Background
Drug-induced liver injury (DILI) refers to liver injury caused by the direct toxicity of a drug or its metabolites or by idiosyncratic reactions after exposure to drugs [
1‐
3]. More than 1,100 drugs have been confirmed to be hepatotoxic worldwide, including non-steroidal anti-inflammatory drugs, anti-infective drugs (including anti-tuberculosis drugs), anti-tumor drugs, cardiovascular system drugs, and biological agents [
3]. The annual incidence of DILI is estimated to be approximately 23.80 per 100,000 in the general population of China and 12 per 100,000 in Korea [
4,
5]. In a prospective study in the US, 11.1% of cases of acute liver failure (ALF) were attributed to DILI [
6]. DILI is also an important cause of failures in pharmaceutical research and development, black box warnings added to post-marketing drugs, and drug withdrawals [
7].
Children are vulnerable to hepatotoxicity owing to the immaturity of their liver, lack of exclusive drugs, shortage of research data, and poor self-reporting [
8‐
12]. A multicenter study conducted 17 years before this study showed that DILI accounts for 19% of ALF cases in children and that it is the leading cause of pediatric ALF [
13]. It is thus essential to conduct studies on pediatric DILI to prevent potential risks for children. In one study, data from 2000 to 2006 were analyzed based on VigiBase and hepatotoxic drugs were compared between children and adults, this information seems insufficient to guide current practice as pediatric medications have been markedly changed [
8]. From the implementation of the
Pediatric Research Equity Act in 2003 to the end of 2018, the Food and Drug Administration (FDA) has accepted over 500 changes to drug labels, most of which expand the population to children [
14]. Moreover, most current evidence of pediatric DILI still focuses on case series, and studies based on a large sample size are scarce. This study aimed to characterize hepatic adverse events (HAEs) in children based on the Food and Drug Administration’s Adverse Event Reporting System (FAERS) and to detect suspected drugs associated with liver injury.
Discussion
In this study, suspected drug signals related to pediatric liver injury were identified by conducting a disproportional analysis of data from FAERS in the period 2004–2020 to provide information on safe medication for children. To the best of our knowledge, this study confirms the issues raised in the previous literature and provides new insights into pediatric liver injury associated with drugs. In 2010, Ferrajolo et al. explored drug-induced hepatic injury in children based on VigiBase from 2000 to 2006 in a case/non-case study [
8]. Both paracetamol and methotrexate were among the top 10 drugs according to the number of reports in previous and current studies. Forty-two suspected drugs identified in the former (e.g., basiliximab, caspofungin, isoniazid) were also identified in our results. Basiliximab, a novel selective anti-human interleukin-2 receptor monoclonal antibody used for immunosuppressive therapy, was the only novel signal detected in children in a previous study, which suggested future monitoring, and this was confirmed herein. Liver injury associated with basiliximab has not been included in the FDA’s prescribing information, but the incidence of abnormal liver biochemical parameters associated with basiliximab, which is greater than 5%, has been indicated by Japanese package inserts [
32,
33]. A randomized controlled trial of adults found an increase in serum alkaline phosphatase within the first month after the administration of basiliximab, but it was considered to be “functional cholestasis” [
34,
35]. No elevated liver enzyme levels or hepatotoxicity were reported in clinical trials among children treated with basiliximab [
36,
37].
Forty-three signals found in Ferrajolo’s study (chlorprothixene, methylphenidate, infliximab, etc.) were not included here. The reasons for this could be as follows. (1) Different data sources. VigiBase is the world’s largest database of individual case safety reports, having collected data on AEs from more than 130 countries since the late 1960s, whereas FAERS has collected data reported to the FDA only since 2004 [
38]. (2) Distinct algorithms have been used for signal detection. Only ROR was adopted in the former, whereas ROR and PRR were combined here. (3) Some drugs were withdrawn from the market in their early days and their AE reporting ratios were diluted. Compared with the 2010 study, 222 suspicious drug signals were newly detected in the current study, including drugs marketed before 2006, such as ganciclovir and amlodipine, and after 2006, such as vandetanib and gemtuzumab.
Anti-infectives, especially antibacterial drugs, are extensively applied, and their irrational use can cause hepatotoxicity. The immature liver function in children and increased hepatotoxicity caused by the concomitant administration of multiple antibacterial drugs make children more vulnerable to hepatic injury than adults. The results of anti-infectives for systemic use, as the most reported class in our study, were consistent with the published literature [
39,
40]. Representative drugs included trimethoprim-sulfamethoxazole (TMP-SMX), ceftriaxone, fluconazole, and isoniazid.
