Biomarkers for the prediction and monitoring of the antipsychotic/antidepressant-induced hepatotoxicity: study protocol

ABSTRACT

Aim

This study is designed to address the connection between antidepressant and antipsychotic-induced hepatotoxicity with pharmacogenetic and epigenetic indicators, using a novel combined approach of CYP450 polymorphism determination and early liver injury detection via microRNA testing.

Methods

The multi-centric retrospective case-control study in Slovakia involves 151 cases with signs of hepatotoxicity and 604 controls without. Participants will be tested for selected CYP450, UGT1A1 polymorphisms, and microRNAs.

Results

Anticipated findings will test if patients with specific CYP450 and UGT1A1 polymorphisms are at higher risk for drug-induced hepatotoxicity and if plasma microRNAs hsa-miR-122-5p and hsa-miR-192-5p, alone or combined, can differentiate patients with abnormal liver function.

Conclusion

The findings could contribute to personalized treatment approach by combining genetic and epigenetic biomarkers.

Plain Language Summary

This study is about making treatment for mental health issues, like depression and anxiety, better and safer. Some patients respond differently to their medications, and some can have serious side effects, such as liver damage. To help with this, we are trying a new approach that uses two tests. The first test looks at certain genes to see how a person might react to medication. The second test checks for small molecules in the blood that can show if there are early signs of liver problems. We want to find out two things: Do people with certain genes have a higher risk of liver problems from their medications? Can these small molecules help us find liver issues better than regular blood tests? We will study about 755 patients from three mental health clinics in Slovakia, who have been taking their medications for at least four weeks. By identifying patients who might struggle with their medications due to their genes, we hope to catch liver problems early. This could lead to more personalized and effective treatments for everyone.

1. Introduction

1.1. Background

1.1.1. Drug-induced hepatotoxicity caused by antidepressants and/or antipsychotics treatment

Issues in the antidepressants and/or antipsychotics treatment include the frequent occurrence of adverse drug reactions, particularly drug-induced hepatotoxicity [1]. The precise terminology related to drug-induced hepatotoxicity, that includes hepatotoxicity, liver injury, and liver damage, remains inadequately defined within the underestimation of real incidence [2,3]. Prevalence of any abnormal liver function tests (LFT) in the general population ranged from 10% to 21.7%, rising to 32% among antipsychotic users [4]. In patients taking antidepressants, only in one third was LFT assessed, revealing abnormalities in 20% [5].

Antidepressant-induced and antipsychotic-induced hepatotoxicity, characterized by elevated liver enzymes, poses risks from asymptomatic to fatal consequences [6]. Routine diagnosis using traditional liver function tests (LFT), i.e., alanine aminotransferase – ALT, aspartate aminotransferase – AST, alkaline phosphatase – ALP, gamma-glutamyltransferase – GMT, total bilirubin – TBIL, albumin, prothrombin time – PT/international normalized ratio – INR; lacks predictive value as changes manifest only after actual liver injury. In consequence, there is a growing interest in exploring microRNAs as early diagnostic biomarkers for liver injury [7].

The surge in psychiatric illnesses, notably depression and anxiety disorders, has led to a rise in psychopharmaceutical prescriptions [8]. Long-term use of antidepressants and antipsychotics presents challenges, including unpredictable patient responses and the risk of adverse effects [9]. To address these issues, a crucial shift toward personalized medicine, employing tools like pharmacogenetic testing, is advocated.

1.1.2. Pharmacogenetic testing in psychiatric practice

Pharmacogenetic testing enhances drug safety and efficacy [10]. Interindividual genetic variations, particularly in drug metabolism pathways primarily located in the liver, intestine, and kidneys, contribute to varied responses to pharmacotherapy. Cytochrome P450 (CYP450) enzymes, responsible for metabolizing 70–80% of drugs [10], show a high rate of polymorphisms, resulting in diverse drug metabolism and therapeutic responses. This interindividual variability can lead to adverse effects, drug toxicity, or ineffective treatment. The clinical effectiveness of pharmacogenomic testing for depressive disorders was supported by multiple studies [11], and several expert groups have developed pharmacogenetic based guidelines for psychotropic drugs, supported by global clinical acceptability of pharmacogenetic testing, that has already been implemented in clinical practice around Europe, North America, and Asia [12]. Given the prevalence of adverse effects, including hepatotoxicity, pharmacogenetic testing becomes crucial for predicting individual responses, enabling the tailored selection of effective and safe treatments.

1.1.3. MicroRNAs (miRNAs) as a biomarker of drug-induced hepatotoxicity

MicroRNAs (miRNAs), small non-coding RNAs inhibiting target gene expression, are present in various tissues, blood, and other body fluids, aiding in detecting or predicting drug toxicity [13,14]. Liver-specific or liver-enriched miRNAs, notably hsa-miR-122-5p and hsa-miR-192-5p, are abundant in liver and have been associated with hepatic function [7,15,16]. Unlike traditional liver enzymes, miR-122 levels are elevated during early-stage liver damage and offer a more accurate prediction value of current liver condition including early damage, demonstrating their potential in assessing hepatic health [17–19].

