Pharmacogenomics in psychiatry – the challenge of cytochrome P450 enzyme phenoconversion and solutions to assist precision dosing

Pharmacogenomic (PGx) testing to assist precision dosing has increased significantly over the past decade. Facilitating factors for PGx include the increased availability and affordability of genotyping, the development of clinical practice PGx guidelines [1,2] and regulatory support [3]. Recommendations from PGx testing about drug and dose selection in psychiatry are primarily based on the genotype-predicted phenotypes of the major drug metabolizing cytochrome P450 (CYP) enzymes [4–8]. However, due to phenoconversion, a central topic of this review article, this does not always provide a reliable prediction of a patient's CYP phenotype [9].

Indeed, prescribing medications in general – and in psychiatry in particular – is complex because it involves the simultaneous consideration of multiple patient (e.g., age and comorbidities) and nonpatient (e.g., cost) factors that influence the benefit versus risk of pharmacotherapy. Pharmacogenomics can sometimes be used for prescribing medications to improve the likelihood of safer and more effective treatment. Undeniably, PGx testing has advanced the field of precision dosing and increased awareness of its potential clinical and public health benefits [10]. However, PGx results are often presented to clinicians as static reports with only gene–drug guideline recommendations. This is acceptable when a single medication and a single gene are concerned. For example, a CYP genotype may help decide the choice of a particular antidepressant in a young adult with no other comorbidities who is taking no other medications. However, the simple addition of any other agent such as medications that inhibit or induce CYP enzymes may change the benefit versus risk of treatment with that antidepressant. Indeed, as the complexity of psychiatric patients increases with multiple medications metabolized by multiple CYPs with or without environmental effects, such as tobacco smoking, rational prescribing becomes increasingly difficult.

In the first half of this article, the problem of genotype–phenotype mismatch with CYP enzymes, known as phenoconversion, is outlined, with an emphasis on the medications used in psychiatry. In the second half of the article, solutions to help precision dosing in patients at risk for phenoconversion are suggested. This includes future perspectives on using model-informed precision dosing (MIPD) tools embedded in dynamic clinical decision support systems (CDSS).

The genotype–phenotype mismatch (phenoconversion)

Current PGx reports provide static guidance on specific gene–drug pairs. This approach is most relevant when patients are not exposed to factors that significantly alter CYP activities post translation of the CYP genes. However, in some cases, there is a mismatch between the patient's genotype-predicted CYP phenotype and their actual CYP phenotype. This phenomenon is called phenoconversion [11,12]. It has been suggested that phenoconversion offers a plausible explanation for the lack of congruence between CYP genotype and phenotype in many PGx studies [9]. In addition, the combined interaction between patient demographic, environmental and PGx covariates is increasingly recognized as important for predicting drug response [11–14].

Several factors can result in phenoconversion. These include the use of medications; inflammatory conditions, such as inflammatory bowel disease; cancer and environmental factors that can inhibit or induce CYP enzymes [9,13,14]. First, we consider the clinical impact of using concurrent medications that are capable of causing CYP phenoconversion. This type of drug-induced phenoconversion is more commonly encountered in practice and is dependent on the dose and washout period, determined by the half-life and inhibition/induction mechanism, of the phenoconverting medication and the substrate medication. For example, a patient predicted by genotype to be a normal metabolizer (NM) via CYP2D6 and CYP2C19 may in fact be phenoconverted to a CYP2D6 poor metabolizer (PM) and a CYP2C19 intermediate metabolizer (IM) phenotype by taking a strong CYP2D6 and moderate CYP2C19 inhibitor such as fluoxetine. The CYP2D6 inhibitory effect has been shown to occur as early as 7 days of continuous daily dosing with fluoxetine (20 mg) [15]. A static PGx report will only offer recommendations based on the CYP2D6 and CYP2C19 NM phenotype and will ignore the fluoxetine inhibition of CYP2D6 and CYP2C19. The clinical consequences of the CYP2D6 phenoconversion are:

• For active drugs (e.g., aripiprazole, risperidone) – higher blood/central nervous system (CNS) concentrations and increased risk of adverse effects.

