Volume 89, Issue 5 p. 662-673
Reports
Free Access

Pharmacogenetics: From Bench to Byte— An Update of Guidelines

JJ Swen

JJ Swen

Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Leiden, The Netherlands

Search for more papers by this author
M Nijenhuis

M Nijenhuis

Division Drug Information Centre, KNMP, The Hague, The Netherlands

Search for more papers by this author
A de Boer

A de Boer

Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands

Search for more papers by this author
L Grandia

L Grandia

Division Drug Information Centre, KNMP, The Hague, The Netherlands

Search for more papers by this author
AH Maitland-van der Zee

AH Maitland-van der Zee

Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands

Search for more papers by this author
H Mulder

H Mulder

Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands

Department of Clinical Pharmacy, Wilhelmina Hospital Assen, Assen, The Netherlands

Search for more papers by this author
GAPJM Rongen

GAPJM Rongen

Department of Pharmacology - Toxicology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands

Department of General Internal Medicine, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands

Nijmegen Centre for Evidence Based Practice, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands

Search for more papers by this author
RHN van Schaik

RHN van Schaik

Department of Clinical Chemistry, Erasmus University Medical Centre, Rotterdam, The Netherlands

Search for more papers by this author
T Schalekamp

T Schalekamp

Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands

Search for more papers by this author
DJ Touw

DJ Touw

Central Hospital Pharmacy, The Hague, The Netherlands

Search for more papers by this author
J van der Weide

J van der Weide

Department of Clinical Chemistry, St Jansdal Hospital, Harderwijk, The Netherlands

Search for more papers by this author
B Wilffert

B Wilffert

Department of Quality and Patient Safety, Zorggroep Noorderbreedte, Leeuwarden, The Netherlands

Search for more papers by this author
VHM Deneer

VHM Deneer

Department of Clinical Pharmacy, St Antonius Hospital, Nieuwegein, The Netherlands

Search for more papers by this author
H-J Guchelaar

Corresponding Author

H-J Guchelaar

Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Leiden, The Netherlands

Search for more papers by this author
First published: 16 March 2011
Citations: 116

Abstract

Currently, there are very few guidelines linking the results of pharmacogenetic tests to specific therapeutic recommendations. Therefore, the Royal Dutch Association for the Advancement of Pharmacy established the Pharmacogenetics Working Group with the objective of developing pharmacogenetics-based therapeutic (dose) recommendations. After systematic review of the literature, recommendations were developed for 53 drugs associated with genes coding for CYP2D6, CYP2C19, CYP2C9, thiopurine-S-methyltransferase (TPMT), dihydropyrimidine dehydrogenase (DPD), vitamin K epoxide reductase (VKORC1), uridine diphosphate glucuronosyltransferase 1A1 (UGT1A1), HLA-B44, HLA-B*5701, CYP3A5, and factor V Leiden (FVL).

Clinical Pharmacology & Therapeutics (2011) 89 5, 662–673. doi:10.1038/clpt.2011.34

In recent years, there has been substantial progress in the field of pharmacogenetics. The number of publications on the subject has risen sharply, and the results of the first randomized clinical trial showing that pharmacogenetics can be used to prevent adverse drug events have been published.1 Meanwhile, an increasing number of pharmacogenetic tests are becoming available.2 However, despite US Food and Drug Administration–approved modifications to more than 30 drug labels to include pharmacogenetic information,3 guidelines that link the result of a pharmacogenetic test to specific dose recommendations are sparse. Therefore, the Royal Dutch Association for the Advancement of Pharmacy established the Pharmacogenetics Working Group with the objectives of developing pharmacogenetics-based therapeutic (dose) recommendations based on systematic review of the literature and assisting physicians and pharmacists by integrating the recommendations into computerized systems for drug prescription, dispensing, and automated medication surveillance. The initial results for 85 genotype/phenotype–drug combinations, comprising 26 drugs, were published in this journal.4 Here we present recommendations for 27 newly assessed drugs and updates of the existing monographs.

