Hot Topic

Cytochrome P450 2D6


Receive notification when new Hot Topics are published:

Click CC to turn on closed captioning.

Published: October 2013

Print Record of Viewing

Several commonly prescribed drugs are metabolized by the CYP2D6 enzyme. Examples include codeine, tamoxifen, tramadol, and metoprolol. Variations in the CYP2D6 gene causes variations in the function of the enzyme, and therefore in drug metabolism. Dr. Black will explain how identification of specific alleles will assist clinicians in selection of appropriate drugs for each individual.

Presenter: John Logan Black, MD

  • Co-director of the Personalized Genomics Laboratory
  • Consultant in the Department of Laboratory Medicine and Pathology
  • Professor in the College of Medicine

Questions and Feedback

Contact us:


Download the PDF

I’m very excited to have an opportunity to talk with all of you about Cytochrome P450 2D6 and to describe the clinical usefulness of this gene and its associated enzyme as well as a new test that we’re going to bring up, which will markedly enhance our ability to provide accurate phenotypes.

I have a disclosure. I have licensed a pharmacogenomic treatment algorithm to AssureX through Mayo Medical Ventures. The information that is reported here has no relationship whatsoever to the license agreement.

As you view this presentation, the following information points regarding the new cytochrome P450 2D6 or CYP2D6 cascade test will be covered. We’ll talk about the clinical relevance of 2D6 or CYP2D6 testing. We’ll talk about test changes that are being implemented and the reasons for these changes. We’ll discuss that we are discontinuing and combining some of our CYP2D6 testing; and we want you to know that even though we’re doing that and we’re combining some tests, that the CYP2D6 cascade will still cover all indications that the discontinued testing used to cover such as tamoxifen, psychotropic drugs, and other drugs.

Well, what about CYP2D6? Why is it important? Well, it’s a member of a super-family of genes, and presently there are about 30 gene members in this super-family. CYP2D6 enzyme is clinically important because it’s one of the major phase I drug metabolizing enzymes. It’s particularly interesting to us in the Personalized Genomics Laboratory because its function is impacted by gene variability, so we have to genotype it to know its predicted phenotype.

CYP2D6 alleles are functionally diverse. Variability is based upon single nucleotide polymorphisms called SNPs, duplications of the gene, deletions of the gene, and recombinants with other CYP2D6 alleles and with the CYP2D7 pseudogene. A star (*) nomenclature is used and the website that is shown here actually links out to the main nomenclature homepage, which is kept by the Karolinska Institutet. And I’ll show you a couple of pages from that nomenclature webpage.

This is a little difficult to see; but on the left, you’ll see that the alleles are listed, and they are listed by star (*) nomenclature. In the middle, you see the nucleotide changes, which make up the haplotype for a particular star allele. You’ll see the effect, where known, under enzyme activity in vivo and in vitro; and finally, you’ll see references that support the claims for each of these star (*) alleles. There are many, many star (*) alleles. In fact, this is just the first page of many;

and the second page here ranges all the way down to CYP2D6*103. And this is one of the challenges when it comes to genotyping is that there are many of these; and understanding what the gene is doing and how these alleles are formed up at the CYP2D locus is important for accurate phenotype prediction, and we’ll get into that a little bit later.

This slide shows some of the CYP2D6 allele functional groups. The most commonly identified alleles are shown on this slide. We have the normal function allele, which is really the wild type or referenced sequence; it’s called CYP2D6*1. We have reduced-function alleles CYP2D6*2, *9, *, 10, *17, and *41 are examples of that. We have absent function in some alleles, so CYP2D6*3-*6 are examples of that. We have one allele which actually has increased function, and that is a CYP2D6*2A; and that’s by virtue of a promoter polymorphism, which turns on gene transcription. And there is a chance that an individual may have either duplications or even multiplication of active alleles, which can lead to increased function as well. Duplications and multiplications are denoted by the symbol “xN” following an allele where the xN actually represents the number of gene copies that are present; and with our new assay, we’ll actually be able to tell you this.

Allele frequency varies according to ethnicity. And so in Caucasians, the major alleles that you’ll see are CYP2D6*2, *2A, *3, *4, and *5. CYP2D6*4 is the most frequent deficiency allele in Caucasians, and it comes in at a rate of about 15% to 20%. In African individuals, the CYP2D6*17 allele virtually replaces the *4 allele and has a frequency that approaches 30% depending upon the region that this individual came from or their ethnicity is derived from; and this causes a reduced but not a deficient enzyme activity. There is some activity still associated with this allele. In Asians, the *10 allele virtually replaces the *4 or the *17 allele in populations noted above. It also has reduced activity and is highly prevalent with a rate approaching to 30%. And in Middle Eastern and Ethiopian individuals, gene duplications, many with increased function, are frequently found with a rate approaching 30% in some groups.

