Introduction

F1000Research

2046-1402

F1000 Research Limited

London, UK

10.12688/f1000research.147794.1

Study Protocol

Articles

Living systematic review and meta-analysis of plasma-concentrations of antipsychotic drugs in carriers and non-carriers of variant CYP450 genotypes: Living systematic review protocol

[version 1; peer review: 2 approved with reservations]

Milosavljević

Filip

Conceptualization Funding Acquisition Methodology Writing – Original Draft Preparation https://orcid.org/0000-0001-6532-4661 a 1 2 Leucht

Stefan

Conceptualization Funding Acquisition Methodology Supervision Writing – Review & Editing https://orcid.org/0000-0002-4934-4352 1 1Section Evidence-Based Medicine in Psychiatry and Psychotherapy, Department of Psychiatry and Psychotherapy, Klinikum rechts der Isar, TUM School of Medicine and Health, Technical University of Munich, Munich, Bavaria, 81675, Germany 2Department of physiology, Faculty of Pharmacy, Univerzitet u Beogradu, Belgrade, 11221, Serbia

a m.filip@tum.de

Competing interests: In the last three years SL has received honoraria for advising/consulting and/or for lectures and/or for educational material from Angelini, Boehringer Ingelheim, Eisai, Ekademia, GedeonRichter, Janssen, Karuna, Kynexis, Lundbeck, Medichem, Medscape, Mitsubishi, Neurotorium, Otsuka, NovoNordisk, Recordati, Rovi, Teva. FM reports no conflict of interests.

7 5 2024

2024

452

26 4 2024

2024

This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Introduction

Carriers of variant alleles of genes that encode liver CYP450 and UGT enzymes may experience abnormal plasma levels of antipsychotics and, consequently, worse efficacy or tolerability. Although pharmacogenomics is a rapidly developing field, current guidelines often rely on limited, underpowered evidence. We have previously demonstrated that meta-analysis is a viable strategy for overcoming this problem. Here, we propose a project that will expand our previous work and create a living systematic review and meta-analysis of drug plasma level differences between carriers and non-carriers of variant genotype-predicted phenotypes for every pharmacokinetic drug-gene interaction relevant to commonly used antipsychotic drugs.

Protocol

First, a baseline systematic review and meta-analysis will be conducted by searching for observational pharmacogenomics-pharmacokinetic studies. Data on dose-adjusted drug plasma levels will be extracted, and participants will be grouped based on their genotype for each drug-gene pair separately. Differences in plasma drug levels between different phenotypes will be compared using a random-effect ratio-of-means meta-analysis. The risk of bias will be assessed using ROBINS-I, and the certainty of evidence will be assessed using GRADE. Following the establishment of baseline results, the literature search will be re-run at least once every six months, and the baseline data will be updated and re-evaluated as new evidence is published. A freely available website will be designated to present up-to-date results and conclusions.

Discussion

This systematic review will provide evidence-based results that are continuously updated with evidence as it emerges in the rapidly developing field of pharmacogenomics. These results may help psychiatrists in their decision-making, as clinicians are becoming increasingly aware of the patients’ genetic data as testing becomes more widespread and cheaper. In addition, the results may serve as a scientific basis for the development of evidence-based pharmacogenomics algorithms for personalized dosing of antipsychotics to mitigate potentially harmful drug-gene interactions.

Living systematic review Pharmacogenomics Antipsychotic drugs Schizophrenia Therapeutic drug monitoring CYP450

Alexander von Humboldt-Stiftung

1235463-HFST-P

This work was supported by Alexander von Humboldt Foundation’s Postdoctoral Fellowship Stipend (awarded to FM, stipend number: 1235463-HFST-P).

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

1. Introduction

Pharmacotherapy for schizophrenia today is challenging because of (I) slow development of new drugs, ¹ (II) treatment resistance, ² (III) frequent need for treatment regimen adjustments ³ (IV) high relapse rates ⁴ and (V) risk of unpleasant adverse drug reactions. ⁴ In such a landscape, approaches such as pharmacogenomics hold great potential for maximizing the effectiveness of currently available psychiatric drugs. It is proposed that mutations in genes that code liver CYP450 (Cytochrome p450) and UGT (UDP-glycosyltransferase) enzymes can cause either too low or too high plasma levels of antipsychotic, ⁵ leading to a lack of efficacy or a higher risk of adverse reactions in some patients. ⁶ The precise extent of these pharmacokinetic drug-gene interactions must be determined to produce a pharmacogenomics-informed dosing algorithm that can mitigate these changes.

Meta-analysis can be an effective tool for overcoming low statistical power, which is a common problem in many pharmacogenomics-pharmacokinetic studies. ⁷ We will build on our previous work on precise quantification of pharmacokinetic drug-gene interactions of antipsychotics ⁸ by analyzing more drug-gene interactions and utilizing a living systematic review approach.

Since pharmacogenomics is a rapidly developing field and the level of evidence in pharmacogenomics is often unsatisfactory, ⁵ ^, ⁷ continuously updated evidence-based data sources regarding pharmacokinetic drug-gene interactions could be very informative to clinicians, as patients’ genomic data are more commonly available to clinicians because commercial pharmacogenomics tools are becoming more frequently used and cheaper over time.

1.1 Research question

Our hypothesis is that polymorphisms in genes encoding CYP450 and/or UGT enzymes are associated with significant changes in dose-corrected plasma concentrations of antipsychotic drugs.

Drug-gene pairs will be formed for every relevant polymorphic gene associated with every relevant antipsychotic drug. For every drug gene pair, our aim is to determine, as precisely as possible, the magnitude of the difference in dose-corrected plasma drug concentrations between carriers and non-carriers of variant genotype-predicted phenotypes for the given gene, and the certainty of this evidence. Table 1 presents the PICO format for the research questions.

Table 1. PICO format for the research question(s).

Both general case of the research question and an example of Risperidone-CYP2D6 drug-gene pair are given.

Criterion	General formulation of the research question	Example: Formulation of the research question for the Risperidone-CYP2D6 drug-gene pair
P – Population	Adults (18-65 years old) who are either healthy volunteers or psychiatric patients, treated with antipsychotic drug of interest.	Adults (18-65 years old) who are either healthy volunteers of psychiatric patients treated with risperidone.
I – Intervention(s)	Carrying of a variant genotype-determined phenotype for the gene of interest.	1) Carrying CYP2D6 poor metabolizer phenotype (PM); 2) Carrying CYP2D6 intermediate metabolizer phenotype (IM); 3) Carrying CYP2D6 ultra-rapid metabolizer phenotype (UM).
C – Control	Carrying of the reference genotype-determined phenotype for the gene of interest.	Carrying CYP2D6 normal metabolizer phenotype (NM).
O – Outcome	Dose-adjusted plasma concentration (C/D) of antipsychotic drug of interest, either after steady state is achieved or after a single drug dose.	Dose-adjusted plasma concentration of risperidone active-moiety (parent drug + active metabolite) either after steady state is achieved or after a single risperidone dose.