1.2. Rationale for pharmacogenetic and epigenetic targets

1.2.1. Pharmacogenetic targets

The selection of pharmacogenetic targets was made according to the following criteria

1. Identification of drugs in ATC groups N06A, N05A with the highest prescription in 2018–2021. Data were based on statistical outputs of human drug consumption of the National Health Information Centre of the Slovak Republic [20]. In ATC group N06A, the following antidepressants were the most prescribed, in descending order: escitalopram (15%), citalopram (15%), sertraline (11%), venlafaxine (10%), mirtazapine (8%), trazodone (7%), paroxetine (7%), duloxetine (5%), tianeptine (4%), vortioxetine (4%), dosulepin (3%), agomelatine (2%), fluvoxamine (2%), clomipramine (2%), amitriptyline (1%), fluoxetine (1%), bupropion (1%). In the ATC group N05A, the following antipsychotics were the most prescribed, in descending order: quetiapine (17%), tiapride (16%), chlorprotixene (14%), olanzapine (10%), haloperidol (8%), risperidone (8%), levomepromazine (7%), sulpiride (5%), clozapine (5%), aripiprazole (3%), zotepine (1.5%), zuclopentixol (1.5%), paliperidone (1.5%), amisulpiride (1%), ziprasidone (1%), cariprazine (0.5%). These drugs represent 99% of antidepressant and antipsychotic usage in Slovakia in the reporting period.

2. Identification of CYP450 enzymes that are significantly involved in the metabolism of identified drugs (Suplementary Table S1). Drugs with significant hepatic metabolism, especially if CYP450 substrates, are prone to dose-independent drug-induced hepatotoxicity, while CYP450 inhibitor drugs show this effect at high doses [21]. Early detection is crucial for these cases. Using the 2020 Transformer database [22], we prioritized CYP2D6 (76%), CYP3A4 (64%), CYP1A2 (45%), CYP2C19 (39%), CYP2C9 (30%), CYP3A5 (27%), CYP2B6 (18%), CYP2C8 (18%), CYP3A7 (18%), and CYP2E1 (15%) as key enzymes based on drug biotransformation proportions. Other enzymes play a role in less than 10% of the metabolism of the most prescribed drugs.

When selecting CYP450 enzymes and gene polymorphisms, we prioritized those with up-to-date data from pharmacogenetic databases: PharmGKB [23], PharmVar [24], gnomAD [25], and expert consensus sources in psychiatric pharmacogenetics [26,27]. Additionally, we considered insights from pharmacogenetic consortia and working groups: Clinical Pharmacogenetics Implementation Consortium (CPIC) [28], Dutch Pharmacogenetics Working Group (DPWG) [29], and Association for Molecular Pathology PGx Working Group (AMP) [30–32].

• (1) Identification of significant polymorphisms of selected CYP450 enzymes with known clinically relevant annotation and expected allele rate ≥1% (Table 1). The requirements for selecting CYP450 target alleles were as follows:

  1. adequately characterized alteration of CYP450 activity/function due to the PharmGKB variant allele.
  2. allele abundance in the European (non-Finnish) patient population ≥1% according to gnomAD.

• (2) Selected polymorphisms of UGT1A1, which is involved in the glucuronidation of selected drugs (especially tricyclic antidepressants) and is a genetic marker of Gilbert’s syndrome associated with hyperbilirubinemia [33], were added.

Table 1.

Selection of significant CYP450 enzyme polymorphisms with known clinically relevant annotation and expected allele frequency AF ≥ 1%.

Allele	Core variant(s)‡	Enzyme activity	AF
CYP2D6*3	rs35742686	null	1.8%
CYP2D6*4	rs3892097	null	18.6%
CYP2D6*5	full gene deletion	null	3.0%#
CYP2D6*6	rs5030655	null	1.2%
CYP2D6*9	rs5030656	decreased	3.1%
CYP2D6*10	rs1065852 + rs1135840	decreased	1.6%#
CYP2D6*41	rs28371725 + rs16947 + rs1135840	decreased	9.2%#
CYP2D6*1×N CYP2D6*2×N	*1: increased number of copies + absence of other monitored alleles. *2: rs16947 + rs1135840 + absence of defining sequences for other alleles + increased number of copies.	increased	1.9%#
CYP2C19*2	rs4244285	null	14.6%
CYP2C19*17	rs12248560	increased	22.0%
CYP2C9*2	rs1799853	decreased	13.2%
CYP2C9*3	rs1057910	null	6.5%
CYP1A2*1C	rs2069514	decreased	1.3%
CYP1A2*1F	rs762551	increased inducibility§	71.2%
CYP3A4*22	rs35599367	decreased	4.9%
UGT1A1*28	rs3064744 (7 TA repeats in the TATA box A(TA)6TAA > A(TA)7TAA)	decreased	30.1%
UGT1A1*80	rs887829	decreased	32.1%

Explanatory notes: AF – frequency of variant allele in the European population (according to gnomAD v4.0.0 [25]; Function – predicted change in CYP450 function for a given variant allele.

^§*1F/*1F homozygotes/smokers associated with increased inducibility.

^#AF according to PharmGKB [23].

^‡Allele defining variant(s) according to the Pharmacogene Variation Consortium website [24].

1.2.2. Customized prioritization of pharmacogenetic panel variants

Tier 1 priority: specific genetic polymorphisms in genes CYP2D6, CYP2C19, CYP2C9 (listed in Table 1) with a confirmed association to change in enzyme function according to the PharmVar database and with a predicted allele frequency in European population ≥1%. For these genes, there are evidence-based pharmacogenetic recommendations of the consortia CPIC (The Clinical Pharmacogenetics Implementation Consortium) or DPWG (The Dutch Pharmacogenetics Working Group) to optimize treatment for selected antidepressants and antipsychotics.

Tier 2 priority: specific genetic polymorphisms in genes CYP1A2 and CYP3A4 (listed in Table 1) with confirmed association to change in enzyme function and allele frequency in European population ≥1%. These alleles may be relevant for selected drugs, especially for agomelatine, fluvoxamine, quetiapine [34–36], and may increase risk of drug-related problems due to drug-drug-gene interactions, including CYP450 inhibitors and inducers. However, variants of these genes currently do not have specific pharmacogenetic recommendations for treatment optimization.

