Introduction

Drug-induced liver injury (DILI) is an adverse reaction to drugs and/or their metabolites that may result in permanent loss of liver function and death. DILI covers a broad spectrum of clinical manifestations, from asymptomatic abnormalities in the liver function tests (prothrombin time, albumin, direct and indirect bilirubin, the level of aspartate, and alanine transaminase in serum) to symptomatic acute liver disease, prolonged jaundice, and disability, or overt acute or subacute liver failure (Chalasani et al. 2008). DILI is commonly divided into two types: intrinsic and idiosyncratic. Intrinsic DILI is dose-dependent, typically has a short latency period and predictable disease course, and is reproducible in animal models. Idiosyncratic DILI is a severe liver injury occurring very rarely and usually is not dose-dependent. It is characterized by unpredictable disease course, more varied manifestations, and is not reproducible in animal models (Ye et al. 2018). Drugs that cause idiosyncratic DILI typically also cause a mild and asymptomatic liver injury—transient, asymptomatic elevations in serum alanine aminotransferase levels (Mosedale and Watkins 2017). The lack of specific serological markers makes DILI hard to diagnose. The clinical symptoms are very diverse and may include tiredness, lack of appetite, nausea, vomiting, fever, joint and muscle pain, rash, and jaundice, while some patients remain asymptomatic. DILI is challenging not only concerning diagnosis but for the management as well. In most cases, the only treatment is to stop the drug administration and provide general supportive care (Navarro and Senior 2006).

The liver is the primary site for the metabolism of drugs, including antidepressants (ADs) and antipsychotics (APs). Thus, it is of great importance to understand how a specific drug and its metabolites affect the structure and function of this organ. Hepatotoxicity may result in hepatic steatosis (accumulation of fat in the liver), steatohepatitis (steatosis with inflammation), fibrosis (excessive accumulation of extracellular matrix proteins), and cirrhosis (formation of bridging fibrous septa and disruption of the standard architecture of the liver) (Aithal et al. 2011). ADs and APs, used in treating different psychiatric disorders such as depression, schizophrenia, and anxiety, may display hepatotoxicity, even at therapeutic doses (reviewed in detail in the following sections). More than 160 psychotropic drugs have been shown to produce hepatic side-effects (Dumortier et al. 2002). A more recent study reported that psychotropic drugs were responsible for 7.6% of DILI cases in a cohort of 185 patients (Licata et al. 2017). Thus, the strategy behind the choice of a psychotropic drug must consider its hepatotoxicity, especially in patients with other risk factors, such as alcoholism, drug abuse, polymedication, obesity, and diabetes.

DILI represents a burden that has been heavily underestimated due to the limitations of clinical trials to identify such rare events. However, it is among the leading causes of late-stage drug development interruption and post-marketing drug withdrawal (Raschi and de Ponti 2015). A systematic review which included 462 medical products withdrawn in the period 1953–2013 revealed that most products were withdrawn from the marketplace due to their hepatotoxicity (18%), followed by immune-related reactions (17%), neurotoxicity (16%), and cardiotoxicity (14%) (Onakpoya et al. 2016). DILI is the most common cause of acute liver failure (50%), ahead of viral infection or autoimmune hepatitis. In developing countries, people are at higher risk of acute liver failure due to viral infections, while in Western Europe and the United States, most cases arise from DILI (Bernal et al. 2010). Epidemiology data regarding DILI are limited, given that the incidence and prevalence of this disorder are underestimated. In China, the incidence of DILI has been rising year by year and now is recognized as a major public health concern (Yu et al. 2017; Shen et al. 2019). According to different national studies, summarized by Ahmad and Odin (2017), DILI incidence varies from 2 to 19 cases per 100,000 per year. However, despite low incidence, DILI should always be considered in acute liver injury cases when other possible aetiologies have been excluded. DILI frequency varies according to age, gender, and socio-economic status. As for the age, the elderly are more affected, probably due to multiple drug use and low tolerability of the therapy. Also, women and wealthier classes are more affected than men and developing societies, respectively (Licata et al. 2017).

Given a large number of new cases of psychiatric patients each year, the need and use of psychotropic drugs are on the constant rise. In the case of depression, the total estimated number of affected people increased by 18.4% between 2005 and 2015 (WHO 2017). Also, it is expected that the current worldwide health emergency caused by COVID-19 adversely affects mental health in many ways. It has already been established that COVID-19 survivors and frontline healthcare engaged with COVID-19 patients presented a high risk of developing symptoms of depression, anxiety, insomnia, fatigue, and post-traumatic stress disorder (Lai et al. 2020; Mazza et al. 2020; Rogers et al. 2020). In addition, fear of infection, as well as unemployment, working from home, home-schooling of children, and lack of contact with family members and friends are risk factors for psychiatric disorders in the general population (WHO 2020). Thus, the current COVID-19 pandemic is expected to increase the use of psychotropic drugs further in the near future. With exploding rates of mental disorders and the growing use of psychotropic drugs, more and more evidences emerge on their hepatotoxic effects.

This review aims to: (1) provide a brief insight into the hepatic metabolism of psychotropic drugs; (2) evaluate the hepatotoxic effects of commonly prescribed ADs and APs; (3) summarize the roles of oxidative stress, inflammation, and lipid accumulation in the pathophysiology of DILI; and (4) review and discuss these pathophysiological mechanisms in the course of AD- and AP-induced hepatotoxicity.

The most important means of assessing a drug’s risk for hepatotoxicity is to review the published case reports, animal studies, and cell-based in vitro studies. We conducted a comprehensive literature search for case reports, original articles, and reviews published from 1960 to 2020 using the search terms “hepatotoxicity,” “DILI,” “liver injury,” “oxidative stress,” “hepatitis,” “steatosis,” and “fatty liver,” cross-referenced with “antidepressant” and “antipsychotic.” This time interval was chosen, because the first ADs and APs were introduced in practice in the 1950s, while the first publications considering their hepatotoxic effects emerged in the 1960s. This approach cannot be used to compare the risks carried by different drugs, since many were introduced at different times and have been prescribed in different doses and rates. It should also be borne in mind that the extent to which the results obtained by animal studies can reasonably be generalised to humans remains open to question. Given that even humans differ in their response to drugs, it is reasonable to suspect if animal models are good predictors of human response (Shanks et al. 2009; Bailey et al. 2014). Thus, relevant in vitro studies with liver microsomes, hepatocytes, liver slices, and recombinant enzymes are very valuable and usually the only source of information reflecting the situation in humans (Martignoni et al. 2006). Taking all this into consideration, we made an effort to highlight the drugs associated with severe DILI cases, explain the potential molecular mechanisms underlying the phenomenon, and point out the ones which are less likely to cause DILI, reviewing relevant case studies, as well as in vivo and in vitro studies.

Hepatic metabolism of psychotropic drugs

Metabolism of psychotropic drugs occurs mainly in the liver, with the cytochrome P450-dependent monooxygenase (CYP) family of isoforms being the most important drug-metabolizing enzymes. CYP450 is a superfamily of heme-containing monooxidases that metabolize xenobiotics, steroids, fatty acids, and vitamins (Furge and Guengerich 2006). The highest expressed isoforms in the liver are CYPs 3A4, 2C9, 2C8, 2E1, and 1A2, while 2A6, 2D6, 2B6, 2C19, and 3A5 are less abundant but still more highly expressed in the liver than in any other organ (Zanger and Schwab 2013). It has been observed that drugs that are substrates of CYP450 enzymes have a higher likelihood of causing DILI in a dose-independent manner, while drugs that are CYP450 inhibitors have a higher likelihood of generating DILI only when they are administered at high daily doses (Yu et al. 2014). The majority of ADs and APs are substrates of CYP450 superfamily isoenzymes, whereby some of them were shown to be their inhibitors. CYP450 expression is influenced by genetic polymorphisms, xenobiotics, cytokines, and hormones, as well as sex and age. CYP3A4 is abundantly expressed in the liver in the majority of individuals and is responsible for the metabolism of > 50% of all clinically used drugs (Zanger and Schwab 2013).