But what we do is we correlate genotype with a phenotype. And really, the clinician is interested in the phenotype that their patient is likely to express. Genetic results are reported as a diplotype, which includes 1 maternal allele and 1 paternally derived allele. So you’ll get a result when the testing is done that looks like this:  CYP2D6*1/*4, for example. So 1 allele is the wild-type allele; it’s normal; it’s a *1. And the other allele has no activity; it’s a null allele, and it’s *4. The phenotype is predicted based upon the number of functional, partially functional, and nonfunctional alleles present in the sample. An example of this would be the ultrarapid metabolizer has more than 2 functional alleles or 2 or more alleles of increased function.

An extensive metabolizer or a normal metabolizer—you’ll hear it both ways or see it both ways in the literature—has 2 normally functioning alleles or 1 normally functioning allele and 2 reduced-function alleles or 2 normally functioning alleles and a nonfunctional allele in the case of a sample with a duplication. An intermediate metabolizer has 1 normally functioning allele and a nonfunctional allele or 2 reduced-function alleles. A poor metabolizer has only nonfunctional alleles. In some cases, we can’t get to a categorical phenotype call such as extensive or intermediate or poor; and we have to give a range such as intermediate to poor metabolizer or, sometimes, an extensive to ultrarapid metabolizer.

Well, as I’ve mentioned, genotype predicts phenotype, and the prevalence of various metabolizer groups will vary by ethnic group as well. So, in Caucasians, we expect about a 5% to 10% of individuals to have a poor metabolizer phenotype. In people of African extraction, this would be 2%, and it’s only about 1% in Asians. The flip side of that is the prevalence of ultrarapid metabolizer also varies by ethnic group. In Caucasians, it’s about 1% to 2%. It’s <1% in Chinese and Japanese, and it’s >15% in people from the Middle East and from Ethiopian populations.

When we talk about CYP2D6 metabolism of medications, there are 2 broad categories. There are drugs that are activated by CYP2D6. These drugs are commonly called prodrugs; and examples of these would include tamoxifen, codeine, and tramadol. There are drugs that are inactivated by CYP2D6; and examples of these are fluoxetine, metoprolol, and many more.

We’ve actually developed a spreadsheet, which is visible in the link out from our reports, which will show you which medications are metabolized by CYP2D6 enzyme, and they are grouped into analgesics and alphabetized, and anesthetics, and psychotropics, and cardiovascular medications and so on. It’s interesting that the codeine, hydrocodone, oxycodone, and tramadol found under the analgesics section, are all prodrugs; and they are activated, to one extent or another, by being metabolized by CYP2D6. And similarly, the antineoplastic category contains tamoxifen, which is activated by CYP2D6, and we’ll discuss this in some detail in a few minutes. Most of the other medications are inactivated by CYP2D6, and it has different implications, again, as we’ll discuss in the future.

CYP2D6 can also be inhibited by medications  - the enzyme can - and here’s a list of medications that do that. We’ve categorized some as strong inhibitors, because they’re known to virtually shut down the enzyme; moderate inhibitors; and weak inhibitors; and the other ones that are listed there fall within the spectrum of inhibitor, but they are not noted to be very, very strong inhibitors for the most part. If you give somebody who is even a normal metabolizer an extensive metabolizer in other words, one of these inhibitors, you can make them behave as though they’re a poor metabolizer simply by the inhibition of the medication on CYP2D6 enzyme.

To complicate matters just a little bit more, there are drugs that will induce the CYP2D6 gene. Dexamethasone and rifampin are 2 examples of that.

I’d like to share a little information with you about some of the medications that are activated by CYP2D6, and we’ll focus on codeine, tramadol, and tamoxifen.

Codeine is activated by being metabolized to morphine by CYP2D6. The efficacy and safety of this analgesic is governed by CYP2D6 polymorphisms. Individuals who are poor metabolizers get little therapeutic effect from codeine because they can’t make morphine out of it, and codeine has essentially no analgesic activity by itself. Individuals who are ultrarapid metabolizers can generate too much morphine even at normal doses and essentially overdose from the medication.

CYP2D6 genotype, therefore, has a great effect on codeine metabolism and on the effectiveness of the medication. The oxidation of codeine to morphine via CYP2D6 is essential for its activity. Therefore, CYP2D6 genotype directly affects efficacy and toxicity. Poor metabolizers, for example, have decreased analgesia due to the lack of bioactivation to an active metabolite. Pharmacokinetic and pharmacodynamic studies show virtually undetectable morphine levels and insufficient analgesia as compared with extensive metabolizers. The frequency and intensity of adverse effects such as sedation, flushing, and dry mouth, actually don’t differ too much between poor and extensive metabolizers. What does differ is the analgesia, in that poor metabolizers will not get pain control and extensive metabolizers will.