Tier 3 priority: specific genetic polymorphisms in UGT1A1 gene (listed in Table 1) involved in limited second-phase biotransformation of tricyclic antidepressants are additionally associated with a predisposition to Gilbert’s syndrome, which may precipitate hepatobiliary problems regardless of the choice of psychopharmacotherapy.

1.2.3. microRNAs – epigenetic targets

To identify microRNA biomarkers for drug-induced hepatotoxicity, we selected two liver-specific microRNAs, hsa-miR-122-5p, and hsa-miR-192-5p, relying on literature data [16,37–40]. These microRNAs, abundant in the liver, are released into the blood circulation in early stages of liver damage, are associated with various liver and biliary diseases, serving as potential early biomarkers for drug-induced toxicity.

1.3. Objectives of the study

1.3.1. Hypotheses

We posit that selected polymorphisms in metabolic enzymes (CYP2D6, CYP2C9, CYP2C19, CYP1A2, CYP3A4 and UGT1A1) (Table 1) influence the pharmacokinetics of antidepressants and antipsychotics in the Slovak population (Supplementary Table S1), impacting the efficacy and safety of affective and psychotic disorder therapies. We hypothesize that patients with these polymorphisms face a higher risk of therapy failure or adverse drug reactions, including drug-induced hepatotoxicity. Consequently, testing the association between hepatic miRNAs, specifically has-miR-122-5p and has-miR-192-5p, and adverse therapy outcomes, could aid in identification of novel early drug-induced hepatotoxicity marker.

Based on these hypotheses, we set the following objectives.

1.3.2. Primary objectives

1. Compare the frequencies of clinically actionable CYP450/UGT1A1 variant alleles in patients with and without drug-induced hepatotoxicity when treated with antidepressants or antipsychotics.

2. Assess the ability of selected plasma microRNAs, individually or in combination, to identify patients with abnormal liver function, compared to established biochemical tests (AST, ALT, ALP, GMT, TBIL), that will be used as a reference diagnostic method.

1.3.3. Secondary objectives

1. Evaluate the impact of clinically actionable CYP450/UGT1A1 enzyme polymorphisms on the efficacy of antidepressant or antipsychotic therapy by analyzing endpoints like clinical improvement.

2. Examine whether clinically relevant CYP450/UGT1A1 enzyme polymorphisms affect the risk of adverse events/effects of antidepressants and antipsychotics other than hepatotoxicity based on measurable safety endpoints, including adverse reactions and drug-induced toxicity.

3. Estimate the allele frequency and calculate Hardy-Weinberg equilibrium of selected CYP450/UGT1A1 polymorphisms in the Slovak population. Compare the estimated frequencies with reported data in the general European population.

1.4. Study originality

To the best of our knowledge, this study marks the first investigation into the pharmacogenetic association of CYP450 enzymes with antidepressant and antipsychotic treatment concerning drug-induced hepatotoxicity. The exception is agomelatine, where previous literature has explored the influence of genetic polymorphisms of the CYP1A2 enzyme on agomelatine-induced liver injury [35,41,42]. No similar studies have been conducted in the Slovak population. Additionally, the potential use of plasma microRNAs for early diagnosis of hepatic toxicity in antidepressant and antipsychotic treatment, mentioned in the literature as potential toxicological markers [7,19,43], remains unexplored. Combining pharmacogenetic (enzyme polymorphisms) and epigenetic (microRNA) testing for drug-induced hepatotoxicity presents a unique molecular-biological approach with considerable clinical benefits for enhancing pharmacotherapy safety.

2. Methods

2.1. Study design

 The project is designed as a multi-centric observational retrospective case-control study in the Slovak population. This protocol corresponds to the items of the STROPS statement (STrengthening the Reporting Of Pharmacogenetic Studies) which define the standard elements of pharmacogenetic clinical study protocols [44]. The target group will consist of patients treated with antidepressants (ATC: N06A) and/or antipsychotics (ATC: N05A) for at least 4 weeks. Patients will be stratified into case/control groups (see Table 3).

Table 3.

Definition of cases and controls.

Control patients	No signs of hepatotoxicity based on laboratory measurements (stage 0). No clinical signs of hepatotoxicity (asymptomatic).
Case patients	Signs of hepatotoxicity based on laboratory measurements (stage 1–4). May or may not have clinical signs of hepatotoxicity.
Unclassified	Patients with signs of hepatotoxicity, but without clear laboratory signs of abnormal LFTs (stage 0). These patients will not be classified into case or control groups, but their samples will be used to evaluate plasma microRNA levels as biomarkers of early liver dysfunction.

2.2. Setting

Patient recruitment, starting in 2024 across psychiatric outpatient clinics in the Slovak Republic, including University Hospital Trnava, University Hospital Nitra, and Centre of Mental Health Matka Bratislava, will conclude upon reaching the required number (755). Data processing, genotyping, microRNA determination, and study evaluation will be conducted at the Department of Pharmacology and Toxicology, Faculty of Pharmacy, Comenius University Bratislava, led by academic pharmacists.

2.3. Study participants

2.3.1. Patients

Patients will be enrolled according to the selection criteria in Table 2, stratification criteria for cases and controls are defined in Table 3.

Table 2.

Selection criteria for patients’ recruitment.

Admission criteria: – Age ≥18 – The patient has been prescribed and consistently takes at least one medication from the antidepressant group (ATC: N06A) or antipsychotic group (ATC: N05A) for a minimum of 4 weeks.

