FormalPara Key Points

This population-based study revealed an increased risk of liver injury among individuals using antidepressants, notably selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), and other non-selective serotonin reuptake inhibitors (NSRIs).

A 1.24-fold higher risk of liver injury was linked to early stages of antidepressant use, emphasizing the need for cautious monitoring during the initial 30 days.

Understanding the heightened risk of liver injury associated with commonly prescribed antidepressants like SSRIs and SNRIs underscores the importance of considering potential hepatic effects in patient care strategies.

1 Introduction

Drug-induced liver injury (DILI) is a relatively rare occurrence, with an incidence rate ranging from 1 per 10,000 to 100,000 individuals [1]. A recent study in France reported that DILI had an incidence rate of approximately 14 per 100,000 person-years, of whom 12% required hospitalization and 6% died [2]. Despite this, there is limited evidence extrapolating the overall incidence of DILI to Taiwan. Prospective studies conducted in Taiwan reported an incidence of 768 cases per 100,000 person-years for statin-induced DILI [3], as well as an incidence of 31.6 cases per 10,000 persons for antifungal agent-induced DILI, exemplified by fluconazole [4]. Given the high prevalence of hepatitis in Taiwan and the clinical presentation similarities with DILI, accurate diagnosis is often complicated, suggesting that reported incidence rates might underestimate the true prevalence.

The risk factors for DILI include age, gender, presence of comorbidities, drug interactions, and drug dosage [5]. Evidence suggests that age can increase sensitivity to and limit the metabolism of hepatotoxic drugs in humans [6], with drug efficacy and metabolism rates significantly differing in children and elderly individuals compared with adults [7]. One study found that females exhibited a higher risk of adverse drug reactions, potentially due to differences in metabolism rates and medication dosage between genders [6], although this was contradicted by another study that reported observing the opposite results [8]. Drug interactions and the presence of comorbidities can also increase the risk of DILI, with Bell and Chalasani demonstrating that patients diagnosed with hepatitis types B and C and human immunodeficiency virus (HIV) exhibited an increased risk of being diagnosed with DILI [9]. Concurrent use of other hepatotoxic drugs was also identified as a risk factor for DILI [10].

In Japan, 60% of DILI occurs within 30 days of commencing medication usage, while 80% occurs within 90 days of commencing usage [11]. Several other studies have also shown that acute DILI may develop within 1–3 months of commencing medication usage [12,13,14]. Some studies have shown that approximately 1.28–4 patients per 100,000 patient-years require hospitalization for antidepressant-induced liver injury (AILI). In particular, the incidence rates of AILI associated with nefazodone and tricyclic/tetracyclic antidepressant usage are approximately 28.96 and 4 patients per 100,000 patient-years, respectively [15,16,17,18]. Selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) have also been shown to lead to asymptomatic mildly abnormal liver function in approximately 0.5–1% of patients, while monoamine oxidase inhibitors (MAOIs) usage resulted in AILI in approximately 3% of patients [17]. However, diagnosis of AILI is challenging as it is primarily based on exclusion, suggesting that the incidence rates may be underestimated [18].

DILI primarily has two pathophysiological types, as follows: (1) idiosyncratic: dose-independent, unpredictable, occurs as a consequence of either immune-mediated liver damage (immuno-allergic idiosyncratic DILI) or direct cellular injury (metabolic idiosyncratic DILI); and (2) intrinsic: dose-dependent, predictable, occurs as a consequence of drug accumulation [17]. Most pathophysiological types of AILI are either immune-allergic or metabolic idiosyncratic DILI, with the majority being of the hepatocellular type and fewer being the cholestatic or mixed types. Moreover, acute hepatocellular injury and jaundice have a mortality rate of approximately 10% [17]. Although there are reports of hepatotoxicity linked to adverse reactions from antidepressants, extensive studies conducted in Taiwan exclusively centered around the occurrence of liver cancer, reporting a low incidence and mortality rate [19, 20]. This contrasts with other research exploring the mechanisms of liver injury, revealing a notable research gap concerning the impact of diverse antidepressant drug treatments and their dose–response relationships on the risk of hepatotoxicity.

Therefore, this nationwide population-based study aims to assess the risk of AILI associated with different pharmacological classifications and the duration of antidepressant use. The investigation relies on data from the Taiwan National Health Insurance Database from 2000 to 2018.