RESULTS

To date, we have compiled therapeutic (dose) recommendations for 163 genotype/phenotype–drug combinations comprising 53 drugs and 11 genes (1; the table's references are provided in the Supplementary References online). The drugs were associated with genes coding for CYP2D6 (n = 25), CYP2C19 (n = 11), CYP2C9 (n = 7), thiopurine-S-methyltransferase (TPMT) (n = 3), dihydropyrimidine dehydrogenase (DPD) (n = 3), vitamin K epoxide reductase (VKORC1) (n = 2), uridine diphosphate glucuronosyltransferase-1A1 (UGT1A1), HLA-B44, HLA-B*5701, CYP3A5, and factor V Leiden (FVL) (all n = 1). Therapeutic (dose) recommendations were formulated for 39 (73.6%) of the drugs. For clozapine, flupenthixol, and olanzapine, a gene–drug interaction with CYP2D6 was considered, but no evidence was found in the literature, and hence no recommendations were required. For 11 of the drugs (20.8%), a gene–drug interaction was present, but no therapeutic (dose) recommendation was deemed necessary.

Table 1. Results for CYP2D6, CYP2C9, CYP2C19, UGT1A1, TPMT, HLA-B44, HLA-B*5701, CYP3A5, VKORC1, factor V Leiden, and DPYD
image

The quality of the retrieved data was scored as category 4 (published controlled studies of “good” quality; see Supplementary Table S1 online for quality criteria) for 49.1% of the data and category 3 (published controlled studies of “moderate” quality) for 37.4%. For 59 (36.2%) of the genotype/phenotype–drug combinations, the clinical relevance of the interaction was classified as category C (long-standing discomfort (48–168 h) without permanent injury) or higher (see Supplementary Table S2 online for details).

CYP2D6

For CYP2D6 poor metabolizers (PMs), defined as patients carrying two defective alleles, dose reductions are recommended for clomipramine, flecainide, haloperidol, zuclopenthixol (all 50%); doxepin, nortriptyline (both 60%); imipramine, propafenone (both 70%); and metoprolol (75%). There were insufficient data to calculate dose adjustments for amitriptyline, oxycodone, risperidone, and venlafaxine. With respect to tamoxifen, an increased risk for breast cancer relapse is present, and it is advised that an aromatase inhibitor be considered for treating postmenopausal women with breast cancer. Other recommendations included the selection of an alternative drug, therapeutic drug monitoring, increased alertness to adverse drug events and to reduced efficacy, and the recording of an electrocardiogram.

For CYP2D6 intermediate metabolizers (IMs), defined as patients carrying two decreased-activity alleles or one active/decreased-activity allele and one inactive allele, dose reductions ranging from 20 to 50% are advised for doxepin, amitriptyline, zuclopenthixol, imipramine, nortriptyline, and metoprolol. There were insufficient data to calculate dose adjustments for clomipramine, oxycodone, propafenone, risperidone, and venlafaxine. For tamoxifen, the use of an aromatase inhibitor for treating postmenopausal women with breast cancer and the avoidance of concomitant use of a CYP2D6 inhibitor are advised. Other recommendations are comparable to the recommendations for PMs.

For CYP2D6 ultrarapid metabolizers (UMs), defined as patients carrying a gene duplication in the absence of inactive or decreased-activity alleles, dose adjustments ranging from 30 to 150% are recommended for doxepin, imipramine, metoprolol, nortriptyline, tramadol, and venlafaxine. For eight of the assessed gene–drug combinations, there were insufficient data to calculate dose adjustments. The metabolic capacity of UMs shows a considerable variability due to the range of gene copy numbers possible within the definition of UM. Also, the impact of the increased concentrations of drug metabolites to which UMs are exposed is often unknown. Therefore, the selection of an alternative drug is frequently advised.

CYP2C9

Seven CYP2C9 substrates were assessed. For phenytoin, dose reductions of 25% (*1/*2, *1/*3) and 50% (*2/*2, *2/*3, *3/*3) are recommended. For acenocoumarol and phenprocoumon, although clinically relevant gene–drug interactions are present, no dose adjustment is recommended because of strict international normalized ratio monitoring by the Dutch Thrombosis Service.5 The need for adjustment of the initial dose is currently under investigation.6 In addition to the CYP2C9 genotype, the VKORC1 genotype is an important determinant of coumarin response. Therefore, the status of both CYP2C9 and VKORC1 should be considered when identifying candidates for intensified international normalized ratio monitoring. Despite a clear pharmacokinetic effect of the gene–drug interaction, no recommendations were formulated for any of the sulfonylureas; the absolute risk for hypoglycemia is low, and the dose is titrated in response to plasma levels of glucose/glycosylated hemoglobin.