In the case of ultrarapid metabolizers, pharmacokinetic studies show increased conversion of codeine to morphine compared to extensive or normal metabolizers, even at low doses, which can result in systemic toxicity. Severe or life-threatening toxicity can happen after normal doses. These have been well documented, and the adverse reactions include respiratory and circulatory depression, respiratory arrest, cardiac arrest and, unfortunately, children are particularly at risk. The FDA has actually issued guidance on this, and most centers are trying to avoid the use of codeine in children unless they know the genotype of that individual. Furthermore, morphine is secreted into breast milk of nursing mothers, and this can lead to neonatal toxicity and its severe adverse reactions including respiratory depression and respiratory arrest and death.

Tramadol is the next medication we’ll consider, and it is extensively metabolized via several pathways including CYP2D6-mediated oxidation to O-desmethyltramadol or ODT. And that has a 200-fold greater affinity for the mu receptor - the mu-opioid receptor - than tramadol does. ODT is principally responsible for opioid receptor-mediated analgesia that’s caused by an individual taking tramadol. And compared with 2D6 extensive metabolizers, poor metabolizers have lower values for the active metabolite; and prospective clinical trials have shown that poor metabolizers often fail to exhibit analgesia in response to tramadol.

Pharmacokinetic studies have also shown higher peak levels of ODT after a dose of tramadol as well as greater analgesia, stronger miosis, and more nausea in the case of ultrarapid metabolizers vs extensive metabolizers. So you can see both sides of the spectrum here—poor metabolizer does not get as many side effects and certainly doesn’t get analgesia; an ultrarapid metabolizer can have more side effects and greater analgesia. There’s one case report of respiratory depression in the 2D6 ultrarapid metabolizer with renal impairment after surgery. Now we believe this to be an exceedingly rare event, and we don’t mean to alarm physicians by this; but on the basis of that and other data as we quoted above, the use of analgesics other than tramadol may be preferable in the CYP2D6 poor metabolizer and also in an ultrarapid metabolizer.

The next medication we’ll consider is tamoxifen. It’s used as an adjunctive therapy in postmenopausal women with early stage breast cancer. It’s a selective estrogen receptor modulator that exerts its effect by binding to the estrogen receptor. Tamoxifen is metabolized by the cytochrome P450 enzyme system to active metabolites. Two enzymes (CYP3A4 and 2D6) are necessary to generate 4-hydroxy-N-desmythyl-tamoxifen, which is generically known as endoxifen and which is the most active metabolite of tamoxifen. 2D6 does the majority of this conversion to endoxifen. Endoxifen is approximately 100 times more potent in estrogen receptor binding than the parent drug, tamoxifen, or another metabolite known as N-desmethyl tamoxifen.

Tamoxifen, again, is metabolized at different rates depending upon the metabolizer status of an individual involved. An ultrarapid or an extensive metabolizer may benefit from tamoxifen therapy because they will convert it to endoxifen at a normal rate. Intermediate metabolizers have a nonstatistically significant trend towards increased risk of breast cancer recurrence on tamoxifen. But again, the research did not show a statistical difference, it showed a trend. Poor metabolizers, however, are at a significantly increased risk of breast cancer recurrence on tamoxifen. They do not convert tamoxifen to endoxifen at a sufficient rate to protect them from breast cancer recurrence.

There are several medications that are inactivated by CYP2D6. There are psychotropic drugs as are shown here and some cardiovascular medications as shown here. I want to go over 2 medications with you that are inactivated by CYP2D6.

The first one is nortriptyline. Poor metabolizers will achieve higher blood levels of the parent drug, nortriptyline, than extensive metabolizers will. This may lead to, effectively, an overdose on normal doses for poor metabolizers. Usually, a different drug not metabolized by CYP2D6 should be used in this situation, but nortriptyline can be used; it’s just that the patient needs to be more closely monitored for toxicity, and therapeutic drug monitoring should be done. At the other extreme, ultrarapid metabolizers will have lower blood levels of the parent drug than extensive metabolizers, and they may not respond to treatment because of that. It’s just difficult to achieve a therapeutic blood level for these patients. Usually, again, a different medication not metabolized by CYP2D6 should be used, or therapeutic drug monitoring should be done.

