A Study of Genetic (CYP2D6 and ABCB1) and Environmental (Drug Inhibitors and Inducers) Variables That May Influence Plasma Risperidone Levels

• Abstract
• Introduction
• CYP2D6
• CYP2D6 and plasma risperidone levels
• CYP3A
• P-glycoprotein
• Plasma risperidone:9-hydroxyrisperidone ratio
• Plasma risperidone total concentration:dose ratio
• Methods
• Sample
• Genotyping
• Co-medications
• Risperidone plasma concentrations
• Statistics
• Linear regression analysis of the log of R:9-OHR ratios
• Logistic regression analysis of R:9-OHR>1
• Linear regression of the log of C:D ratio
• Logistic regression analysis of C:D >14 and C:D <3.5
• Results
• Linear regression of the log of R:9-OHR concentration ratio
• Logistic regression of R:9-OHR >1
• Linear regression of the log of total concentration:dose ratio
• Logistic regression of concentration:dose ratio >14
• Logistic regression of concentration:dose ratio <3.5
• Discussion
• Non-significant variables
• CYP2D6 polymorphism
• ABCB1 polymorphism
• CYP inhibitors
• CYP3A inducers
• Age and sex
• Limitations
• Future studies
• Acknowledgements
• References

Abstract

Risperidone (R) is metabolized to 9-hydroxyrisperidone (9-OHR) by cytochrome P450 2D6 (CYP2D6). The main objective of this naturalistic study was to investigate the variables associated with two plasma ratios: the plasma R:9-OHR concentration ratio and the total concentration-to-dose (C:D) ratio. These ratios were studied as continuous measures by linear regression analyses and as three dichotomous variables in logistic regression analyses: R:9-OHR ratio >1 (indicative of lack of CYP2D6 activity), C:D ratio >14 (indicative of diminished R elimination), and C:D ratio <3.5 (indicative of increased R elimination). Plasma R levels; genotypes for CYP2D6, CYP3A5; and ABCB1 genes, and co-medication, including CYP inhibitors and CYP3A inducers, were studied in 277 patients. Almost all CYP2D6 poor metabolizers (PMs) had an inverted R:9-OHR ratio (>1). Having a CYP2D6 PM phenotype was strongly associated with a C:D ratio >14 (OR=8.2; 95% confidence interval [CI]=2.0-32.7), indicating diminished R elimination. CYP2D6 ultrarapid metabolizers (UMs) did not exhibit an increased R elimination. Some ABCB1 (or MDR1) variants were significantly associated with increased R:9-OHR ratios and decreased C:D ratios, but the results were neither consistent nor robust. Taking CYP inhibitors was significantly associated with a C:D ratio >14 (OR=3.8; CI=1.7-8.7) and with an inverted R:9-OHR ratio. Taking CYP3A inducers was significantly associated with a C:D ratio <3.5 (OR=41.8; CI=12.7-138), indicating increased R elimination. Female gender and old age appeared to be associated with a lower R elimination. Our study indicated that the CYP2D6 PM phenotype may have a major role in personalizing R doses, whereas the CYP3A5 PM phenotype probably has no role. CYP inducers and inhibitors appear to be relevant to R dosing. New studies are needed, particularly to further assess the role of the CYP2D6 UM phenotype and ABCB1 variants in R pharmacokinetics.

Introduction

CYP2D6

Cytochrome P450 2D6 (CYP2D6) metabolizes several antipsychotic and antidepressant drugs, including risperidone (R). The CYP2D6 enzyme is expressed constitutively in several tissues, particularly the liver; therefore, the type of CYP2D6 alleles expressed in a subject is predictive of his/her CYP2D6 enzyme activity. In addition, certain drugs that may be co-administered with R (e.g., paroxetine, bupropion, and fluoxetine) are potent CYP2D6 inhibitors and are expected to affect R elimination.

The activity of the CYP2D6 enzyme is extremely variable in subjects due to its absent expression in some subjects and its overexpression in others. The CYP2D6 gene exhibits substantial genetic variation, with 59 alleles (www.cypalleles.ki.se) reported to date. CYP2D6 enzyme activity is generally described based upon one of four phenotypes that can be predicted from its genotype [9]. The ultrarapid metabolizer (UM) phenotype is associated with three or more functional copies of the active CYP2D6 gene and overexpresses the CYP2D6 enzyme. The extensive metabolizer (EM) phenotype is associated with one or two functional copies of the CYP2D6 gene and expresses normal or typical CYP2D6 activity. The intermediate metabolizer (IM) phenotype is associated with a nonfunctional CYP2D6 allele and a second CYP2D6 allele with a low level of expression and/or activity [9]. The poor metabolizer (PM) phenotype is associated with two nonfunctional CYP2D6 alleles that cannot be transcribed into a functional CYP2D6 enzyme.

The importance of the CYP2D6 enzyme in drug metabolism was recognized by the Food and Drug Administration (FDA) when it granted market approval for the AmpliChip CYP450 Test, which is the first pharmacogenetic test using a DNA microarray device and the only FDA-approved device for CYP2D6 genotyping [10] [13].

CYP2D6 and plasma risperidone levels

The studies from R's marketer suggest that R and its main metabolite, 9-hydroxyrisperidone (9-OHR), have similar pharmacodynamic activity. This implies that the total plasma R moiety (sum of plasma R and 9-OHR concentrations) determines R pharmacological activity [19]. If this is correct, there should be no concern about the CYP2D6 metabolism of R, because a decrease in plasma 9-OHR concentration is almost perfectly countered by a corresponding increase in plasma R concentration in CYP2D6 PMs. The published information supporting the concept that having a CYP2D6 PM phenotype is irrelevant for R treatment was based on an R study in healthy volunteers; the study used single R doses and measured prolactin and plasma R and 9-OHR levels [20]. PMs and EMs exhibited similar total plasma R moieties, although only 11 subjects (2 PMs and 9 EMs classified by CYP2D6 phenotyping) were studied. A study using single R doses administered in various ways can hardly be considered equivalent to clinical practice, which involves chronic therapy, nor does it give consideration to the racemic nature of its metabolite, 9-OHR. Nonetheless, this initial study led to R's package insert suggesting that CYP2D6 polymorphic expression and CYP drug-drug interactions were therapeutically unimportant in R therapy.

In contrast, an initial pilot study found that all five CYP2D6 PMs identified in routine clinical practice experienced adverse drug reactions (ADRs) while taking R [5]. Moreover, case reports suggest that CYP2D6 UMs may need high R doses. In this more recent study of 360 patients on R treatment and 252 patients who were discontinued from R, the odds, adjusted for confounding variables, that a CYP2D6 PM would develop R ADRs was found to be more than three times higher than the odds for a non-PM (odds ratio [OR]=3.4) [14]. When the analysis focused on those patients who had discontinued R treatment, the adjusted odds that a CYP2D6 PM would discontinue R due to an ADR were found to be six times higher than the odds for a non-PM. (OR=6.0) [14]. These two ORs reflect the individual perspective and suggest that the CYP2D6 PM phenotype's effects were consistent, powerful, and relevant for the individual patient. The way to avoid R ADRs in CYP2D6 PMs was to prescribe small R doses. From the public health or patient population perspective, being a CYP2D6 PM explained only 16% of the R ADRs and 9% of all discontinuations due to ADRs [13].

