Adverse event profiles of drug-induced liver injury caused by antidepressant drugs: a disproportionality analysis

Abstract

Background:

Antidepressants are widely used to manage depression and other psychiatric diseases. A previous study revealed that hepatotoxicity was the main adverse event related to antidepressants. Therefore, drug-induced liver injury (DILI) caused by antidepressants deserves more attention.

Objectives:

To investigate DILI adverse events reported due to antidepressant use in the United States Food and Drug Administration Adverse Events Reporting System (FAERS) database.

Research design:

A disproportionality analysis of spontaneously reported adverse events was conducted to assess the association between antidepressant drugs and DILI.

Methods:

FAERS data from 1 January 2004 to 31 December 2021 were compiled and analyzed using the reporting odds ratio (ROR) and information component (IC).

Results:

As per the FAERS database, of the 324,588 cases that were administered antidepressants, 10,355 were identified as cases with DILI. Among the identified 42 antidepressants, nefazodone (n = 47, ROR = 7.79, IC = 2.91), fluvoxamine (n = 29, ROR = 4.69, IC = 2.20), and clomipramine (n = 24, ROR = 3.97, IC = 1.96) had the highest ROR for cholestatic injury; mianserin (n = 3, ROR = 21.46, IC = 3.99), nefazodone (n = 264, ROR = 18.67, IC = 3.84), and maprotiline (n = 15, ROR = 5.65, IC = 2.39) for hepatocellular injury; and nefazodone (n = 187, ROR = 12.71, IC = 0.48), clomipramine (n = 35, ROR = 2.07, IC = 0.26), and mirtazapine (n = 483, ROR = 1.96, IC = 0.94) for severe drug-related hepatic disorders. Only nefazodone elicited hepatic failure signals (n = 48, ROR = 18.64, IC = 4.16). There are limited reports on the adverse reactions of relatively new antidepressant drugs, such as milnacipran, viloxazine, esketamine, and tianeptine, and those not approved by the Food and Drugs Administration, such as reboxetine and agomelatine.

Conclusion:

A significant association was observed between DILI and nefazodone. Duloxetine and clomipramine were associated with three DILI categories, except hepatic failure. The disproportionality analysis cannot conclude on a definite causal link between antidepressants and DILI. Additional research is required to assess new-generation antidepressants for their propensity to cause DILI.

Plain language summary

Adverse events reported on drug-induced liver injury caused by antidepressants

Introduction: Adverse drug events (ADEs) refer to all harmful events related to medications, including adverse drug reactions (ADRs) and other unexpected events. ADEs encompass a wider range and are very important for the post-market surveillance of drugs. This study investigated the voluntary reporting of drug-induced liver injury (DILI) adverse events associated with antidepressant drugs.

Methods: We retrieved data on DILI and related terms submitted between 2004 and 2021 from the United States Food and Drug Administration Adverse Events Reporting System (FAERS) database. We analyzed the data for the detection of DILI signals associated with antidepressants.

Results: We retrieved and analyzed 324,588 reports on antidepressant drugs. A total of 10,355 reports were associated with DILI.

The three drugs with the highest reporting odds ratio (ROR) in each DILI category were as follows:

• cholestatic injury (nefazodone, fluvoxamine, and clomipramine)

• hepatocellular injury (mianserin, nefazodone, and maprotiline)

• hepatic failure (nefazodone)

• drug related hepatic disorders-severe events (nefazodone, clomipramine, and mirtazapine)

The absence of signals from some drugs may be due to:

• non-association with DILI

• novelty of the drug in the market

• non-approval from the Food and Drugs Administration (FDA)

• lack of voluntary reporting of adverse events due to other reasons

Conclusion: Drug safety studies utilizing publicly available large databases allowed the evaluation of the safety profile of widely used antidepressant drugs in clinical practice. Nefazodone, duloxetine, and clomipramine were associated with significant DILI signals. Further research is needed to determine the safety concerns of new-generation antidepressants.