Enzymes CYPs 3A4, 1A2, 2C9, 2C19, and 2D6 account for the metabolism of the majority of ADs and APs. The lipophilic property, which enables ADs to pass the cell membranes, might be one of the reasons why these drugs undergo extensive metabolism in the liver and show the first-pass effect, leading to variable bioavailability ranging from 30 to 80%. The linear relationship between dose and plasma concentrations exists for most ADs, except for paroxetine and fluvoxamine, whereby the time until peak plasma concentration is reached varies between 1 and 12 h (Mauri et al. 2014). Concerning APs, the relationship between dose and effect of several drugs vary considerably between patients, mainly due to pharmacokinetic differences influenced by age, changes in the first-pass effect, and the induction or inhibition of the metabolic system (Mauri et al. 2018). Most of the APs are also lipophilic and move freely across the lipoidal membranes. When administered orally, APs are well absorbed and undergo substantial pre-systemic elimination (bioavailability: 10–70%). They are highly bound to plasma proteins (75–99%) and tissues, and are extensively distributed (Javaid 1994). However, co-medication can influence pharmacokinetics. The patients with multiple comorbidities, which take many medications, are exposed to the risk of drug–drug interactions. Drug interactions are classified as either pharmacodynamic or pharmacokinetic. Pharmacodynamic drug interactions suggest that mechanisms of actions or adverse effects of two drugs used concomitantly are altered in nature, magnitude, or duration. Pharmacokinetic drug interactions occur where one drug alters the absorption, distribution, metabolism, or elimination of another (Preskorn and Werder 2006). Significant pharmacokinetic interactions with ADs and APs predominantly involve drug-induced changes in hepatic metabolism. Pharmacokinetic of these drugs can be altered by drugs that inhibit or induce metabolizing routes of appropriate CYP enzymes. Besides, some ADs and APs are inhibitors of the certain CYP enzymes themselves, so they can increase levels of other medications (Bleakley 2016). Psychotropic drugs that act as CYP450 inhibitors may cause adverse drug–drug interaction if co-administered with another drug which is CYP450 substrate.

Even though drug metabolism usually yields an inactive metabolite (detoxication), in some cases, metabolites can be more reactive than the parent drug (bioactivation) (Park et al. 2011). This is the case with some tricyclic antidepressants (TCAs) such as amitriptyline and imipramine, and APs like chlorpromazine and risperidone (Telles-Correia et al. 2017). Reactive metabolites may covalently bind cellular proteins, lipids, and nucleic acids, causing cell structure and function disruption. Binding to mitochondrial proteins may impair mitochondrial respiration and cause the leakage of superoxide radical anion (O2·−), which is one of the respiration mediators. Apart from producing reactive metabolites, CYP-mediated reactions during drug metabolism may also generate reactive oxygen species (ROS), including O2·−, hydrogen peroxide (H2O2), hydroxyl radical (.OH), hydroperoxyl radical (HOO.), and singlet oxygen (1O2) (Hrycay and Bandiera 2015).

Hepatotoxicity of ADs

ADs are drugs that relieve symptoms of depression, but are also used in treatments of anxiety disorders, panic, and obsessive–compulsive disorders. Monoamine oxidase inhibitors (MAOIs) and TCAs, the pioneering classes of ADs, were introduced in the clinical practice in the 1950s. MAOIs and TCAs were often related to liver toxicity (Table 1), but since their current use is infrequent, most of these cases were reported during the 1980s and ’90s. In the late 1990s, MAOIs and TCAs were largely replaced by safer and better tolerated new generations of ADs: selective serotonin reuptake inhibitors (SSRIs), serotonin and norepinephrine reuptake inhibitors (SNRIs), and serotonin antagonist and reuptake inhibitors (SARIs). MAOIs raised safety concerns due to their dietary and drug interactions (serotonin syndrome/toxicity, sympathomimetic drug interactions, and dietary tyramine interactions) (Grady and Stahl 2012), while both MAOIs and TCAs were associated with higher cardiac side-effects comparing to other ADs (Spindelegger et al. 2014). A major concern regarding TCAs is the extensive side effect profile, including anticholinergic symptoms, orthostatic hypotension, cardiac dysrhythmia, and neurological side-effects (including sedation and seizures) (Mullish et al. 2014). Thus, they are reserved primarily for depressed patients unresponsive to more commonly prescribed ADs such as the SSRIs or SNRIs. Fluoxetine, paroxetine, sertraline, citalopram, and fluvoxamine are SSRIs mostly linked with hepatotoxicity (Table 1). SNRIs venlafaxine and duloxetine, as well as SARI trazodone, are also strongly associated with hepatotoxic side-effects. All these drugs mostly demonstrate an idiosyncratic, unpredictable, and reversible hepatic injury. The injury may onset as early as after several days or after up to 6 months after drug administration and generally ends after the drug is withdrawn. Voican et al. (2014) reviewed relevant clinical data related to AD-induced liver injury published from 1965 onward and concluded that 0.5–3% of patients taking AD may develop asymptomatic mild elevation of serum aminotransferase levels.

Table 1 DILI cases with type of lesion and suspected mechanism, as well as hepatic effect reported in animal studies for listed ADs
Full size table

AD-induced DILI is generally of hepatocellular pattern, and less frequently of the cholestatic or mixed pattern. Regarding the pathophysiology, it is usually immune-allergic or metabolic (Voican et al. 2014). The observational study of Friedrich et al. (2016), based on data from a multicentre drug surveillance program conducted from 1993 to 2011 in German-speaking countries, reported that ADs caused DILI in 0.08% of patients. Their findings also suggested that SSRIs are less likely to contribute to DILI development compared to other classes of ADs. However, a recent study of Billioti de Gage et al. (2018) is not in line with this conclusion. These authors conducted a cohort study using data from the French National Health Insurance Database which included almost 5 million patients initiated on an AD from 2010 to 2015. They identified 382 severe DILI cases, but did not find evidence that SNRIs have a higher risk of serious liver injury than SSRIs. Interestingly, their analysis showed that duloxetine displays lower hepatotoxicity than SSRIs. However, one should bear in mind that duloxetine has previously been identified, in pre-marketing clinical trials and post-marketing monitoring, as a causative agent of liver injury (DeSanty and Amabile 2007) and thereby it had not been prescribed to patients believed to present higher susceptibility to DILI (e.g., elderly, obese patients, individuals with diabetes, chronic renal failure, etc.). Taking both studies into consideration, it is safe to assume that SSRIs are less likely to precipitate DILI comparing to TCAs and MAOIs, but not comparing to SNRIs. With regard to SARI, the incidence of severe liver failure leading to death or liver transplantation due to nefazodone is 1 per 250,000–300,000 patient-years of treatment. This drug carries a Food and Drug Administration (FDA) “black box” warning for hepatotoxicity and is withdrawn from use in many countries, so it will not be further discussed in this review. Fulminant hepatic failure leading to liver transplantation or death has also been reported for another SARI trazodone, but to a lesser extent comparing to nefazodone. Trazodone is still in wide use and will be discussed (Voican et al. 2014).