CYP3A

We were puzzled by the observation that carbamazepine seemed to induce R metabolism in patients co-medicated with this drug. Because the CYP2D6 enzyme is known not to be inducible, it was hypothesized that cytochrome P450 3A (CYP3A) also metabolizes R [12]. This was subsequently confirmed by in vitro studies [17]. R metabolism by CYP3A seems to explain why CYP3A inhibitors and inducers influence R metabolism. Cytochrome P450 3A4 (CYP3A4) is the most important enzyme of the CYP3A isoenzyme subfamily. Another member of this subfamily, CYP3A5, exhibits polymorphic variations, shares high homology with CYP3A4, and is reported to metabolize many of the same substrates. However, the clinical relevance of the CYP3A5 PM phenotype remains unclear.

P-glycoprotein

We proposed that the plasma profile for CYP2D6 PMs (characterized by higher R than 9-OHR concentrations) may be more “toxic” than that of other phenotypes [3]. A recent study using rats demonstrated that R may cross the blood-brain barrier more easily than 9-OHR [26]. Although this has not been studied in humans, enhanced brain penetration of R relative to 9-OHR may explain why plasma R levels have greater toxicity than comparable plasma 9-OHR levels. Diminished entry of 9-OHR into the brain is reportedly due to the presence of a transporter in the blood-brain barrier, the P-glycoprotein (PgP), which appears to have greater affinity for 9-OHR than for R [26]. A recent study in 12 male volunteers suggested that verapamil, a PgP inhibitor, had only modest effects on R bioavailability [22].

PgP is an ATP-dependent efflux pump located in the small intestine, brain, kidney, and other organs where it may influence movement through tissues and exposure to drugs [28]. PgP and CYP3A share similar specificity toward substrates, inducers, and inhibitors. The gene that controls PgP is called ABCB1 or MDR1. A human variant of the ABCB1 gene (C3435T in exon 26) has attracted particular interest [8]. This particular ABCB1 variant is silent but is linked with a non-synonymous variation in the ABCB1 gene, including G2677 (A, T) in exon 21, that exhibits two possible mutations, T and A [24]. It is believed that the T variation in exon 26 may be associated with lower PgP activity because of its linkage with the variation in exon 21. More recently, researchers have focused on ABCB1 haplotypes. Haplotype 22 (2677T/3435T) and haplotype 12 (2677G/3435 T) are frequently involved in pharmacokinetic drug studies with positive results [31]. In a study of 75 Japanese patients taking R, the T/T variants in exons 21 and 26 were not associated with R or 9-OHR levels [30]. Far more research is required for understanding the functional relevance of ABCB1 variations before it is possible to truly relate the ABCB1 genotype to its phenotypic relevance in vivo.

Plasma risperidone:9-hydroxyrisperidone ratio

Clinicians can use two major ratios to interpret plasma R and 9-OHR concentrations: the R:9-OHR ratio and the total concentration:dose ratio (C:D) [11]. The R:9-OHR concentration ratio is an index of CYP2D6 activity, as the aliphatic hydroxylation of R is performed mainly by CYP2D6. Usually, plasma R:9-OHR ratios are <1 (with an average of about 0.1-0.2) [2] [16]. An R:9-OHR ratio >1, also called an inverted plasma ratio (R concentration > 9-OHR concentration), indicates a CYP2D6 PM phenotype or the presence of a powerful CYP2D6 inhibitor [16].

Plasma risperidone total concentration:dose ratio

The sum of plasma R and 9-OHR concentrations (called total R concentration or total moiety) divided by the dose is called the C:D ratio. The multi-center study from R's marketer indicates that R follows linear kinetics and has an average C:D of about 7 [2]. We calculated a mean C:D ratio of 7.1 for a dose of 2 mg/day, 7.2 for 6 mg/day, 7.3 for 10 mg/day, and 7.0 for 16 mg/day, using the data from the manufacturer's multi-center study [2]. This suggests that there is a linear relationship between the total concentration and the dose. We proposed that an increase or decrease in R elimination (or in the C:D ratio) by a factor of 2 is probably meaningful from the clinician's point of view [5]. The R C:D ratio is reportedly decreased by CYP3A inducers and increased by CYP3A inhibitors [5] [11].

The main objective of the current naturalistic study was to investigate the variables associated with the R:9-OHR and C:D ratios. These ratios were studied as continuous measures. In addition, these ratios were used to define three dichotomous variables that were believed to reflect R metabolic activity: an R:9-OHR ratio >1 was considered to be indicative of lack of CYP2D6 activity, and C:D ratios >14 or <3.5 were considered to be indicative of impaired and increased R elimination from the body, respectively. A CYP2D6 PM phenotype and CYP inhibitor intake were hypothesized to be associated with both an R:9-OHR ratio >1 and a C:D ratio >14, while taking CYP3A inducers was hypothesized to be associated with a C:D ratio <3.5. In addition to the CYP2D6 gene, the variants of two other genes (CYP3A5 and ABCB1) were explored.

Methods

Sample

In the context of our previously published R pharmacogenetic study [14] of the relationship between CYP2D6 phenotypes and R side effects, we studied 325 patients on R who were inpatients or outpatients in central Kentucky psychiatric facilities. After completely describing the study to the subjects, written informed consent was obtained. Of the 325 patients, 2% (7/325) did not provide enough blood, 2% (7/325) had undetectable R levels, 9% (30/325) had interfering substances, 1% (3/325) provided R levels but their doses were changed before their levels were drawn, and <1% (1/325) had renal insufficiency. The remaining 277 patients (85%) constituted the sample investigated in the current study. Each of these patients provided a trough R level, which was measured from an early-morning blood sample that was collected before the morning R dose and after at least five days under a constant R dose ([Table 1]).

Genotyping

DNA was extracted from whole blood according to a previously described method [9]. Up to 34 CYP2D6 alleles (^*2, ^*3, ^*4, ^*5, ^*6, ^*7, ^*8, ^*9, ^*10, ^*11, ^*14, ^*15, ^*17, ^*18, ^*19, ^*20, ^*25, ^*26, ^*29, ^*30, ^*31, ^*35, ^*36, ^*37, ^*40, ^*41, ^*43, ^*45, and the following duplicated CYP2D6 alleles: ^*1xn, ^*2xn, ^*4xn, ^*10xn, ^*17xn, and ^*35xn) were tested by allele-specific PCR in our laboratory [9] and/or by a prior version of the AmpliChip P450 microarray system from Roche Molecular Systems [10] [13]. All samples were tested by both systems. Patient CYP2D6 genotypes were converted into four CYP2D6 phenotypes predicted by the genotypes PM, IM, EM, and UM, using a method previously described [9].

Prior published methods to identify the CYP3A5^*3 and CYP3A5^*6 alleles were modified in our laboratory [14]. Subjects with a CYP3A5^*3/^*3, ^*3/^*6, or ^*6/^*6 genotype were considered deficient in CYP3A5 expression and thus classified as CYP3A5 PMs. Using published methods [8], patients were classified as having two (T/T), one (C/T), or no (C/C) variations in the ABCB1 exon 26 gene region and were divided into one of six ABCB1 exon 21 genotypes (G/G, G/T, T/T, A/A, G/A, and T/A). Based on published literature [24], it was hypothesized that T/T subjects had lower activity, and thus they were compared with all remaining subjects. Two ABCB1 haplotypes were also investigated: haplotype 12 (2677G/3435 T) and haplotype 22 (2677T/3435 T) [31].