TMP-SMX exerts its antimicrobial effect by blocking the metabolism of folic acid through a dual pathway and can cause DILI via allergic reactions (fever, rash, etc.) [
41]. Bell et al. [
42] reported a case of DILI induced by TMP-SMX; specifically, a 9-year-old boy with a community-acquired methicillin-resistant
Staphylococcus aureus skin and soft tissue infection received TMP-SMX, and 14 days after administration, he developed fever, vomiting, and abdominal pain, with poor appetite and mentality. He was diagnosed with TMP-SMX-induced liver injury based on biochemical tests and the exclusion of other factors contributing to hepatic injury.
Ceftriaxone is a semi-synthetic, third-generation cephalosporin that mainly causes cholelithiasis or bile stasis [
43]. One retrospective study showed that 3.2% of patients treated with ceftriaxone might develop liver injury [
44]. Liver injury caused by ketoconazole in children mostly presents as elevated levels of direct bilirubin and liver enzymes [
45]. Liver functions usually recover after drug cessation, but severe cases can also develop into liver failure [
46,
47]. Animal studies have confirmed the dose-dependent liver injury induced by ketoconazole [
48].
Isoniazid is a first-line anti-tuberculosis drug. Approximately 10–20% of patients administered isoniazid develop the transient elevation of alanine aminotransferase, and fewer than 1–3% develop severe liver injury or even liver failure [
49,
50]. A greater risk of isoniazid- and rifampicin-related liver injury in children (6.9% in children and 2.7% in adults) was reported in a retrospective study [
51]. In the US, the incidence of isoniazid-associated hepatotoxicity was determined to be 1% in children treated for latent tuberculosis infection, and there was a synergistic harmful effect on the liver when isoniazid and rifampin were used in one prescription [
52,
53]. It is generally believed that isoniazid-induced liver injury is attributed to the toxicity of the metabolites acetylhydrazine and hydrazine [
54]. Moreover, mitochondrial damage caused by isoniazid-induced oxidative stress and lipid peroxidation stimulated by isoniazid and its metabolites has also been discussed [
55].
In this study, six signals were found to be disproportionally associated with pediatric liver injury for the first time, namely acetylcysteine, thiopental, temazepam, nefopam, primaquine, and pyrimethamine. Acetylcysteine (also known as N-acetylcysteine, NAC) was detected as a disproportional signal associated with pediatric hepatic injury (ROR, 3.0; 95% CI 1.9–4.8). None of these HAEs had been reported in either package inserts or the published literature since NAC was marketed. NAC is the only drug approved by the FDA for the treatment of DILI caused by acetaminophen (APAP) overdose and might prevent hepatic injury by restoring glutathione levels [
56]. A dosage of NAC for children greater than 5 kg is set based on clinical practice; however, the safety and effectiveness of NAC have not been verified in adequate and well-controlled studies. The
ACG Clinical Guideline: Diagnosis and Management of Idiosyncratic Drug-Induced Liver Injury suggests the use of NAC for children with ALF caused by severe DILI, as a significantly lower 1-year spontaneous survival rate was associated with the intravenous infusion of NAC in children with non-APAP ALF in a placebo-controlled clinical trial [
2,
57]. In our study, we found that NAC was a secondary suspected or concomitant drug in most reports and a primary suspicious drug in one report where APAP was secondary. Therefore, as a disproportional signal, NAC is used to treat or prevent HAEs, especially APAP-induced HAEs. However, whether NAC has a negative effect on children with non-APAP ALF is unclear.
We also found an association between temazepam use and pediatric liver injury (ROR, 2.4; 95% CI, 1.2–4.7). No hepatotoxic risk is indicated in the prescribing information, and there is a lack of evidence to ensure the effectiveness and safety of temazepam among children. Temazepam is a benzodiazepine, and it is uncommon to observe elevated hepatic enzymes with benzodiazepine used and to report hepatoxic cases in practice. However, it has been reported that benzodiazepines (alprazolam, chlordiazepoxide, clonazepam, diazepam, flurazepam, and triazolam) are associated with rare cholestatic liver injuries [
58,
59]. We have not retrieved any literature directly reporting temazepam-induced liver injury and postulated that this is due to its lower frequency of medication and shorter duration [
2]. In addition, there is a possibility that cross-sensitivity to temazepam might occur with other benzodiazepines [
58].