Exclusion criteria: – Diagnosis of liver disease (ICD-10 codes K70-K77) or disorders of gallbladder and biliary tract (ICD-10 codes K80-K83), including viral hepatitis (ICD-10 codes B15-B19), except in patients with diagnosed or suspected drug-induced toxic liver disease (ICD-10 code K71) and diagnosed adverse effects of the therapeutic use of psychotropic drugs (ICD-10 codes Y49.0-Y49.5). – HIV positive patients. – Diagnosis of kidney disease (ICD-10 codes N00-N29, including kidney diseases classified elsewhere). – Evidence of impaired renal function within the past 12 months as assessed by a physician, including abnormal findings on SONO/CT/MRI scans of the kidneys, abnormal biochemical parameters indicating renal dysfunction – especially serum creatinine, creatinine clearance, GFR, urea, and urinary values of albuminuria and proteinuria. – Pregnancy or lactation. – Malnourishment with a BMI < 18.5. – Reported use of narcotic drugs and psychotropic substances without a prescription within the past 12 months prior to recruitment. – Alcoholism or suspected alcoholism according to physician assessment.

2.3.2. Role of doctors/physicians

Practicing psychiatrists at study sites will assess patients’ eligibility based on inclusion and exclusion criteria, ensuring anonymization, and managing biological sample and data collection, evaluation, interpretation, and overall patient care.

2.3.3. Role of pharmacists

Pharmacists at the study sites have primary responsibilities for conducting this study.

They will be responsible for project management, genotyping, and plasma microRNA determination. They will collaborate with academic pharmacists in evaluating laboratory and clinical data, interpreting data connections, summarizing the study design, and communicating findings.

2.4. Expected outcomes of the study

2.4.1. Primary indicators

Indicators of abnormal liver function or hepatotoxicity based on standardly used liver markers (ALT, AST, ALP, GMT, TBIL) will be evaluated for individual drug or group of drugs in the selected ATC groups. The evaluation will be performed both independently of genetic polymorphisms and in subgroups according to presence/absence of CYP450/UGT1A1 variant alleles:

1. Number/prevalence of patients with abnormal results of standard liver markers in the total of enrolled patients.

2. Mean value of measured liver function tests (ALT, ADT, ALP, GMT, TBIL) according to the presence/absence of CYP450/UGT1A1 variant alleles compared in both case and control groups.

3. Genetic association between specific CYP450/UGT1A1 variant alleles and hepatotoxicity will be expressed as Odds ratio (OR), 95% confidence interval of OR (95% CI) and p values.

Evaluation of plasma microRNA target levels hsa-miR-122-5p and hsa-miR-192-5p as potential markers of abnormal liver function:

• (4) Mean difference of Normalized Relative Quantity (NRQ) of target microRNAs or alternatively their absolute quantities between case and control group.

• (5) Comparison of the mean NRQ or alternatively absolute quantities of individual microRNAs or their combination based on severity of hepatotoxicity in grades defined by classical liver markers (ALT, AST, ALP, GMT, TBIL).

• (6) Correlation of NRQ of individual microRNAs or their absolute quantities with standard liver markers.

• (7) Determination of the NRQ cutoff value for target microRNAs or the cutoff of their absolute quantities (or a combination of microRNAs) with the highest specificity and sensitivity to distinguish between control and case patients using ROC analysis.

• (8) Comparison of the mean NRQ or alternatively absolute quantities of individual microRNAs or their combination according to the presence/absence of CYP450/UGT1A1 variant alleles compared in both case and control groups.

2.4.2. Secondary indicators

The quality of life indicators, symptomatology, severity of psychiatric illness, and efficacy of pharmacotherapy will be evaluated by comparing cases and controls to assess the effect of drug-induced hepatotoxicity. Additionally, parameters will be assessed according to the genotype of CYP450/UGT1A1 genes by comparing three genotype groups – minor allele homozygotes, heterozygotes, and major allele homozygotes:

1. (9) Difference in the degree of self-assessment of quality of life, defined as the average changes in scores of the WHODAS 2.0 scale.

2. (10) Difference in self-rated measures for depressive disorder, defined as the average changes in scores of PHQ-9, BDI-II scales.

3. (11) Difference in the severity rate of depression, defined as the average changes in scores of the HAMD and MARDS scale.

4. (12) Difference in psychopathology score for psychotic diseases, defined as the average changes in scores of the BPRS, PANSS scales.

5. (13) Difference in the severity of a psychotic disorder, defined as the average changes in scores of the CGI-S scale.

Safety endpoints for antidepressant and/or antipsychotic pharmacotherapy will be evaluated by comparing cases and controls to assess the effect of drug-induced hepatotoxicity. Additionally, parameters will be assessed according to the genotype of CYP450/UGT1A1 genes by comparing three genotype groups – minor allele homozygotes, heterozygotes, and major allele homozygotes:

(14) Difference in the rate of adverse events, defined as the mean change in scales of ASEC (for antidepressant therapy), GASS (for antipsychotic therapy).

Evaluation of allele frequency of established CYP450/UGT1A1 polymorphisms

(15) The allele frequency of the selected polymorphisms listed in Table 1, defined as the incidence of polymorphism analyzed in the total sample of patients, independently of group/subgroup, will be compared with the expected allele frequency reported in the Table 1.

2.5. Clinical data collection

Data that will be collected at the time of patients’ recruitment are stated in Table 4.

Table 4.

List of clinical data obtained from patients at the time of enrollment.