• For prodrugs (e.g., codeine) – lower blood/CNS concentration of the active metabolite (morphine) and increased risk of therapeutic failure.

Phenoconversion of CYP enzymes may also be caused by factors in the patient's lifestyle. For example, smoking induces the expression of CYP1A2, resulting in increased metabolic clearance and decreased concentrations of medications predominantly eliminated by CYP1A2 [16]. In this example, the time course of phenoconversion is dependent on the amount of cigarettes smoked, the turnover of the CYP1A2 enzyme and the half-life of the interacting medication. It is important to note that the turnover of the CYP1A2 enzyme is approximately 2 days [17] and is increased by smoking-induced phenoconversion over a 2- to 4-week period. The increased CYP1A2 enzyme abundance results in more efficient metabolism of the CYP1A2 substrate drug and a shorter half-life. The classic example is clozapine, where smoking-induced CYP1A2 phenoconversion results in increased clozapine dose requirements [18]. On the contrary, smoking cessation can increase clozapine concentrations and the risk of toxicity as the inducing effect of smoking on CYP1A2 is lost over a 2- to 4-week period [18]. Chronic alcohol consumption, a common comorbidity in patients with mental health diagnoses, can also contribute to the phenoconversion of CYP2C19 to a lower metabolic activity level; however, the exact mechanism causing this is not fully elucidated [19]. Regardless, the impact of this CYP2C19-phenoconversion is clinically significant as CYP2C19 is responsible for the metabolism of many psychotropics (Table 1).

Changes in physiological or pathophysiological states also cause CYP phenoconversion [13,20]. Indeed, CYP2D6 is induced in pregnancy (e.g., potential phenoconversion of CYP2D6 from the NM phenotype to the ultrarapid metabolizer phenotype), which may lower concentrations of many psychotropics extensively metabolized by CYP2D6 (Table 1). Higher concentrations of cytokines in inflammatory conditions downregulate the expression of CYPs resulting in phenoconversion of genotype predicted NMs to lower metabolizer levels [21]. Patients with cancer classified as CYP2C19 NMs by genotype are more frequently phenoconverted into CYP2C19 PMs compared with non-cancer patients [20]. Although these lifestyle and disease-related phenoconversion factors are important, there are currently no expert consensus guidelines on how to test and quantify the clinical impact of these factors. Therefore, further research is required to guide the implementation of these phenoconversion factors in clinical practice.

Precision dosing with pharmacogenomics is challenging in psychiatry

Pharmacogenomics has particular appeal for precision dosing in psychiatry because therapeutic failure is common, adverse effects occur frequently and are sometimes severe, poor adherence and/or discontinuation of treatment is common, healthcare costs for psychiatric patients are high and many psychotropics are metabolized by the highly polymorphic CYP2D6 and CYP2C19 enzymes [22] (Table 1).

Although PGx testing has particular appeal in psychiatry, the clinical implementation is faced with a number of practical challenges.

Prescribing tools that assist clinicians with CYP phenoconversion at the point of care are being developed. Sequence2Script, a free web-based tool, requires the CYP genotypes and current medications to be entered before it provides guidance on appropriate PGx-based recommendations for a particular patient [34]. Another example is The PROP Pharmacogenetics Calculator, produced by University of Florida Health [35], which requires the clinician to input the patient's CYP2D6 genotype result and any interacting medication to receive the phenoconverted CYP2D6 phenotype. However, the latter example does not provide the clinician with PGx guideline recommendations based on the phenoconverted CYP2D6 phenotype.