2 Methods

2.1 Data Sources

This population-based, case–control study used the Taiwan National Health Insurance (NHI) database, which was established in 1995 and covers more than 99.6% of the Taiwanese population. The National Health Insurance Research Database (NHIRD) contains claims data (e.g., registry of beneficiaries, clinical and hospital care records, drug prescriptions, and other medical service records) from the NHI database and is reviewed annually. Additionally, data from the Longitudinal Generation Tracking Database (LGTD 2005), which contains approximately two million individuals randomly selected from the NHIRD, was also used. The clinics and hospitals recorded diagnoses using the International Classification of Diseases, Ninth and Tenth Revision, and Clinical Modification (ICD-9-CM and ICD-10-CM). Patient confidentiality was protected through data pseudonymization (wherein all original identification numbers were replaced with encoded serial numbers to enable linkage to claims data) implemented by the Taiwanese government prior to the release of data for research.

2.2 Study Population

The study population included individuals aged 20 years or over who did not have any history of relevant code from ICD-9-CM [21] of malignant neoplasms (140–239); viral hepatitis A, B, or C infections, other viral hepatitis infections (070.0–070.9), or carriers of the hepatitis virus (V026.1–V026.9); infectious mononucleosis and cytomegalovirus and coxsackie virus infections (573.1–573.2); cholelithiasis (574.0–574.9); chronic liver disease; cirrhosis; alcoholic liver disease; liver abscess; portal pyemia; hepatic coma; portal hypertension; hepatorenal syndrome; chronic passive congestion of the liver (571.0–573.0); and liver disorders during pregnancy (646.7) between January 1, 2000 and December 31, 2018.

Cases were patients hospitalized with the diagnosis code of suspected DILI during the study period based on the ICD-9-CM, ICD-10-CM, and the previous study, including disorders of bilirubin excretion, acute and subacute necrosis of liver, liver cirrhosis, chronic liver disease (nonalcoholic), hepatic coma, hepatitis (non-viral), other disorders of the liver, and other specified disorders of the biliary tract [3, 4, 21]. A control individual free from liver injury was randomly selected for each case, matching (1:1) on age, gender, and the index date (the first observed diagnosis of liver injury in the case).

2.3 Exposure to Antidepressants

The antidepressants included in this study were monoamine oxidase A inhibitors (MAO-AIs), non-selective serotonin reuptake inhibitors (NSRIs), SSRIs, and SNRIs. Medications not included in the aforementioned classes were classified as other antidepressants (i.e., trazodone hydrochloride, oxitriptan, mirtazapine, and agomelatine, etc.). The cumulative exposure to antidepressants was defined as the total number of days before the index date that the patient was prescribed antidepressants. Patients were categorized into groups (i.e., 1–30 days, 31–90 days, and 91+ days) based on their cumulative exposure to antidepressants to allow examination of the association with the risk of liver injury. Non-users were those who had never used antidepressants before the index date.

2.4 Covariates

Factors such as the use of other medications and a history of comorbidities have been shown to be related to liver injury and, therefore, were considered confounders in the current analysis (Table 1). Medications causing liver damage were prescribed for >90 days a year before the index date, including anti-tuberculosis agents, statins (or fibrates), antibiotics, antifungals, anti-epileptics, non-steroidal anti-inflammatory drugs (NSAIDs), and other potential hepatotoxic agents (i.e., acetaminophen, amiodarone, and methotrexate) [22]. Comorbidities were defined as the presence of chronic renal failure, hyperthyroidism, diabetes mellitus, malnutrition, hyperlipidemia, heart failure, hypertension, ischemic heart disease, arrhythmia, chronic kidney disease, and renal failure.

Table 1 Demographics of the study population after matching

2.5 Statistical Analyses

Descriptive statistics were compared between the cases and controls using the Student’s t-test or Mann–Whitney U test for continuous variables and the Chi-square test for categorical variables, respectively. Crude and multivariate logistic regressions were used to compare the risk of liver injury between antidepressant users and non-users, and the findings were reported as odds ratios (ORs) and 95% confidence intervals (CIs). The multivariate models adjusted for age, gender, history of comorbidities, and use of liver damage-related drugs prior to the index date of 1 year. All data analyses were carried out using SAS 9.4 (SAS Institute, Cary, NC, USA), and a two-tailed p value of <0.05 was considered statistically significant.