1. Basic characteristics:  – age, sex, BMI, ethnicity reported by the patient  – marital status, employment status  – substance abuses (alcohol, smoking, other relevant abuses)

2. Current diseases:  – current psychiatric diagnosis (ICD-10 code)  – patients’ age at first diagnosis of the main psychiatric disease (alternatively: duration of disease in weeks/months, categorization of the onset of disease: <1 year, 1–5 years, >5 years)  – number of hospital admissions due to psychiatric reasons  – non-psychiatric diagnoses of the patient  – psychotherapy

3. Evaluation of the presence/absence of signs of hepatotoxicity:  – fatigue, nausea, vomiting, pain in the right upper abdominal quadrant, itching, rash, jaundice, weakness, anorexia, or weight loss

4. Complete medical history – The best possible medication history:  – name and dosage of any medications the patient is taking  – previous psychopharmacotherapy, or OTC self-medication (if available)  – suspected adverse drug reactions reported  – duration of current psychopharmacotherapy  – therapeutic monitoring of drug plasma concentration, if indicated  – evaluation of potential adverse drug effects in the form of questionnaires (to be filled in by the patient with/without the assistance of a physician)  – for patients taking antidepressants: ASEC – The Antidepressant Side-Effect Checklist  – for patients taking antipsychotics: GASS – Glasgow Antipsychotic Side-effect Scale

5a. Rating scales for patients with depressive disorder: Self-assessment scales (to be completed by the patient):  – Patient Health Questionnaire, Depression Module (PHQ-9)  – Beck Depression Inventory (BDI-II) Objective rating scales – one of the following assessments of the severity of depression (to be completed by a physician):  – Hamilton Depression Rating Scale (HAMD-17)  – Montgomery and Asberg Depression Rating Scale (MARDS) Quality of life assessment (to be completed by the patient with/without the help of a physician)  – WHODAS 2.0 (WHO Disability Assessment Schedule)

5b. Rating scales of patients with psychotic disease: Evaluation of the severity of the disease (completed by a physician):  – Global clinical impression of severity (CGI-S) Assessment of psychopathology (filled out by the physician, interview with the patient):  – Short Psychiatric Assessment Scale (BPRS)  – Positive and Negative Syndromes Scale (PANSS) Quality of life assessment (to be completed by the patient with/without the assistance of a physician)  – WHODAS 2.0 (WHO Disability Assessment Schedule)

2.6. Laboratory measurements

Two blood samples will be obtained from each patient once at the time of recruitment.

2.6.1. Blood collection and initial processing

Two venous blood samples will be taken from each patient during scheduled regular appointment with the psychiatrist: one in an EDTA tube (BD Vacutainer EDTA 6 ml) and one in a serum analysis tube (BD Vacutainer SST 6 ml). EDTA blood will be split into two fractions: one for genotyping and the other processed into plasma by centrifugation (15 min, 1200 × G, at room temperature) on the collection day within four hours after collection and then immediately frozen until further processing. Plasma and whole blood tubes will be stored at −70°C to −80°C securely. The serum will be sent for routine commercial laboratory tests for liver markers.

2.6.2. MicroRNA extraction and analysis

RNA isolation from plasma will be performed by using TRI Reagent BD (Merck/Sigma-Aldrich), incorporating carrier RNA (MS2 Carrier RNA; Roche) and synthetic microRNA (cel-miR-39-3p; Merck/Sigma-Aldrich). Selected target and control microRNAs will be detected by RT-qPCR on QuantStudio 3 (Thermo Fisher Scientific) using specific probes. Results analyzed using Pfaffl method [45] and LinregPCR software [46] will be expressed as normalized relative quantity (NRQ). Alternatively, we will use the method of absolute quantification using digital PCR: QuantStudio Absolute Q (Thermo Fisher Scientific).

2.6.3. Genomic DNA extraction and genotyping

Genomic DNA will be isolated from whole blood using the QIAamp® DNA Blood Mini Kit (Qiagen). Single nucleotide polymorphisms (SNPs) and copy number variations (CNVs) analysis will be performed by quantitative or digital polymerase chain reaction (qPCR/dPCR) using relevant TaqMan assays (Thermo Fisher Scientific) for specific polymorphisms listed in Table 1. PCR with high-resolution melting (HRM) analysis will be used to detect TA repeats in the UGT1A1 promoter region.

2.6.4. Routine commercial determination of basic liver markers

Blood serum sample will be commercially analyzed for the basic markers of liver function: ALT, AST, ALP, GMT and TBIL. The results of the determination will be used to assess the severity of hepatotoxicity as described in Supplementary table S2 and Table 5.

Table 5.

Assessment of severity of hepatotoxicity based on laboratory measurements.

Stage 0	Liver enzymes < ULN Total bilirubin < ULN
Stage 1	Liver enzymes > ULN; but at the same time ≤ 3 × ULN Total bilirubin > ULN; but at the same time ≤ 1.5 × ULN
Stage 2	Liver enzymes > 3 × ULN; but at the same time ≤ 5 × ULN Total bilirubin > 1.5 × ULN; but at the same time ≤ 3 × ULN
Stage 3	Liver enzymes > 5 × ULN; but at the same time ≤ 20 × ULN Total bilirubin > 3 × ULN; but at the same time ≤ 10 × ULN
Stage 4	Liver enzymes > 20 × ULN Total bilirubin > 10 × ULN

ULN – upper limits of normal; liver enzymes: ALT, AST, ALP, GMT.

The criteria for assessing hepatotoxicity have been formulated based on the Common terminology criteria for adverse events (CTCAE, version 5.0) [47].

2.7. Quality control

2.7.1. Genotyping quality

All samples will be analyzed in duplicate. In the event of inconsistent results, the analysis will be repeated in duplicate. If the second result is also inconsistent, it will not be included. In the event of no result, the analysis will be repeated in duplicate. If the issue persists, the entire procedure will be repeated, including DNA extraction. Positive and No Template Controls will be included during the measurements.

2.7.2. Quality of microRNA determination

Control of RNA isolation and target microRNA identification in blood plasma will involve adding a standard amount of synthetic control microRNA (cel-miR-39-3p), typically absent in human samples. MicroRNA determination in all samples requires three technical replicates with the following criteria: cycle of quantification of all replicates Cq <35 with coefficient of variation of the triplicate CV < 35% [48]. Target microRNA results will be normalized to the control synthetic microRNA values.