2. Protocol

This protocol was written in accordance with the PRISMA-P (Preferred reporting items for systematic review and meta-analysis protocols) guideline. ⁹ A checklist is presented in the extended data. The final report of this living systematic review will be written in accordance with MOOSE (Meta-analysis of observational studies) reporting guideline. ¹⁰ This protocol is also registered with PROSPERO (CRD42024485626).

2.1 Literature search

The following databases will be searched: ClinicalTrials.gov, Cochrane Library (Cochrane Database of Systematic Reviews=CDSR and Cochrane Central Register of Controlled Trials=CENTRAL), Embase, MEDLINE, PsycINFO and WHO ICTRP. No studies that were published prior to 1.1.1995. will be searched because of the extremely limited availability of PCR genotyping technology. Both published and unpublished clinical studies will be searched, without language restrictions. The search will be re-run every time the resulting paper is submitted for peer review, as well as continuously once every 6 months during the living meta-analysis maintenance.

Search strategy draft is presented in the extended data.

2.2 Drug-gene pairs of interest

Based on the information from CPIC, ¹¹ PharmVar ¹² and PharmGKB ¹³ resources, drug-gene pairs of interest were identified and are presented in Table 2. If new relevant data emerge during the maintenance of this living meta-analysis, this list will be updated, and a new drug-gene pair will be included.

Table 2. Drug-gene pairs of interest.

Drug-gene pairs that will be considered for analysis are formed by pairing most clinically relevant antipsychotic drugs and liver enzymes that are coded by polymorphic genes and are substantially involved in the metabolism of said drugs.

	CYP2D6	CYP3A4/5	CYP2C19	CYP1A2	UGT1A4	UGT2B7
Aripiprazole	X	X
Asenapine				X	X
Brexpiprazole	X	X
Clozapine	X	X	X	X
Cariprazine	X	X
Iloperidone	X	X
Lumateperone		X		X
Olanzapine	X	X		X	X
Paliperidone	X	X
Quetiapine	X	X
Risperidone	X	X
Sertindole	X
Zotepine		X		X
Haloperidol	X	X		X		X
Chlorpromazine	X	X		X
Perphenazine	X			X
Zuclopentixol	X	X

Amisulpride will not be analyzed, as it is mainly extracted unchanged via the kidneys. Lurasidone will not be analyzed because of the uncertain importance of drug-blood level measurements. ⁶ Ziprasidone will not be analyzed because of its low inter-individual variability in plasma drug levels. ¹⁴ First-generation antipsychotic drugs, other than chlorpromazine, perphenazine, haloperidol, and zuclopentixol, were not included because of their limited clinical utility today.

2.3 Definition of intervention and control groups

For every drug-gene pair of interest, the exposure group(s) (i.e., variant group; experimental group) and control group (i.e., wild-type group; reference group) will be defined based on the genotype-predicted phenotypes for the gene in question. In the cases of CYP2C19 and CYP2D6, genotype-predicted phenotypes are also called “Metabolizer groups” where “Normal” metabolizer group is considered as a control, and “Intermediate,” “Poor” and “Ultra-rapid” metabolizer groups are considered as 3 possible exposure groups. Definitions of the variant genotypes were adopted from consensus guidelines for every individual gene. ¹¹ ^– ¹³ Metabolizer groups defined by phenotyping with a probe drug will not be considered valid for the purpose of this review. Detailed definitions of the control and experimental groups based on CYP2D6, CYP3A4/5, CYP2C19, CYP1A2, UGT1A4 and UGT2B7 genotypes are presented in the extended data.

2.4 Definition of the outcome

Pharmacokinetic parameters used to represent the drug plasma level will be included only if they are dose adjusted or if the entire cohort is administered the same dose. If necessary and possible, dose-adjusted pharmacokinetic parameters will be manually estimated by dividing the mean drug plasma level and mean drug dose using the Taylor expansion method. ¹⁵ The potential impact of this estimation will be explored in a sensitivity analysis, where manually estimated dose-adjusted data will be excluded. In addition, for this calculation, a correlation between drug dose and drug plasma levels of ρ=0.5 will be assumed, and two alternative values, ρ=0.2 and ρ=0.8 will be used instead in the sensitivity analysis.

Drug plasma pharmacokinetic parameters are often presented as body-weight-adjusted. To increase the scope of the review, body-weight-adjusted and unadjusted data will be treated as interchangeable. If body-weight-adjusted and unadjusted data are presented simultaneously, body-weight-adjusted data will be prioritized. This decision will be tested using sensitivity analysis, in which only body-weight-adjusted data will be included.

Pharmacokinetic parameters that will be used to assess drug plasma levels will be: I).

Dose-adjusted steady-state drug plasma concentration (Css/D): Measured after five half-elimination times have elapsed since drug initiation or therapy regimen change, during which patients took stable doses of the drug with good compliance. Ideally, measurements were taken in the morning, just before the new dose was taken.

II).

Dose-adjusted area under the steady-state plasma concentration-time curve (AUC/D): Measured after five half-elimination times had elapsed since drug initiation or therapy regimen change, during which patients took stable doses of the drug with good compliance. Measurements were taken during the same period for all participants.

III).

Single-dose plasma concentration (C) or area under the plasma concentration curve (AUC): Even though drug plasma level measurements are not usually considered valid before the steady-state has been achieved owing to unpredictable variability, single-dose pharmacokinetic studies will also be included owing to their high level of confounding control and strict study design. These studies often use homogeneous study groups of healthy volunteers with controlled intake of food, water, and other substances and drugs, with standardized blood sampling times. Thus, well-controlled single-dose pharmacokinetic parameters will be used interchangeably with steady-state parameters, and this decision will be tested in a sensitivity analysis where single-dose studies are excluded.

IV).

Reciprocal value of the total drug clearance (1/Cl _total): For drugs with a linear pharmacokinetic profile, the reciprocal value of the total clearance is equal to the AUC/D metric. If clearance is presented as median with interquartile range and/or minimum and maximum values, these values will be transformed into their reciprocal values, which will be further transformed into mean and standard deviation. ¹⁶ This decision will be tested in the sensitivity analysis, where reciprocal value data are excluded.

If the blood sample for the Css/D measurement was not taken in the morning before the next dose, but the blood sampling time was consistent in the entire cohort, the study will be included; however, this decision will be tested in the sensitivity analysis where only studies with the ideal sampling times are included.

If the drug possesses one or more active metabolites that are present in the plasma in non-negligible amounts ¹¹ compared to the parent drugs, drug plasma levels will be expressed as “Active moiety,” i.e. the sum of the plasma parent drug levels and the plasma active metabolite levels.

If multiple pharmacokinetic parameters are present within the same study for the same cohort, Css/D will be prioritized because it is the most common readout. ⁸ If the same pharmacokinetic parameter is presented for multiple time-points, the longest time point will be selected because of the lower chance that some of the included participants have not yet reached steady-state blood drug levels.