Introduction

Drug-induced liver injury (DILI) is a challenging clinical problem because it is often underdiagnosed and is an underestimated cause of liver injury. The incidence of DILI is approximately 14–19 cases per 100,000 persons, 30% of whom are diagnosed with jaundice. ¹ It also majorly accounts for the market withdrawal or removal of first-line therapeutic drugs. ² In developed countries, non-acetaminophen idiosyncratic DILI (IDILI) is the second most common etiology of acute liver failure (ALF) and accounts for 10–20% of ALF, second only to acetaminophen. ³ In recent years, DILI has received increasing attention from the public and medical community. In the last 5 years, new guidelines on the management of DILI have been released and updated. The European Association for the Study of the Liver, ⁴ Russian Association, ⁵ Asian Pacific Association for the Study of the Liver, ⁶ and the American Association for the Study of Liver Diseases ⁷ have issued guidelines or expert consensus on DILI. The American College of Gastroenterology ⁸ clinical guidelines for IDILI were also republished in 2021.

DILI can be classified into three groups, hepatocellular, cholestatic, and mixed, according to the pattern of elevation of liver enzymes evaluated by the first available biochemical tests. The type of liver injury is categorized using the R value, a standard and reliable tool calculated by dividing serum alanine aminotransferase/upper limit of normal (ULN) value with serum alkaline phosphatase/ULN value. An R value of ⩾5 indicates hepatocellular, ⩽2 indicates cholestatic, and 2–5 indicates mixed injury.^4,6,7 The risk factors for DILI include host- and drug-dependent risk factors. Age, sex, race, alcohol consumption, pregnancy, and underlying diseases are the host-dependent risk factors. Drug dose, lipophilicity, concomitant drugs, and special chemical moieties involved in hepatic metabolism are drug-dependent risk factors. Some drugs have specific injury patterns, such as the role of acetaminophen in hepatocellular injury, whereas others may cause multiple patterns of liver injury. These factors warrant more exploration related to DILI.

Depression involves various mood-related symptoms and is one of the most common mental health conditions. The prevalence of depression in the young population (aged 10–24 years) has sharply increased over the past decade. ⁹ The shortest duration of adequate treatment for unipolar major depression is 6–12 weeks. Moreover, in patients with persistent depressive disorder, antidepressant drug maintenance therapy is usually administered for approximately 2 years or more. ¹⁰ In addition to treating depression, most of the newer antidepressants are also used as the first-line pharmacotherapy in anxiety disorders, bipolar disorders, obsessive-compulsive disorders, and other psychiatric diseases. A review highlighted that hepatotoxicity was the main adverse event related to the use of newer-generation antidepressants, which are considered to be safer than tricyclic agents. ¹¹ Considering the wide clinical use and long duration of antidepressant therapy, antidepressant-related adverse events should be given more attention, especially regarding the possibility of ALF and DILI. Therefore, understanding DILI associated with antidepressants and safety monitoring is essential to optimize antidepressant treatment outcomes. Previous reports were based on retrospective multicenter studies,^12,13 case reports,^14–17 or case series. ¹⁸ To the best of our knowledge, there has been no comprehensive evaluation of the association between individual antidepressant drugs and DILI.

To provide more information on DILI and antidepressant drugs, we conducted a pharmacovigilance analysis using United States Food and Drug Administration Adverse Event Reporting System (FAERS) data. This study may provide information to clinicians about the DILI profile of antidepressant drugs in real-world clinical practice.

Methods

FAERS data

Adverse drug event cases were obtained from post-marketing adverse drug reactions reported to the MedWatch program enacted by the FAERS. ¹⁹ The FAERS data included patient demographic and administrative information (DEMO), drug information (DRUG), adverse events (REAC), patient outcomes (OUTC), report sources (RPSR), therapy start and end dates for reported drugs (THER), and indications for drug administration (INDI). ²⁰

We performed a retrospective pharmacovigilance study using data obtained from the FAERS database from the first quarter of 2004 to the fourth quarter of 2021. We excluded the ‘duplicated’ and ‘deleted’ reported cases by following the Food and Drugs Administration (FDA) recommendation in the DEMO table. We removed the same records from the same ‘CASEID’ column and retained the earliest record. If the ‘CASEID’ and ‘FDA_DT’ were the same, the higher ‘PRIMARYID’ was selected. We managed the FAERS data using the Microsoft Office Excel 2017 software.