2.8. Adherence to pharmacotherapy

Nonadherence to pharmacotherapy may bias study outcomes. Patients with suspected significant nonadherence to psychiatric medication will be excluded based on physicians’ assessment.

2.9. Sample size

Estimated sample size for was calculated with QUANTO version 1.2.4 (2009) based on following assumptions: 1) unmatched Case-Control design for discrete traits; 2) 10% prevalence of elevated liver enzymes [4]; 3) 1:4 ratio of cases:controls; 4) 80% power, α = 0.05 with two-sided test; 5) four genetic models, from which the sample size estimates decrease in the following order: dominant > additive > multiplicative ≈ allelic; recessive model was excluded due to unfeasible estimate.

Finally, two factors were considered: minor allele frequency (MAF) 1% − 10% in the European population and odds ratio (OR) 2–4. Total number of patients for the dominant model strategy with MAF = 1% and OR = 4 is expected to be 755, of which 151 cases and 604 controls. In case of lower OR (MAF = 1%, OR = 2) the recruitment of patients will continue up to 3845 individuals in total.

2.10. Potential sources of bias

The occurrence of hepatotoxicity may be influenced by several confounding factors, particularly: body mass index (BMI), age, sex, smoking status, the number of drugs (individual active substances, both prescription and non-prescription) used by an individual, total defined daily doses (DDDs as defined by the WHO Collaborating Centre for Drug Statistics Methodology) of combined pharmacotherapy, and the presence of potentially hepatotoxic drugs identified by the LiverTox database with a likelihood score of A or B. Modifiers of CYP450 activity – specifically inhibitors and inducers that can lead to drug-drug-gene interactions – will be monitored and also considered as confounding factors, including: (i) drugs such as antiepileptics, macrolide antibiotics, fluoroquinolones, azole antifungals, and SSRIs like fluvoxamine and paroxetine; and (ii) other common modifiers, including smoking, alcohol, grapefruit juice, and St. John’s wort. All these factors may impact the plasma levels and elimination of antidepressants and antipsychotics, potentially contributing to adverse drug reactions, including hepatobiliary problems. These covariates will be used to calculate adjusted odds ratios (ORs) when assessing genetic associations.

This study will employ a targeted panel of gene variants for enzymes involved in drug biotransformation (CYP450, UGT1A1) that influence the availability and plasma levels of antipsychotics and antidepressants, which thus far have not been adequately associated with hepatobiliary complications in the context of the medications studied. However, other polymorphisms known to be associated with drug-induced liver injury (DILI), especially those in HLA genes, PTPN22, ERAP2, and others, have been linked with certain medications. Nonetheless, it is not feasible to account for this residual confounding factor in the current analysis.

The measurement of allele frequencies in the Slovak population will be limited to psychiatric patients rather than a randomized population, which may also serve as a potential source of bias.

2.11. Missing data management

In statistical computations, an available-case analysis approach will be employed if missing data are encountered; however, only variables with less than 5% missing data will be considered.

2.12. Statistical methods

All statistical analyses will be conducted using GraphPad Prism, R Software, and Microsoft Excel. Data will be presented using descriptive statistics, including arithmetic mean with sample standard deviation (SD) or counts with percentages (%). Statistical differences between two groups (controls, cases) of continuous data will be assessed using the Mann-Whitney U test or Welch’s t-test, depending on the estimated data distribution determined by the Shapiro–Wilk test. The Fisher exact test will be employed for comparing proportions. The allele frequencies of individual variants in the sample will be compared to the allele frequency in the European (non-Finnish) population using the one proportion Z test. Genetic associations between variant alleles and hepatotoxicity will be evaluated using multiple genetic models, including dominant, additive, multiplicative, and allelic models. These models will undergo testing using logistic regression analysis and the Fisher exact test. Genetic association will be expressed as Odds Ratio (OR) with a 95% confidence interval and related measures of accuracy and predictive power, including receiver-operating characteristics derived AUC, sensitivity, specificity, and others. The OR will be adjusted for body mass index (BMI), age, sex, smoking status, the number of drugs used by an individual, total defined daily doses (DDDs) of combined pharmacotherapy, the presence of potentially hepatotoxic drugs identified by the LiverTox database with a likelihood score of A or B, and concomitant use of CYP450 activity modifiers – enzyme inhibitors and inducers. Multiple comparison correction will be performed using the Benjamini–Hochberg method where appropriate. A result will be considered statistically significant if the achieved p-value or adjusted p-value is <0.05.

2.13. Hardy-Weinberg equilibrium

For each identified polymorphism, agreement with expected Hardy–Weinberg equilibrium (HWE) will be assessed using Fisher’s exact test.

2.14. Ethical approval

2.14.1. Approval by ethics committees

Biomedical research has already been approved by the Ethics Committee of the University Hospital Bratislava, Hospital of Academic L. Dérer, Limbová 5, 833 05 Bratislava, Slovak Republic on 20 October 2020. The requirements of Good Clinical Practice (ICH-GCP), the State Institute for Drug Control and valid legislation for clinical trials in the Slovak Republic were considered.

2.14.2. Informed consent of study participants

Study participants will be instructed and will provide approved written informed consent to participate in the study and provide a sample for research purposes in accordance with the laws of the Slovak Republic and the declaration of Helsinki.

2.14.3. Anonymisation and access to samples and data

All patient data will be anonymized using a unique, non-retrospectively identifiable code, overseen by the responsible physician in healthcare facilities. Results from clinical data and biological samples will be processed in a coded format at the Department of Pharmacology and Toxicology, Faculty of Pharmacy, Comenius University Bratislava, preventing patient re-identification. Biological samples will be securely stored at −70°C and disposed of appropriately once the study is complete.