Co-medications

Based on the literature and our study of R ADRs [14], we decided to consider fluoxetine, paroxetine, bupropion, fluvoxamine, sertraline, nefazodone, and celecoxib as relevant inhibitors of R metabolism. This definition provided better significance in predicting R ADRs than did narrower or wider inhibitor definitions [14]. Carbamazepine, phenytoin, and phenobarbital were the CYP3A inducers that increased CYP3A activity and were co-prescribed in the sample. Because these clinically relevant inhibitors and inducers may cause relevant drug-drug interactions, patients were classified according to whether they were taking CYP inhibitors and/or CYP3A inducers. In addition, on the basis of an extensive review of the literature, patients were classified according to whether they were taking CYP2D6 substrates and/or CYP3A substrates. There is some potential for these substrates to cause drug-drug interactions due to competitive inhibition. No patient was taking verapamil, a PgP inhibitor.

Risperidone plasma concentrations

Plasma R and 9-OHR concentrations were quantified at the Nathan Kline Institute using a published liquid chromatography method [27] with minor modifications. The two minor changes involved the extraction procedure from plasma and the chromatographic column employed. The calibration curve comprised seven levels of R and 9-OHR concentrations in the range of 2.5 to 150 ng/mL, which were applied to each batch of samples, and three levels of quality controls (in duplicate). Interassay variability for R and 9-OHR for the high (60 and 120 ng/mL), medium (25 and 50 ng/mL), and low (5 and 10 ng/mL) quality controls were 7.2% and 6.3%, 6.0% and 5.4%, and 8.0% and 8.9%, respectively (n=32 days). Intraassay variabilities for R and 9-OHR concentrations of 150, 100, 50, 25, 10, 5, and 2.5 ng/mL were 7.4% and 6.4%, 5.0% and 4.8%, 8.5% and 7.9%, 7.5% and 6.6%, 10.2% and 9.8%, 5.3% and 5.6%, and 11.4% and 13.0%, respectively (n=12 for each of the seven calibration standards). The limits of quantification and detection were 2 ng/mL and 1 ng/mL, respectively, for R and 9-OHR.

Statistics

Statistical analyses were performed with SPSS [25]. The R administration pattern and the time from the last dose to blood sampling were not significantly correlated with R:9-OHR or C:D ratios, suggesting that R levels probably reflected trough levels. As expected, R dose was not significantly associated with R:9-OHR ratio.

The R:9-OHR and C:D ratios were studied as continuous variables by linear regressions. A regression coefficient (B) was used because this provides information on both the direction of the association between a response and an independent variable and the magnitude of a change in the response variable due to a change in the independent variable. In addition, R:9-OHR and C:D ratios were used to define the following three dichotomous indexes of R elimination that were studied through logistic regressions: (1) the presence or absence of an R:9-OHR ratio >1, (2) the presence or absence of a C:D ratio >14, and (3) the presence or absence of a C:D ratio <3.5. The rationale for the use of these dichotomous indexes is that clinicians need to know the clinical and demographic variables that may cause dramatic changes in R levels, as these types of marked changes are most likely to have the greatest relevance when treating patients with R.

Based on a literature review [16] and our prior work [5], we hypothesized that variations in the CYP2D6 genotype would influence R levels and that the most relevant effect would result from the CYP2D6 PM phenotype. To examine this hypothesis, each patient was dichotomously classified as being a CYP2D6 PM or not. This dichotomous variable was used as an independent variable in statistical analyses. In addition, an approximately continuous classification was performed based on the literature information on CYP2D6 allele activity (see the footnote to [Fig. 1]). The variable thus created was called the number of CYP2D6 active alleles and was used as an independent variable and as an alternative to the dichotomous CYP2D6 PM classification. ([Figs. 1] and [2] describe the log of R:9-OHR and C:D ratios according to the number of CYP2D6 active alleles, respectively.) The CYP2D6 PM phenotype was the main independent variable. The effects of having CYP3A5 and ABCB1 variations were also examined in an exploratory fashion, due to our limited knowledge of their relevance and genotype-phenotype relationships. Two alternative, dichotomous CYP3A5 classifications (PM with no activity versus the others; EM with two active alleles versus those with zero or one active allele) were explored in the multivariate analyses. Similarly, the ABCB1 gene variations - T/T genotype in exon 21, T/T in exon 26, and haplotypes 12 and 21 ([Table 1]) - were explored as alternative dichotomous independent variables in multivariate analyses.

Linear regression analysis of the log of R:9-OHR ratios

A linear regression analysis that investigated the effect of a number of dichotomous independent variables on the R:9-OHR concentration ratio was performed ([Table 2]). In this analysis, only the 171 patients with detectable R levels (≥2 ng/mL) were used. A stepwise procedure was implemented to find the independent variables that had significant effects on the R:9-OHR ratio, using a 0.05 level of significance [29]. Because the distribution of the R:9-OHR ratio was skewed and caused heteroscedasticity in regression models, a natural log transformation of the R:9-OHR ratio was used as the dependent variable in the regression models [29]. Residual analyses suggested that the log transformation of the R:9-OHR ratio and regression models were appropriate. A similar stepwise linear regression analysis was performed by using the number of CYP2D6 active alleles as an independent variable instead of the dichotomous CYP2D6 PM variable.

A further advantage of using the natural log of the R:9-OHR ratio as the dependent variable in linear regression analyses is that a useful measure of the effect size of any dichotomous independent variable on plasma 9-OHR levels can be computed. This measure is based on the concept of relative percentile [21]. If B is the regression coefficient for the dichotomous variable, then the effect size was computed with the formula E=(e ^-B-1) × 100.

Logistic regression analysis of R:9-OHR>1

A stepwise logistic regression analysis was implemented to find the independent variables that were significantly associated with an R:9-OHR ratio >1, using a 0.05 level of significance ([Table 3]). ORs and their 95% confidence intervals (CIs) were computed.

Linear regression of the log of C:D ratio

A stepwise linear regression analysis was used to find the independent variables that were significantly associated with the C:D ratio ([Table 4]). The natural log of the C:D ratio was used as the dependent variable because the distribution of the C:D ratio was skewed and yielded heteroscedastic models [29]. According to residual analyses, the log transformation was appropriate. The effect size of any dichotomous independent variable on R total concentrations was computed as E=(e ^B-1) × 100, where B is the regression coefficient of the dichotomous variable.

Logistic regression analysis of C:D >14 and C:D <3.5

Stepwise logistic regression analyses were implemented to find the independent variables that were significantly associated with a C:D ratio >14 ([Table 5]) and a C:D ratio <3.5, using a 0.05 level of significance.

Results

Linear regression of the log of R:9-OHR concentration ratio

As expected, according to a linear regression analysis, the CYP2D6 PM phenotype significantly increased the R:9-OHR concentration ratio ([Table 2]). Also, the presence of a T/T genotype in ABCB1 exon 21 and the intake of CYP inhibitors were associated with increased R:9-OHR ratios ([Table 2]). The above analyses were repeated by using only the subsample of Caucasian patients. The results were essentially the same as those in [Table 2].