Identification of antidepressant drugs

For drugs, generic names, trade names, or abbreviations were all included. Before identifying target drugs, we standardized different names of the same drug into a ‘generic name’ using the MedEx software (MedEx UIMA 1.3.8, Vanderbilt University, Nashville, TN, USA). We identified the target antidepressants according to the World Health Organization (WHO) Anatomical Therapeutic Chemical (ATC) classification (ATC code: N06A) in the DEMO_DRUG table (DEMO and DRUG table linked by a primary ID), including the non-selective monoamine reuptake inhibitors (NSMRis) (ATC code: N06AA), selective serotonin reuptake inhibitors (SSRis) (ATC code: N06AB), non-selective monoamine oxidase inhibitors (NMOis) (ATC code: N06AF), monoamine oxidase A inhibitors (MOAis) (ATC code: N06AG), and other antidepressants (OTAs) (ATC code: N06AX). A total of 65 single antidepressants were identified. Finally, we included cases in which antidepressants were reported as ‘primary suspect’ drugs; 42 antidepressants were involved.

Identification of cases with DILI events

The preferred terms (PTs) refer to the recommended medical terminology that describes an event. The Medical Dictionary for Regulatory Activities (MedDRA) version 23.1 was used to code the adverse events Medical Dictionary for Regulatory Activities (MedDRA), developed by the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH). You can see more in https://www.meddra.org/how-to-use/tools/smqs. We identified cases of hepatic impairment in the DEMO_REAC table (DEMO and REAC table linked by primary ID). DILI events were identified in two ways. First, severe DILI cases were identified using a narrow search of Standardized MedDRA Queries (SMQs) (version 23.1, SMQ code: 20,000,007), including 47 PTs (Supplemental Table S1). Duplicate records were excluded from the analysis. Second, DILI cases were grouped into three categories based on evidence-based classification, which was also used in our previous studies. ²¹ Finally, after combining the identified DILI events, DILI cases were grouped into four categories: hepatocellular injury (24 PTs), cholestatic injury (12 PTs), hepatic failure (6 PTs), and drug-related severe hepatic disorder events (47 PTs) (Supplemental Table S1). The data collected included case ID, drug indications, suspected drugs, adverse events, serious outcomes, country of occurrence, reporter type, sex, age, treatment date, and event date. We further excluded patients with hepatic impairment complications. The reported number of DILI cases from 2004 to 2021 determined the sample size.

Statistical analyses

We used the reporting odds ratio (ROR, frequency method) and the information component (IC, Bayesian approach) to detect the ‘signal’ (the association between liver injury and antidepressants). For ROR, a case number ⩾3 and the lower limit of the two-sided 95% confidence interval (95% CI) > 1 implied a signal detection. A ROR (95% CI) crossing a value of 1 was interpreted as statistically significant. An IC > 0 and the lower limit of the 95% CI > 0 implied signal detection. Both methods were used for signal detection, and liver injury was considered drug-associated if both ROR and IC met the above-mentioned criteria. ²¹ Data were analyzed using Microsoft Excel 2020 (Microsoft Corp., Redmond, WA, USA) and Statistical Package for the Social Sciences (SPSS) version 26.0 (IBM Corp., Armonk, NY, USA). The ROR, IC, and 95% CI calculation methods are presented in Supplemental Table S2. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement ²² and the reporting of a disproportionality analysis for drug safety (READUS) statement. ²³

Results

Characteristics of included FAERS cases

After standardization and deduplication, we retrieved 324,455 individual case safety reports with information on age, sex, and other individual characteristics from the FAERS database between Q1 2004 and Q4 2021. The antidepressant drugs were classified based on the drug name and ATC codes, which is a WHO international classification system. We attempted to identify all the antidepressant drugs. However, only 42 drugs were identified in the present study. A flow diagram depicting the selection of antidepressant drugs from the FAERS database is shown in Figure 1. We retrieved data from 324,588 patients who were administered antidepressants as primary suspected drugs. After excluding hepatic impairment complication cases, a total of 324,455 individual cases were included, out of which 10,355 (3.19%) were DILI cases and 314,100 (96.81%) were non-DILI cases. The details are provided in Supplemental Table S3.

Figure 1.

Flow chart of the identification of antidepressant-induced liver injury cases from the FAERS database.

FAERS, United States Food and Drug Administration Adverse Event Reporting System; MOAi, monoamine oxidase A inhibitor; NMOi, non-selective monoamine oxidase inhibitor; NSMRi, non-selective monoamine reuptake inhibitor; OTAs, other antidepressants; SSRi, selective serotonin reuptake inhibitor.

