Have we successfully implemented CYP2D6 genotyping in psychiatry?

KEYWORDS:

• Biological markers
• cytochrome P-450 enzyme system/genetics
• drug monitoring
• genotype
• mental disorders/drug therapy/genetics
• pharmacogenetics
• psychotropic drugs/therapeutic use
• United States Food and Drug Administration

1. Introduction

‘Have we successfully implemented CYP2D6 genotyping in psychiatry?’ This editorial describes the evidence that we have not. Following the introduction, six brief sections identify the 20-year journey of the author’s involvement in CYP2D6 genotyping in psychiatry; recent advances in the field with the development of guidelines for psychiatric drugs; three areas of difficulty, namely the laboratory, the clinician, and the patient; and finally, the seventh section opines on the current and future states of CYP2D6 genotyping.

Treating psychiatric patients is complex; it cannot be described in a simple marketing message of 140 characters, many psychiatric patients do not take their medication, and psychiatric science is relatively underdeveloped. Any reader willing to deal with this level of complexity and wanting to learn from the mistakes of this author’s 20-year journey may access the Supplementary Material which includes three boxes on (1) his research on CYP2D6 genotyping, (2) his views on CYP2D6 pharmacology, and (3) his opinions of pharmacogenetic tests.

2. Is it possible that after 20 years of supporting this field the author has the guts to answer ‘no’?

In the mid-1990s, the author became aware that he and other psychiatrists were not paying attention to the polymorphic variations of the CYP2D6 gene which lead to two main phenotypes, poor metabolizers (PMs) with no CYP2D6 and ultrarapid metabolizers (UMs) with too much CYP2D6 enzyme. Since then, pharmacogenetic testing has gone through three stages: (1) the 1990s, ‘the fear stage’; (2) the 2000s, ‘the failure stage’; and (3) the 2010s, ‘the hype stage’ [1]. Supplementary Box 1 describes more of the author’s CYP2D6 genotyping journey, including (1) a 1996 pilot study of 100 psychiatric patients, using the rudimentary genetic technology of the time [2]; (2) an extension of the pilot study [3] conducted by re-genotyping the samples with the Affymetrix GeneChip and exploring its cost-effectiveness, leading to the hypothesis that treatment for CYP2D6 PMs and UMs costs more when using first-generation psychiatric drugs; (3) the 2006 US FDA approval of the AmpliChip CYP450 test [4]; (4) the study genotyping 4500 patients for CYP2D6 and CYP2C19 using this test between 2003 and 2005 [5] when the testing was no longer clinically relevant for most second-generation psychiatric drugs; (5) the 2008 literature review of the first wave of pharmacogenetic testing in psychiatry, which failed [6]; and finally, (6) the current frustration with the second-wave pharmacogenetic tests, which were based on ‘hype’ rather than science [1].

3. Recent advances in CYP2D6 genotyping guidelines in psychiatry

A group of pharmacists, scientists, and physicians, the Clinical Pharmacogenetics Implementation Consortium (CPIC), developed CYP2D6 genotyping guidelines for tricyclic antidepressants (TCAs) [7] and selective serotonin reuptake inhibitors (SSRIs) [8]. Unfortunately, fluoxetine and paroxetine, the two CYP2D6 SSRIs, are not ideal drugs for using CYP2D6 genotyping, since their CYP2D6 inhibition may somewhat mask the effects of PM status (Supplementary Box 2). Other CYP2D6 drugs not yet reviewed by the CPIC (Supplementary Box 2) are two newer antidepressants (venlafaxine and vortioxetine), four second-generation antipsychotics (aripiprazole, brexpiprazole, iloperidone, and risperidone), and atomoxetine [9]. Atomoxetine is a drug for treating attention deficit hyperactivity disorder and, after reviewing its limited literature published by the pharmaceutical company, the author [10] proposed that atomoxetine is a wide-window therapeutic drug and that the dosages approved by the FDA were definitively too low for CYP2D6 UMs (approximately 1.5% of the US population) and probably too low for normal subjects with two active CYP2D6 alleles (approximately one third of the US population).

4. Difficulties with laboratories

A decade ago, US scientists agreed that for appropriately marketing genetic tests in the clinical environment, it is necessary to first establish their analytical validity, clinical validity, and clinical utility [1]. As there is a legal vacuum because the FDA cannot regulate diagnostic tests, any US lab can use a genetic test if that lab meets minimal requirements for analytical validity [1]. Therefore, many commercial companies have introduced pharmacogenetic panels testing multiple genes for psychiatric use in the US and other countries without demonstrating their clinical validity and utility.