I’ll give you the example of pimozide also, which is often used for Tourette’s syndrome and some other psychiatric conditions. The drug label actually notes that CYP2D6 genotype issues exist and says that, at doses above 0.05 mg/kg/day, CYP2D6 genotyping should be performed. And if that patient is a poor metabolizer, the dose should not exceed 0.05 mg/kg/day, and dose changes should not be made earlier than 14 days.

Well this brings us to a discussion about CYP2D6 genotyping and why we are changing our test. This slide depicts the CYP2D locus. The first image (image A) shows a normal CYP2D6 locus arrangement. You have the orange CYP2D6 gene; and then to the left of that, you have CYP2D7 pseudogene; and then surrounding both the pseudogene and the gene, you have some repetitive elements, which are actually quite important to understanding how this gene recombines and detecting recombinant events. The second image (image B) shows the arrangement of the 2D locus when there is either a duplication or a multiplication. The part between the brackets here, here, and here actually will be multiplied in a multiplication situation. And you’ll notice that there is a section called REP DUP, which is a hybrid between the REP region normally found after 2D6 and the REP region found behind CYP2D7 pseudogene, and that’s where the recombinant event has allegedly happened. Finally, on (image) C, we have a deletion event. If you can have a duplication, you can certainly have a deletion of a gene, and so that’s occurred; and each individual with a deletion has this type of a signal or this type of a structure that is used to generate the deletion signal on our assays. In this diagram, you’ll see there are some dumbbells located underneath the gene structures. Those are actual amplicons that we generate in order to do our advanced testing when necessary, and you can see that there are several of them that we had to validate in order to do this next series of testing that I’ll describe. Above, you see some symbols, and these are the location of TaqMan assay probes that allow us to check for copy number variation across the gene. Well, you might think—well, why is this important? This looks relatively simple. One should be able to decipher this gene rather easily.

Well the fact is that on the next slide, you’ll see that we have problems with hybrid events occurring within CYP2D6 and 2D7, which give rise to what we call single or hybrid tandem structural arrangements, and these are difficult to test for. They have to be specifically amplified, and we’ve developed an assay that allows us to do that. To make matters worse, if you can have a CYP2D6-2D7 hybrid, you can certainly have the opposite, and that’s a CYP2D7-2D6 hybrid.

These also need to be detected. This one, in particular, is a bit of a problem because, under normal circumstances, we’ll actually get a duplication signal out of our assay and we’ll think—ah-a, this is fine; we have a duplication of this gene when in fact we don’t have a traditional duplication. We have a hybrid tandem and what is duplicated is part of 2D6, but it’s not enough to make a functional gene. So as a result of that, you don’t get increased function like you would expect. You won’t get, in most cases, an ultrarapid metabolizer if that’s what you were expecting, and we need to understand in more detail what this patient actually has in order to be able to predict phenotype.

So, we’ve developed an algorithm of testing that’s shown here. It’s a cascade and we start at the top. When the sample comes in and somebody orders the Cytochrome P450 2D6 Cascade, that’s our new test name—cascade, we will begin with the use of a Luminex kit, and this is the kit that we’ve used all along to do our genotyping. The good news is that about 94% of the samples that we receive can be genotyped just fine using the Luminex kit alone, but it’s the 6% that we cannot get a definite answer on using the Luminex kit, which we need to test further. So in those cases, we will actually proceed down the cascade to the next step, which is to do a TaqMan assay, which looks for copy number variation across the gene in 3 locations, and that will tell us, as a first step, whether we have, for example, a duplication or a multiplication event; and if that’s enough information, we’re done. We result the test and the results go back to the physician who ordered it. If this isn’t enough information, then we proceed down to the next level, which would be to do one or several of these Sanger sequencing-based assays. And that then allows us to give a definitive genotype in up to 99% of cases. What would trigger the cascade? Well, if we end up with a no-call on 1 of the single nucleotide polymorphisms in the Luminex kit, which means the kit failed to be able to detect accurately whether there is a variation there or not, then we have to do something else, and we’ll do this cascade. If there’s a series of single nucleotide polymorphisms that don’t yield a star (*) allele, we can’t recognize it as a star (*) allele, that’s a very good example of a recombinant event probably happening in that sampling, and we have to resolve that. If we have a very rare homozygous allele, say it’s a *14B/*14B, well we don’t see many *14Bs, but if we see a homozygous one, there’s a very high probability that it’s really not homozygous but that the Luminex kit failed to amplify 1 allele, and we have simply a heterozygous but we haven’t detected what’s on the other chromosome. If we have a duplication, we always have to proceed to the cascade. We have to do copy number variation if it will make a difference in the phenotype call. If we have a duplication of null alleles, well, you know, that isn’t going to matter. It’s still going to be a poor metabolizer sample. But if we have a duplication involving active alleles, we have to know if there’s a hybrid present upstream of the main allele and if that hybrid is active or not. So we have to proceed with copy number variation testing and then on to Sanger sequencing in most cases. Occasionally, we run into problems with genotype/phenotype discord. We’ll give a physician results. The physician will say—look, I have clear evidence that this patient, even though you say this patient is an extensive metabolizer, I’m quite certain that this patient is behaving like a poor metabolizer. Is there anything else we can do to find out what’s going on? Yes. In those circumstances, we would exercise the cascade. If the laboratory director has concerns about the result, and most of the time, we’re happy with what comes off of the Luminex and we don’t reflex further down the line, but, of course, that’s up to the laboratory director. And finally, if the client wants to have copy number variation assay done or additional testing done, we would certainly run down the cascade with that.