The demographic and drug-class exposure characteristics of the DILI and non-DILI cases are presented in Table 1. This population included 6291 female (60.80%) and 3387 male patients (32.70%). DILI occurrence was majorly observed in the 18–64 years age group (56.70%). Physicians, pharmacists, and other health professionals reported 66.4% of the DILI cases. Most patients were from Europe (4426 cases, 42.70%), followed by North America (3645 cases, 35.20%) and Asia (1004 cases, 9.70%). The proportion of life-threatening events and hospitalizations (initial or prolonged) was much higher in the DILI group than in the non-DILI group.

Table 1.

Primary characteristics of antidepressant cases included from FAERS.

Characteristics	DILI cases		Non-DILI cases
	Number (n)	Proportion (%)	Number (n)	Proportion (%)
Drug classification
OTA	6382	61.60	167,409	53.30
SSRi	3687	35.60	136,310	43.40
NSMRi	267	2.60	8983	2.90
NMOi	18	0.20	1390	0.40
MOAi	1	0.00	8	0.00
Age group
<18 years	361	3.50	150,46	4.80
18–64 years	5873	56.70	142,962	45.50
⩾65 years	1737	16.80	41,198	13.10
Unknown	2384	23.00	114,894	36.60
Gender
Female	6291	60.80	190,322	60.60
Male	3387	32.70	90,344	28.80
Unknown	677	6.50	33,434	10.60
Reporter occupation
Health professionals	6871	66.40	147,491	47.00
Non-health professionals	2778	26.80	145,766	46.40
Unknown	706	6.80	20,843	6.60
Reported region
Europe	4426	42.70	91,032	29.00
North America	3645	35.20	182,018	57.90
Asian	1004	9.70	10,593	3.40
Oceania	173	1.70	2934	0.90
South America	83	0.80	2480	0.80
Africa	35	0.30	590	0.20
Unknown	989	9.60	24,453	7.80
Outcomes
HO	3400	32.80	56,935	18.10
OT	3091	29.90	88,439	28.20
DE	808	7.80	27,197	8.70
LT	764	7.40	13,908	4.40
DS	121	1.20	8838	2.80
RI	54	0.50	1745	0.60
CA	33	0.30	8794	2.80
UK	2084	20.10	108,244	34.50

CA, congenital anomaly; DE, death; DS, disability; FAERS, United States Food and Drug Administration Adverse Event Reporting System; HO, hospitalization-initial or prolonged; LT, life-threatening; MOAi, monoamine oxidase A inhibitor; NMOi, non-selective monoamine oxidase inhibitor; NSMRi, non-selective monoamine reuptake inhibitor; OT, other serious (important medical event); OTA, other antidepressant; RI, required intervention to prevent permanent impairment or damage; SSRi, selective serotonin reuptake inhibitor; UK, unknown.

The number of adverse reaction cases related to antidepressant drugs increased slowly after 2007; however, there was a sudden increase in adverse event reports in 2015 (Figure 2). The number of DILI cases has fluctuated slightly over the past 18 years. However, the number of DILI cases related to SSRi increased dramatically between 2018 and 2019. By contrast, the number of DILI cases related to OTA decreased after 2013.

Figure 2.

Annual number of reported DILI and non-DILI cases associated with antidepressant drugs in FAERS.

DILI, drug-induced liver injury; FAERS, United States Food and Drug Administration Adverse Event Reporting System.

DILI signal detection

We first analyzed severe drug-related hepatic disorders identified using a narrow SMQ search. OTA, SSRi, NSMRi, and NMOi did not detect significant signals. The signal detection results for hepatocellular injury, cholestatic injury, and liver failure are presented in Table 2. OTAs were significantly associated with cholestatic injury (ROR: 1.20, 95% CI: 1.13–1.27; IC: 0.26, 95% CI: 0.07–0.45) and hepatocellular injury (ROR: 1.67, 95% CI: 1.62–1.72; IC: 0.71, 95% CI: 0.62–0.81), while SSRis were significantly associated with cholestatic injury (ROR: 1.34, 95% CI: 1.26–1.42; IC: 0.41, 95% CI: 0.21–0.61).

Table 2.

RORs and ICs of antidepressants.