CYP2C19

The number of CYP2C19 substrates assessed increased from 1 to 12, and the CYP2C19*17 allele (resulting in UMs) was added. Recommendations have been made with respect to all drugs except moclobemide and rabeprazole. Several articles have reported that the use of proton pump inhibitors results in better clinical efficacy in PMs and IMs as compared to extensive metabolizers. These results were scored as clinical relevance category AA# (AA: no statistically significant kinetic or clinical effect; “#” indicates a positive effect). Because of the risk of undertreatment, dose increases ranging from 50 to 400% are advised for UMs who are receiving treatment with proton pump inhibitors. In the case of voriconazole, because of its nonlinear pharmacokinetics, no dose adjustment is recommended.

UGT1A1

The UGT1A1*28 allele is associated with irinotecan toxicity. Although results are not consistent, there is sufficient evidence that a reduction in the initial dose by 30% is required for regimens containing >250 mg/m2 of irinotecan prescribed to homozygous carriers of the UGT1A1*28 allele. This is in agreement with the Food and Drug Administration–mandated label change. No dose reduction is recommended for heterozygous carriers of the UGT1A1*28 allele because dose reduction might result in undertreatment.

TPMT

TPMT catalyzes the S-methylation of the thiopurine drugs 6-mercaptopurine, azathioprine, and thioguanine. Selection of an alternative drug is advised for IMs and PMs. If this is not possible, the dose should be reduced by 50 and 90%, respectively. The data for thioguanine were insufficient for calculating dose adjustments.

HLA-B44

There was some evidence that HLA-B44-negative patients show less response to treatment with ribavirine. However, given that ~90% of the population is HLA-B44-negative and that no alternative treatment is available, no action is advised.

HLA-B*5701

To date, the association between HLA-B*5701 genotype and the hypersensitivity reaction to abacavir remains the only example of a randomized clinical trial of pharmacogenetics. The advice regarding selection of an alternative drug for treating HLA-B*5701-positive patients is in agreement with the recommendations of the Food and Drug Administration and the European Medicines Agency.

CYP3A5

Because of the large number of publications, studies limited to healthy volunteers, pharmacokinetic end points, or liver transplantations were excluded. Although an interaction between CYP3A5 genotype and tacrolimus metabolism exists, no action is advised because in Dutch transplantation hospitals the tacrolimus dose is titrated in response to therapeutic drug monitoring.

VKORC1

The VKORC1 genotype appears to contribute more to the variability in coumarin dose requirements than the CYP2C9 genotype does. The presence of the VKORC1 C1173T polymorphism results in a decrease in dose requirements of acenocoumarol and phenprocoumon. However, for reasons identical to those related to the coumarin–CYP2C9 interaction, it was decided not to advise a dose reduction.

FVL

Patients with a positive (family) history of thrombotic events, and who are also carriers of the FVL allele, are advised to avoid the use of estrogen-containing oral contraceptives.

DPYD

Three DPD substrates were evaluated: 5-fluorouracil, its oral prodrug capecitabine, and tegafur. Selection of an alternative drug is advised for PMs, defined as homozygous carriers of a nonfunctional allele. For IMs, defined as heterozygous carriers of a nonfunctional allele, a dose reduction of 50% is advised for 5-fluorouracil and capecitabine.

DISCUSSION

We have developed pharmacogenetics-based therapeutic (dose) recommendations for 163 genotype/phenotype–drug combinations comprising 53 drugs and 11 genes. These recommendations include updates on the 26 existing therapeutic (dose) recommendations as well as recommendations for 27 new gene–drug combinations. The recommendations issued since October 2006 are available through most automated drug prescription, dispensing, and medication surveillance systems in the Netherlands.

The Pharmacogenetics Working Group initiative is not the first to develop guidelines with pharmacogenetics-based dose recommendations. A 2001 paper on CYP2D6 phenotype–based dose recommendations for antidepressants represents an early step.7 A more recent example is the inclusion of pharmacogenetic information in coumarin dosing algorithms.6,8 Furthermore, several groups have developed databases that are devoted to disseminating knowledge in the area of pharmacogenetics, e.g., PharmGKB (http://www.pharmgkb.org/). However, our recommendations are the first to be available nationwide during the process of drug prescribing and dispensing.