It is also important to note that SSRIs have the potential to cause drug–drug interactions through inhibition of CYP isoforms, which can contribute to DILI (Crewe et al. 1992). Fluoxetine and paroxetine are potent CYP2D6 inhibitors, whereas norfluoxetine, the main metabolite of fluoxetine, has a moderate inhibitory effect on CYP3A4 isoenzyme. Sertraline is a moderate inhibitor of CYP2D6, while citalopram has little effect on the major CYP isoforms (Hemeryck and Belpaire 2002). Thus, these SSRIs can substantially elevate the levels of co-prescribed drugs, which are dependent on these specific CYP enzymes. Consequent high exposure to the co-prescribed drug may result in both hepatic and non-hepatic complications (Preskorn and Werder 2006). Given the long half-life of both fluoxetine (2–6 days) and norfluoxetine (7–15 days), a special attention should be paid to fluoxetine administration, as its inhibitory effects on CYP- isoenzymes can persist for weeks after drug discontinuation (Bergstrom et al. 1993). Also, co-treatments with fluvoxamine and clozapine resulted in elevated plasma levels of clozapine, most likely due to drug–drug interaction. Of the SSRIs, fluvoxamine is one of the most potent inhibitors of the CYP1A2, isoenzyme predominantly involved in clozapine metabolism (Heeringa et al. 1999). Fluvoxamine also inhibits CYP3A4 and CYP2D6 (van Harten 1995). Co-administration of aripiprazole with fluvoxamine was shown to cause hepatic damage mediated by CYP3A4 inhibition and aripiprazole accumulation (Shastry et al. 2013). Therefore, co-administration of fluvoxamine with clozapine or aripiprazole should be avoided if possible or co-prescribed with caution.

Growing evidence suggests that obesity and diabetes mellitus are both risk factors for DILI. It is important to point out that a reciprocal link between depression and obesity has been noted. The meta-analysis on the longitudinal relationship between depression, overweight, and obesity, which included 15 studies, confirmed that the risk of depression is increased in obese people, as well as vice versa; depression is predictive for developing obesity (Luppino et al. 2010). This, together with the fact that obesity increases the risk of DILI, highlights the need to recognize the hepatotoxic potential of ADs.

The available data show that all listed ADs are associated with the risk of hepatotoxicity. Life-threatening or severe DILI has been reported for MAOI phenelzine, TCA imipramine, SSRI sertraline, SNRIs venlafaxine and duloxetine, and SARI trazodone, while SSRIs citalopram and fluvoxamine are characterized by lower risk.

Hepatotoxicity of APs

APs are drugs primarily prescribed to control schizophrenia and bipolar disorders, but are also used in the treatment of dementia and major depression. Typical, classical, or first-generation antipsychotics (FGA) were introduced more than 60 years ago. Those drugs, strong dopamine D2 receptor antagonists, are effective against hallucinations and delusions, but cause heavy extra-pyramidal side-effects (Miyamoto et al. 2005). In addition, commonly prescribed FGA, such as chlorpromazine and haloperidol, frequently resulted in elevated serum levels of the liver enzymes but also with severe liver damage (Table 2).

Table 2 DILI cases with type of lesion and suspected mechanism, as well as hepatic effect reported in animal studies for listed APs
Full size table

Atypical, second-generation antipsychotics (SGA), including clozapine, risperidone, olanzapine, and quetiapine, are available since the 1990s. These drugs antagonize not only D2 receptors but serotonin 5-HT2A/5-HT2C, histaminergic H1, and muscarinic M3 receptors as well (Meltzer 2013). Besides managing hallucinations and delusions, they are effective against anhedonia and apathy. Therefore, they are used in the treatment of depressive disorders alone or in combination with ADs (Wang and Si 2013). Unlike FGAs, SGAs produce minimal extra-pyramidal side-effects. On the other hand, they are strongly associated with adverse metabolic effects like weight gain, obesity, hyperlipidaemia, insulin resistance, hyperglycaemia, and diabetes (De Hert et al. 2012).

Clozapine is mostly associated with hepatotoxicity among SGAs (Telles-Correia et al. 2017). The asymptomatic rise in serum transaminase level is noticed in up to 60% of patients treated with clozapine, whereby 15–30% experienced twofold or threefold elevation (Wu Chou et al. 2014). The increase in serum levels of the liver enzymes is most often transient and asymptomatic. Still, many cases reported clozapine-induced liver damage and even fulminant liver failure accompanying moderate doses of clozapine (Table 2). A case report of hepatotoxicity from a low dose of clozapine (125 mg per day) was also reported (Wu Chou et al. 2014). However, in most cases, patients would recover from clozapine-induced DILI after stopping the treatment. Although knowledge regarding side-effects mostly comes from case studies, the relationship between clozapine and the hepatic injury was also unequivocally demonstrated in animal studies (Table 2). Both case and animal studies provide solid evidence on risperidone hepatotoxicity, as well (Table 2). Risperidone has been associated with clinically non-significant abnormalities in liver function tests, mostly with serum alkaline phosphatase increase, in 52.5% of cases. Marked elevation of serum levels of the liver enzymes was reported in 0.8% of cases, in a patient cohort encompassing 120 children and adolescents (Erdogan et al. 2008). Also, several case studies reported olanzapine-induced hepatotoxicity. In line with that, a number of animal studies demonstrated various hepatic biochemical and histological changes due to olanzapine (Table 2, discussed in more detail in further text). Quetiapine, another SGA, has been associated with cholestasis and rare fulminant hepatic failure (Table 2). Ilgin et al. (2018) demonstrated dose-dependent hepatotoxic effects of this drug in rats. Histopathology of the animal liver revealed necrotic regions accompanied by sinusoidal dilatation.

Most APs are hepatically metabolized by CYP2D6 and CYP3A4, although clozapine and olanzapine are predominantly metabolized via CYP1A2. Clozapine was shown to inhibit CYPs 2C19, 2D6, and 3A4, and therefore has the potential to cause pharmacokinetic drug–drug interactions (Urichuk et al. 2008). This is of particular importance for schizophrenia patients who are likely to take other drugs concomitantly.

The presented data show that both FGAs and SGAs are associated with a risk of hepatotoxicity. Life-threatening or severe DILI has been reported for FGA chlorpromazine and SGAs clozapine and quetiapine. Aripiprazole bears lower risk compared to other APs listed in Table 2. However, it should be borne in mind that this SGA was last introduced into practice, so it has been available for a shorter time for pharmacovigilance (pharmacological science related to the collection, detection, assessment, monitoring, and prevention of adverse effects of pharmaceutical products) to detect post-marketing signals of DILI.

Major pathophysiological pathways in DILI

Drugs may induce damage in the liver through multiple molecular mechanisms, including direct hepatic toxicity, as well as innate and adaptive immune response. Drugs and/or their metabolites can cause direct hepatic toxicity through oxidative stress by accumulating ROS and inducing mitochondrial dysfunction. They can also trigger inflammation and lipid accumulation, causing impairments in hepatic structure and function (Ye et al. 2018). Oxidative stress, inflammation, and lipid accumulation, as major pathophysiological pathways in DILI, will be discussed in more detail further in the text. Overview of molecular mechanisms will be followed by review and discussion of knowledge regarding AD- and AP-related case reports, animal, and in vitro studies. Figures summarizing oxidative, inflammatory, and, finally, steatogenic pathways demonstrated for listed ADs (Table 1) and APs (Table 2) will be presented at the end of each section dedicated to respective phenomena.