DILI category	Drug classification	DILI cases (n)	All AE cases (n)	DILI proportion (%)	ROR	95% CI for ROR	IC	95% CI for IC
Drug-related severe hepatic disorders	OTA	2698	17,3791	1.55	0.95	0.91–0.99	−0.07	−0.20−0.05
	SSRi	2054	13,9997	1.47	0.90	0.86–0.94	−0.15	−0.30−0.01
	NSMRi	144	9250	1.56	0.95	0.81–1.12	−0.07	−0.61−0.48
	NMOi	7	1408	0.50	0.30	0.14–0.63	−1.72	−3.89−0.72
Cholestatic injury	OTA	1215	17,3791	0.70	1.20	1.13–1.27	0.26	0.07–0.45
	SSRi	1089	13,9997	0.78	1.34	1.26–1.42	0.41	0.21–0.61
	NSMRi	76	9250	0.82	1.41	1.13–1.77	0.49	−0.26−1.23
	NMOi	5	1408	0.36	0.61	0.25–1.46	−0.72	−3.29−2.04
Hepatocellular injury	OTA	4894	17,3791	2.82	1.67	1.62–1.72	0.71	0.62–0.81
	SSRi	2469	13,9997	1.76	1.03	0.99–1.07	0.04	−0.09−0.17
	NSMRi	194	9250	2.10	1.23	1.06–1.41	0.29	−0.18−0.76
	NMOi	14	1408	0.99	0.57	0.34–0.97	−0.79	−2.44−0.94
	MOAi	1	9	11.11	7.15	0.89–57.21	2.69	−4.27−5.81
Hepatic failure	OTA	371	17,3791	0.21	0.85	0.77–0.94	−0.23	−0.57−0.11
	SSRi	263	13,9997	0.19	0.75	0.66–0.84	−0.42	−0.82−0.01
	NSMRi	19	9250	0.21	0.82	0.52–1.28	−0.29	−1.73−1.18
	NMOi	2	1408	0.14	0.57	0.14–2.27	−0.82	−4.36−3.17

AE, adverse event; 95% CI, 95% confidence interval; DILI, drug-induced liver injury; IC, information component; MOAi, monoamine oxidase A inhibitor; NMOi, non-selective monoamine oxidase inhibitor; NSMRi, non-selective monoamine reuptake inhibitor; OTA, other antidepressant; ROR, reporting odds ratio; SSRi, selective serotonin reuptake inhibitor.

Association between drug-related severe hepatic disorders and individual antidepressant drugs

In all, 29 drugs have been reported to cause severe hepatic disorders. Three drugs were detected with significant signals: nefazodone, mirtazapine, and duloxetine (Supplemental Table S4). Forest plots illustrating the ROR and 95% CI of antidepressants ranked from the highest to lowest ROR of drug-related severe hepatic disorders reported to the FAERS database are presented in Figure 3.

Figure 3.

Association of drug-related severe hepatic disorders with antidepressant drugs.

Association between cholestatic injury and individual antidepressant drugs

In all, 29 drugs were reported to cause cholestatic injury. Eight drugs with significant signals were identified: nefazodone, fluvoxamine, clomipramine, imipramine, mirtazapine, duloxetine, escitalopram, and venlafaxine. The details are presented in Supplemental Table S5. Forest plots illustrating the ROR and 95% CI of antidepressants ranked from the highest to lowest ROR of drug-related cholestatic injuries reported to the FAERS database are presented in Figure 4.

Figure 4.

Association of cholestatic injury with antidepressant drugs.

Association between hepatocellular injury and individual antidepressant drugs

In all, 35 drugs were reported to cause cholestatic injury. Eight drugs with significant signals were detected: mianserin, nefazodone, maprotiline, fluvoxamine, duloxetine, clomipramine, mirtazapine, and venlafaxine (Supplemental Table S6). Forest plots illustrating the ROR and 95% CI of antidepressants ranked from the highest to lowest ROR of drug-related hepatocellular injuries reported to the FAERS database are presented in Figure 5.

Figure 5.

Association of hepatocellular injury with antidepressant drugs.

Association between hepatic failure and individual antidepressant drugs

In all, 24 drugs have been reported to cause hepatic failure. Only nefazodone elicited a significant signal (Supplemental Table S7). Forest plots illustrating the ROR and 95% CI of antidepressants ranked from the highest to lowest ROR of drug-related hepatic failures reported to the FAERS database are presented in Figure 6.

Figure 6.

Association of hepatic failure with antidepressant drugs.