Our approach has some limitations, though. First, pharmacogenetics was not the primary objective for most of the studies we assessed; therefore, many of the studies were underpowered, with insufficient sample size per genotype or phenotype. Second, the end points assessed were often pharmacokinetic ones and the result of single-dose experiments in healthy volunteers—not representative of the conditions in daily clinical practice. However, since our previous report, the number of studies with pharmacogenetics as the primary objective has increased significantly.4

In our opinion, there is currently only limited evidence to justify population-wide prospective pharmacogenetic screening. A pharmacogenetic test prior to drug prescription is obligatory only for trastuzumab. Yet there are indications that patients with a non-wild-type genotype may be at increased risk for an aberrant drug response. Therefore, we formulated recommendations for patients with a previously determined genotype. In current clinical practice, the number of such patients is limited and consists mainly of subjects who were genotyped after unexplained adverse drug events or lack of response to “normal” drug dose. However, with the continuous decline in the costs of pharmacogenetic tests and the increasing number of laboratories with genotyping infrastructure, this number is bound to increase.

The recommendations of the Pharmacogenetics Working Group focus on the combination of a single gene with a single drug. However, the predictive value of a single genetic variant with regard to drug response is often limited, and combinations of multiple genetic variants may be involved. For example, only 5–18% and 15–37% of the variation in warfarin dose requirements are explained by CYP2C9 and VKORC1 genotypes, respectively.9,10,11,12,13 Models that combine information on both genetic and nongenetic factors are able to explain up to 50% of the variation in warfarin dose requirements.8 The formulation of recommendations that consider combinations of multiple genes presents a significant challenge for the future, given that very large study populations will be required to gather significant numbers of patients with combinations of rare genotypes. A second challenge is the integration of gene–drug and drug–drug interactions. To date, drug–drug interactions have been considered characteristic only of the drugs involved. However, in the light of current knowledge of pharmacogenetics, this might no longer be valid. For example, the interaction between a CYP2D6 inhibitor and a CYP2D6 substrate requires different management for CYP2D6 IMs than for CYP2D6 PMs. Therefore, the combination of gene–drug and drug–drug interactions may have major implications for drug prescribing and dispensing. Research in this field is only starting to evolve.14

In conclusion, we have developed pharmacogenetics-based therapeutic (dose) recommendations for 53 drugs. The recommendations are available nationwide during the process of drug prescribing and dispensing. We believe that the availability of the therapeutic (dose) recommendations during the process of therapeutic decision making represents an important step in the clinical use of pharmacogenetic information.

METHODS

A detailed description of the methods used for data collection, data assessment, and preparation of gene–drug monographs has previously been provided in this journal.4 In brief, a list of genetic polymorphisms affecting pharmacokinetics and pharmacodynamics, including an overview of drug substrates, was compiled. For each drug, a systematic search of the literature was performed. Review articles and studies involving nonhuman subjects and in vitro experiments were excluded. Each gene–drug interaction was scored on two parameters. First, the quality of evidence for the gene–drug interaction was scored on a five-point scale ranging from 0 (lowest evidence) to 4 (highest evidence) (Supplementary Table S1). Population size was not included as a parameter for assessing the quality of evidence, but dose adjustments were calculated as the population size–weighted mean. Second, the clinical relevance of the potential gene–drug interaction was scored on a seven-point scale ranging from AA (lowest impact) to F (highest impact) (Supplementary Table S2). For each gene–drug interaction, a risk analysis containing a review of the selected articles, their assigned levels of evidence and clinical relevance, and a therapeutic (dose) recommendation were compiled. Recommendations included those related to dose adjustments as well as advice on therapeutic strategy (e.g., therapeutic drug monitoring, selection of alternative drugs, and warning for adverse drug events).

SUPPLEMENTARY MATERIAL is linked to the online version of the paper at http://www.nature.com/cpt

ACKNOWLEDGMENTS

No additional funding was received. We thank Jean Conemans and Ingeborg Wilting for their valuable contributions as former members of the Pharmacogenetics Working Group.

CONFLICT OF INTEREST

The authors declared no conflict of interest.