3 Results

The study sample included 59,989 cases hospitalized with a primary diagnosis of liver injury and 59,989 controls matched by age, gender, and index date. Approximately half of the cases exhibited hepatitis (49%), and were statistically significantly more likely than controls to have a higher prevalence of antidepressant usage (20.3% vs 13.9%), longer durations of antidepressant exposure (33 days vs 30 days), and use of other liver damage-related medications for >90 days (i.e., statins, antifungals, anti-tuberculosis agents, antibiotics, anti-epileptics, NSAIDs, acetaminophen, amiodarone, and methotrexate). Moreover, a greater proportion of cases were likely to have a history of comorbidities, including hypertension (33.2% vs 24.4%), diabetes mellitus (17.7% vs 10.1%), ischemic heart disease (17.0% vs 11.6%), arrhythmia (7.0% vs 4.6%), heart failure (4.8% vs 2.8%), hyperlipidemia (29.5% vs 15.6%), chronic kidney disease and renal failure (5.8% vs 2.7%), hyperthyroidism (2.6% vs 1.4%), and fasting or malnutrition (0.6% vs 0.2%) than controls (Table 1).

Table 2 presents the distribution and odds of antidepressant users among the study population. Compared with the controls, cases had a significantly higher proportion of patients using any classes of antidepressants, which was associated with an increased risk of liver injury across all durations of use (the range of crude ORs was 1.44–2.35). After adjusting for covariates, antidepressant users exhibited a significantly higher risk of liver injury (adjusted OR [aOR] 1.16, 95% CI 1.12–1.20). This risk was particularly elevated when individuals were prescribed NSRIs (aOR 1.05; 95% CI 1.01–1.10), SSRIs (aOR 1.22; 95% CI 1.16–1.29), SNRIs (aOR 1.18; 95% CI 1.13–1.24), or other types of antidepressants including trazodone hydrochloride, oxitriptan, mirtazapine, and agomelatine (aOR 1.27; 95% CI 1.14–1.42). However, all MAO-AIs users had no significant result compared with non-users.

Table 2 The crude and adjusted odds ratios for liver injury by antidepressant usage

Subgroup analyses presented in Table 2, stratified based on days of drug usage, revealed that individuals using any class of antidepressants faced a higher risk of developing liver injury within the initial 1–30 days compared with non-users (aOR 1.24; 95% CI 1.03–1.54). Furthermore, these findings remained consistent across all classes, including MAO-AI (aOR 1.26; 95% CI 1.03–1.54), NSRI (aOR 1.14; 95% CI 1.08–1.20), SSRI (aOR 1.34; 95% CI 1.24–1.45), and SNRI (aOR 1.21; 95% CI 1.14–1.30) users. Notably, there was a discernible declining trend in risk with longer durations of usage, especially among patients using NSRIs for >90 days, who exhibited a significantly reduced risk when compared with individuals who had not utilized any form of NSRI (aOR 0.86; 95% CI 0.79–0.94).

4 Discussion

To the best of our knowledge, this is the first Asian population-based study to illustrate the association between antidepressant use and the risk of DILI in the general population. The current study first demonstrates the nationwide risk of AILI in Taiwan over 18 years. Systematically, the study collected data encompassing the risk of liver injury linked to all classes of antidepressants, analyzing the cumulative days of exposure. The findings of the current study showed an association between antidepressant usage and risk of liver injury, with a significantly larger proportion of patients with liver injury using antidepressants compared with those without liver injury. Additionally, the median duration of antidepressant use was significantly longer in patients with liver injury compared with those without liver injury. Noteworthy is the observation that nearly 50% of suspected AILI patients presented with non-viral hepatitis. Furthermore, a higher prevalence of AILI was observed in patients with comorbidities, including cardiovascular disease, diabetes, renal disease, and others.

The risk of AILI in this study is lower than in the previous research, which may be due to differences in drug category distribution and other demographics [23]. Further analysis showed that certain pharmaceutical classes of antidepressants (e.g., SSRIs and other classes of antidepressants) were associated with an overall higher risk of liver injury compared with other classes, except for MAOIs. In addition, certain antidepressants may carry an increased risk of liver injury, particularly in the first 30 days of use, as suggested by research findings [23]. Patients using these medications should exercise caution, and healthcare providers should monitor for potential liver-related adverse effects during the initial month of treatment.

The prevalence of liver injury was also relatively high among those who had used anti-tuberculosis agents, statins (or fibrates), antibiotics, antifungals, anti-epileptics, NSAIDs, and other potential hepatotoxic agents (i.e., acetaminophen, amiodarone, and methotrexate) [22] for >90 days compared with those who had not, consistent with previous studies [12]. This could be attributed to the independent association of these drugs with the risk of liver injury. When combined with antidepressants, they may collectively contribute to an increased burden on the liver, potentially leading to damage.