2.5 Study inclusion criteria

I).

For every drug-gene pair of interest, the study will be included if it reports drug plasma levels for control genotype-predicted phenotype and at least one variant genotype-predicted phenotype for the given gene.

II).

Studies involving participants of all ethnicities will be included. However, since the genotype frequencies vary drastically based on ethnic background, subgroup analysis will be performed to compare the results between European, East Asian, South Asian, African, and Admixed American origin cohorts. ¹⁷

III).

Studies in which the participants were treated with antipsychotic drugs were included. This includes studies on patients with schizophrenia, other psychiatric illnesses, and healthy volunteers. To assess the effect of this decision, studies on healthy volunteers will be excluded as a part of the sensitivity analysis.

IV).

All age groups were eligible for inclusion in principle, but the main results will be presented only for adults between 18 and 65 years of age. Results on participants of all ages, results only in adolescents (12-18 years old), and the results only for the elderly (>65 years old) will be explored in the sensitivity analysis.

V).

Studies will be included regardless of whether the participants were on a stable dosing regimen on enrolment, started drug regimen on enrolment, or switched to a new drug on enrolment.

VI).

Studies on both responders and treatment resistant patients will be included.

VII).

Studies on participants not exposed to co-medications, foods, or smoking, in cases where these factors are known to interfere with the given antipsychotic drug, are preferred because of the lower risk of confounding. Still, studies where participants are exposed to these confounders will be eligible for inclusion only if these factors are well balanced across different phenotype groups or corrected by the authors during data analysis. In addition, studies with co-medication/food/smoking confounding factors will be excluded as a part of the sensitivity analysis.

VIII).

Studies on participants taking any drug formulation (immediate release tablets, extended release tablets, long-acting injections, intravenous administration, etc.) will be eligible, and the data for all formulations will be used interchangeably in the main meta-analysis. To test this decision, every formulation that is not an immediate release tablet will be excluded as a part of the sensitivity analysis.

IX).

Prospective studies will be preferred because of the higher reliability of the pre-planned experiments. ¹⁸ Nevertheless, retrospective studies are very common in pharmacogenomics and will be included to increase the scope of this review. To test this decision, retrospective studies will be excluded as a part of the sensitivity analysis.

X).

To increase the scope of this living systematic review, both published and unpublished studies will be included.

2.6 Study exclusion criteria

I).

Studies on pregnant women will be excluded due to unpredictable pharmacokinetics during pregnancy.

II).

Studies with a critical overall risk-of-bias grade based on the ROBINS-I (Risk of bias in non-randomised studies of interventions) tool ¹⁹ will be excluded (four possible risk-of-bias grades in the ROBINS-I tool are low, moderate, serious, and critical risk of bias, ranked from the best to the worst grade).

III).

Studies with a substantial proportion of the population consisting of non-compliers will be excluded because of the risk of erroneous drug plasma level measurements.

IV).

Studies on participants with severe kidney or liver impairments or other severe conditions that interfere with drug pharmacokinetics (Malabsorption, Hypoproteinemia, etc.) will be excluded because of unpredictable changes in the pharmacokinetics of these participants.

V).

Studies with unbalanced known confounding factors (co-medications, food, cigarette smoke) for drug plasma levels between normal and variant genotype-predicted phenotype groups will be excluded.

VI).

Population pharmacokinetic studies will be excluded due to incompatibility with other included studies.

2.7 Data screening and selection

Both the screening and selection of the results of the systematic search will be performed independently by two researchers, and any disputes will be resolved by consensus with a third researcher.

Screening will be performed based on the titles and abstracts of the search results. During the screening, search results will be labels as “not relevant,” “relevant” or “unclear relevance.”

For the search results labelled as “relevant” and “unclear relevance,” full text retrieval will be attempted. If the full text could not be retrieved, the corresponding author will be contacted to provide the full text. Next, the retrieved full texts will be assessed for eligibility based on the inclusion and exclusion criteria.

Process of screening and selection will be visually reported using PRISMA 2020 flowchart. ²⁰

2.8 Data collection and missing data imputation

Data will be extracted using an internally developed standardized form.

Means and standard deviations of drug plasma levels, as well as the number of participants, will be extracted. Means and standard deviations will be estimated using standard procedures if the data are presented as geometric mean and 95%CI, ²¹ as median, interquartile range, minimum, and maximum ¹⁶ or as mean values and the p-value of the mean difference. ²¹ If the Plasma levels are presented graphically but not numerically, numerical data will be extracted from the digital image of the given graph. ²²

If patients were not grouped into genotype-predicted phenotypes according to our desired criteria, manual regrouping will be performed when possible. For example, if plasma levels in participants carrying CYP2D6*1/*10 and *1/*5 are presented separately, but there are no data for CYP2D6 IM collectively, the mean plasma levels in *1/*10 and *1/*5 carriers will be summed manually to form the mean IM plasma levels. ²¹

Regarding demographics, data on ethnicity, sex, and age distribution will be collected. Regarding clinical characteristics, data on the participants’ diagnosis, drug dose, body weight, duration of treatment, blood sampling time, co-medications, and other drug-specific confounders will be collected. Finally, data on study design and the characteristics of reported pharmacokinetic parameters will be collected.

If cohorts of two or more clinical studies overlap, inclusion of the studies will be performed to maximize the inclusion of as many unique participants as possible while not having any repeating participants.

If the included manuscript suggests that the data of interest were measured but not presented, the corresponding author will be contacted to provide the data. If this is not possible, the study will be labelled as “Awaiting assessment”. Studies with missing data will be reported as a list and considered during the interpretation of publication bias using funnel plots.

2.9 Risk of bias analysis

The ROBINS-I tool ¹⁹ will be used to assess the seven domains of bias by two researchers independently. Disputes will be resolved by consensus with a third researcher. Based on the semi-quantitative grades for all seven domains, the overall grade for each study will be assigned. Due to the nature of observational studies that will be searched, it is expected for a small number of included studies, if any, to receive the best grade: “Low” risk-of-bias. Instead, it is expected that studies with relatively better design to be graded as having “Moderate” risk-of-bias, while the studies with relatively worse design to receive “Serious” of “Critical” risk-of-bias grade. “No information” overall risk-of-bias grade will be used sparingly, i.e. everywhere where reasonable assumptions regarding the signalling question cannot be made. Studies with a risk of bias grade other than “moderate” or “low” will be excluded as a part of the sensitivity analysis.

2.10 Certainty of evidence analysis

Certainty of evidence will be assessed using the GRADE tool ²³ across five domains of evidence certainty; overall certainty of evidence will also be assessed for every drug-gene pair of interest. GRADE scoring will be performed independently by two researchers, and disputes will be resolved by consensus with a third researcher.