Among patients who were not taking CYP inhibitors, after controlling for confounding variables and R levels, 9-OHR levels in CYP2D6 PMs were 92% lower than those in patients who were not PMs (E=(e ^-2.5-1) × 100=-92%). In contrast, among patients who were taking CYP inhibitors, 9-OHR levels in patients who were CYP2D6 PMs were 81% lower than those in patients who were not PMs (E=(e ^-2.5+0.83-1) × 100=-81%). After controlling for confounding variables and R levels, 9-OHR levels in patients who had a T/T genotype in ABCB1 exon 21 were 37% lower than those in patients who did not have this genotype (E=(e ^-0.47-1) × 100=-37%).

The stepwise linear regression analysis of the log R:9-OHR was repeated by using the number of CYP2D6 active alleles instead of the dichotomous CYP2D6 PM phenotype ([Table 2]). Only two independent variables had significant effects on the R:9-OHR ratio: the number of CYP2D6 active alleles and the intake of CYP inhibitors. No ABCB1 genotype was found to be significantly associated with the R:9-OHR ratio when the number of CYP2D6 active alleles, rather than the dichotomous CYP2D6 PM-predicted phenotype, was used as an independent variable. When the above linear regression analyses were restricted to the subpopulation of Caucasian patients, the results were essentially the same as above.

In the total sample of 277 patients, the Spearman correlation between the R:9-OHR ratio and the number of CYP2D6 active alleles was r=-0.57 (P<0.001). (When the analysis was restricted to Caucasians, Spearman r=-0.59, P<0.001.) The median R:9-OHR ratio (and 25^th, 75^th percentiles) for eight predicted CYP2D6 UMs was 0.04 (0.02, 0.08), for 219 EMs it was 0.13 (0.05, 0.30), for 30 IMs it was 0.86 (0.35, 1.6), and for 20 PMs it was 2.6 (1.9, 4.1). Thus, as expected, the median R:9-OHR ratio increased when moving from UM toward PM phenotype status.

Logistic regression of R:9-OHR >1

Almost every CYP2D6 PM had an inverted ratio: 95% (19/20) versus 13% (34/257) for the rest of the patients. These two percentages were significantly different (χ²=80.2, df=1, P<0.001). According to a logistic regression that adjusted for taking CYP inhibitors and for weight, the OR comparing the presence of inverted ratios in PMs versus non-PMs was very high (OR=313; 95% CI=35-2815). Also, the percentage of patients with inverted ratios was 37% (11/30) for IMs, 11% (23/219) for EMs, and 0% (0/8) for UMs. The OR comparing IMs versus EMs was significantly higher than 1 (OR=4.9; CI=2.1-11.6). After excluding the patients who were taking CYP inhibitors, the percentage of patients with inverted ratios was 27% (4/15) for IMs and 2% (3/159) for EMs (OR=18.9; CI=3.8-95.3).

[Table 3] describes the logistic regression analysis of R:9-OHR >1 performed after excluding the CYP2D6 PMs, using the number of CYP2D6 active alleles as an independent variable. The analysis suggested that the higher the number of CYP2D6 active alleles, the lower the odds were of having an inverted R:9-OHR ratio in patients without a CYP2D6 PM phenotype. Taking CYP inhibitors and having a high body weight were also significantly associated with inverted ratios in these patients.

Linear regression of the log of total concentration:dose ratio

According to the linear regression analysis of the log of the C:D ratio, a predicted CYP2D6 PM phenotype, female sex, and the intake of CYP inhibitors significantly increased the C:D ratio ([Table 4]). On the other hand, taking CYP3A inducers significantly decreased the C:D ratio. Thus, after controlling for sex, CYP3A inducers, CYP inhibitors, and dose, R total concentrations in CYP2D6 PMs were 27% higher than those in non-PMs (E=(e ^0.24-1)-100=+27%). Also, after controlling for potential confounding variables and dose, R total concentrations in patients who were taking CYP3A inducers were 59% lower than those in patients who were not (E=(e ^-0.90-1)×100=-59%); R total concen-trations in patients who were taking CYP inhibitors were 27% higher than those in patients who were not (E=(e ^0.24-1)× 100=+27%); and R total concentrations in female patients were 28% higher than those in male patients (E=(e ^0.26-1)×100=30%).

A stepwise linear regression analysis of the log of the C:D ratio was also performed, using the number of CYP2D6 active alleles instead of the dichotomous CYP2D6 PM phenotype as an independent variable and using a 0.05 level of significance. The number of active alleles had a significant effect on the log of the C:D ratio (B=-0.09, P=0.04) after adjusting for CYP3A inducers (B=-0.90, P<0.001), CYP inhibitors (B=0.24, P<0.001), and female sex (B=0.26, P<0.001). The median C:D ratio for the eight CYP2D6 UMs was 6.0, for the 219 EMs it was 7.0, for the 30 IMs it was 7.8, and for the 20 PMs it was 11.0. The Spearman correlation between the number of CYP2D6 active alleles and the C:D ratio was r=-0.12 (P=0.04) (in Caucasians, r=-0.12, P=0.07). A partial correlation between the number of CYP2D6 active alleles and the C:D ratio, which controlled for potential confounders, was r=-0.12 (P=0.04). The Spearman correlation between the number of CYP3A5 active alleles and the C:D ratio was r=-0.05 (P=0.38).

Logistic regression of concentration:dose ratio >14

According to the logistic regression analysis, a predicted CYP2D6 PM phenotype, female sex, intake of CYP inhibitors, and the interaction between CYP2D6 PM phenotype and the intake of CYP inhibitors were associated with a C:D ratio >14 ([Table 5]). The interaction suggested that patients with the CYP2D6 PM phenotype exhibited a higher frequency of C:D ratios >14 than did the patients without the PM phenotype, but only in the absence of CYP inhibitors (OR adjusted for sex=9.0; 95% CI=2.1-38.5; Wald χ²=8.8; df=1, P=0.003). This association between the predicted CYP2D6 PM phenotype and a C:D ratio >14 was not observed in patients taking CYP inhibitors (OR adjusted for sex=0.49, Wald χ²=0.42, df=1, P=0.5).

Logistic regression of concentration:dose ratio <3.5

According to the stepwise logistic regression analysis, only the intake of CYP3A inducers was significantly associated with a C:D ratio <3.5 (OR=41.8; 95% CI=12.7-138.0; Wald χ²=37.6; df=1; P<0.001; n=277). The OR of 41.8 reflected that 63% (10/16) of patients taking CYP3A inducers versus 4% (10/261) of patients not taking CYP3A inducers had a C:D ratio <3.5. Among the six patients who were taking CYP3A inducers and had C:D ratios ≥3.5, there were three patients who had relatively low C:D ratios (≤4.5). Only one patient had a normal C:D ratio (7.3). This patient was a CYP2D6 PM.

After adjusting for intake of CYP3A inducers, a borderline significant association between not taking CYP inhibitors and a C:D ratio <3.5 was observed (OR=4.1; CI=(0.89, 18.6); χ²=3.3; df=1; P=0.07). Surprisingly, none of the patients with a predicted CYP2D6 UM phenotype had a C:D ratio lower than 3.5.