2.14.4. Communication of study results

The results of the study will be disseminated in line with good scientific practice. The study findings will be published in both national and international scientific journals and presented at scientific events. The result concerning genotype of patients is intended only for research purposes and would not be communicated to them.

3. Discussion

The incidence of psychiatric disorders in Slovakia and worldwide has risen, leading to increased antidepressant and antipsychotic use, often lifelong, posing hepatotoxic risks. Early liver damage detection is crucial for safe treatment. Plasma levels of these drugs depend on CYP450 enzyme metabolism, influenced by genetic variability. Pharmacogenetic tailoring of the therapy can enhance treatment effectiveness, reduce risks, and optimize resource use in the public health system [49]. Several nations are adopting pharmacogenetic testing-guided therapy, including the UK pharmacogenomics program [50]. This study in Slovak psychiatric patients will explore pharmacogenetic-guided therapy’s potential to prevent therapeutic failures and hepatotoxicity due to drug-drug-gene interactions and interindividual variability. The study will also provide valuable insights into CYP450/UGT1A1 enzyme polymorphism allele frequencies in Slovakia.

3.1. Drug-induced hepatotoxicity

Drug-induced hepatotoxicity can result from known or unknown causes, either dose-related or idiosyncratic. Expected damage occurs predictably and swiftly due to direct drug toxicity, while idiosyncratic injuries, more common in practice, have unpredictable onsets [51,52]. Genetics play a role in specific drug-induced hepatoxicity, as shown in Genome Wide Association Studies and Candidate Gene Studies. Variations in injury incidence and severity are linked to individual genetic makeup and polymorphisms. Antibiotics, antivirals, antituberculotics, theophylline, and valproic acid are among the drugs extensively studied. However, limited data exist on drug-induced liver injury during psychiatric therapy and its association with drug-metabolizing enzymes` polymorphisms [52].

3.2. Incidence of adverse drug reactions associated with pharmacotherapy of affective and psychotic disorders

Some antidepressants and antipsychotics should be used cautiously or avoided altogether due to their effects on CYP450 enzymes, impacting drug metabolism. For example, agomelatine is primarily metabolized by CYP1A2 and in lesser extent by CYP2C9/19, with potent inhibitors like fluvoxamine and ciprofloxacin significantly hindering its metabolism, raising the risk of liver damage, including failure, hepatitis, and jaundice. Patients with risk factors such as obesity, fatty liver disease, diabetes, heavy alcohol consumption, or use of hepatotoxic drugs should be carefully assessed before treatment initiation [53]. However, Summaries of Product Characteristics of agomelatine lack information on alleles affecting function of CYP1A2 and CYP2C9 enzymes, which could predispose patients to adverse reactions or treatment failure. Consequently, a trial-and-error approach often leads to prolonged treatment adjustments, impacting patients’ quality of life, particularly in depression cases. Agomelatine is the only antidepressant with documented associations between CYP1A2 and CYP2C9 polymorphisms and plasma levels, but these findings are limited to studies in healthy volunteers receiving single doses [35,41,42]. However, systematic studies on CYP450 polymorphisms in antidepressant/antipsychotic users, especially regarding liver function safety, are lacking in the literature.

Selected variants of the UGT1A1 gene, involved in drug glucuronidation, will also be studied. UGT1A1*28, associated with Gilbert syndrome, may predispose individuals to drug toxicities due to impaired glucuronidation. Although not all homozygous individuals will show symptoms, those with Gilbert syndrome, with a prevalence rate between 5 and 16%, will be susceptible to medication-induced hyperbilirubinemia. Additionally, variant UGT1A1*80, which is in almost 100% linkage disequilibrium with UGT1A1*28 in European populations, will be methodically easier to measure, less expensive with a shorter time to decision-making than UGT1A1*28 genotyping, since there is commercially available TaqMan probe [33,54].

3.3. Prevalence of CYP450/UGT1A1 polymorphisms in the Slovak population

Current data on CYP450 genetic polymorphisms in Slovakia are limited. In the European population CYP2D6 and CYP2C19 allele frequencies were studied, noting that CYP2C19*17 allele is the most common in Slovakia, at 33% compared to 23% in the wider European population [55]. However, this study included only 26 Slovak individuals. Our project aims to provide a more comprehensive analysis of these polymorphisms in Slovakia. Assuming similar allele frequencies to the European average [25], adjustments may be needed if Slovak frequencies diverge significantly. This will inform the selection of relevant polymorphisms for future pharmacogenetic testing panels in Slovakia.

3.4. Pharmacogenetic panel – allele selection

In the literature, the authorities, such as the Association for Molecular Pathology Pharmacogenomics (PGx) Working Group, already recommend a minimal set of variants, that should be included in clinical pharmacogenetic genotyping assays to ensure uniformity across clinical laboratories [30–32]. However, in our case, modifications to these minimum sets were necessary, aligning with specific study purposes and the environment outlined in the methodology section. These adjustments aim to address the study’s objectives more effectively, with a particular focus on drug-induced hepatotoxicity in psychiatric patients. Additionally, they consider factors such as estimated allele frequency in the Slovak population.

3.5. Abnormal liver function tests (LFTs)

In patients taking antidepressants or antipsychotics, mildly abnormal liver function tests may signify potential medication-induced liver hepatotoxicity/liver injury, necessitating closer monitoring and possible adjustments to the medication regimen. While short-term consequences include the need for increased vigilance and potential dose modifications, long-term implications encompass the risk of progression to liver disease and associated complications, such as nonalcoholic fatty liver disease (NAFLD), hepatitis, even liver failure [4,5]. Estimating the risk of complications in patients with mildly elevated LFTs involves considering factors such as underlying liver health, severity, and duration of LFT elevation, concomitant medications, comorbidities, and patient-specific factors, such as age, diet, obesity and genetics [3,4].