Discussion

The choice of an appropriate antidepressant drug depends not only on its efficacy in different indications but also on its tolerability and potential for serious toxicity in individual patients. The FAERS database contains extensive data and is publicly available. Many previous studies on DILI have used the FAERS database; however, there have been no studies on DILI cases related to antidepressant drugs. We performed the pharmacovigilance analysis of FAERS data over the past 18 years to explore the relationship between individual antidepressant drugs and DILI. Since 2007, there has been a slow increase in the number of reaction cases associated with antidepressants. However, in 2015, adverse reports showed a sharp increase, especially for OTA. Reports of duloxetine-induced DILI increased by 6.7 times, 96% of which were reported in North America, which may have been due to organizational oversights (e.g. the data were not updated for a long period, and the delay in updating may have contributed to the increase in the DILI cases). Overall, the main drugs reported for DILI were SSRi and OTAs, which accounted for >94% of the reported cases. From 2016 to 2021, the number of SSRi-related DILI cases increased significantly, most notably between 2018 and 2019, with the main contributors being escitalopram, sertraline, and fluoxetine. In addition, in the last 5 years (2016–2021), DILI cases due to NSMRi have markedly increased. Further analysis revealed that amitriptyline and clomipramine were the major contributors. Notably, in the past 5 years, amitriptyline and clomipramine have contributed to more than half of the total number of reported DILI cases from 2004 to 2021 (amitriptyline, 61/110; clomipramine, 33/63).

Previous studies have shown that females are twice as likely to be diagnosed with depression and experience higher symptom severity than males. ²⁴ In our study, 60% of the reported cases involved females. Among the reported DILI cases, the number of female cases is almost twice that of male cases.

For older drugs, notably selective MOAis, NMOis, and noradrenergic and selective serotonergic antidepressants (NaSSAs), such as mianserin, data are scarce because clinical trial results are not available, and only case reports have been published. In the previous studies, the antidepressants associated with greater risks of hepatotoxicity were iproniazid, nefazodone, phenelzine, imipramine, amitriptyline, duloxetine, bupropion, trazodone, tianeptine, and agomelatine. ²⁵ Citalopram, escitalopram, paroxetine, and fluvoxamine had the least potential for hepatotoxicity.^12,25 However, the present study results are not entirely consistent with these conclusions.

Cross-toxicity has mainly been reported for tricyclic and tetracyclic antidepressants. In this study, tricyclic and tetracyclic antidepressants were categorized as NSMRis. For cholestatic injury, the tricyclic antidepressants, imipramine and clomipramine, produced significant signals. Notably, more than half of the reported imipramine-induced cholestatic injury cases were also female patients.^26,27 The mechanism of imipramine-induced hepatotoxicity may involve a direct toxic effect or a hypersensitivity reaction, but this has not yet been determined. ²⁸ Imipramine-induced fulminant hepatic failure and the patient subsequently underwent liver transplantation. ²⁹ Clomipramine-induced cholestatic injury has not been mentioned in previous studies; only allergic hepatitis has been reported, ³⁰ but our study is the first to detect these signals. Although the tetracyclic antidepressant maprotiline had a high ROR of 3.03, its IC was <0. A case report described a female patient with elevated liver enzymes shortly after the initiation of maprotiline treatment, ³¹ which may have been due to the metabolism by cytochrome P-450 enzymes. However, the exact mechanism of maprotiline-induced hepatotoxicity is still unknown and rarely reported. Other tricyclic and tetracyclic antidepressants did not produce DILI signals.

SSRis have revolutionized the development of antidepressant drugs. SSRis have advantages in terms of tolerability. A previous study reported that SSRIs have the lowest probability of causing DILI (0.03%). ¹¹ However, in our study, significant signals were detected. Furthermore, differences in hepatotoxicity exist among individual drugs. SSRi-induced DILI is generally characterized by a hepatocellular pattern and less frequently by a cholestatic or mixed pattern. ²⁵ However, as per ROR results, we found that all the FDA SSRIs were associated with cholestatic injury. Furthermore, a relatively balanced proportion of male and female patients were observed with DILI due to the above drugs, with a metabolic or immune-allergic pathophysiology. ²⁵ In many cases, the onset of DILI occurs within several days to 6 months after the beginning of treatment with antidepressant drugs. Overall, considering both the ROR and IC criteria, fluvoxamine and escitalopram showed significant signals associated with DILI. SSRIs are the most widely used and well-known drugs in clinical practice. Although case reports on SSRi-induced liver injury have never reported severe events or death, monitoring needs to be strengthened during treatment, especially in the first 6 months.