CYP2D6 is a complex, highly polymorphic gene, has pseudogenes, and is known to form hybrid gene structures with CYP2D7 [11]. Most of the 100 CYP2D6 alleles described are very rare, while the prevalences of the frequent ones are influenced by patient ancestry [12]. Therefore, it is not surprising that there is no agreement in the literature concerning how many CYP2D6 alleles should be included in a commercial test. In a recent comprehensive review, Bousman et al. [13] identified 20 pharmacogenetic panels that include CYP2D6 genotyping from various countries. Compared with the no-longer-available AmpliChip CYP450 which included 27 alleles [5], many of these 20 tests appeared unsatisfactory because they only include a mean of 20 CYP2D6 alleles (range 12–36). CYP2D6 PM genotyping ideally needs to test for all relatively frequent CYP2D6 null alleles. The AmpliChip CYP450 test [5] included 13 null alleles while only 3 null alleles (*3, *4, and *6) were included in all 20 commercial tests [13]. In our sample with 4265 patients, the AmpliChip CYP450 test identified 320 CYP2D6 PMs [5] while a laboratory testing only these three null alleles would find 257 PMs and miss 63 PMs (or 20%, 63/320). Three-quarters of the 20 commercial tests [13] included seven null alleles (*3, *4, *5, *6, *7, *8, and *12). In our sample, a laboratory testing only seven null alleles would find 312 PMs and miss 8 PMs (or 2.5%, 8/320). To identify CYP2D6 UMs, the AmpliChip CYP450 distinguished between duplications/multiplications of active (*1xn, *2xn and *35xn) versus deficient (*10xn, *17xn, *29xn and *41xn) or null alleles (*4xn and *6xn), but the newer tests only identify duplicate alleles with no distinction concerning their activity [13], potentially misclassifying CYP2D6 UMs. In the 4265 patients [5], the AmpliChip CYP450 test identified 65 UMs and another 25 patients with duplications from deficient or null alleles (including 2 PMs) that would be misclassified as UMs. In summary, some but not all of these CYP2D6 tests are reasonable for identifying PMs, but they are very deficient regarding UMs.

5. Difficulties with clinicians

Psychiatrists trained prior to or during the 1980s have difficulty venturing into the ‘brave new world’ of CYP2D6 genotyping, but are familiar with TCAs. Moreover, they are usually comfortable using therapeutic drug monitoring (TDM) to avoid TCA dose-dependent adverse drug reactions including potentially lethal arrhythmias. Psychiatrists trained more recently are more open (some too open) to pharmacogenetic testing in psychiatry, but frequently have no expertise in TCAs, which are the ideal psychiatric drugs for self-training in CYP2D6 genotyping. These younger psychiatrists rely more on SSRIs for antidepressant therapy.

6. Difficulties with web-empowered but ill-informed patients and families

The marketing of the various pharmacogenetic panels in psychiatry has led many of them to summarize gene variants using the traffic light patterns of green, yellow, and red to classify drugs [1]. Empowered patients and families searching the Internet can easily become convinced that these pharmacogenetic panels can provide the right drug (‘green’) for a specific patient; however, none of these webpages explain that drug response is not only controlled by genes but also by environment (other drugs) and personal variables [1].

7. Expert opinion

Based on his clinical experience, research studies, and literature reviews, the author concludes that, in theory, CYP2D6 genotyping is (1) very beneficial for TCAs, aripiprazole, risperidone, and venlafaxine, and (2) possibly beneficial for fluoxetine, fluvoxamine, paroxetine, vortioxetine, brexpiprazole, haloperidol, iloperidone, and atomoxetine (Supplementary Box 2). The answer ‘no’ to the title question is explained by the difficulties in implementing CYP2D6 genotyping in the clinical environment in the US. None of the commercial laboratories reviewed by Bousman et al. [13] and none of the commercial laboratories that the author knows appear suitable to him for recommending to US clinicians as a cheap and reliable laboratory for CYP2D6 genotyping. The practical difficulties of selecting a laboratory for pharmacogenetics testing have recently been reviewed [14].

Because the use of inhibitors, along with personal variables, may mimic PM or UM phenotypes, CYP2D6 genotyping should be interpreted in combination with TDM. Finally, CYP2D6 is a single gene; it is not unreasonable to think that in the future CYP2D6 may become one of the multiple elements forming a complex system for selecting psychiatric drugs [15]. In the view of the author (Supplementary Box 3), it is not easy to transition from this current unsatisfactory level of CYP2D6 genotyping to a future of personalized prescription [15]. The main limitation is the lack of regulation in the US and other countries; this deficit allows the marketing of non-validated pharmacogenetic panels, including unsatisfactory versions of CYP2D6 genotyping. This could lead to skepticism if pharmacogenetic panels with better proven clinical utility in psychiatry become available in the future.

During this uneasy transition, the author plans to continue to (1) encourage pharmacists with expertise in the science of CYP2D6 to get more involved in helping clinicians with CYP2D6 genotyping, (2) educate skeptical psychiatrists and other prescribers about the limited benefits of CYP2D6 genotyping in psychiatry, and (3) educate those (patients, families or some prescribers) who have been brainwashed by commercial marketing of the lack of utility of the large pharmacogenetic test panels currently used in psychiatry.

Declaration of interest

J de Leon has received researcher-initiated grants from Eli Lilly (one ended in 2003 and the other, as co-investigator, ended in 2007); from Roche Molecular Systems, Inc. (ended in 2007); and, in a collaboration with Genomas, Inc., from the NIH Small Business Innovation Research program (ended in 2010). He was on the advisory boards of Bristol-Myers Squibb (2003/04) and AstraZeneca (2003). Roche Molecular Systems supported one of his educational presentations, which was published in a peer-reviewed journal (2005). His lectures have been supported once by Sandoz (1997), twice by Lundbeck (1999 and 1999), twice by Pfizer (2001 and 2001), three times by Eli Lilly (2003, 2006, and 2006), twice by Janssen (2000 and 2006), once by Bristol-Myers Squibb (2006), and seven times by Roche Molecular Systems, Inc. (once in 2005 and six times in 2006). The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Acknowledgments

The author acknowledges Lorraine Maw at the Mental Health Research Center at Eastern State Hospital, Lexington, KY, USA, who helped in editing the text and the reviewers who helped him improve this editorial.

Supplementary Material

Supplemental data for this article can be accessed here.

Additional information

Funding

This paper was not funded.