Proposed solutions to assist precision dosing with pharmacogenomics in psychiatry

Stepwise approach

Figure 1 suggests a stepwise approach for implementing PGx testing while considering CYP phenoconversion as part of an advanced precision dosing approach in psychiatry. For a patient prescribed a psychotropic with no actionable PGx recommendations (Step 1), the standard prescribing recommendations based on the medication product label or current prescribing guidelines apply. In a drug-naive patient without any environmental factors causing phenoconversion (e.g., smoking) who is prescribed a psychotropic with actionable PGx recommendations (Step 2), a static PGx report with relevant interpretation and clinical prescribing recommendations could be effectively used. However, if this patient requires pharmacotherapy for other health conditions and one or more of the prescribed medications have actionable PGx recommendations (Step 3), then a clinical decision support system (CDSS) is typically required, supported by clinical review assessing phenoconversion, to provide prescribing guidance for multiple medications. Components required for this stepwise approach are described here and a later section describes Step 4 of Figure 1, in which MIPD could be used to guide dosing in complex patients.

Although Figure 1 may oversimplify the current state of PGx implementation, it is important to note that the PGx and pharmacology knowledge of the prescriber and the PGx readiness of their healthcare system must be considered.

Improved CDSSs at the point of care

Point-of-care solutions that efficiently deliver clinically actionable PGx recommendations to clinicians are needed. Recently, CDSSs have been implemented in practice to deliver PGx information [36]. These systems typically provide PGx-based prescribing information to the clinician and flag medications where PGx testing might be indicated.

Although current CDSSs deliver PGx recommendations at the point of prescribing, they often do not have the ability to adjust recommendations based on phenoconversion. It is impractical for the busy psychiatrist to interrogate separate standalone CDSS and phenoconversion systems to determine which PGx guideline recommendation should be followed. As a first step, contemporary CDSSs should consider integration with validated phenoconversion tools to account for basic drug-induced phenoconversion. Therefore, when a patient's PGx result is available, the CDSS can facilitate a clinically actionable prescribing alert in real time for the psychiatrist.

The past decade has seen a significant increase in newly discovered CYP alleles, notably CYP2D6 [22], enabled by more advanced genotyping technology with enhanced coverage. Retesting patient samples with more contemporary technologies may result in new CYP2D6 genotypes. Therefore, it is important that CDSSs allow patient PGx results to be updated over time to ensure up-to-date clinical utility of PGx results.

Although there are examples of CDSSs used successfully to implement PGx testing in practice, a key limitation to their widespread adoption is the lack of standardization in the reporting of PGx data across service providers [37,38]. Ongoing efforts are needed by PGx guideline authorities such as CPIC and DPWG to establish a standardized system for CYP genotype to phenotype translation [39].

Improved PGx education

Many surveys show that doctors and pharmacists have positive attitudes toward PGx testing but feel inadequately trained to implement PGx in a clinically meaningful way [40]. While the inclusion of PGx education in the curriculum of tertiary institutions offering pharmacy and medical programs has improved, PGx education for practicing clinicians remains suboptimal. In psychiatry, many consultants lack an understanding about the high variability in CYP activities within phenotypes; prescribers take CYP genotype as an absolute – “you are a CYP2C19 NM and so should be OK to take the standard dose of citalopram”. This interpretation may be problematic if the patient is taking a medication that can inhibit CYP2C19 and result in phenoconversion to a lower level of activity. Professional bodies and continuing professional development accrediting organizations should develop strategies to integrate PGx education into practice via ‘just-in-time’ or ‘on-demand’ learning, which could be delivered via a CDSS. In addition, PGx education can also be embedded in stewardship programs (described next) in which prescribers are educated on the potential benefits and limitations of PGx testing for the medications they routinely prescribe.