In this study, most of the new-generation antidepressant drugs were categorized as OTAs. After the approval of SSRIs, numerous other categories followed, identified by their main mechanisms of activity, such as serotonin and norepinephrine reuptake inhibitors (SNRIs); serotonin modulators and stimulators; serotonin antagonists and reuptake inhibitors (SARIs); NaSSAs; norepinephrine reuptake inhibitors (NeRIs); serotonin, norepinephrine, and dopamine reuptake inhibitors (SNDRIs) or triple reuptake inhibitors; and melatonin and serotonin agonists. ³²

The SNRIs (venlafaxine and duloxetine) and NaSSAs (mirtazapine) were strongly associated with DILI. In severe drug-related hepatic disorders, such as cholestatic injury and hepatocellular injury, duloxetine and mirtazapine showed significant signals. Venlafaxine showed signals in both cholestatic and hepatocellular injuries. Mirtazapine increases the release of serotonin and norepinephrine and shows activity after only a few weeks of administration. Although previous studies have noted hepatotoxicity due to mirtazapine,^33,34 it is very rare. Hepatotoxicity due to mirtazapine has only been reported in isolated cases.^35,36 Our results may increase physicians’ awareness of this complication and warrant more attention to the regular monitoring of the liver in patients taking mirtazapine.

A recent study ³⁴ showed that duloxetine has lower hepatotoxicity than SSRIs. Inconsistent with previous reports, our study revealed that duloxetine-induced liver injury signals were much stronger than those induced by SSRIs. Considering previous studies,^37,38 duloxetine may be associated with an increased risk of certain less severe hepatic events; however, it was not significantly associated with hepatic-related death or ALF. Duloxetine accounts for a considerable number of hepatotoxicity cases in the FAERS database. Notably, in the label of duloxetine for the FDA and European Medicines Agency (EMA), hepatotoxicity has been listed separately. Hepatic failure, which is sometimes fatal, and cholestatic jaundice with minimal elevation in transaminase levels have also been reported in patients without a history of hepatic diseases.^16,39 Venlafaxine causes a threefold (0.4%) increase in liver enzymes. ⁴⁰ However, venlafaxine also causes liver failure, ⁴¹ with most patients experiencing it as early as several days or up to 6 months after drug administration. The effects generally cease after drug withdrawal. ⁴⁰ Physicians should be aware of the potential for liver injury when prescribing these medications. Another SNRI, milnacipran, approved by the FDA for the same indications as duloxetine for fibromyalgia syndrome and depression, detected no signals. Levomilnacipran was approved in 2013, but only eight DILI-related events were reported; neither showed any signals in the present study.

Owing to the complex receptor and transporter interaction profile, nefazodone can be classified as either SARI or SNDRI. ⁴² It was approved in 1994 but was voluntarily discontinued in many countries in less than a decade. Many European countries have discontinued its use due to severe and potentially fatal liver toxicity; however, it is used in the United States. Thus, 96.94% (348/359) of events on nefazodone-induced liver injury were reported between 2004 and 2006. The incidence of severe liver failure leading to death or liver transplantation due to nefazodone is 1 in 250,000–300,000 patients. These results are consistent with its FDA ‘black box’ warning for hepatotoxicity. ⁴³ Nefazodone showed the strongest signal among the 42 antidepressant drugs in the four categories of DILI. Careful analysis of the reporting years for nefazodone showed that 83.33% (895/1074) of the cases were reported between 2004 and 2006. Since 2007, the number of adverse effects reported has sharply decreased, which may be related to its withdrawal in some countries and the introduction of new-generation antidepressants. Trazodone, another SARI, is still widely used. It is mostly used at low doses in polypharmacy, along with OTAs. Although a few DILI-related events have been reported, no DILI signals were detected in the present study. However, physicians should consider the probability of ALF, which may warrant liver transplantation in patients. ¹⁴

Reboxetine, the first NeRI, is currently marketed in several countries around the world, including the United States. Therefore, only 13 adverse events have been reported in Europe or Asia, which were not associated with DILI. Another NeRI, viloxazine, was developed into immediate-release capsules for the treatment of depression in the 1960s and 1970s and marketed in the United Kingdom and several other countries in Europe but was withdrawn from the market due to its large dosage and low efficacy. Several years later, new sustained-release viloxazine capsules were developed, focusing on the treatment of attention deficit hyperactivity disorders in children and adolescents. ⁴⁴ The FDA officially approved the use of viloxazine sustained-release capsules for marketing in April 2021. In all, 50 adverse events in our study were reported in 2021, and there were no cases of viloxazine-induced liver injury. Therefore, more post-marketing adverse events monitoring is required.

