D6 genotype (number of active genes, 0, 1, 2, or 3,
`with the latter coding for ultrafast metabolizers); CYP2D6 inhibitors (CYP2D6
`inhibitory comedication was indicated with 2 and other or no comedication
`with 0); CYP3A4 interactors (inducers of CYP3A4 were indicated with -2,
`inhibitors with 2, and no CYP3A4-interacting comedication with 0. If more
`than one comedication was given, the sum was used). Significant correlations
`are shown in boldface.
`
`Fig 4. Dependence of reduced haloperidol serum trough
`levels (normalized for haloperidol dose) on CYP2D6 geno-
`type. Grouping according to CYP2D6 genotype was per-
`formed as described in Fig 3.
`
`individuals with one or more active CYP2D6 genes
`(P ⫽ .02, Mann-Whitney U test). The median EPS sum
`score versus CYP2D6 activity is depicted in Fig 5. As
`an anticholinergic agent, only biperiden was applied in
`our sample. The median EPS sum scores when patients
`were stratified for comedication with biperiden were
`12, 5, 6, and 7 for poor metabolizers, intermediate
`metabolizers, extensive metabolizers, and ultrafast me-
`tabolizers, respectively, compared with median scores
`without biperiden of 8, 5, 4, and 9, for these groups,
`respectively.
`Neither akathisia, as measured by the Barnes akathi-
`sia scale, nor the scores for tardive dyskinesia, as mea-
`sured by the abnormal involuntary movement scale,
`depended significantly on the CYP2D6 genotype (Table
`IV).
`The correlation between serum haloperidol concen-
`trations and acute dyskinesia was not statistically sig-
`nificant (tested with Spearman rank correlation analy-
`sis). Patients were assigned to 4 groups (first to fourth
`quartiles) according to their haloperidol serum trough
`concentrations. Medians of the sums of the EPS scores
`were calculated for each group. These EPS scores,
`reflecting pseudoparkinsonism, were measured on the
`day the blood samples were taken. As shown in Table
`IV, median EPS scores were the same in all 4 groups.
`
`Vanda Exhibit 2026 - Page 7
`
`

`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 72, NUMBER 4
`
`Brockmo¨ller et al 445
`
`Fig 5. Extrapyramidal symptoms (EPS) in individuals with 0, 1, 2, or more than 2 active CYP2D6
`genes. The differences were statistically significant according to the Kruskal-Wallis test (P ⫽ .02).
`
`Table IV. Median sum scores of adverse side effect rating scales for haloperidol and reduced haloperidol
`concentration ranges
`
`AIMS score
`(median and range)
`
`BARS score
`(median and range)
`
`EPS score
`(median and range)
`
`Mean haloperidol
`Group 1 (0-2.9 ␮g/L)
`Group 2 (2.9–5.2 ␮g/L)
`Group 3 (5.2–9.3 ␮g/L)
`Group 4 (⬎9.3 ␮g/L)
`Mean reduced haloperidol
`Group 1 (0–0.6 ␮g/L)
`Group 2 (0.6–1.3 ␮g/L)
`Group 3 (1.3–2.5 ␮g/L)
`Group 4 (⬎2.5 ␮g/L)
`RH/H ratio
`⬍1
`⬎1
`The concentration ranges among the 4 quartiles were indicated as groups 1 to 4. Data were taken from day 3 only. AIMS, Abnormal involuntary movement scale;
`BARS, Barnes akathisia scale; EPS, extrapyramidal symptoms; RH/H, reduced haloperidol/haloperidol concentration ratio.
`
`4 (0–15)
`4 (0–31)
`4 (0–18)
`4 (0–21)
`
`4 (0–14)
`3 (0–18)
`5 (0–31)
`5 (0–21)
`
`4 (0–31)
`10 (0–16)
`
`0 (0–15)
`0 (0–16)
`0 (0–14)
`0 (0–14)
`
`0 (0–16)
`0 (0–11)
`0 (0–13)
`0 (0–14)
`
`0 (0–16)
`0 (0–2)
`
`3 (0–12)
`1 (0–7)
`2 (0–7)
`1.5 (0–10)
`
`3 (0–12)
`2 (0–7)
`2.5 (0–7)
`0.5 (0–10)
`
`2 (0–12)
`0 (0–4)
`
`Also, akathisia and tardive dyskinesia did not correlate
`with haloperidol trough concentrations.
`However, trough concentrations of reduced haloper-
`idol showed a trend toward correlation with pseudopar-
`kinsonism (EPS) (P ⫽ .06, Spearman rank correlation
`analysis). The same trend was seen for the ratio of
`reduced haloperidol to haloperidol (P ⫽ .075, Spear-
`man rank correlation analysis), with a median EPS sum
`
`score of 4 in those patients with a ratio less than 1 and
`a median score of 10 in those with a ratio greater than
`1 (Table IV).
`Other complaints were assessed by adverse event
`rating scales at 3 predefined times (days 3, 14, and 28
`after hospitalization). The severity of the effects was
`graded as 1 (no change in medication), 2 (additional
`drug or dose reduction), or 3 (discontinuation of drug
`
`Vanda Exhibit 2026 - Page 8
`
`