Discussion

Non-significant variables

The intake of CYP2D6 or CYP3A substrates at the same time as R intake was not significantly associated with the C:D or R:9-OHR ratios in these analyses, although the intake of CYP3A4 substrates had a small and borderline significant effect on the C:D ratio (E=+11%, P=0. 07; [Table 4]). The literature suggests that smoking is not an inducer of the CYP2D6 or CYP3A4 enzymes [1] [11]. Therefore, it is not surprising that this study did not find a significant influence of smoking on R:9-OHR or C:D ratios. The investigated CYP3A5 polymorphic variations also did not have a significant effect on these ratios. Patients both with and without the CYP3A5 PM phenotype had the same median R:9-OHR ratio: 0.16. The median C:D ratios in patients with and without the CYP3A5 PM phenotype were 7.3 and 6.6, respectively. Moreover, the extremely low Spearman correlation (r=-0.05) between the number of CYP3A5 active alleles and the C:D ratio suggests that these variations are likely to be irrelevant to R dosing. This further supports our prior conclusion that the CYP3A5 PM phenotype was not associated with R ADRs in these patients [14].

CYP2D6 polymorphism

In this study, the CYP2D6 PM phenotype predicted R ADRs and R discontinuation due to ADRs [14]. The pharmacokinetic analyses showed inverted ratios for nearly every predicted CYP2D6 PM, an observation consistent with other studies. A new finding from the current pharmacokinetic analyses is that the CYP2D6 PM phenotype was strongly associated with a C:D ratio >14 (OR=8.2), suggesting a low capacity for eliminating R. In summary, CYP2D6 PMs appear to have a more toxic profile (inverted R:9-OHR ratios) and lower capacity for eliminating R from their bodies. This marked effect on R handling probably explains why the CYP2D6 PM phenotype has clinical relevance during R treatment. The effects from the predicted CYP2D6 PM phenotype on R pharmacokinetics are powerful, but some results suggest that they can be somewhat influenced by co-medication. Two interactions between the CYP2D6 PM phenotype and CYP inhibitors were observed (a significant interaction in the logistic regression of C:D ratio >14 and a borderline significant interaction in the linear regression of the log of the R:9-OHR ratio; [Tables 5] and [2], respectively).

Surprisingly, the predicted CYP2D6 UM phenotype did not appear to be associated with an increased R elimination (C:D <3.5). In this study, we did not quantify the number of active copies, but most of our CYP2D6 UMs probably had duplications (three active copies) [7]. As expected, the number of CYP2D6 active alleles strongly predicted the values of the R:9-OHR ratio and weakly predicted the C:D ratio. This suggests that the PM is the main CYP2D6 phenotype that is relevant to the R C:D ratio.

ABCB1 polymorphism

In the linear regression of the log of the R:9-OHR ratio, the T/T variant of exon 21 in the ABCB1 gene exhibited a significant though modest effect on this ratio. In the logistic regression of R:9-OHR >1 and of C:D >14, no ABCB1 variant or haplotype was significant.

The median R:9-OHR ratios in patients with and without a T/T genotype in exon 21 were 0.20 and 0.15, respectively. After controlling for confounding variables and R levels, 9-OHR levels in patients who had a T/T genotype were 37% lower than those in patients who did not. The median C:D ratios in patients with and without a T/T genotype in exon 21 were 9.9 and 7.0, respectively.

Although it cannot be ruled out that these significant associations may be false positives caused by the multiple comparisons performed, we believe that PgP activity may have some relevance to R pharmacokinetics, although we may not be measuring the appropriate ABCB1 variants. It is also possible that even if some ABCB1 variants influence some R pharmacokinetic parameters, these variants still may have little clinical relevance. The ABCB1 variants described in this article were not found to have relevant influence on R ADRs in this study [14], and the observed pharmacokinetic influences were rather modest. Identifying a genetic polymorphism with relevance to drug disposition does not imply that the polymorphism has relevance to clinical practice. Typically, changes in drug disposition have relevance to clinical practice when the changes in drug plasma levels are considerable and/or the drug has a narrow therapeutic window.

CYP inhibitors

As expected, taking CYP inhibitors was strongly associated with inverted R:9-OHR ratios (OR=16.3;. [Table 3]). Also, taking CYP inhibitors increased the odds of having a C:D ratio >14 by a factor of about 4. The median R:9-OHR ratios in patients taking and not taking CYP inhibitors were 0.84 and 0.096, respectively. The median C:D ratios were 9.0 and 6.5, respectively. These comparisons between medians do not control for confounding variables. A logistic regression analysis indicated that R total concentrations in patients who were taking CYP inhibitors were 27% higher than those in patients who were not, after controlling for confounding variables including dose.

These analyses explain why CYP inhibitors may have clinical relevance. As a matter of fact, taking CYP inhibitors appears to explain 38% of R discontinuations due to ADRs in this study (versus 9% due to the CYP2D6 phenotype) [13]. However, the CYP inhibitor effect appears to be weaker than the CYP2D6 PM phenotype effect. Thus, with the variability in dose, type of inhibitor, extent of inhibition superimposed on the number of active CYP2D6 alleles, and the usual “noise” in a routine clinical environment, we failed to detect a significant effect of CYP inhibitors on ADRs in patients taking R [14]. Taking CYP inhibitors may have clinical relevance to R treatment, but the CYP2D6 PM phenotype resulted in more predictable clinical effects.

CYP3A inducers

As expected, taking CYP3A inducers was associated with low C:D ratios (<3.5), indicating that these compounds increase R elimination. The median R:9-OHR ratios in patients taking and not taking CYP3A inducers were 0.40 and 0.16, respectively. The median C:D ratios were 3.1 and 7.5, respectively. These comparisons between medians do not control for confounding variables. The effect size based on relative percentiles, computed from the linear regression of the log of the C:D ratio, indicated that these compounds reduced total R levels to less than half (E=-59%). This result is consistent with those from other studies [11].

CYP3A inducers appear to have clinical relevance in this sample because they significantly decreased R discontinuations due to ADRs [14]. However, maybe because of the “noise” in naturalistic studies and the limited number of patients taking inducers, we could not detect a significant decrease in ADRs in R patients taking CYP3A inducers [14]. In summary, taking CYP3A inducers may have clinical relevance to R treatment.

Age and sex

As predicted by the literature, age increased total R plasma levels (R+9-OHR), although the increase was of only borderline significance (P=0.06). The effect size from the linear regression of the log of the C:D ratio indicated that being older than 60 increased total levels by 20%. Although this effect size is a little lower than that of the CYP2D6 PM phenotype (+27%) and lower than that of CYP inhibitors (+27%), it supports the recommendation that R doses should be reduced in elderly patients due to their decreased R elimination. Another reason that older patients may need lower R doses is that aging is associated with pharmacodynamic changes in the brain due to a decrease in dopamine receptors.

Eighteen percent (23/127) of female patients versus 7% (10/150) of male patients had a C:D ratio >14. The odds of having a C:D ratio >14 in females were 2.8 times higher than the odds in males ([Table 5]). Assuming that this association is replicated, its mechanism needs further investigation. This association does not appear to be explained by the relatively lower body weights among females, since the weight variable was not significantly associated with a C:D ratio >14 after adjusting for sex. According to our data, C:D ratios in females were 30% higher than in males. Aichhorn et al. [1] also found that females had significantly higher C:D ratios (median, 7.2 versus 5.6 in males), but the difference disappeared after correcting for weight (median, 6.1 in females versus 6.2 in males). A recent trial found that reported higher dropouts in females after relatively high R doses may be consistent with lower R elimination in females [23].