The mechanism for AILI is not yet fully understood. Studies suggest that DILI may be related to allergic reactions, oxidative stress, lipid peroxidation, cytochrome-P450 enzyme (CYP450) metabolites, or immune-mediated reactions [17, 24]. SSRIs and SNRIs are metabolized in the liver, especially through the CYP450 enzyme pathway, which includes CYP2C19, CYP2C9, and CYP2D6 [25]. In elderly patients or CYP2C19 poor metabolizers, the oral clearance rate is diminished, leading to a higher likelihood of adverse reactions and hepatotoxicity due to dose accumulation [26]. In clinical management, the dosage or medication regimen may be altered when an elevation in the patient’s liver indices is detected. This could explain the slight reduction in the observed risk of liver injury after a treatment duration exceeding 30 days.

MAO-AIs inhibit monoamine oxidase in the liver, reducing the breakdown of neurotransmitters and increasing their concentration in the synaptic cleft. In this study, the overall risk of DILI in users of MAO-AIs was not significantly higher than that in non-users, perhaps due to the known liver toxicity of these drugs, leading clinicians to avoid prescribing them to high-risk patients. However, MAO-AIs demonstrate multiple interactions with other medications, food, or overdose, leading to an elevated risk of hepatotoxicity [27]. A significantly higher risk observed within the first 30 days with MAO-AI use in this study may be attributed to the potential omission of these interaction factors during the initial administration period.

On the other hand, using NSRIs involves liver metabolism, either through a direct toxic effect or a hypersensitivity reaction, and is influenced by other drugs that induce or inhibit microsomal enzymes [24]. Long-term NSRI users typically administer low doses. Hence, the risk of hepatotoxicity is less obvious. In particular, our findings revealed that patients using NSRIs (e.g., tricyclic antidepressants [TCAs] like imipramine or amitriptyline) for more than 90 days had a significantly reduced risk compared with non-users. This can be explained by the fact that most prescriptions for NSRIs with a low dose and long duration are given to nocturia sufferers. Nocturia is often managed by addressing underlying causes or conditions that contribute to increased nighttime urination, such as age, diabetes, and cardiovascular-related conditions [28, 29]. NSRIs, with their anticholinergic effects that can affect bladder function, have been used off-label for managing nocturia. Overall, it is important to note that there may be residual confounding from long-term antidepressant usage in mitigating the risk of certain adverse outcomes.

Antidepressants, particularly SSRIs and SNRIs, are commonly used to treat depression and other psychological disorders, and awareness of their potential effects on the liver is essential. Clinicians should evaluate the patient’s condition when considering the risks and benefits of prescribing antidepressants and take appropriate measures where necessary.

5 Strengths and Limitations

The current study had several strengths, including a large sample size and longitudinal follow-up, which facilitated robust investigation of rare events such as liver injury. The effects of the duration of and cumulative exposure to antidepressant use were also examined to supplement existing evidence on the association between long-term antidepressant use and liver injury. Strict statistical analysis, including subgroup comparisons, was carried out to reduce the risk of bias.

This study also had several limitations. Firstly, this nationwide case–control study lacked detailed laboratory information such as hepatic fibrosis stage, baseline liver biochemical tests (e.g., alanine aminotransferase [ALT], alkaline phosphatase [ALP], total bilirubin [TB]), and lipid profile (e.g., triglycerides, low density lipoprotein cholesterol). Secondly, lack of access to medical records prevented use of a standardized definition for liver damage (e.g., Roussel Uclaf Causality Assessment Method [RUCAM]). Therefore, the lack of a validation study specifically conducted for the diagnosis code of liver injury in the NHIRD and the inclusion of only patients hospitalized with a primary diagnosis of liver injury (defined using ICD codes) may have led to selection bias. Thirdly, the NHIRD did not include information on the patients’ body index (e.g., height and weight), behavioral factors (e.g., alcohol consumption and smoking), and use of alternative medications (e.g., over-the-counter products and herbal products). These limitations were addressed by imposing strict criteria to exclude non-DILI patients.

6 Conclusion

This population-based study adds to existing evidence by exploring the association between antidepressant use and the risk of liver injury. Overall, the findings showed that antidepressant usage was consistently associated with an increased risk of liver injury, and this was particularly applicable to SSRI and SNRI users as well as those using antidepressants for 1–30 days.