2.11 Statistical methods

Quantitative data needed for the meta-analysis includes the mean plasma level, standard deviation, and number of participants in the given study group. For every drug-gene pair analyzed, all experimental groups will be compared pairwise with their corresponding control group. Drug plasma levels will be compared between the groups using the ratio-of-means (RoM) effect measure ²⁴ and its 95% confidence interval. This will allow for an intuitive interpretation of the results, as well as the comparison of different pharmacokinetic parameters of drug plasma levels within one analysis. Our previous research showed a high probability of high heterogeneity within the included studies; therefore, a random-effect meta-analysis will be performed. Specifically, RoM=1.15 will be interpreted as 15% higher, and RoM=0.85, which will be interpreted as a 15% lower drug plasma level in the experimental group compared to the control group. RoM values for individual values will be log-transformed and pooled using the inverse-variance method to produce the grand-mean.

As a secondary analysis, all meta-analysis results calculated as RoM will also be calculated as the Standardized Mean Difference (SMD, Hedges’ g). This is aimed at providing an alternative interpretation of the results, which is important considering the rare use of the RoM effect measure.

Heterogeneity will be assessed using the I ² metric, where I ²>50% will be considered substantial heterogeneity. ²¹ In the case of substantial heterogeneity, the effect of outlier results on the overall results will be assessed using a sensitivity analysis. Further, heterogeneity will be explored using other sensitivity and subgroup analyses (previously described) to explore what clinical or demographic factors may be the cause of study misalignment.

Potential publication bias and small-study effects will be assessed using contour-enhanced funnel plots ²⁵ and Egger’s test. ²⁶ If fewer than 10 studies are included for a certain drug-gene pair, publication bias will not be performed, as recommended. ²⁵ The critical p-value for Egger’s test will be set at p<0.10, as suggested. ²⁶

2.12 Sensitivity and sub-group analyses

Sensitivity analysis will be used to test the robustness of the obtained results with regard to changes to the inclusion/exclusion criteria. The following sensitivity analyses will be performed: 1) Exclusion of the manually dose-adjustment plasma level data; 2) Usage of ρ=0.2 and ρ=0.8 as an assumed correlations levels between drug dose and drug plasma levels for manual dose-adjustment of plasma levels; 3) Exclusion of body-weight-unadjusted data; 4) Exclusion of the data estimated from reciprocal clearance values; 5) Exclusion of studies with suboptimal time of plasma sampling; 6) Inclusion of data on non-adults; 7) Exclusion of the studies with suboptimal confounding control; 8) Exclusion of “Single drug dose” studies; 9) Exclusion of the studies with “Serious” risk-of-bias overall score; 10) Exclusion of healthy volunteer data; 11) Exclusion of retrospective studies and 12) Exclusion of the data on drug formulations other than immediate release tablets.

Subgroup analyses will be used to compare the meta-analysis results only between participants of different ethnic backgrounds.

2.13 Maintenance of the living systematic review and meta-analysis

Following an initial literature search and data analysis, the baseline results and conclusions of the living systematic review and meta-analysis will be established. As these results are being published, designated website will be created to display these results, and thus the maintenance phase of the living systematic review will begin. In this phase, once every six months, a systematic literature search will be repeated, and two designated researchers will be tasked with screening and testing for eligibility of the new search results. Alternatively, if a sudden increase in the number of new publications is detected during the first year of maintenance, updates from that point onwards will be performed more often once every three months. The data from the eligible studies will be extracted in the same manner as during the establishment of the baseline results and will be added to the database of the baseline results. Updated results will then be published on the website, together with the change log. If a new discovery regarding pharmacogenomics of liver enzymes involved in the metabolism of antipsychotics emerges, the entire database will be updated accordingly in the shortest possible notice. This includes newly detected drug-gene pairs of interest and new consensus for genotype-determined phenotype classification.

For pragmatic reasons, once all analyzed drug–gene pairs either achieve satisfactory strength of evidence or experience no new publications for an extended period of time, updates of the living systematic review will be performed less frequently, and the update frequency will depend on the probability of new evidence emerging in the future.

2.14 Dissemination

As previously mentioned, the baseline results will be published in separate papers, while the up-to-date results and conclusions will be presented on a designated website. Papers published in scientific journals are planned from time to time during the living systematic review maintenance phase.

2.15 Study status

At the time of writing the manuscript (29. January 2024), the literature search strategy was designed, and a systematic search was not performed for either of the drug-genes of interest. No study screening, selection, or data extraction was performed at the time of writing.

3. Discussion

This project has the potential to provide much-needed clinical information to the psychiatry community. Psychiatrists are becoming increasingly aware of the CYP450 genetic status of their patients, and the implementation of evidence-based knowledge of drug-gene interactions in psychiatrists’ decision-making will improve the quality of pharmacotherapy in psychiatry. This interaction will be facilitated by up-to-date reporting of the systematic review results on a designated website, which will be freely available to any interested clinician.

Furthermore, several attempts have been made to create and test clinical tools for pharmacogenomics-guided pharmacotherapy for schizophrenia. ²⁷ ^– ²⁹ Such tools aim to detect and mitigate potentially dangerous drug-gene interactions, and thus improve the efficacy and/or safety of antipsychotic drugs. One common approach is to completely avoid drugs known to interact negatively with the genetic makeup of a patient. This approach is criticized as potentially dangerous because it can overly restrict the list of effective options available to the patient. ³⁰ Another promising approach is to use genetic information to guide the dosing regimen of antipsychotic drugs, but due to limited and often conflicting evidence, genotype-guided dose regimen adjustments are often arbitrary. The results of this systematic review may facilitate the development of new and the improvement of existing pharmacogenomic tools by providing precise estimates of expected blood drug concentration changes in carriers of different genetic variants. This may help in the transition from arbitrary to evidence-based algorithms for the mitigation of potentially harmful drug-gene interactions related to the CYP450 and UGT genotypes.

3.1 Limitations

One of the most substantial limitations of this project was the expected high risk of bias in the included studies. This is because of the very common naturalistic approach with insufficient confounder control in the eligible studies observed in our previous work. ⁸ In addition, carriers of variant CYP450 or UGT alleles may be more likely to drop out due to their atypical drug plasma levels during observational studies, potentially biasing the results. Finally, there is a risk that some of the results of this systematic review may become outdated with the discovery of new alleles for some genes of interest. This could cause participants to become misclassified in the studies published prior to this discovery without the option to re-classify them post hoc, as this would require repeated genetic testing.

Ethics and consent

Ethical approval and written informed consent were not required.

Data availability Underlying data

No data are associated with this article.