Limitations

This study has the typical limitations of naturalistic studies, which are not designed to identify variables with small effect sizes. To identify variables with small effects, more controlled clinical designs are needed. In spite of this, this naturalistic study provided further evidence that the CYP2D6 PM phenotype is a major clinical determinant of R ADRs and that patients with this genotype require dosing adjustments. The effects of additional variables were also detected in this “noisy” clinical environment, including CYP inhibitors, CYP3A inducers, sex, and possibly age. Because of our limited knowledge about the ABCB1 gene, it is difficult to make recommendations concerning its potential for individualizing R dosing. More studies of functional variants of this gene are needed.

The study did not investigate the effects of R treatment on psychotic symptoms and focused only on R ADRs [14]. The reason for this is that ADRs of many drugs can be directly associated with high drug levels and altered metabolic profiles that typically result from genetic polymorphisms in metabolic enzymes and transporters. On the other hand, drug response (or drug efficacy) is a complex phenomenon that is presently poorly defined for antipsychotics and is unlikely to show a simple dependence on drug levels or the still poorly characterized receptor genetic variants. Therefore, no therapeutic window was explored in this study. The focus of our analyses was not on R response but on how pharmacogenetics and R levels may modify R dosing and influence ADRs.

The goal of our analyses of the C:D ratio was to study the sum of R and 9-OHR plasma concentrations, which probably reflects overall R elimination. A study of the ability of total concentrations to reflect ADRs or therapeutic response should consider the weaker brain effects of plasma 9-OHR concentrations, using a weighted average of R and 9-OHR concentrations [18].

Future studies

Our study indicates that the CYP3A5 PM phenotype does not appear to have an important role in personalizing R doses. Future studies of the CYP2D6 UM phenotype need to quantify the number of active copies to determine whether four or more copies of this gene may have more pronounced effects on R elimination. Greater knowledge of the polymorphic variations in the ABCB1 gene is needed to better understand its potential for personalizing R dosing. Prospective randomized clinical studies of R dosing according to CYP2D6 phenotypes predicted by the genotype are needed. Any study focusing on this or any other gene needs to determine environmental factors contributing to R dosing, such as CYP inducers and inhibitors. The observed effects of the CYP2D6 PM phenotype on R pharmacokinetics were very strong, and our data suggest that R ADRs may be influenced by co-medication in some cases.

Acknowledgements

The original study [14] was supported by several sources: a researcher-initiated grant from the Eli Lilly Research Foundation to Jose de Leon, M.D. (24% of direct costs); a NARSAD Independent Award to Jose de Leon, M.D. (11% of direct costs); internal funding (37% of direct costs); and Roche Molecular Systems, Inc., which provided free genotyping and laboratory supplies (equivalent to 28% of direct costs). The statistical analyses were conducted by Francisco J. Diaz, Ph.D. (partially supported by Grant 30802940 from the Dirección de Investigaciones de la Universidad Nacional, Medellín); by Jose de Leon, M.D. (without additional external support); and by Martha Orrego, M.S. (without additional external support). Tom Cooper, M.A., (Nathan Kline Institute, Orangeburg, New York, and New York University School of Medicine, New York, New York) supervised the laboratory analysis of the plasma risperidone concentrations.

During the past two years, Dr. de Leon has been on the advisory boards of Bristol-Myers Squibb and Roche Molecular Systems, Inc. He has received investigator-initiated grants from Roche Molecular Systems, Inc., and the Eli Lilly Research Foundation and has held one lecture supported by Eli Lilly, one supported by Bristol-Myers Squibb, two supported by Janssen, and six supported by Roche Molecular Systems, Inc.

Dr. Wedlund is a consultant for Roche Molecular Systems, Inc., which markets the AmpliChip CYP450 microarray that detects CYP2D6 and CYP2C19 gene variations. The authors thank Lorraine Maw, M.A., for editorial assistance.