Rollout of PGx stewardship programs

Pharmacogenomic stewardship programs that review PGx results and the potential for CYP phenoconversion could significantly improve prescribing for at-risk polypharmacy patients [11]. A PGx stewardship team should be multidisciplinary with expertise in medicine, with clinical pharmacology specialist training, pharmacy and PGx. The team would ensure correct diagnoses, that accurate medication histories are taken and maintained and that PGx information is used at the point-of-care. Such an approach would use PGx, medication and lifestyle factors to provide initial first doses, which could then be further adjusted using therapeutic drug monitoring, if required and available. A similar approach was recently endorsed by the Royal College of Physicians and British Pharmacological Society joint working party on personalized prescribing [41]. The Australian government has recently announced that funding will be provided to embed pharmacists in residential aged care facilities to help improve medication safety in older Australians [42]. Indeed, recent data from Australia show the use of psychotropic medicines in older Australians increases after their admission to residential aged care facilities. Embedding pharmacists in these facilities for timely medication reviews could support prescribers with the delivery of PGx, potentially unlocking healthcare benefits for these vulnerable patients. Such initiatives could provide a fertile landscape to implement PGx stewardship programs, offering best-practice clinical governance for the delivery of PGx within precision dosing initiatives.

Future perspectives: CDSSs powered by MIPD

MIPD is defined as “the use of computer modelling and simulation to predict a medication dose regimen that is most likely to yield a better benefit to harm balance than traditional dosing, based on individual characteristics” [43]. The combined interaction of patient demographic, environmental and PGx covariates is becoming more widely recognized as important in predicting medication response [43–46]. While catering for the interplay between all of these interaction terms in a CDSS is technically challenging, a CDSS powered by a MIPD platform might offer a pragmatic solution (Figure 2). In this design, the CDSS is responsible for the translation of patient genotype to phenotype, ingesting the MIPD dosage recommendation and delivering a complete PGx recommendation with precision dosing to the electronic medical record (EMR) for use by the psychiatrist. The MIPD platform will require the following data integrations:

• Patient data from the EMR to provide demographic, disease and medication information.

• CDSS information (patient genotypes and phenotypes).

• Data from latest clinical studies and literature (e.g. to provide updated data on enzyme abundance).

• Environmental and BioSensor information from various devices (e.g. wearables).

The precision dosing recommendation predicted by the modelling can be sent back to the CDSS, which then renders this data into a Fast Healthcare Interoperability Resources (FHIR) file for ingestion by an EMR that is FHIR aware and then integrated into the patient management system for use at the point of prescribing (Figure 2).

In psychiatry, the proposed CDSS powered by MIPD would apply precision dosing of psychotropics by accounting for the influence of genetic variations in CYPs; the presence of CYP phenoconversion; and coexisting lifestyle (smoking), pregnancy or disease (cancer) factors. Indeed, a recent study of clozapine showed that MIPD using a physiologically based pharmacokinetic model improved the prediction of clozapine's PK compared with population-based statistics [47]. In this study, clozapine concentrations were better predicted by MIPD accounting for the CYP1A2 inducing effect in smokers homozygous for the CYP1A2*1F allele. This is an example of where environmental (smoking) and PGx (CYP1A2 genotype) factors were used to optimize the MIPD model, resulting in improved predictions of clozapine plasma concentrations. In principle, this approach can be applied across other psychotropics, especially those with a high risk of toxicity in overdose (e.g., tricyclic antidepressants). Although there are many technical challenges (some examples are given in Table 2), these preliminary data warrant further investigation with other psychotropics to understand how incorporating gene–environment effects may improve MIPD predictions of pharmacokinetics and whether this translates into meaningful predictions of medication responses. The exact regulatory framework to ensure adequate validation of such models before clinical use is currently being addressed [43].

Conclusion

 The potential of PGx testing to assist medication prescribing in psychiatry is limited by CYP phenoconversion. Several factors may cause phenoconversion, but inhibition or induction of CYP2C19 and/or CYP2D6 by concurrently prescribed medications is the main mechanism of CYP phenoconversion in psychiatry. As described in the executive summary, there are mitigation strategies that clinicians and healthcare workers can implement to optimize the choice and dose of medicines. As such, realizing the full potential of PGx testing requires innovative education strategies, particularly at the postgraduate level, PGx stewardship programs embedded in clinical practice, user-friendly CDSS and, in the future, more advanced technologies such as MIPD that aid precision dosing in complex patients.