Extended data

Figshare, LivingCYP - Living CYP - Extended data - F1000 25mar2024.docx, https://doi.org/10.6084/m9.figshare.25109675.v2. ³¹

Reporting guidelines

Figshare: PRISMA –P checklist for Living systematic review and meta-analysis of plasma-concentrations of antipsychotic drugs in carriers and non-carriers of variant CYP450 genotypes: Living systematic review protocol, https://doi.org/10.6084/m9.figshare.25109675.v2. ³¹

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

References 1

Miller

Raison

: Burning down the house: reinventing drug discovery in psychiatry for the development of targeted therapies. Mol. Psychiatry. Jan 2023;28(1):68–75. 36460725

10.1038/s41380-022-01887-y

Howes

Thase

Pillinger

: Treatment resistance in psychiatry: state of the art and new directions. Mol. Psychiatry. Jan 2022;27(1):58–72. 34257409

10.1038/s41380-021-01200-3

PMC8960394

Dold

Leucht

: Pharmacotherapy of treatment-resistant schizophrenia: a clinical perspective. Evid. Based Ment. Health. May 2014;17(2):33–37. 24713315

10.1136/eb-2014-101813

Kaar

Natesan

McCutcheon

: Antipsychotics: Mechanisms underlying clinical response and side-effects and novel treatment approaches based on pathophysiology. Neuropharmacology. Aug 1 2020;172:107704. 31299229

10.1016/j.neuropharm.2019.107704

Jukic

Milosavljevic

Molden

: Pharmacogenomics in treatment of depression and psychosis: an update. Trends Pharmacol. Sci. Dec 2022;43(12):1055–1069. 36307251

10.1016/j.tips.2022.09.011

Mauri

Paletta

Di Pace

: Clinical Pharmacokinetics of Atypical Antipsychotics: An Update. Clin. Pharmacokinet. Dec 2018;57(12):1493–1528. 29915922

10.1007/s40262-018-0664-3

Shekhani

Steinacher

Swen

: Evaluation of Current Regulation and Guidelines of Pharmacogenomic Drug Labels: Opportunities for Improvements. Clin. Pharmacol. Ther. May 2020;107(5):1240–1255. 31715018

10.1002/cpt.1720

PMC7232863

Milosavljevic

Bukvic

Pavlovic

: Association of CYP2C19 and CYP2D6 Poor and Intermediate Metabolizer Status With Antidepressant and Antipsychotic Exposure: A Systematic Review and Meta-analysis. JAMA Psychiatry. Mar 1 2021;78(3):270–280. 33237321

10.1001/jamapsychiatry.2020.3643

PMC7702196

Shamseer

Moher

Clarke

: Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. Jan 2 2015;349:g7647. 25555855

10.1136/bmj.g7647

Stroup

Berlin

Morton

: Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA. Apr 19 2000;283(15):2008–2012. 10789670

10.1001/jama.283.15.2008

Hiemke

Bergemann

Clement

: Consensus Guidelines for Therapeutic Drug Monitoring in Neuropsychopharmacology: Update 2017. Pharmacopsychiatry. Jan 2018;51(1-02):9–62. 28910830

10.1055/s-0043-116492

Gaedigk

Casey

Whirl-Carrillo

: Pharmacogene Variation Consortium: A Global Resource and Repository for Pharmacogene Variation. Clin. Pharmacol. Ther. Sep 2021;110(3):542–545. 34091888

10.1002/cpt.2321

PMC8725060

Whirl-Carrillo

Huddart

Gong

: An Evidence-Based Framework for Evaluating Pharmacogenomics Knowledge for Personalized Medicine. Clin. Pharmacol. Ther. Sep 2021;110(3):563–572. 34216021

10.1002/cpt.2350

PMC8457105

Beedham

Miceli

Obach

: Ziprasidone metabolism, aldehyde oxidase, and clinical implications. J. Clin. Psychopharmacol. Jun 2003;23(3):229–232. 12826984

10.1097/01.jcp.0000084028.22282.f2

Stuart

Ord

: Kendall’s Advanced Theory of Statistics. 6th ed. Arnold;1998; vol.1.

Wan

Wang

Liu

: Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med. Res. Methodol. Dec 19 2014;14:135. 25524443

10.1186/1471-2288-14-135

PMC4383202

Zhou

Ingelman-Sundberg

Lauschke

: Worldwide Distribution of Cytochrome P450 Alleles: A Meta-analysis of Population-scale Sequencing Projects. Clin. Pharmacol. Ther. Oct 2017;102(4):688–700. 28378927

10.1002/cpt.690

PMC5600063

Norvell

: Study types and bias-Don’t judge a study by the abstract’s conclusion alone. Evid. Based Spine Care J. Aug 2010;1(2):7–10. 23637661

10.1055/s-0028-1100908

PMC3623096

Sterne

Hernan

Reeves

: ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ. Oct 12 2016;355:i4919. 27733354

10.1136/bmj.i4919

PMC5062054

Page

McKenzie

Bossuyt

: The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. Mar 29 2021;372:n71. 33782057

10.1136/bmj.n71

PMC8005924

Nasser

: Cochrane Handbook for Systematic Reviews of Interventions. Am. J. Public Health. Jun 2020;110(6):753–754. 10.2105/Ajph.2020.305609

Rohatgi

: WebPlotDigitizer 4.6.

Reference Source

2022;

Schünemann

Brozek

Guyatt

: GRADE Handbook. UK: Cochrane collaboration London;2013.

Friedrich

Adhikari

NKJ

Beyene

: The ratio of means method as an alternative to mean differences for analyzing continuous outcome variables in meta-analysis: A simulation study. BMC Med. Res. Methodol. May 21 2008;8:8. 18492289

10.1186/1471-2288-8-32

PMC2430201

Sterne

Sutton

Ioannidis

: Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ. Jul 22 2011;343:d4002. 21784880

10.1136/bmj.d4002

Egger

Davey Smith

Schneider

: Bias in meta-analysis detected by a simple, graphical test. BMJ. Sep 13 1997;315(7109):629–634. 9310563

10.1136/bmj.315.7109.629

PMC2127453

Carrascal-Laso

Franco-Martín

García-Berrocal

: Application of a Pharmacogenetics-Based Precision Medicine Model (5SPM) to Psychotic Patients That Presented Poor Response to Neuroleptic Therapy. J. Pers. Med. Dec 2020;10(4). 33352925

10.3390/jpm10040289

PMC7767089

Jurgens

Andersen

Rasmussen

: Effect of Routine Cytochrome P450 2D6 and 2C19 Genotyping on Antipsychotic Drug Persistence in Patients With Schizophrenia: A Randomized Clinical Trial. JAMA Netw. Open. Dec 1 2020;3(12):e2027909. 33284338

10.1001/jamanetworkopen.2020.27909

Kang

Qin

Sun

: Multigenetic Pharmacogenomics-Guided Treatment vs Treatment As Usual Among Hospitalized Men With Schizophrenia. JAMA Netw. Open. Oct 6 2023;6(10):e2335518. 37801319

10.1001/jamanetworkopen.2023.35518

PMC10559185

Goldberg

Rosenblat

McIntyre

: Letter to the Editor: Clinical versus statistical significance of pharmacogenomic-guided antidepressant therapy: What’s really being measured and marketed?. J. Psychiatr. Res. Jul 2019;114:208–209. 31064655

10.1016/j.jpsychires.2019.04.002

Milosavljević

: Living CYP - Extended data - F1000 25mar2024.docx. figshare. 2024. 10.6084/m9.figshare.25109675.v2

10.5256/f1000research.162029.r280401

Reviewer response for version 1

Baumann

Pierre

1 Referee 1Department of Psychiatry (DP-CHUV), Universite de Lausanne, Lausanne, Vaud, Switzerland

Competing interests: No competing interests were disclosed.