Oxidative stress in DILI

Oxidative stress is strongly associated with DILI. Various drugs induce oxidative stress in the liver by promoting ROS production and lipid peroxidation, as well as by reducing antioxidants levels (Li et al. 2015). The main sources of ROS in the liver are mitochondria and CYP450 enzymes in hepatocytes. Besides, significant sources of ROS generation are immune cells, primarily Kupffer cells (resident hepatic macrophages), and neutrophils (Cesaratto et al. 2004). ROS introduce carbonyl groups into amino acid side chains, affecting protein structure and function (Stadtman and Levine 2000). Furthermore, ROS may oxidize polyunsaturated fatty acids in the process of lipid peroxidation. Lipid peroxidation within the membrane bilayer increases membrane permeability and violates the integrity of the cells and organelles (Wong-Ekkabut et al. 2007). ROS may also cause different types of mutation of the DNA and oxidative histone modifications, and affect DNA methylation and DNA-associated proteins, interfering with gene expression profiles (Kreuz and Fischle 2016).

Hepatocytes are provided with an antioxidant system consisting of several enzymatic (superoxide dismutase—SOD, catalase—CAT, and glutathione peroxidase—GPx) and non-enzymatic (glutathione—GSH, vitamin E, vitamin C) antioxidant molecules. GSH, the most abundant thiol in cells, is capable of reacting directly with ROS and prevent damage to macromolecules. Besides, GSH is involved in peroxide reduction and drug detoxification in reactions that are catalysed by GPx and GST, respectively (Gupta et al. 2005). This is why, the deficit in GSH is the main indicator of a pro-oxidant state and is commonly associated with liver injury (Han et al. 2006). Antioxidant enzymes in the liver are influenced by various factors, including stress (Filipović et al. 2010), hormones (Pajović and Saicić 2008), as well as xenobiotics (Durašević et al. 2010; Djurasevic et al. 2019).

Pro-oxidant effects of ADs in the liver

Oxidative stress in the brain plays a major role in the pathophysiology of depression, and the antioxidant effects of ADs may result in the remission of depression (Xu et al. 2014). However, ADs seem to have antioxidant or pro-oxidant effects, depending on the tissue and the dose applied. Thus, in the case of SSRIs, the majority of data highlight an antioxidant effect in brain tissues, while in the liver, these drugs mainly show a pro-oxidant effect (Ștefan et al. 2020). The influence of ADs on hepatic mitochondria homeostasis and redox balance, in general, was mostly studied in the case of SSRI fluoxetine. Souza et al. (1994) showed that fluoxetine, as well as its main metabolite norfluoxetine, disturbed oxidative phosphorylation and increased pro-oxidant production in isolated hepatic mitochondria, while our results showed that fluoxetine cause oxidative damage of lipids and proteins in rat liver (Zlatković et al. 2014). In line with that, increased total oxidant status paralleled with a decrease in total antioxidant capacity were reported in the liver of fluoxetine-treated rats (Yilmaz et al. 2016). Accordingly, Elgebaly et al. (2018) reported a decreased level of GSH and activity of SOD, CAT, and GPx, accompanied by increased lipid peroxidation and NO production in rat liver due to fluoxetine. The same study revealed the up-regulation of pro-apoptotic Bax and caspase-3 and down-regulation of anti-apoptotic Bcl-2, indicating pro-apoptotic events in hepatocytes (Fig. 1). The authors speculated that an imbalance between Bax and Bcl-2 was a consequence of oxidative stress. It is known that ROS down-regulates Bcl-2 and originates the pro-apoptotic mitochondrial signalling (Herrera et al. 2001). Besides, fluoxetine easily crosses the mitochondrial membrane and interferes with membrane-bound proteins causing pro-apoptotic events (de Oliveira 2016). Hepatic oxidative stress due to fluoxetine is also, at least in part, a result of increased O2•− production coming from intensive CYP-mediated metabolism (Bondy and Naderi 1994). It is known that CYP-derived ROS can cause oxidative damage of macromolecules, impair mitochondrial function, and trigger signalling resulting in cell death (Begriche et al. 2011; Du et al. 2016).

Fig. 1
figure 1

Oxidative stress in AD- and AP-induced DILI. Summarizing hepatic oxidative stress pathways demonstrated for listed antidepressants (ADs) and antipsychotics (APs). a Drug bioactivation by CYP450 enzymes into reactive metabolite that can bind macromolecules covalently and impair cell structure and function: amitriptyline, imipramine, tianeptine, citalopram, duloxetine (ADs); chlorpromazine, risperidone, clozapine, olanzapine (APs); b CYP-mediated generation of reactive oxygen species (ROS) and oxidative damage of macromolecules: fluoxetine, sertraline, citalopram (ADs); chlorpromazine, haloperidol, clozapine (APs); c Covalent binding of reactive drug metabolites to mitochondrial proteins, compromising mitochondrial respiration and causing O2·− hyperproduction: fluoxetine, sertraline (ADs); d Initiating apoptosis by down-regulation of anti-apoptotic Bcl-2, up-regulation of pro-apoptotic Bax, cytochrome C release from mitochondria and caspase-3 activation: fluoxetine (AD); e Compromising antioxidant defence by decreasing glutathione (GSH) level and activity of glutathione-S transferase (GST), glutathione peroxidase (GPx) and catalase (CAT): fluoxetine, citalopram (ADs); chlorpromazine, haloperidol, clozapine (APs)

Full size image

Regarding SSRIs other than fluoxetine, sertraline and citalopram have been associated with oxidative stress as well. Sertraline was demonstrated to inhibit mitochondrial respiration complexes by uncoupling oxidative phosphorylation in isolated hepatic mitochondria (Li et al. 2012) (Fig. 1). In addition, 10-day sertraline treatment was found to increase lipid peroxidation and NO production in the mice liver (Abdel-Salam et al. 2013). It has been shown recently that DNA damage plays an important role in sertraline-induced hepatic cytotoxicity. The same study has shown that the parental form of sertraline is likely the main cause of liver toxicity, while CYP2D6, 2C19, 2B6, and 2C9-mediated metabolism is responsible for its detoxification (Chen et al. 2020). Citalopram has also been associated with ROS overproduction and antioxidant defence depletion in rat liver. Ahmadian et al. (2017) indicate that metabolic activation plays a crucial role in citalopram-induced hepatotoxicity. TCAs imipramine (Duda et al. 2016) and amitriptyline (Bautista-Ferrufino et al. 2011) were also shown to increase ROS production and lipid oxidation in rat and mice liver.

Concerning SNRIs, venlafaxine and duloxetine were demonstrated to induce oxidative stress in hepatocytes. Mitochondria and lysosomes are proposed as the primary targets in venlafaxine-mediated cell damage (Ahmadian et al. 2016). Mitochondrial dysfunction may be triggered by the drug itself or its reactive drug metabolite, generated through the activity of CYP450 enzymes. In the case of duloxetine, hepatotoxicity might come from oxidative stress triggered by hydroxylated and epoxide metabolites, which can release free radicals or ROS (Mishra et al. 2016). This is corroborated by our findings that duloxetine increase GPx and GST activity in the rat liver (Đorđević et al. 2016). SARI trazodone also demonstrated pro-oxidant action by inducing oxidative stress in isolated rat liver mitochondria (Dykens et al. 2008).