`
`446 Brockmo¨ller et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`OCTOBER 2002
`
`Fig 6. Improvement of psychotic symptoms in individuals with 0, 1, 2, or more than 2 active
`CYP2D6 genes as measured by the positive and negative symptoms scale (PANSS). The difference
`in the total PANSS score between days 14 and 3 is shown. The trend was not statistically significant.
`
`therapy). We evaluated the relationship between the
`number of grade 2 and grade 3 side effects and the
`CYP2D6 metabolizer group. The highest percentage of
`patients with at least one side effect with a severity of
`grade 2 or 3 was seen in the ultrafast metabolizer group
`(100%), whereas the extensive and intermediate me-
`tabolizer groups did not differ very much (66% and
`73%, respectively). This may indicate particular thera-
`peutic problems in the ultrafast metabolizer group.
`Illness severity and therapeutic improvement were
`monitored by the PANSS scale ratings at days 3, 14,
`and 28. The differences in scores between day 3 and
`day 28 and between day 3 and day 14 were taken as a
`measure of therapeutic success, with negative values
`indicating a decline of symptoms. In Fig 6 the changes
`in total PANSS ratings between day 14 and day 3 are
`depicted for the different metabolizer groups. Table V
`shows the overall decline in the PANSS item scores
`(positive, negative, general) in the different CYP2D6
`metabolizer groups, as well as for different haloperidol
`concentration ranges and different dose ranges, during
`the 28-day study period. The improvements in the total
`score (Fig 5), the general item score, and the positive
`item score (Table V) were slightly superior in poor
`metabolizers, but this trend did not reach statistical
`significance. It is interesting that ultrafast metabolizers
`
`experienced the smallest improvement between day 3
`and day 28 (Fig 6, Table V) and even a worsening of
`symptoms between day 3 and day 14.
`The lowest doses of haloperidol (0-8 mg/d) resulted
`in no or only minimal improvement of positive and
`negative schizophrenic symptoms (Table V). No corre-
`lation was found between PANSS scores and mean
`drug concentration levels (mean concentrations from
`days 3 and 28), and there was no trend toward greater
`improvement of symptoms in the patients with higher
`haloperidol levels.
`
`DISCUSSION
`In an unselected sample of inpatients with schizo-
`phrenia, the impact of the CYP2D6 genotype on halo-
`peridol pharmacokinetics, adverse events, and outcome
`of treatment was prospectively studied. Patients re-
`ceived standard clinical
`treatment according to the
`guidelines established at
`the participating hospitals.
`This study was initiated to investigate whether patients
`benefit
`from a dosage regimen adjusted to their
`CYP2D6 genotype and to therapeutic drug monitoring
`(TDM) of their haloperidol trough levels. We found
`that the pharmacokinetics of haloperidol and reduced
`haloperidol, as well as the occurrence of EPS (pseu-
`
`Vanda Exhibit 2026 - Page 9
`
`

`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 72, NUMBER 4
`
`Brockmo¨ller et al 447
`
`Table V. Haloperidol serum concentrations, CYP2D6 genotype, and haloperidol mean dose as possible predictors
`of therapeutic success in acute psychosis
`
`Median
`difference and range of
`PANSS (day 28 – day 3),
`general items
`
`Median
`difference and range of
`PANSS (day 28 – day 3),
`positive items
`
`Median
`difference and range of
`PANSS (day 28 – day 3),
`negative items
`
`Mean haloperidol serum
`concentration
`Group 1 (0–2.9 ␮g/L)
`Group 2 (2.9–5.2 ␮g/L)
`Group 3 (5.2–9.3 ␮g/L)
`Group 4 (⬎9.3 ␮g/L)
`CYP2D6 active genes
`0
`1
`2
`3
`Mean daily dose of
`haloperidol (range)*
`0-8 mg
`8-13 mg
`13-17 mg
`⬎ 17 mg
`
`⫺5 (⫺20 to ⫹5)
`⫺3 (⫺13 to 0)
`⫺13 (⫺32 to ⫹3)
`⫺10 (⫺25 to ⫹25)
`
`⫺17 (⫺33 to ⫺4)
`⫺9 (⫺41 t ⫹25)
`⫺9.5 (⫺33 to ⫹17)
`⫺8 (⫺31 to ⫹1)
`
`⫺10 (⫺53 to ⫹14)
`⫺9 (⫺33 to ⫹1)
`⫺8 (⫺41 to ⫹17)
`⫺12 (⫺47 to ⫹25)
`
`⫺14 (⫺18 to ⫺9)
`⫺1 (⫺10 to ⫹1)
`⫺13 (⫺28 to ⫺3)
`⫺10 (⫺22 to ⫹13)
`
`⫺13 (⫺15 to ⫺3)
`⫺7 (⫺29 to ⫹13)
`⫺9 (⫺25 to ⫹18)
`⫺10 (⫺15 to ⫺8)
`
`0 (⫺12 to ⫹2)
`⫺9 (⫺29 to ⫹2)
`⫺7 (⫺22 to ⫹13)
`⫺11 (⫺28 to ⫹18)
`
`⫺4 (⫺7 to⫺1)
`⫺1 (⫺1 to 0)
`⫺4 (⫺18 to ⫹3)
`⫺3 (⫺15 to ⫹9)
`
`⫺9 (⫺11 to ⫺4)
`⫺5 (⫺18 to ⫹7)
`⫺3 (⫺27 to ⫹27)
`⫺2 (⫺8 to⫹5)
`
`⫺3 (⫺41 to ⫹9)
`⫺4 (⫺22 to ⫹7)
`⫺2 (