6 6 2024

2024

This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

recommendation

approve-with-reservations

General comment:

The authors propose a protocol for future studies on the relationship between drug plasma concentrations in patients treated with antipsychotics and their pharmacogenetic status regarding genetic polymorphisms of cytochrome P-450 isoforms and/or UGT enzymes (cf Table 1, criterion O (outcome)). It would have been of interest to include a part which concerns the clinical outcome (efficacy, effectiveness, adverse effects) of the patients, as the present protocol will not allow to clearly show the clinical relevance of pharmacokinetic-pharmacogenetic exams. However, the present project is already ambitious and the inclusion of a regularly occurring update is highly promising and novel.

Other comments and questions:

The issues dealing with pharmacokinetics and pharmacogenomics presented in the Introduction should be presented in more details as some references are not adequate. Indeed, some citations lack precision, e.g. ref 5 and 6, in the sentence: “It is proposed that mutations in genes that code liver CYP450 (Cytochrome p450) and UGT (UDP-glycosyltransferase) enzymes can cause either too low or too high plasma levels of antipsychotic, ⁵ leading to a lack of efficacy or a higher risk of adverse reactions in some patients. ⁶ The citation of reference 5 suggests that it presents data about the importance of UGT in the pharmacogenomics of antipsychotics. However, UGT is only mentioned as a key word (and only as UGT-2B.. but not as UGT-1A..). In ref 6, UGT is not even mentioned. Therefore, the reader with limited knowledge in this field may not get a full picture of the importance of the subject. In summary, the Introduction should be extended!

Table 1 presents in “Intervention I” the different phenotypes using the example risperidone. The reader is then supposed to presume that this presentation is also valid for UGT enzymes (eg gene-drug pair: UGT-olanzapine), but this classification (poor, intermediate, extensive, ultrarapid) is seemingly not introduced for these polymorphisms. If however this is the case, this should be presented in some details and corresponding references should be presented.

Indeed, the average reader will not be familiar with the genetic polymorphisms of conjugating enzymes, and in particular, with their relevance for antipsychotics. In other words: are there PM, EM, UM, IM in patients treated with e.g. asenapine or olanzapine, with regard to the metabolism by UGT-enzymes? (Admittedly, in the suppl. material is some information, but the protocol should be more precise, especially in Introduction). It is a fact that among cytochrome P-450 isozymes, only about half a dozen isozymes play a role in the metabolism of antipsychotics. In contrast, there are considerably more isozymes which contribute to the conjugation of drugs, but their individual role in the metabolism of antipsychotics is poorly characterized.

The pioneering study of Haslemo et al. (CPT, 2012, 221) (besides that of Erickson et al., Pharmacogen Genom 2011, 539) shows that plasma concentrations of conjugated olanzapine (10-N-ola-glucuronide) differ in olanzapine treated patients classified according to their UGT1A genotypes, but the difference is negligible for (unconjugated) olanzapine concentrations. This rises the question whether in the present meta-analysis study both ola and conjugated ola concentrations will be considered. It could be then of interest to examine the relationships between the UGT genotypes and the ratios parent compound/metabolite (Ola/10-N-ola-glucoronide).

This approach may also be valuable for all other antipsychotics, whether they are metabolized by cytochrome P-450 or by polymorphic conjugating enzymes, or by both types (e.g. risp/9-OH-risp; 9-OH-risp/conjug. 9-OH-risp; etc).

Probably of little importance for the establishment of the protocol, it is to mention that in the case of risperidone, its hydroxylation to 9-OH-risperidone is stereoselective, in that CYP2D6 preferentially forms (+)-9-OH-risperidone rather than (-)-9-OH-risperidone (Yasui-Furukori et al. Drug Metab. Disp 2001, 1263). It should be considered, as rarely, stereoselective procedures were used in the published studies, the net effect of the genotype may be masked when non-stereoselective procedures are used.

The authors exclude some antipsychotics, but the mentioned reasons are not always evident, and again, the citations do not always correspond to the statements. Therefore, the presentation of the reference list demands to be improved. Eg, 2.2. Drug gene pairs …:

“Lurasidone will not be analyzed because of the uncertain importance of drug-blood level measurements.” Ref 6 is cited, but this ref does apparently not deal with this subject. Actually, lurasidone is mainly a substrate of CYP3A4/5. As its blood levels strongly depend on dietary conditions, this may be an important reason for recommending TDM (cf Preskorn 2013: Lurasidone should be administered with food-at least 350 kcal-to ensure maximum exposure). However, one can agree with the decision of the authors, as this drug belongs to the group of drugs mentioned in 2.5. VII (study inclusion citeria) resp. 2.6. V.

2.4. Definition of the outcome

As pointed out by the authors, weight corrected drug concentrations may be advantageous, but they are often not available.

Steady-state plasma concentrations: In the case of risperidone, there may be a problem in that in extensive metabolizers (CYP2D6), its T1/2 = 2 - 4h, but in poor metabolizers, considerably longer, comparable to that of its metabolite 9-OH-risperidone. This means that in EM, risp. steady-state concentrations will hardly be reached in EM but only in PM.

Extended data: March :

6. UGT1A4

“At the time of writing, there is no consensus regarding the classification of different UGT1A4 diplotypes into metabolizer groups. Therefore, pragmatically, alleles known to affect CYP1A2 activity with Minor allelic frequencies >1% in the world population will be considered for this analysis.” Not clear: this suggests that there is a relationship CYP1A2 – UGT1A4 (?).

Is the study design appropriate for the research question?

Partly

Is the rationale for, and objectives of, the study clearly described?

Partly

Are sufficient details of the methods provided to allow replication by others?

Yes

Are the datasets clearly presented in a useable and accessible format?

Not applicable

Reviewer Expertise:

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Milosavljević

Filip

EBMpp, Technical University of Munich, Munich, Bavaria, Germany

Competing interests: We declare no competing interest with regards to this response.

1 7 2024

"The authors propose a protocol for future studies on the relationship between drug plasma concentrations in patients treated with antipsychotics and their pharmacogenetic status regarding genetic polymorphisms of cytochrome P-450 isoforms and/or UGT enzymes (cf Table 1, criterion O (outcome)). It would have been of interest to include a part which concerns the clinical outcome (efficacy, effectiveness, adverse effects) of the patients, as the present protocol will not allow to clearly show the clinical relevance of pharmacokinetic-pharmacogenetic exams. However, the present project is already ambitious and the inclusion of a regularly occurring update is highly promising and novel."

We thank the reviewer for positive comment and for recognizing the scale of this project.