So our CYP2D6 testing offered will change. We’re going to implement the CYP2D6 cascade for blood - only for blood because this requires very high-quality DNA, and we’re simply not able to get high enough quality and high enough quantity DNA from saliva. So, we can only run this on blood. So, that will be available - should start first part of September of this year. We will continue to do the CYP2D6 saliva test, however, because clearly in some physician’s offices, this is the best way to get a sample to us. And as I’ve said, that’s fine; in 94% of cases, we can provide you with a fine genotype that will tell you what to predict for phenotype. If we run into problems with a saliva sample, we will indicate that you may wish to order up the 2D6 cascade and send us a blood sample. We’re going to stop offering a couple of tests. We used to offer the CYP2D6 genotype-tamoxifen therapy test for both blood and saliva. We’ve come to know that we can’t offer a specific gene test for every drug. We could test 2D6 for codeine. We could test 2D6 from tramadol. We could test 2D6 for nortriptyline, but it becomes inefficient; and we’d have to make a new test build for each and every one of those, and it’s a lot of work. So, we’ve decided to collate all of our 2D6 testing together in the cascade test and the saliva test. But don’t worry! The comments are written for tamoxifen in all of our 2D6 tests. So if that’s what you’re used to ordering this test for, you’ll still get high-quality results that will tell you what to expect from your patient; and we’re also adding some specific comments for codeine and tramadol and other medications that happen to be inactivated by CYP2D6.

Here’s an example of a report for a poor metabolizer. You’ll notice that the first paragraph says that prodrugs are converted to their active metabolite at a much reduced rate, if at all, which is expected to reduce efficacy. Alternatively, drugs that are inactivated by CYP2D6 are metabolized at a much reduced rate, if at all, which may lead to increased side effects. Then we have a specific comment for tamoxifen. In this instance, you would be careful with using tamoxifen in a postmenopausal woman who is receiving tamoxifen for breast cancer adjuvant therapy because she is less likely to respond, and you may wish to treat with some other medication such as an aromatase inhibitor. For codeine and tramadol, the same situation exists. These drugs are not likely to be helpful in controlling pain for these patients who are poor metabolizers, so we make comments along those lines. The final paragraph states that drugs that are inactivated by CYP2D6 are inactivated at a much reduced rate, if at all, in alternative medications not metabolized by CYP2D6 or should be used or therapeutic drug monitoring should be considered.

This is a description of the cascade again. We start with a PCR allele-specific primer extension assay followed by bead hybridization with fluorescent detection that is a Luminex-based kit. If needed, we’ll then reflex to a Taqman assay for copy number variation. And if that does not provide sufficient information, we proceed to Sanger sequencing. The specimen type is a 3 mL whole blood EDTA sample, and the test classification is that this is a laboratory-developed test.

This test is useful for determination of genotype for phenotype prediction. It is used to aid in clinical decision making for drug selection of CYP2D6-metablized medications including codeine and tramadol, tamoxifen, psychotropics as listed, and cardiovascular medications. An interpretive report is provided, but no dosing recommendations are given.

Some cautions: There are limitations to every test, and this one has some as well. It won’t detect all possible variations of the gene, although we believe we’ll detect 99% of them. An absence of a detectable variation doesn’t rule out the possibility that one is present. Regardless of the predicted phenotype, use of inhibitors of 2D6 can change the actual function of the enzyme, so that needs to be kept in mind. Drug selection, of course, is left to the discretion of the care provider, and we will not make specific recommendations on what drug to use. And the report will include other cautions, so always make sure that you read the entire report.

If you have questions or requests regarding this topic, please feel free to email us at the address shown; and for more information, please visit or call Mayo Laboratory Inquiry at 800-533-1710. Thank you for your attention, and I look forward to your questions.