Pro-oxidant effects of APs in the liver

There is a large body of data concerning the roles of oxidative stress in psychiatric disorders and the redox status in patients treated with FGA and SGA, which is reviewed elsewhere (Padurariu et al. 2010; Hendouei et al. 2018; Lin and Lane 2019). However, far less data are available on how APs influence redox balance in the liver, as most studies were conducted on brain tissue and blood serum.

Chlorpromazine was shown to impair the antioxidant system by decreasing SOD and CAT activity and reducing GSH-dependent defence, as well as to cause lipid peroxidation and nitrosative damage in the rat liver (Dejanovic et al. 2017). Chlorpromazine-induced hepatotoxicity is the result of its bioactivation in hepatocytes, which can be done by (1) CYP-catalysed pathway, which yields highly reactive quinone imine metabolite; or (2) peroxidase-catalysed oxidation, which generates toxic radicals (MacAllister et al. 2013). Haloperidol, another FGA, was also shown to decrease GSH level and produce lipid peroxidation in rat and mice liver (Andreazza et al. 2015; El-Awdan et al. 2015; Abdel-Salam et al. 2018).

Regarding SGAs, clozapine was shown to cause severe adverse effects, including metabolic syndrome (Pillinger et al. 2020) and agranulocytosis (Mijovic 2020), as well as hepatotoxicity (Table 2). Both metabolic syndrome and agranulocytosis were previously linked to oxidative stress (Fischer et al. 1991; Ando and Fujita 2009). Oxidative bioactivation of clozapine by CYP450 generates highly reactive nitrenium ion, clozapine-N-oxide. This reactive metabolite causes ROS formation, imposes oxidative stress, and requires GSH-mediated neutralization (Dragovic et al. 2013) (Fig. 1). In line with that, we found that clozapine increased GST activity, decreased GSH level, and caused oxidative damage of lipids and proteins in the rat liver (Zlatković et al. 2014). Due to noted severe adverse effects caused by highly reactive nitrenium metabolite, the use of clozapine is restricted. In an effort to modify the clozapine structure to minimize the nitrenium ion formation, another SGA was designed: quetiapine. This AP has not been associated with hepatic oxidative stress so far. On the other hand, risperidone-induced hepatotoxicity, associated with oxidative stress, mitochondria, and lysosome damage, was reported in rats (Eftekhari et al. 2016). Our results also revealed that olanzapine, like clozapine, increases hepatic GST activity, probably due to the elimination of reactive olanzapine metabolites (Todorović et al. 2016). Four major oxidative metabolites of olanzapine, 4′-N-desmethyl olanzapine, 7-OH olanzapine, 2-OH olanzapine, and N–O olanzapine were formed in vitro by human liver microsomes (Wrighton and Ring 1999).

Liver inflammation in DILI

Activation of the immune response in DILI involves activation of hepatic resident immune cells such as Kupffer cells, natural killer, and natural killer T cells, as well as recruitment of peripheral immune cells like lymphocytes and neutrophils (Barnes et al. 2013).

Reactive drug metabolites may trigger apoptosis and necrosis and initiate an inflammatory response through the release of damage-associated molecular patterns (DAMP) from damaged cells. Kupffer cells express a plethora of receptors necessary for the detection of DAMP (e.g., Toll-like receptors), and therefore mediate the initial response to acute cell injury, which may include the release of inflammatory cytokines and chemokines (Woolbright and Jaeschke 2018). Toll-like receptor family activates nuclear factor-kappa b (NF-κB), a pro-inflammatory transcription factor that promotes production not only of cytokines but ROS as well (Morgan and Liu 2011). Thus, activated Kupffer cells, a major source of inflammatory mediators, including cytokines, O2·−, NO, eicosanoids, chemokines, as well as lysosomal and proteolytic enzymes, demonstrate increased cytotoxicity and chemotaxis (Kolios et al. 2006).

DILI is often associated with immune cell infiltrates, and the nature and size of these infiltrates may serve as a prognosis of the progress and severity of tissue damage. The liver is a site where lymphocytes undergo apoptosis during the resolution phase of the peripheral immune response. However, infiltrating activated B and T lymphocytes have been associated with hepatic histopathology (Ikeda and Yoshikawa 2003). Cytotoxic (CD8 +) T cells were found to be dominant lymphocytes present in liver biopsies of subjects with DILI (Foureau et al. 2014). Besides direct toxicity, reactive drug metabolite can act as a hapten which covalently binds to protein and form an immunogenic hapten–carrier adduct. These adduct molecules activate antigen-presenting cells, which further activate T cells that may be cytotoxic to other cells, including hepatocytes. Also, covalent modification of self-proteins by reactive drug haptens may trigger the loss of self-recognition. Consequently, immune cells whose primary role is to fight against non-self-cells, such as bacteria, would then attack self-cells, leading to autoimmune-mediated hepatitis (Pohl et al. 1989; Woolbright and Jaeschke 2018).

Liver eosinophilia has often been associated with DILI. It was reported in 40% of 570 case reports reviewed by Bjornsson et al. (2007). Eosinophils have been associated with detrimental effects on one hand, but also with remodelling and repairing ones on the other. Although the role played by eosinophils in DILI is ambiguous, in the clinical context, the presence of eosinophilia, in most cases, indicates a good prognosis (Bjornsson et al. 2007).

Finally, neutrophils in an infiltrate are common in the early response to tissue damage (Jaeschke and Tadashi 2006). Neutrophils are essential mediators of drug hepatotoxicity, since they are a significant source of ROS, mainly O2·− produced by NADPH oxidase 2. Inside neutrophils O2·− dismutates to H2O2, which may diffuse into hepatocytes and provoke intracellular oxidative stress, or be used by myeloperoxidase to generate hypochlorous acid, a potent oxidant and chlorinating agent (Jaeschke and Tadashi 2006). Thus, neutrophil-derived ROS mediate further damage to liver tissue (Fig. 2).

Fig. 2
figure 2

Inflammation in AD- and AP-induced DILI. Summarizing hepatic inflammatory pathways demonstrated for listed antidepressants (ADs) and antipsychotics (APs). a Necrosis and apoptosis caused by reactive drug metabolite and/or CYP-derived reactive oxygen species (ROS) and release of damage-associated molecular patterns (DAMP): amitryptiline, fluoxetine, duloxetine (ADs); haloperidol, clozapine, olanzapine, quetiapine (APs); b activation of Kupffer cells through DAMP-mediated toll-like receptors (TLR) activation, and initiation of inflammatory response—activation of nuclear factor-kappa B (NF-κB); increase in production and release of cytokines (TNF-α and IL-1β), ROS, NO and chemokines: fluoxetine, duloxetine, sertraline (ADs); haloperidol (AP); c infiltration of inflammatory cells, mainly lymphocytes and eosinophils, triggered by chemokines/cytokines leakage: amitriptyline, fluoxetine, paroxetine, sertraline, citalopram, venlafaxine, duloxetine (ADs); haloperidol, clozapine, risperidone, olanzapine, quetiapine (APs); d Neutrophil infiltration and activation; increased production and release of ROS; ROS diffusion to hepatocytes and reinforcement of oxidative damage: fluoxetine (AD); clozapine, olanzapine (APs)

Full size image

Pro-inflammatory effects of ADs in the liver

AD-associated DILI is generally immuno-allergic or metabolic, regarding pathophysiology. Hepatitis, inflammatory liver disease, was reported in patients treated with various TCA drugs. A case of disturbed hepatic architecture due to massive liver necrosis with portal areas containing lymphocytes and a significant amount of eosinophils was described in an imipramine-treated patient (Ilan et al. 1996). Amitriptyline-induced hepatitis with infiltration of mononuclear cells, neutrophils, and eosinophils (Larrey et al. 1988), as well as with massive necrosis and an intense inflammatory reaction (Danan et al. 1984) were also demonstrated. Recently, another acute hepatitis case due to amitriptyline was reported (Chen et al. 2019). In addition, acute hepatitis was described in a patient 8 weeks after beginning tianeptine administration (Bricquir et al. 1994).