"Other comments and questions:

The issues dealing with pharmacokinetics and pharmacogenomics presented in the Introduction should be presented in more details as some references are not adequate. Indeed, some citations lack precision, e.g. ref 5 and 6, in the sentence: “It is proposed that mutations in genes that code liver CYP450 (Cytochrome p450) and UGT (UDP-glycosyltransferase) enzymes can cause either too low or too high plasma levels of antipsychotic,5 leading to a lack of efficacy or a higher risk of adverse reactions in some patients.6 The citation of reference 5 suggests that it presents data about the importance of UGT in the pharmacogenomics of antipsychotics. However, UGT is only mentioned as a key word (and only as UGT-2B.. but not as UGT-1A..). In ref 6, UGT is not even mentioned. Therefore, the reader with limited knowledge in this field may not get a full picture of the importance of the subject. In summary, the Introduction should be extended!"

We sincerely apologize for inaccurate citations. In accordance with this comment and the comment from the other reviewer, introduction section is now expended to shortly summarize all gene candidates potentially important for the pharmacogenomics of antipsychotic drug pharmacokinetics.

"Table 1 presents in “Intervention I” the different phenotypes using the example risperidone. The reader is then supposed to presume that this presentation is also valid for UGT enzymes (eg gene-drug pair: UGT-olanzapine), but this classification (poor, intermediate, extensive, ultrarapid) is seemingly not introduced for these polymorphisms. If however this is the case, this should be presented in some details and corresponding references should be presented."

We agree, Table 1 can be misinterpreted. In the introduction section, it is now described which genes have established metabolizer status definitions and which genes are instead analyzed on the basis of genotypes and haplotypes. Also, Table 1 is now updated to include another example that involves UGT isoenzyme to further eliminate confusion.

"Indeed, the average reader will not be familiar with the genetic polymorphisms of conjugating enzymes, and in particular, with their relevance for antipsychotics. In other words: are there PM, EM, UM, IM in patients treated with e.g. asenapine or olanzapine, with regard to the metabolism by UGT-enzymes? (Admittedly, in the suppl. material is some information, but the protocol should be more precise, especially in Introduction). It is a fact that among cytochrome P-450 isozymes, only about half a dozen isozymes play a role in the metabolism of antipsychotics. In contrast, there are considerably more isozymes which contribute to the conjugation of drugs, but their individual role in the metabolism of antipsychotics is poorly characterized."

In line with this and previous comments, introduction section is now extended to also include this information. Also, our justification for the choice of UGT isoenzymes included in the first iteration of our living systematic review is included in the section 2.2. If the Reviewer finds that some crucial UGT or CYP450 isoenzymes are omitted, we will gladly consider them for the inclusion.

"The pioneering study of Haslemo et al. (CPT, 2012, 221) (besides that of Erickson et al., Pharmacogen Genom 2011, 539) shows that plasma concentrations of conjugated olanzapine (10-N-ola-glucuronide) differ in olanzapine treated patients classified according to their UGT1A genotypes, but the difference is negligible for (unconjugated) olanzapine concentrations. This rises the question whether in the present meta-analysis study both ola and conjugated ola concentrations will be considered. It could be then of interest to examine the relationships between the UGT genotypes and the ratios parent compound/metabolite (Ola/10-N-ola-glucoronide).This approach may also be valuable for all other antipsychotics, whether they are metabolized by cytochrome P-450 or by polymorphic conjugating enzymes, or by both types (e.g. risp/9-OH-risp; 9-OH-risp/conjug. 9-OH-risp; etc)."

To our knowledge, Olanzapine metabolites including 10-N-Olanzapine-Glucuronide have marginal activity and it is not recommended to measure them during therapeutic drug monitoring [PMID: 1683407]. Even though the pharmacokinetics of inactive drug metabolites is very relevant and important topic, we find this data to be of limited usefulness for our living systematic review as we plan to focus on the clinical aspects of therapeutic drug monitoring.

Regarding risperidone, there is evidence that risperidone and 9-OH-risperidone have different efficacy and safety profiles due to different blood-brain-barriers permeability [PMID: 33054667]. Therefore, we agree for that analysis of riperidone/9-OH-risperidone ratio may be very useful for the interpretation of meta-analysis of risperidone active moiety (Risperidone + 9-OH-Risperidone) plasma levels.

Section 2.4 is now updated to reflect this.

"Probably of little importance for the establishment of the protocol, it is to mention that in the case of risperidone, its hydroxylation to 9-OH-risperidone is stereoselective, in that CYP2D6 preferentially forms (+)-9-OH-risperidone rather than (-)-9-OH-risperidone (Yasui-Furukori et al. Drug Metab. Disp 2001, 1263). It should be considered, as rarely, stereoselective procedures were used in the published studies, the net effect of the genotype may be masked when non-stereoselective procedures are used."

As we mentioned previously, we would like to focus on clinical aspects of therapeutic drug monitoring, i.e. we will try to use drug blood levels as a biomarkers of drugs biological effects. Therefore, even though stereo-selective analysis highlights the influence of CYP2D6, we find very limited clinical relevance of this information for our review as two enantiomers of 9-OH-Risperidone have been previously described as equipotent (PMID: 17344104).

"The authors exclude some antipsychotics, but the mentioned reasons are not always evident, and again, the citations do not always correspond to the statements. Therefore, the presentation of the reference list demands to be improved. Eg, 2.2. Drug gene pairs …:

We agree with the reviewer, Lurasidone is the substrate for CYP3A4/5 and should therefore be included. Also, unrelated to reviewer’s comment, we have noticed that we listed Paliperidone as CYP3A4/5 and CYP2D6 substrate even though paliperidone is minimally metabolized in the liver and is mostly eliminated by kidneys [PMID: 20118446].

We sincerely apologize for these errors; they are now corrected in the revised version of the manuscript.

"2.4. Definition of the outcome

As pointed out by the authors, weight corrected drug concentrations may be advantageous, but they are often not available.

We agree, people with lower enzyme metabolism capacities might take longer time to reach steady state. Therefore, we changed the criterion for the time needed to reach steady state to be 5x the conservative t _1/2estimate. In other words, whenever the t _1/2 is presented as the range in AGNP guidelines (PMID: 28910830) we will use the higher limit of this range (longest time). This is now reflected in section 2.4.

"Extended data: March :

6. UGT1A4

This was a typing error, and we sincerely apologize for it. We also noticed several other typing errors in the Extended data and they are now corrected in the new version.

We again thank the Reviewer for the constructive criticism. We have no further comments.

10.5256/f1000research.162029.r275387

Reviewer response for version 1

de Lima Benzi

Jhohann Richard

1 Referee https://orcid.org/0000-0001-6986-6193 1University of São Paulo, São Paulo, Brazil

Competing interests: No competing interests were disclosed.

31 5 2024

2024

recommendation

approve-with-reservations

The manuscript under review presents a comprehensive systematic review protocol into the relevance of CYP pharmacogenetics in antipsychotic (AP) therapy, offering a clear and well-written discourse on the topic. The study's goal is undeniably pertinent, addressing a critical aspect of personalized medicine in psychiatric care. I commend the authors for their meticulous approach and the significance of their endeavors. However, I would like to suggest some considerations to enrich the manuscript.