Insufficient monitoring of liver functions in psychiatric patients and inadequate reporting of hepatobiliary adverse effects of psychopharmaceuticals represent significant gaps in clinical care and pharmacovigilance. Studies indicate that a substantial proportion of psychiatric patients, particularly those on antidepressants or antipsychotics, do not undergo routine liver function testing despite the potential hepatotoxicity of these medications. However, determining the true incidence is complicated by underestimation, with discrepancies in terminology and diagnostic criteria leading to difficulties in recognizing and documenting cases of hepatotoxicity. Additionally, underperformance of liver function tests by physicians contributes to underreporting and challenges in estimating the real prevalence [2,3].

In up to one third of individuals using antipsychotics, abnormal liver function tests (LFTs) were detected [8], with 4% showing clinically significant elevations, usually within 6 weeks and frequently without symptoms [7,9]. The situation is similarly concerning for patients taking antidepressants. Despite LFT testing in only one-third of antidepressant users, up to 20% of them exhibited abnormal values, and 1% were confirmed to have antidepressant-induced liver injury. Furthermore, patients on antidepressants tend to display higher rates of early LFT abnormalities, and those with major depressive disorder have a notably increased prevalence of chronic liver disease compared to the general population [56]. This underscores the importance of thorough monitoring and management of LFT abnormalities in individuals prescribed psychiatric medications.

Drug-induced hepatotoxicity or liver injury (DILI) is of concern with every drug class. There are multiple databases, such as LiverTox [56] or DILIrank [57] which classify the evidence-based risk of hepatotoxicity. In these sources are listed both antidepressants and antipsychotics with a range of likelihoods of hepatotoxicity, from unlikely to highly likely or unknown.

3.6. New biomarkers of drug-induced hepatotoxicity

Exploring new biomarkers for early detection of drug-induced hepatotoxicity is vital to minimize patient harm. MicroRNAs like hsa-miR-122-5p and hsa-miR-192-5p show promise in this regard, potentially outperforming traditional liver function tests [17]. Identifying liver damage promptly is crucial to prevent irreversible harm, as late detection can lead to fatal consequences. Furthermore, testing microRNAs could help connect treatment with existing adverse effects or predict their development in predisposed patients. Notably, individuals with CYP450 polymorphisms, particularly those with the Poor/Intermediate metabolizer phenotype, face heightened risk of liver injury, underscoring the clinical importance of early detection markers [17].

In conclusion, combining pharmacogenetic testing for CYP450/UGT1A1 polymorphisms with epigenetic testing using microRNAs as indicators of drug-induced hepatotoxicity offers a novel approach to the optimization of antidepressant/antipsychotics therapy. While demanding in laboratory expertise, its clinical potential is significant. Analysing adverse drug reactions alongside patient clinical data and genetic variations could help prevent drug-induced hepatobiliary complications and determine safe dosages tailored to individual patients. Patients identified as poor drug metabolisers based on CYP450 polymorphisms could benefit from routine microRNA testing for early liver injury detection. This combined approach holds great promise for personalized pharmacotherapy.

3.7. Limitations of the study

Study limitations include incomplete data availability and patient medical records, potentially impacting results. Treatment adherence, particularly challenging in psychiatric patients, may also influence study outcomes. Questionnaire completion relies on subjective assessment by physicians or patients, introducing potential bias. Difficulty in reaching required patient numbers due to factors like declined consent or low hepatotoxicity incidence is another risk. The study focuses on a targeted pharmacogenetic panel, which limits the detection and genetic association analysis of rare or novel alleles and polymorphisms in genes not included in the panel, even if they have been identified in the literature as potentially associated with DILI. Additionally, the retrospective design is a study limitation because it does not enable tailoring therapy in advance based on pharmacogenetic results. Nevertheless, retrospective design is suitable for miRNA testing for monitoring of possible hepatoxicity.

3.8. Relatedness among study participants

Only patients who, according to the doctor’s and patients’ understanding, are not blood-related to other patients, will be included in the study.

Funding Statement

The study is financially supported by grants [UK/276/2021, FaF/16/2022, UK/131/2023, VEGA 1/0203/19]. Grant Univerzity Komenského [UK/131/2023]; Grant Farmaceutickej fakulty Univerzity Komenského [FaF/16/2022]; Vedecká Grantová Agentúra MŠVVaŠ SR a SAV [VEGA 1/0203/19].

Article highlights

• The genetic association between specific CYP450/UGT1A1 variant alleles and hepatotoxicity in Slovak patients undergoing antidepressant or antipsychotic therapy will be evaluated.

• The study will assess whether plasma levels of microRNAs hsa-miR-122-5p and hsa-miR-192-5p could serve as potential early markers of abnormal liver function.

• Combining pharmacogenetic testing for CYP450/UGT1A1 polymorphisms with epigenetic testing using microRNAs as indicators of drug-induced hepatotoxicity offers a novel approach to optimizing antidepressant and antipsychotic therapy.

• Analysing adverse drug reactions alongside patient clinical data and genetic variations could help prevent drug-induced hepatotoxicity and tailor drug dosages to individual patients.

Disclosure statement

The authors have no 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. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Administrative note

The manuscript utilized AI-driven language processing tools for British English correction, ensuring consistency and accuracy. Furthermore, AI-powered text summarization techniques were employed to condense the original primary text, maintaining integrity while enhancing readability and clarity.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14622416.2025.2456449.