Leflunomide-Induced Liver Injury: Differences in Characteristics and Outcomes in Indian and US Registries

Abstract

Background:

Leflunomide, a disease-modifying anti-rheumatic drug, has been associated with elevations of serum aminotransferases. Herein, we describe the clinical, laboratory features, and outcomes of 17 patients with leflunomide/teriflunomide hepatotoxicity from two large drug-induced liver injury (DILI) registries.

Methods:

Consecutive, adjudicated cases of leflunomide (n=16)-or teriflunomide (n=1)-related DILI from a single center in Bangalore, India and the multicenter US Drug-Induced Liver Injury Network (DILIN) were reviewed.

Results:

Nine (0.8%) of the 1070 Indian patients and 8 (0.5%) of the 1400 DILIN patients fulfilled criteria for DILI due to leflunomide- or teriflunomide. 89% of the Indian cases were women and all were associated with severe cutaneous adverse reaction (SCAR) and a median drug latency of 49 days, whereas 37.5% of the DILIN cases were female, none exhibited SCAR, and the median drug latency was 166 days. Hepatocellular injury (70%) was more common in women than men (92% vs. 20%) and was associated with younger mean age (41 vs. 59 years), higher peak INR (2.3 vs. 1.2), and higher mortality (58% vs. 0%). Mortality was observed in 6 patients from India (2 of the three with myocarditis) and 1 received liver transplantation from the USA

Conclusion:

Leflunomide-induced liver injury is predominantly hepatocellular. Leflunomide hepatotoxicity is more likely accompanied by SCAR, a short latency, and a higher mortality in the Indian cohort, with a predominance of females, compared to US DILIN patients. The differences in skin involvement, immunoallergic features, and outcomes among subjects from India vs. the US suggest that genetic or environmental factors are important in the pathogenesis of liver injury.

Lay Summary

Leflunomide an immunomodulatory drug, widely used in medicine. Although liver injury is known to be associated with leflunomide use, the characteristics of liver injury are surprisingly lacking. Our study across two large registries, the drug-induced liver injury network from USA and another one from India, illustrate the spectrum of liver injury which is often severe with characteristic differences: all patients in the Indian registry exhibited immunoallergic hypersensitivity phenotype with liver injury playing no or contributory role in mortality while none presented with immunoallergic phenotype in the US series with all surviving except one that needed a liver transplantation. The disparate presentations and outcome point to the role of genetic and environmental factors in disease dynamics.

Leflunomide, an inhibitor of pyrimidine synthesis, is an increasingly used, orally administered disease-modifying anti-rheumatic drug with of approximately 1 million prescriptions in the US per year from 2016–2019 (https://clincalc.com/DrugStats/Drugs/Leflunomide). It is used predominantly in the treatment of rheumatoid or psoriatic arthritis.¹ Although considered safe and used widely, leflunomide therapy is known to be associated with asymptomatic and reversible mild elevation of serum aminotransferase levels that usually resolve with continuation of therapy or decrease in dose.²

Since its approval in 1998, leflunomide has been reported to cause rare instances of clinically apparent acute liver injury as well as severe cutaneous adverse reactions (SCAR). This prompted both the US Food and Drug Administration (FDA) and European Medicines Evaluation Agency (EMEA) to label leflunomide with a black box warning regarding potential hepatotoxicity in 2011 along with the recommendation to monitor serum aminotransferases frequently during the first 6 months and periodically thereafter. However, a detailed characterization of liver and skin injury from leflunomide is not well described and information is limited to biochemical abnormalities without clinical jaundice. The US Drug Induced Liver Injury Network (DILIN) has been prospectively enrolling patients with DILI since 2004. Similarly, the Indian DILI registry in Bangalore has been enrolling patients since 1997. The aim of this study is to describe the clinical and laboratory features as well as clinical outcomes of prospectively enrolled patients with liver injury ascribed to leflunomide or its active metabolite teriflunomide. In addition, information regarding the clinical management, including the use of corticosteroids for patients with severe hypersensitivity features and cholestyramine were reviewed. Cholestyramine is given to accelerate the elimination of leflunomide and its metabolites due to enterohepatic recirculation, which can maintain elevated blood levels for up to 8 months.

Methods:

Consecutive cases of suspected leflunomide - or teriflunomide-related DILI in 2 large registries, including a single center registry from Bangalore, India³ and the multicenter Drug-Induced Liver Injury Network (DILIN) from the US were reviewed.⁴ The details of the registries and definition of DILI have been described before.^{3, 4} For uniformity of reporting, the DILIN study criteria were applied to patients in the Bangalore, India cohort, and included one or more of the following: an ALT > 5 × ULN, alkaline phosphatase > 2 × ULN, total bilirubin > 2.5 mg/dl, and INR > 1.5.⁴ DILI severity scales (mild, moderate, severe and fatal/liver transplantation) were based on the DILIN scoring system,⁴ and was similarly applied to the Bangalore, India, cohort. Written informed consent was obtained from all participants per local IRB regulations.