References

• 1 Aichhorn W, Weiss U, Marksteiner J, Kemmeler G, Walch T, Zernig G. et al . Influence of age and gender on risperidone plasma concentrations.  J Psychopharmacol. 2005;  19 395-401
  CrossrefPubMedSearch in Google Scholar
• 2 Anderson CB, True JE, Ereshefsky L, Miller AL, Peters BL, Velligan DI. Risperidone dose, plasma levels and response. 1993 New Research Program and Abstracts: American Psychiatric Association Annual Meeting. San Francisco CA 1993 (NR 217): 113
• 3 Barnhill J, Susce MT, Diaz FJ, de Leon J. Risperidone half-life in a patient taking paroxetine: a case report.  Pharmacopsychiatry. 2005;  38 223-225
  Thieme ConnectPubMedSearch in Google Scholar
• 4 Baumann P, Hiemke C, Ulrich S, Eckermann G, Gaertner I, Gerlach M. et al . The AGNP-TDM expert group consensus guidelines: therapeutic drug monitoring in psychiatry.  Pharmacopsychiatry. 2004;  37 243-256
  Thieme ConnectPubMedSearch in Google Scholar
• 5 Bork J, Rogers T, Wedlund P, de Leon J. A pilot study of risperidone metabolism: the role of cytochrome P450 2D6 and 3A.  J Clin Psychiatry. 1999;  60 469-476
  CrossrefPubMedSearch in Google Scholar
• 6 Cai WM, Nikoloff DM, Pan RM, de Leon J, Fanti P, Fairchild M. et al . CYP2D6 genetic variations in healthy adults and psychiatric African-American subjects: implications for clinical practice and genetic testing.  Pharmacogenomics J. 2006;  6 343-350
  PubMedSearch in Google Scholar
• 7 Candiotti KA, Birnbach DJ, Lubarsky DA, Nhuch F, Kamat A, Koch WH. et al . The impact of pharmacogenomics on postoperative nausea and vomiting: do CYP2D6 allele copy number and polymorphisms affect the success or failure of ondansetron prophylaxis?.  Anesthesiology. 2005;  102 543-549
  CrossrefPubMedSearch in Google Scholar
• 8 Cascorbi I, Gerloff T, Johne A, Meisel C, Hoffmeyer S, Schwab M. et al . Frequency of single nucleotide polymorphisms in the P-glycoprotein drug transporter MDR1 gene in white subjects.  Clin Pharmacol Ther. 2001;  69 169-174
  PubMedSearch in Google Scholar
• 9 Chou WH, Yan FX, Robbins-Weilert DK, Ryder TB, Liu WW, Perbost C. et al . Comparison of two CYP2D6 genotyping methods and assessments of genotype-phenotype relationships.  Clin Chemistry. 2003;  49 542-551
  PubMedSearch in Google Scholar
• 10 de Leon J. The AmpliChip CYP450 Test: personalized medicine has arrived in psychiatry.  Expert Rev Mol Diagn. 2006;  6 277-286
  CrossrefPubMedSearch in Google Scholar
• 11 de Leon J, Armstrong SC, Cozza KL. The dosing of atypical antipsychotics.  Psychosomatics. 2005;  46 262-273
  CrossrefPubMedSearch in Google Scholar
• 12 de Leon J, Bork JA. Risperidone and the cytochrome P4503A (letter).  J. Clin Psychiatry. 1997;  58 450
  PubMedSearch in Google Scholar
• 13 de Leon J, Susce MT, Murray-Carmichael E. The AmpliChip CYP450 genotyping test: integrating a new clinical tool.  Mol Diag Ther. 2006;  10 135-151
  PubMedSearch in Google Scholar
• 14 de Leon J, Susce MT, Pan RM, Fairchild M, Koch W, Wedlund PJ. The CYP2D6 poor metabolizer phenotype may be associated with risperidone adverse drug reactions and discontinuation.  J Clin Psychiatry. 2005;  66 15-27
  CrossrefPubMedSearch in Google Scholar
• 15 de Leon J, Susce MT, Pan RM, Fairchild M, Koch W, Wedlund PJ. Polymorphic variations in GSTM1, GSTT1, PgP, CYP2D6, CYP3A5, and dopamine D2 and D3 receptors and their association with tardive dyskinesia in severe mental illness.  J Clin Psychopharmacol. 2005;  25 448-456
  CrossrefPubMedSearch in Google Scholar
• 16 Ereshefsky L. Pharmacokinetics and drug interactions: update for new antipsychotics.  J Clin Psychiatry. 1996;  57 ((Suppl. 11)) 12-25
  PubMedSearch in Google Scholar
• 17 Fang J, Bourin M, Baker GB. Metabolism of risperidone to 9-hydroxyrisperidone by human cytochromes P450 2D6 and 3A4.  Naunyn Schmiedebergs Arch Pharmacol. 1999;  359 147-151
  PubMedSearch in Google Scholar
• 18 Hendset M, Haslemo T, Rudberg I, Refsum H, Molden E. The complexity of active metabolites in therapeutic drug monitoring of psychotropic drugs.  Pharmacopsychiatry. 2006;  39 121-127
  Thieme ConnectPubMedSearch in Google Scholar
• 19 Heykants J, Huang ML, Mannens G, Meudermans W, Snoeck E, Van Beijsterveldt L. et al . The pharmacokinetics of risperidone in humans: a summary.  J Clin Psychiatry. 1994;  55 ((Suppl. 5)) 13-17
  PubMedSearch in Google Scholar
• 20 Huang M, Van Peer A, Woestenborghs R, De Coster RD, Heykants J, Jansen AAI. et al . Pharmacokinetics of the novel antipsychotic agent risperidone and the prolactin response in healthy subjects.  Clin Pharmacol Ther. 1993;  54 257-268
  PubMedSearch in Google Scholar
• 21 Muñoz A, Xu J. Models for the incubation of AIDS and variations according to age and period.  Stat Med. 1996;  15 2459-2473
  CrossrefPubMedSearch in Google Scholar
• 22 Nakagami T, Yasui-Furukori N, Saito M, Tateishi T, Kaneo S. Effect of verapamil on pharmacokinetics and pharmacodynamics of risperidone: in vivo evidence of involvement of P-glycoprotein in risperidone disposition.  Clin Pharmacol Ther. 2005;  78 43-51
  CrossrefPubMedSearch in Google Scholar
• 23 Raedler TJ, Schreiner A, Naber D, Wiedemann K. Gender-specific effects in the treatment of acute schizophrenia with risperidone.  Pharmacopsychiatry. 2006;  39 171-174
  Thieme ConnectPubMedSearch in Google Scholar
• 24 Schwab M, Eichelbaum M, Fromm MF. Genetic polymorphisms of the human MDR1 drug transporter.  Annu Rev Pharmacol Toxicol. 2003;  43 285-307
  CrossrefPubMedSearch in Google Scholar
• 25 SPSS Inc .SPSS 9.0, Applications Guide. Chicago, IL: SPPS Inc 1999
  Search in Google Scholar
• 26 Wang JS, Ruan Y, Taylor RM, Donovan JL, Markowitz JS, DeVane CL. The brain entry of risperidone and 9-hydroxyrisperidone is greatly limited by P-glycoprotein.  Int J Neuropsychopharmacology. 2004;  7 415-419
  PubMedSearch in Google Scholar
• 27 Woestenborghs R, Lorreyne W, Van Rompaey F, Heykants J. Determination of risperidone and 9-hydroxyrisperidone in plasma, urine and animal tissues by high-performance liquid chromatography.  J Chromatogr. 1992;  583 223-230
  PubMedSearch in Google Scholar
• 28 Weber CC, Kressmann S, Ott M, Fricker G, Muller WE. Inhibition of P-glycoprotein function by several antidepressants may not contribute to clinical efficacy.  Pharmacopsychiatry. 2005;  38 293-300
  Thieme ConnectPubMedSearch in Google Scholar
• 29 Woodward M. Epidemiology. Study design and data analysis. Boca Raton, FL: Chapman and Hall/CRC 1999
• 30 Yasui-Fururoki N, Mihara K, Takahata T, Suzuki A, Nakagami T, De Vries R. et al . Effect of various factors on steady-plasma concentration of risperidone and 9-hydroxyrisperidone: lack of impact of MDR-1 genotypes.  Br J Clin Pharmacol. 2003;  57 569-575
  PubMedSearch in Google Scholar
• 31 Zheng H, Schuetz E, Zeevi A, Zhang J, McCurry K, Webber S. et al . Sequential analysis of tacrolimus dosing in adult lung transplant patients with ABCB1 haplotypes.  J Clin Pharmacol. 2005;  45 404-410
  PubMedSearch in Google Scholar