SSRI-induced hepatitis is rare, but can be serious. Chronic hepatitis has been rarely attributed to fluoxetine therapy, although acute hepatitis was reported in several cases (Table 1). A patient with moderately expanded portal tracts, infiltrated by lymphocytes and plasma cells, was described in a case of hepatitis associated with paroxetine (Colakoglu et al. 2005) (Fig. 2). Curiously, fatal hepatitis was reported in a patient treated with sertraline (Fartoux-Heymann et al. 2001), a drug that was considered less dangerous for the liver, comparing to other SSRIs. The authors speculated that sertraline-induced hepatitis was due to immunization against self-proteins covalently modified by a reactive metabolite generated in situ in the liver. However, the reactive metabolites have not been identified yet, although it is known that sertraline is extensively metabolized by hepatic CYP450 (Fartoux-Heymann et al. 2001). In addition, lobular inflammation and a mild prominence of eosinophils were reported in another patient treated with sertraline (Suen et al. 2013) (Fig. 2).

The other group of the new generation ADs, SNRIs, also rarely causes hepatitis. Venlafaxine-induced hepatitis was reported only in a few cases. Surprisingly, the first case was described in a patient who received venlafaxine in a low dose (75 mg per day) (Phillips et al. 2006). However, fulminant hepatic failure was reported in a patient receiving venlafaxine combined with trazodone at normal therapeutic doses (Detry et al. 2009). Cholestatic hepatitis involving mixed portal inflammatory infiltrates along with eosinophils was also described in a patient taking venlafaxine (Stadlmann et al. 2012). The post-marketing reports of hepatitis and cholestatic jaundice in patients receiving duloxetine emerged in late 2005. Fulminant hepatic failure in a duloxetine-treated patient reported by Hanje et al. (2006) was associated with a mixed inflammatory infiltrate within the portal and lobular areas composed mainly of neutrophils with occasional eosinophils and lymphocytes. Neutrophils are important mediators of drug hepatotoxicity, since they are a significant source of ROS, which further damage liver tissue (Jaeschke and Tadashi 2006). Vuppalanchi et al. (2010) described in detail a case of duloxetine-associated hepatitis. The histological analysis showed a pronounced portal, and sinusoidal infiltrates consisted of activated lymphocytes. Besides, Kupffer cells appeared activated, and rare foci of necrosis were detected, while endothelitis was present in central and portal venules. SARI trazodone, combined with diazepam, was demonstrated to cause hepatocellular liver injury with confluent necrosis and modest inflammation (Carvalhana et al. 2016).

Regarding animal studies, there are little data about inflammatory changes in the liver under AD treatment. Most studies concerning the immuno-modulatory effects of ADs were conducted in blood and brain tissues. However, few studies described the pro-inflammatory effects of fluoxetine in rat liver. Özden et al. (2005) described lobular inflammation, focal necrosis, apoptosis, portal area inflammation, and Kupffer cell hyperplasia. In line with that, we found the infiltrates of neutrophils, lymphocytes, and macrophages in portal triads; Kupffer and inflammatory cells in sinusoids, and focal necrosis in the liver of fluoxetine-treated rats (Zlatković et al. 2014) (Fig. 2). Like fluoxetine, sertraline was found to induce hepatic inflammation. Analysing the liver of rabbits exposed to sertraline, Almansour et al. (2018) noted sinusoidal dilatation; lobular and periportal hepatic spaces infiltrated mostly by lymphocytes and plasma cells; portal vessel congestion, as well as Kupffer cell increased number and hyperplasia. Similarly, a mild inflammatory cell infiltrate, composed mainly of lymphocytes, was detected in citalopram-treated rats (Ahmadian et al. 2017). SSRIs were also reported to increase the production of pro-inflammatory cytokines TNF-α, IL-6, and IL-1β in an in vitro whole blood assay (Munzer et al. 2013). Thus, scarce data regarding liver inflammatory changes under ADs are mostly related to SSRI drugs. However, it is noteworthy that liver inflammation, including infiltration of inflammatory cells and dilated sinusoids, was observed in rats treated with an SNRI venlafaxine, as well (Paulis et al. 2018). Here is also important to address the question concerning the applicability of rodent models for understanding hepatic pro-inflammatory effects of drugs. A recent study of Ölander et al. (2020) showed that human and murine hepatocytes were fairly similar, whereas non-parenchymal cells, including Kupffer cells, tended towards larger species differences. Namely, they found substantial proteomic differences between human and murine non-parenchymal cell types. This interspecies variability in protein expression contributes to the debatable predictive value of animal models in toxicity of drugs, including their hepatic pro-inflammatory effects.

Pro-inflammatory effects of APs in the liver

Hepatitis has been associated with antipsychotic medication, since it was reported in patients treated with both FGAs and SGAs, including haloperidol, clozapine, risperidone, olanzapine, quetiapine, and aripiprazole (Table 2). Severe cases of hepatitis, even fatal in a small minority, have been observed. Massive infiltration of inflammatory cells, predominantly eosinophils, in the vicinity of portal bile ducts was noticed in a patient with haloperidol-induced hepatitis (Fuller et al. 1977) (Fig. 2). With regard to SGA, acute necrotic hepatitis, as well as cholestatic hepatitis with focal necrosis and eosinophil infiltration were reported after clozapine treatment (Slim et al. 2016). Also, immune-allergic hepatitis with eosinophilia and high levels of anti-smooth-muscle antibodies was described in a patient taking risperidone (Esposito et al. 2005). Hepatocellular damage, centrilobular necrosis, and infiltrates in the portal areas containing mononuclear granulocytes, lymphocytes, and eosinophils were observed in some patients treated with olanzapine (Slim et al. 2016) (Fig. 2). Extensive hepatocyte necrosis with nonspecific inflammatory infiltrates was found after quetiapine treatment (Shpaner et al. 2008; Slim et al. 2016). Finally, portal inflammatory cellular reaction with eosinophilia was reported in a patient using aripiprazole (Kornischka and Cordes 2016).

Animal studies have also shown hepatic inflammatory changes in response to AP treatment. Haloperidol-treated rats manifested centrilobular necrosis and inflammatory cell infiltration, especially in the periportal area, as well as high TNF-α immunoreactivity in the liver (Abdel-Salam et al. 2018). Our study reported infiltrates of neutrophils, lymphocytes, and macrophages in portal triads; Kupffer and inflammatory cells in sinusoids, and focal necrosis in the liver of clozapine-treated rats (Zlatković et al. 2014) (Fig. 2). We also described portal inflammation and focal hepatocyte necrosis in olanzapine-treated rats (Todorović et al. 2016). In line with that, Elbakary (2017) detected various histological changes, including dilatation and congestion of central veins, inflammatory cellular infiltration in the portal areas, cytoplasmic vacuolation of hepatocytes, mitochondrial degeneration, bile ducts dilatation, and excessive deposition of lipid droplets in olanzapine-treated rats. Noted cellular infiltration was described as the consequence of ROS, cytokines, and chemokines released from activated Kupffer cells. Besides, Mahmoud and El-deek (2019) (Fig. 2) recently reported lobular hepatitis, marked dilation and congestion of central vein and blood sinusoids, spotty necrosis, and portal tract expansion by inflammatory cells in rat liver due to olanzapine.