In our study, nefopam was disproportionally associated with liver injury in children (ROR, 18.7; 95% CI, 5.7–61.3). It is unclear whether nefopam is effective and safe for children as a painkiller. In this study, we identified five nefopam-related pediatric hepatotoxic AEs. Four patients reported nefopam as a secondary suspected drug, and one reported nefopam as a concomitant drug. The primary suspected drugs in these reports were ketoprofen, esomeprazole, and pregabalin, all of which have adverse effects on the liver, as specified in the package inserts. Whether it is safe to use nefopam alone in children merits attention in clinical practice.
In this study, thiopental was detected as a disproportional signal (ROR, 3.5; 95% CI, 2.0–5.9). Thiopental is an intravenous anesthetic without hepatotoxic information provided in the inserts. In our findings, most reports of HAEs in children associated with thiopental recorded it as a concomitant drug, and the suspected drugs were mainly propofol, lamotrigine, phenytoin, propranolol, valproic acid, carbamazepine, lacosamide, and clonazepam, all of which have hepatotoxicity labeled in their package inserts. The risk of hepatotoxicity associated with this drug in children when applied alone is unclear. Bedir et al. indicated that oxidative stress and inflammation develop in the liver tissue of rats injected with thiopental alone, but further clinical studies are needed to explore this effect in humans [
60].
Pyrimethamine was also detected as a disproportional signal (ROR, 3.5; 95% CI, 1.2–9.9). Four reports of pyrimethamine-associated hepatic injury in children have been identified. One patient recorded pyrimethamine as the primary suspected drug and sulfadiazine as the secondary drug, of which liver injury was a definite adverse effect. In the other three reports, the primary suspected drugs associated with liver injury were clindamycin, calcium folinate, and isotretinoin, all of which were reported to have a risk of hepatotoxicity. Pyrimethamine is typically combined with sulfonamides in practice. Many case reports of liver injury, such as granulomatous hepatitis, hypersensitivity to liver injury, and fatal hepatic necrosis, were determined to be induced by pyrimethamine-sulfadoxine [
61‐
63]. Whether pyrimethamine alone can cause liver injury requires further investigation.
We found that primaquine is associated with pediatric HAEs (ROR, 14.0; 95% CI, 4.6–42.9). Among the five liver injury records associated with primaquine, one reported primaquine as the primary suspected drug, and four reported primaquine as a secondary suspected or concomitant drug, with chloroquine, tocilizumab, and malarone as the primary suspected drugs, all of which are toxic to the liver. The safety of primaquine when applied alone in children is not clear, as current studies have only presented an elevation of biochemical indicators of the liver when primaquine is used with chloroquine [
64,
65].
This study has both limitations and strengths. Firstly, due to the differences in physiological characteristics, biochemical parameters, and disease spectrum between children and adults, and the lack of prospective data on cases in FAERS, such as viral antibody tests and re-exposure results, it is not appropriate to use tools that aid in the diagnosis of DILI to establish a robust causal relationship indicating that a drug ‘caused’ the liver injury in children. For example, the Roussel-Uclaf Causality Assessment Method is the most commonly used tool to assess DILI in adults, but it is not available as an independent assessment for children and is more appropriate for prospective data [
66]. In this study, the association between suspected drugs and HAEs was derived from a statistical perspective. Secondly, HAEs in children may be under-reported due to the wide variation in the clinical presentation of HAEs, ranging from asymptomatic elevated liver enzymes to ALF, and the weak autonomic expression in children. For example, specific hepatotoxic drugs for children may be ignored if liver injury is not manifest. Despite its limitations, this study has the following non-negligible advantages. Firstly, this study uses a robustly validated and easily accessible real-world database, combined with advanced tools and commonly recommended pharmacovigilance methods, to statistically detect the association between drugs and HAEs in children, and to supplement clinical information with prescribing information and published literature. Secondly, unlike previous studies in children with small sample sizes, such as case reports and case series, this study reports the characteristics of HAEs and suspected drugs in a specific population of children based on a large volume of data (~ 14,000 cases) spanning a long time period (2004–2020), which can provide information for clinical drug safety in children and guide future independent studies in this important area of DILI for the pediatric population.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.