Correspondence

J. de LeonM.D. 

Mental Health Research Center at Eastern State Hospital

627 West Fourth St.

Lexington

KY 40508

Phone: +1/859/246 74 87

Fax: +1/859/246 70 19

Email: jdeleon@uky.edu

References

• 1 Aichhorn W, Weiss U, Marksteiner J, Kemmeler G, Walch T, Zernig G. et al . Influence of age and gender on risperidone plasma concentrations.  J Psychopharmacol. 2005;  19 395-401
  CrossrefPubMedSearch in Google Scholar
• 2 Anderson CB, True JE, Ereshefsky L, Miller AL, Peters BL, Velligan DI. Risperidone dose, plasma levels and response. 1993 New Research Program and Abstracts: American Psychiatric Association Annual Meeting. San Francisco CA 1993 (NR 217): 113
• 3 Barnhill J, Susce MT, Diaz FJ, de Leon J. Risperidone half-life in a patient taking paroxetine: a case report.  Pharmacopsychiatry. 2005;  38 223-225
  Thieme ConnectPubMedSearch in Google Scholar
• 4 Baumann P, Hiemke C, Ulrich S, Eckermann G, Gaertner I, Gerlach M. et al . The AGNP-TDM expert group consensus guidelines: therapeutic drug monitoring in psychiatry.  Pharmacopsychiatry. 2004;  37 243-256
  Thieme ConnectPubMedSearch in Google Scholar
• 5 Bork J, Rogers T, Wedlund P, de Leon J. A pilot study of risperidone metabolism: the role of cytochrome P450 2D6 and 3A.  J Clin Psychiatry. 1999;  60 469-476
  CrossrefPubMedSearch in Google Scholar
• 6 Cai WM, Nikoloff DM, Pan RM, de Leon J, Fanti P, Fairchild M. et al . CYP2D6 genetic variations in healthy adults and psychiatric African-American subjects: implications for clinical practice and genetic testing.  Pharmacogenomics J. 2006;  6 343-350
  PubMedSearch in Google Scholar
• 7 Candiotti KA, Birnbach DJ, Lubarsky DA, Nhuch F, Kamat A, Koch WH. et al . The impact of pharmacogenomics on postoperative nausea and vomiting: do CYP2D6 allele copy number and polymorphisms affect the success or failure of ondansetron prophylaxis?.  Anesthesiology. 2005;  102 543-549
  CrossrefPubMedSearch in Google Scholar
• 8 Cascorbi I, Gerloff T, Johne A, Meisel C, Hoffmeyer S, Schwab M. et al . Frequency of single nucleotide polymorphisms in the P-glycoprotein drug transporter MDR1 gene in white subjects.  Clin Pharmacol Ther. 2001;  69 169-174
  PubMedSearch in Google Scholar
• 9 Chou WH, Yan FX, Robbins-Weilert DK, Ryder TB, Liu WW, Perbost C. et al . Comparison of two CYP2D6 genotyping methods and assessments of genotype-phenotype relationships.  Clin Chemistry. 2003;  49 542-551
  PubMedSearch in Google Scholar
• 10 de Leon J. The AmpliChip CYP450 Test: personalized medicine has arrived in psychiatry.  Expert Rev Mol Diagn. 2006;  6 277-286
  CrossrefPubMedSearch in Google Scholar
• 11 de Leon J, Armstrong SC, Cozza KL. The dosing of atypical antipsychotics.  Psychosomatics. 2005;  46 262-273
  CrossrefPubMedSearch in Google Scholar
• 12 de Leon J, Bork JA. Risperidone and the cytochrome P4503A (letter).  J. Clin Psychiatry. 1997;  58 450
  PubMedSearch in Google Scholar
• 13 de Leon J, Susce MT, Murray-Carmichael E. The AmpliChip CYP450 genotyping test: integrating a new clinical tool.  Mol Diag Ther. 2006;  10 135-151
  PubMedSearch in Google Scholar
• 14 de Leon J, Susce MT, Pan RM, Fairchild M, Koch W, Wedlund PJ. The CYP2D6 poor metabolizer phenotype may be associated with risperidone adverse drug reactions and discontinuation.  J Clin Psychiatry. 2005;  66 15-27
  CrossrefPubMedSearch in Google Scholar
• 15 de Leon J, Susce MT, Pan RM, Fairchild M, Koch W, Wedlund PJ. Polymorphic variations in GSTM1, GSTT1, PgP, CYP2D6, CYP3A5, and dopamine D2 and D3 receptors and their association with tardive dyskinesia in severe mental illness.  J Clin Psychopharmacol. 2005;  25 448-456
  CrossrefPubMedSearch in Google Scholar
• 16 Ereshefsky L. Pharmacokinetics and drug interactions: update for new antipsychotics.  J Clin Psychiatry. 1996;  57 ((Suppl. 11)) 12-25
  PubMedSearch in Google Scholar
• 17 Fang J, Bourin M, Baker GB. Metabolism of risperidone to 9-hydroxyrisperidone by human cytochromes P450 2D6 and 3A4.  Naunyn Schmiedebergs Arch Pharmacol. 1999;  359 147-151
  PubMedSearch in Google Scholar
• 18 Hendset M, Haslemo T, Rudberg I, Refsum H, Molden E. The complexity of active metabolites in therapeutic drug monitoring of psychotropic drugs.  Pharmacopsychiatry. 2006;  39 121-127
  Thieme ConnectPubMedSearch in Google Scholar
• 19 Heykants J, Huang ML, Mannens G, Meudermans W, Snoeck E, Van Beijsterveldt L. et al . The pharmacokinetics of risperidone in humans: a summary.  J Clin Psychiatry. 1994;  55 ((Suppl. 5)) 13-17
  PubMedSearch in Google Scholar
• 20 Huang M, Van Peer A, Woestenborghs R, De Coster RD, Heykants J, Jansen AAI. et al . Pharmacokinetics of the novel antipsychotic agent risperidone and the prolactin response in healthy subjects.  Clin Pharmacol Ther. 1993;  54 257-268
  PubMedSearch in Google Scholar
• 21 Muñoz A, Xu J. Models for the incubation of AIDS and variations according to age and period.  Stat Med. 1996;  15 2459-2473
  CrossrefPubMedSearch in Google Scholar
• 22 Nakagami T, Yasui-Furukori N, Saito M, Tateishi T, Kaneo S. Effect of verapamil on pharmacokinetics and pharmacodynamics of risperidone: in vivo evidence of involvement of P-glycoprotein in risperidone disposition.  Clin Pharmacol Ther. 2005;  78 43-51
  CrossrefPubMedSearch in Google Scholar
• 23 Raedler TJ, Schreiner A, Naber D, Wiedemann K. Gender-specific effects in the treatment of acute schizophrenia with risperidone.  Pharmacopsychiatry. 2006;  39 171-174
  Thieme ConnectPubMedSearch in Google Scholar
• 24 Schwab M, Eichelbaum M, Fromm MF. Genetic polymorphisms of the human MDR1 drug transporter.  Annu Rev Pharmacol Toxicol. 2003;  43 285-307
  CrossrefPubMedSearch in Google Scholar
• 25 SPSS Inc .SPSS 9.0, Applications Guide. Chicago, IL: SPPS Inc 1999
  Search in Google Scholar
• 26 Wang JS, Ruan Y, Taylor RM, Donovan JL, Markowitz JS, DeVane CL. The brain entry of risperidone and 9-hydroxyrisperidone is greatly limited by P-glycoprotein.  Int J Neuropsychopharmacology. 2004;  7 415-419
  PubMedSearch in Google Scholar
• 27 Woestenborghs R, Lorreyne W, Van Rompaey F, Heykants J. Determination of risperidone and 9-hydroxyrisperidone in plasma, urine and animal tissues by high-performance liquid chromatography.  J Chromatogr. 1992;  583 223-230
  PubMedSearch in Google Scholar
• 28 Weber CC, Kressmann S, Ott M, Fricker G, Muller WE. Inhibition of P-glycoprotein function by several antidepressants may not contribute to clinical efficacy.  Pharmacopsychiatry. 2005;  38 293-300
  Thieme ConnectPubMedSearch in Google Scholar
• 29 Woodward M. Epidemiology. Study design and data analysis. Boca Raton, FL: Chapman and Hall/CRC 1999
• 30 Yasui-Fururoki N, Mihara K, Takahata T, Suzuki A, Nakagami T, De Vries R. et al . Effect of various factors on steady-plasma concentration of risperidone and 9-hydroxyrisperidone: lack of impact of MDR-1 genotypes.  Br J Clin Pharmacol. 2003;  57 569-575
  PubMedSearch in Google Scholar
• 31 Zheng H, Schuetz E, Zeevi A, Zhang J, McCurry K, Webber S. et al . Sequential analysis of tacrolimus dosing in adult lung transplant patients with ABCB1 haplotypes.  J Clin Pharmacol. 2005;  45 404-410
  PubMedSearch in Google Scholar

Correspondence

J. de LeonM.D. 

Mental Health Research Center at Eastern State Hospital

627 West Fourth St.

Lexington

KY 40508

Phone: +1/859/246 74 87

Fax: +1/859/246 70 19

Email: jdeleon@uky.edu