Fatty liver changes in DILI

Hepatic steatosis, or fatty liver, is a reversible condition in which fat accumulates in hepatocytes due to an imbalance between lipid deposition and removal. An increase in fatty acid influx from the diet and adipose tissue lipolysis, or de novo lipogenesis, and a decrease in fatty acid removal through β-oxidation or very-low-density lipoprotein (VLDL) secretion, can all lead to the intracellular lipid droplets accumulation and hepatic steatosis (Marchesini et al. 2016). All of the noted mechanisms can be affected by different ADs and APs.

Steatosis occurs in two primary forms, macrovesicular and microvesicular. Macrovesicular steatosis or macrosteatosis is characterized by a single, large fat vacuole per hepatocyte that fills up most of the cytoplasm and displaces the nucleus to the periphery of the cell. Macrosteatosis has a good long-term prognosis with rare progression to fibrosis or cirrhosis (Bessone et al. 2018). On the other hand, diffuse, microvesicular steatosis, or microsteatosis consists of enlarged hepatocytes with multiple small droplets in the cytoplasm, while the nucleus holds a central location. It is often associated with chronic lipid peroxidation, development of steatohepatitis, liver insufficiency, and encephalopathy, and can be life-threatening when extensive or long-lasting (Amacher and Chalasani 2014).

Steatogenic effects of ADs in the liver

Hepatic steatogenic effects of different AD classes were shown in patients, animal models, as well as in in vitro studies. Regarding case studies, moderate steatosis due to imipramine treatment was reported (Horst et al. 1980). Moderate macro- and microvesicular fat accumulation, involving 40% of hepatocytes, was detected in a patient who developed acute hepatitis after taking tianeptine for 8 weeks (Bricquir et al. 1994). Recently, a case of fluoxetine-induced hepatomegaly (enlarged liver) with steatosis was described (Agrawal et al. 2019). In the case of SNRI, moderate-to-marked steatosis was detected in patients who received venlafaxine (Levine et al. 1996), as well as duloxetine (Hanje et al. 2006; Vuppalanchi et al. 2010).

In vitro and animal studies provided substantial evidence that various ADs have the steatogenic effect. TCAs amitriptyline and imipramine were shown to promote the activation of sterol regulatory element-binding proteins (SREBP), the most important regulators of cellular cholesterol and free fatty acids’ (FFAs) biosynthesis (Fig. 3). This effect was observed in cultured human liver cells (Raeder et al. 2006). Besides, the steatogenic effect of amitriptyline was shown in vivo in rat liver (Sahini et al. 2014; Kampa et al. 2020). Interestingly, the other two TCAs, amineptine and tianeptine, were shown to cause mild hepatic steatosis due to heptanoic chain attached to a tricyclic moiety. This heptanoic chain undergoes mitochondrial β-oxidation, which shortens it to the 5-carbon and 3-carbon derivatives. Therefore, hepatic mitochondria of the patients receiving amineptine or tianeptine are exposed to C7, C5, and C3 analogues of natural FFAs, which reversibly inhibit β-oxidation of medium- and short-chain fatty acids (Fromenty et al. 1989; Pessayre et al. 2012). As a result, FFAs increasingly undergo esterification into triglycerides that accumulate in hepatocytes as benign large lipid droplets (Bessone et al. 2018). As opposed to the steatogenic effects of ADs described so far, a recent study demonstrated that MAOI phenelzine reduced hepatic lipid content in mice fed with a high-fat diet (Mercader et al. 2019).

Fig. 3
figure 3

Mechanisms of fat accumulation in hepatocytes due to AD- and AP-induced DILI. Summarizing hepatic steatogenic pathways demonstrated for listed antidepressants (ADs) and antipsychotics (APs). a Activation of sterol regulatory element-binding proteins (SREBR) and up-regulation of lipogenic enzymes resulting in increased triglycerides’ (TG) synthesis and accumulation in fat vacuole: amitryptiline, imipramine (ADs); clozapine (AP); b down-regulation of lipolytic enzymes and consequential accumulation of TG in fat vacuole: fluoxetine (AD); clozapine (AP); c reduced β-oxidation of free fatty acids (FFAs) in mitochondria: amineptine, tianeptine (ADs); clozapine (AP); d increased de novo lipogenesis induced by hyperglycaemia and hyperinsulinemia: fluoxetine (AD); clozapine, olanzapine (APs). e Reduced transport of long-chain fatty acids into the mitochondria due to decreased levels of available carrier l-carnitine: clozapine, olanzapine (APs); f decreased extrahepatic transport of TG due to down-regulation of very-low-density lipoprotein (VLDL) secretion-related proteins apoB and apoE: olanzapine (AP)

Full size image

The long-term use of SSRIs is associated with obesity/weight gain, diabetes mellitus, and dyslipidemia in children and adults (Jerrell 2010). Also, prenatal exposure to SSRIs, the most common ADs used in pregnancy, increases the risk of childhood overweight (Grzeskowiak et al. 2013). The rodent model demonstrated that fetal and neonatal exposure to fluoxetine cause adiposity, fatty liver, and insulin resistance (De Long et al. 2015). Indeed, the relationship between obesity, hepatic lipid accumulation, insulin resistance, and hepatic inflammation was described in both clinical and animal studies (Gruben et al. 2014). Obesity and fatty liver changes have been associated with increased production of pro-inflammatory cytokines, including TNF-α, IL-6, and IL-1β (Ma et al. 2008). As mentioned in the section “Pro-inflammatory effects of ADs in the liver,” SSRIs were shown to increase the production of these pro-inflammatory mediators (Munzer et al. 2013). In vitro studies have found that fluoxetine regulates lipogenic and lipolytic genes to promote hepatic lipid accumulation (Fig. 3). Specifically, fluoxetine was reported to induce hepatic lipid accumulation in primary mouse hepatocytes via both promotion of de novo lipogenesis and reduction of lipolysis (Feng et al. 2012; Xiong et al. 2014). These findings are supported by the results of a recent in vivo study of Pan et al. (2018), who showed that fluoxetine increased hepatic triglyceride level by increasing the expression of lipogenic and decreasing the expression of lipolytic enzymes in the mice liver (Fig. 3). In line with that, we noticed macrovesicular fatty changes in the liver of rats treated with fluoxetine (Zlatković et al. 2014). Chronic exposure to sertraline was shown to down-regulate expression of drug- and arachidonic acid-metabolizing enzymes in hepatic tissue, indicating that patients using this drug may be at risk of hepatotoxicity with reduced capacity to metabolize drugs and fatty acids (Almansour et al. 2018). SSRI fluvoxamine was also demonstrated to cause fat accumulation in the liver of mice by shifting the metabolism towards fatty acid synthesis (Rozenblit-Susan et al. 2016).

Steatogenic effects of APs in the liver

Many AP drugs, particularly SGAs, are strongly associated with adverse metabolic effects, including steatosis. Even though some FGAs demonstrated steatogenic effects, like chlorpromazine did in the rat liver (Yang et al. 2015), steatosis is primarily associated with SGAs. Fatty liver was identified in two cases of pediatric patients treated with risperidone. Both cases were accompanied by obesity and abnormal serum levels of the liver enzymes (Kumra et al. 1997). Risperidone-associated steatohepatitis, accompanied by excessive weight gain, was also described (Holtmann et al. 2003). A study of Morlán-Coarasa et al. (2016) aimed to investigate the development of fatty liver during the first 3 years of SGA treatment (aripiprazole, risperidone, quetiapine, and ziprasidone) in 191 non-affective psychosis patients. After 3 years, 25.1% of individuals scored ≥ 60 on a fatty liver index, representing a strong prediction of steatosis. The fatty liver index predicts fatty liver disease based on the body mass index, waist circumference, and blood levels of triglyceride and ɣ-glutamyltransferase, an enzyme involved in drug detoxification.