Bupropion is a norepinephrine and dopamine reuptake inhibitor. Bupropion was approved by the FDA in the 1980s for the treatment of depression. Due to the high incidence of adverse seizures, it was taken off the market and reintroduced at a lower dose. ⁴⁵ In our study, bupropion had 38,648 reported adverse events, but no signals were detected in DILI. Based on the literature, we found less than 10 reported cases ⁴⁶ of hepatotoxicity considered to be induced by bupropion. Hepatotoxicity may be a rare adverse effect; however, physicians should be aware that bupropion has caused a fatal case ⁴⁷ of hepatotoxicity with autoimmune features.

Agomelatine has been approved in Europe and Australia but not in the United States. The EMA has revised the new contraindications for hepatic toxicity. ⁴⁸ Due to the few reports in the FAERS database, this study did not detect significant agomelatine signals. Esketamine, an antagonist of N-methyl-D-aspartate (NMDA) glutamatergic receptors, ⁴² was approved by the FDA until 2019. To date, only eight DILI-related events have been reported. Tianeptine is an ‘atypical’ tricyclic antidepressant (TCA) with a benzothiazepine structure. As it has not obtained marketing authorization from the United States, United Kingdom, Canada, Australia, and New Zealand, there are less than 10 adverse reaction reports from the FAERS database and only one event associated with DILI.

DILI signals have been associated with many antidepressants, including not only older-generation drugs but also some newer-generation antidepressants, such as escitalopram, venlafaxine, and mirtazapine. Antidepressants may not be as safe as expected due to their association with DILI, and this factor has not been widely studied. Antidepressants with no significant signal detected did not indicate any risk; however, they indicated the need for wider monitoring and evaluation of antidepressant use in the future. We categorized these associations according to positive signals from the strongest to weakest. However, disproportionality analysis can neither quantify the risk nor prove causality. The detection of risk signals only indicates a statistical association between the drug and the risk of DILI, which can provide hypotheses and directions for further research. Future large-sample and prospective studies on antidepressants can provide high-quality safety evidence and guide the long-term clinical applications of antidepressants. However, our results can provide physicians with a reference to optimize drug benefits, specifically for patients with liver dysfunction.

Study limitations

This study had some limitations. First, although the FAERS database supports published research on many individual drugs, it is dependent on the submission and integrity of information; hence, it cannot be used to calculate the incidence of an adverse event, such as DILI, or the risk of DILI. Disproportionality analysis can neither quantify the risk nor prove causality. Assessing causality requires a prospective study design. Furthermore, the term DILI may have been reported and recorded in multitudinous PTs; although two methods of DILI event identification have been applied, the risk of omission could not be completely avoided. Meanwhile, the FAERS database also lacks drug dose, combination drug, laboratory, and imaging data, precluding a complete assessment of alternative causes of hepatotoxicity. Thus, conclusions drawn from the spontaneous surveillance FAERS database represent only safety signals that require further investigation to confirm the events of antidepressant-associated liver injury. Therefore, further prospective clinical studies are needed to confirm our findings. In addition, future efforts should focus on clinically meaningful cases, such as those requiring hospitalization or other medical interventions.

Conclusion

In conclusion, we present a cost-effective method for analyzing the FAERS database for antidepressant-related liver injury adverse events. In this study, we confirmed the association between nefazodone and duloxetine and DILI, which concurred with the findings of previous studies. We also detected a potential association between DILI and newer-generation antidepressants, including SSRis, such as fluvoxamine, escitalopram, and fluoxetine, and other drugs, such as mirtazapine, venlafaxine, and clomipramine, which require further investigation. Therefore, although clinical guidelines do not require regular monitoring of liver function with newer-generation antidepressants, more attention should be paid to the safety profile of these drugs, particularly in patients with hepatic insufficiency. This study provides an informative ranking of the odds of DILI among 42 antidepressant drugs with a wide array of clinical indications. However, the disproportionality analysis cannot conclude on a definite causal link between antidepressants and DILI. Furthermore, ongoing surveillance using the FAERS database or other data sources will play a key role in assessing antidepressant-related liver injury adverse events, particularly with the introduction of new drugs in the market.