Firstly, the inclusion or mention of the importance of drug transporters in AP therapy warrants attention, particularly those expressed in the blood-brain barrier. Given their pivotal role in drug disposition and response, their omission from the review raises questions about their relevance in AP pharmacotherapy. Exploring the available pharmacogenetic data pertaining to these transporters could provide valuable insights into their impact on treatment outcomes.

Secondly, the decision to exclude data based on probe phenotyping drug data when defining metabolizer groups requires clarification. Elaborating on the rationale behind this exclusion would enhance the transparency and reproducibility of the study, ensuring a more robust analysis of CYP pharmacogenetics in AP therapy.

Thirdly, I respectfully disagree with the assertion that single-dose pharmacokinetic parameters can be used interchangeably with steady-state parameters, particularly in the context of AP drugs with non-linear pharmacokinetics. Given the complexities inherent in non-linear PK profiles, a sensitivity analysis excluding single-dose studies may not adequately capture the dynamic nature of drug metabolism. It is imperative to devise alternative strategies to handle such data, perhaps through modeling techniques or subgroup analyses, to ensure the integrity and reliability of the findings.

In conclusion, while the manuscript admirably addresses the relevance of CYP pharmacogenetics in AP therapy, the incorporation of drug transporter considerations, clarification on data exclusion criteria, and careful handling of non-linear PK data are essential for a more comprehensive and nuanced analysis.

Is the study design appropriate for the research question?

Partly

Is the rationale for, and objectives of, the study clearly described?

Yes

Are sufficient details of the methods provided to allow replication by others?

Partly

Are the datasets clearly presented in a useable and accessible format?

Yes

Reviewer Expertise:

Clinical pharmacology, pharmacokinetics, pharmacometrics

Milosavljević

Filip

EBMpp, Technical University of Munich, Munich, Bavaria, Germany

Competing interests: We declare no competing interests with regards to our response.

1 7 2024

"The manuscript under review presents a comprehensive systematic review protocol into the relevance of CYP pharmacogenetics in antipsychotic (AP) therapy, offering a clear and well-written discourse on the topic. The study's goal is undeniably pertinent, addressing a critical aspect of personalized medicine in psychiatric care. I commend the authors for their meticulous approach and the significance of their endeavors. However, I would like to suggest some considerations to enrich the manuscript."

We thank the reviewer for positive feedback.

"Firstly, the inclusion or mention of the importance of drug transporters in AP therapy warrants attention, particularly those expressed in the blood-brain barrier. Given their pivotal role in drug disposition and response, their omission from the review raises questions about their relevance in AP pharmacotherapy. Exploring the available pharmacogenetic data pertaining to these transporters could provide valuable insights into their impact on treatment outcomes."

Indeed, drug efflux pumps are coded by highly polymorphous genes (ABCB1, ABCG2, ABCC2…) and the research of their pharmacogenomics has been very active in the recent years. We initially omitted to mention drug transporter in an effort to keep our report concise, but we agree with the reviewer that these genes have to be addressed in order to provide comprehensive overview of pharmacogenomics of antipsychotic drug pharmacokinetics, and that the omission of drug transporter genes from our analysis has to be justified in the manuscript.

The main reason to exnclude pharmacogenomic studies on efflux pumps from our review is potential inappropriateness of our main outcome to capture the impact of pharmacogenomics of drug transporters on antipsychotic therapy. Our review is focused on plasma levels of antipsychotics, and in our analysis we will assume that cerebrospinal-fluid-to-plasma ratios of antipsychotic drugs are similar between people with different CYP450/UGT genotypes. Since drugs transporters are expressed on blood-brain barrier, in some antipsychotic drugs it has been demonstrated that mutations of drug transporter genes such as ABCB1 can change cerebrospinal-fluid-to-plasma ratios [PMID: 21148022, PMID: 20599439, PMID: 24666334]. This means that, for example, if two groups of patients with different ABCB1 mutations have the same antipsychotic plasma levels, they may not have the same antipsychotic cerebrospinal fluid levels, which puts into question the appropriateness of drug plasma level as a biomarker. Therefore, we believe that analyzing drug efficacy and safety directly, or analyzing cerebrospinal fluid drug levels are more appropriate outcomes for the analysis of drug transporters pharmacogenomics, all of which are out of the scope of this review.

Introduction section now includes short summary of the potential significance of efflux pumps, and Section 2.2. now includes the justification of our justification not to include efflux pump pharmacogenomic data.

"Secondly, the decision to exclude data based on probe phenotyping drug data when defining metabolizer groups requires clarification. Elaborating on the rationale behind this exclusion would enhance the transparency and reproducibility of the study, ensuring a more robust analysis of CYP pharmacogenetics in AP therapy."

We agree with the reviewer that this decision needs more clarification. Also, upon revision of our manuscript, we noticed that our previous wording “for the purpose of this review, phenotyping will not be considered valid” can be interpreted as us questioning the validity of phenotyping method, which is not the case. Instead, section 2.3 is now updated to introduce pros and cons of both phenotyping and genotyping approaches and the justification why we opted not to use them interchangeably in this review.

"Thirdly, I respectfully disagree with the assertion that single-dose pharmacokinetic parameters can be used interchangeably with steady-state parameters, particularly in the context of AP drugs with non-linear pharmacokinetics. Given the complexities inherent in non-linear PK profiles, a sensitivity analysis excluding single-dose studies may not adequately capture the dynamic nature of drug metabolism. It is imperative to devise alternative strategies to handle such data, perhaps through modeling techniques or subgroup analyses, to ensure the integrity and reliability of the findings."

We thank the reviewer for raising this concern. Even though the majority of antipsychotics have linear pharmacokinetic profile, sublingual asenapine, oral lumateperone, oral chlorpromazine and oral perphenazine possess non-linear pharmacokinetic profiles (PMID: 29915922, PMID: 32060882, PMID: 8157044, PMID: 3933033). We agree with the reviewer that for the given examples, data from single dose studies and steady state studies cannot be used interchangeably and we apologize for our omission. As we are currently not aware of any model published in the literature designed to predict steady state levels of given four drugs based on single-dose data, we will, for the time being, opt to exclude single-dose studies for asenapine, lumateperone, chlorpromazine and perphenazine. We expect very small (if any) impact of this decision on the results of our meta-analyses, as we, based on our previous experience, expect to detect very few pharamcogenomic studies in the literature on these 4 drugs. If instead we happen to detect many eligible single-dose studies for these 4 drugs, we will post-hoc find appropriate model developed for other drugs that may fit the drug in question.

This clarification is now reflected in the manuscript.

"In conclusion, while the manuscript admirably addresses the relevance of CYP pharmacogenetics in AP therapy, the incorporation of drug transporter considerations, clarification on data exclusion criteria, and careful handling of non-linear PK data are essential for a more comprehensive and nuanced analysis."

We thank the reviewer again for the constructive criticism that will improve the quality of this paper. We have no further comments.