SGAs, primarily clozapine and olanzapine and, to a lesser extent, quetiapine, risperidone, and aripiprazole can trigger body weight gain, insulin resistance, and metabolic syndrome. All these metabolic disorders promote liver lipogenesis and can subsequently lead to fatty liver and steatohepatitis (Slim et al. 2016). Insulin resistance exposes hepatocytes to an overload of glucose and insulin due to decreased glucose uptake by peripheral tissues and compensatory hyperinsulinemia, respectively. The resulting hyperglycaemia and hyperinsulinemia can promote de novo lipogenesis (Matsuzaka and Shimano 2011). Besides, insulin represses the synthesis of apolipoprotein B (apoB), the main VLDL protein cofactor, reducing the exportation of hepatic lipids via VLDL (Charlton et al. 2002). Thus, high insulin levels further increase hepatic triglyceride levels by blocking VLDL-mediated lipid secretion. Although AP-induced steatosis may be the indirect result of drug-induced obesity and insulin resistance, an experimental study on primary rat hepatocytes showed that clozapine and olanzapine can directly increase de novo lipogenesis, as well (Lauressergues et al. 2010) (Fig. 3). A later study on the human hepatocyte cell culture demonstrated that other APs like haloperidol, olanzapine, and risperidone activated SREBP and increased expression of target genes involved in FFAs biosynthesis, resulting in intracellular lipids accumulation (Lauressergues et al. 2012). These results corroborated the findings of Raeder et al. (2006) that clozapine activates SREBP-mediated lipogenic gene expression in cultured human liver cells. These findings were confirmed in our in vivo study, where clozapine was shown to induce rat liver steatosis (Zlatković et al. 2014). Liver steatosis, as well as an increase in plasma triglycerides and FFAs, due to clozapine treatment, was confirmed by later studies (Li et al. 2017; Liu et al. 2017). Moreover, a single intraperitoneal injection of clozapine (50 mg/kg) was demonstrated to down-regulate genes involved in FFA β-oxidation and lipolysis, leading to lipid accumulation in rat liver (Fernø et al. 2009) (Fig. 3). It was found recently that clozapine can contribute to the fatty liver by reducing renal reabsorption of l-carnitine, an amino acid that acts as a carrier for transporting long-chain fatty acids into the mitochondria for β-oxidation (Wang et al. 2019). Similar to clozapine, olanzapine is strongly associated with fatty liver changes. This AP dysregulated glucose and lipid metabolism, caused mild glucose intolerance, increased body weight, and caused hepatic steatosis in the mouse model after only 4 weeks. When combined with a high-fat diet, olanzapine caused an even greater increase in weight gain, dysregulation of glucose and lipid metabolism, and more pronounced liver damage (Isaacson et al. 2020). These results indicate that obesity may aggravate olanzapine-induced side-effects. Soliman et al. (2013) reported that olanzapine- and aripiprazole-treated rats developed steatosis in the liver, whereby more prominent changes were caused by olanzapine. Hepatic abnormalities of lipid and glucose metabolism due to olanzapine have been intensively studied lately. All these studies accumulated evidence that olanzapine has multifaceted effects on hepatic lipid metabolism, including (1) increasing de novo lipogenesis; (2) decreasing lipoprotein internalization and cholesterol clearance; and (3) reducing VLDL secretion through down-regulation of VLDL secretion-related proteins, including apoB and apoE; which all together result with lipid accumulation in the liver (Chen et al. 2018; Liu et al. 2019; Mahmoud and El-deek 2019; Ren et al. 2019a, b) (Fig. 3). In addition, a recent study demonstrated that olanzapine, like clozapine, decreases the level of available l-carnitine and, therefore, negatively affects β-oxidation of FFAs in mice liver (Jiang et al. 2019). Thus, l-carnitine supplementation could be a promising strategy to prevent or ameliorate olanzapine- and clozapine-induced steatosis. Further on, visceral adiposity associated with hepatic steatosis in mice after long-term risperidone treatment was demonstrated, as well (Auger et al. 2014). Risperidone was also shown to significantly increase gene expression of major lipogenic factors, including fatty acid synthase, in the mouse and rat liver (Auger et al. 2018; Azirak et al. 2019) (Fig. 3).

Conclusions

All ADs and APs listed in this review have the potential to cause hepatotoxicity, even at therapeutic doses. AD- and AP-induced DILI can be mediated through multiple molecular mechanisms, including unwanted side reactions and products of drug metabolism, hepatic oxidative stress, inflammation, and steatosis. Although fulminant and irreversible liver injury due to ADs or APs is a rare event, hepatotoxicity of these drugs should be recognized and carefully considered. These drugs carry a potential risk for liver damage, particularly in patients with polypharmacy or reduced drug metabolism. Besides, patients suffering from disorders that share pathogenic mechanisms with DILI, such as diabetes and metabolic syndrome, bear a higher risk of developing psychotropics-induced hepatotoxicity. In this respect, therapy must be carefully planned, tailored, and monitored, not just in terms of the effect on a psychiatric disorder of patient, but hepatic homeostasis, as well. For example, patients prone to DILI may benefit from lower doses and slower upwards titration to avoid the hepatotoxic effects. When using ADs and APs associated with a greater risk of hepatotoxicity (nefazodone, phenelzine, imipramine, duloxetine, and trazodone as well as chlorpromazine and clozapine, respectively), the serum levels of the liver enzymes should be monitored regularly. Also, these drugs should not be initiated in individuals with pre-existing liver disease, as well as in elderly patients, and alcohol and illegal drug users. Even though there is no clear evidence that pre-existing chronic liver disease increases the likelihood of developing DILI, baseline abnormalities can complicate patient monitoring. Also, a special attention should be paid when implementing therapy combining SGA and AD, which is often the case in treatment-resistant depressive patients. Even though SGA combined with an SSRI can ameliorate depressive symptoms, one should always bear in mind that drug–drug interactions may have serious, even life-threatening consequences. Aripiprazole, quetiapine, and olanzapine have US FDA approval as an augmentation therapy to ADs, but one should always consider their association with metabolic disorders, which may promote liver lipogenesis. For example, patients could benefit from monitoring for signs of the fatty liver when taking these drugs together with fluoxetine; SSRI also related to adiposity and steatosis. Besides, SSRI fluvoxamine co-administration with SGAs clozapine or aripiprazole should be co-prescribed with special caution due to drug–drug interactions described in section “Hepatotoxicity of ADs.” Thus, clinicians should carefully assess the potential risk-to-benefit ratio when planning to implement adjunctive pharmacotherapy. Personalized medicine should be the preferred approach whenever possible and mandatory for risk groups of patients. Finally, patients should be informed that ADs and APs can be associated with liver damage, even liver failure, and encouraged to refer to a primary care physician or psychiatrist if any DILI-related symptoms emerge. In addition, consumption of alcohol, illegal drugs, and over-the-counter medications during ADs and APs treatment may increase the risk of severe liver abnormalities, so it should be strongly discouraged.