Causality assessment was carried out by the expert consensus of the DILIN Causality Committee.⁵ Clinical narratives and laboratory, imaging and biopsy data on every patient from the Bangalore, India, cohort was reviewed by three DILIN investigators for expert consensus adjudication of DILI. All DILIN patients enrolled in the prospective study were followed for a minimum of 6 months while subjects from India were followed until LFT normalization or death. Competing causes such as viral hepatitis A, B, C, D, E, CMV, EBV, and autoimmune hepatitis were ruled out by serological and virological tests. Pancreaticobiliary diseases, were excluded by ultrasonography, contrast enhanced computerized tomography, or magnetic resonance imaging (MRI)/magnetic resonance cholangiopancreatography (MRCP). The pattern of liver injury was classified based on the R value.

Drug-induced skin injury and drug reaction with eosinophilia and systemic symptoms (DRESS) and Stevens-Johnson syndrome/ toxic epidermal necrolysis (SJS/TEN) were defined according to expert working group criteria and International Registry of Severe Cutaneous Adverse Reactions (RegiSCAR) criteria.^{6, 7 8}

The study was approved by the institutional ethical review board of St. John’s Medical College Hospital and by institution review boards (IRB) of the individual centers of DILIN.

Descriptive statistics are presented as mean with standard deviations or medians with interquartile range for continuous variables and frequency with percentages for categorical variables. Non-parametric Wilcoxon test was used for continuous variable, while Fisher’s exact test was used for categorical variables to compare two groups of Bangalore and DILIN cases. Kaplan-Meier curves were used to estimate the media time to lab normalization where log rank test was used for comparison between the two groups.

Results:

We identified 16 patients with definite (n=8), highly likely (n=5), probable (n=3), leflunomide-related DILI and 1 with definite teriflunomide-related DILI. Nine (0.8%) cases among 1070 patients with DILI were identified from the Bangalore, India, registry, and 8 cases (0.5%) were among 1400 patients from DILIN. Selected clinical and biochemical variables are shown in Table 1. All patients from Bangalore were Asians, 7 of 8 patients from USA were Caucasian and one subject was African-American. The predominant underlying conditions for which leflunomide was used were seronegative arthritis (in 6 patients in the Bangalore cohort, 2 in the DILIN cohort) and rheumatoid arthritis (in 2 patients in the Bangalore cohort and 6 in the DILIN cohort), and multiple sclerosis in 1 patient in the Bangalore cohort. The median latency, the time from starting leflunomide to DILI onset, was significantly shorter in the Indian patients (49 vs 166 days; p=0.016).

Table 1:

Selected clinical characteristics and outcomes of individuals with DILI due to leflunomide

	Bangalore cases (N=9)	DILIN cases (N=8)	P-value
Age (years)			0.001
Mean (SD)	34 (14)	63 (12)
Median (Q1, Q3)	35 (22, 40)	61 (54, 70)
Female (%)	8/9 (89%)	3/8 (37%)	0.050
Self-reported race (%)			<0.001
- Caucasian	0/9	7/8 (87%)
- African American	0/9	1/8 (12%)
- Asian	9/9 (100%)	0/8
BMI (kg/m2)			0.089
Mean (SD)	20 (2)	27 (8)
Median (Q1, Q3)	20 (18, 22)	24 (21, 34)
Prior drug allergies (%)	0	6 (75%)	0.002
Alcohol use (%)	0	3 (37.5%)	0.082
Latency (days)			0.016
Mean (SD)	51 (19)	180 (126)
Median (Q1, Q3)	49 (34, 64)	166 (65,270)
Fever (%)	9/9 (100%)	1(12%)	<0.001
Rash (%)	9/9 (100%)	1(12%)	<0.001
Eosinophilia>500/uL (%)	3/7 (43%)	1/7 (14%)	0.559
DRESS (%)	6 (67%)	0	0.009
SJS or TEN (%)	3 (33%)	0	0.21
Liver tests at presentation
ALT (U/L)
Mean (SD)	713 (450)	770 (834)	0.501
Median (Q1, Q3)	517 (397, 1046)	380 (176, 1228)
Alk P (U/L)
Mean (SD)	292 (135)	289 (195)	0.641
Median (Q1, Q3)	270 (189, 336)	212 (195, 252)
Total bilirubin (mg/dL)
Mean (SD)	9 (10)	6 (3)	0.463
Median (Q1, Q3)	5 (3, 7)	6 (6, 8)
INR
Mean (SD)	1.5 (0.5)	1.1 (0.1)	0.029
Median (Q1, Q3)	1.3 (1.1, 1.9)	1.1 (1.0, 1.2)
Pattern of liver injury (%)
Cholestatic	0/9	1/8 (12.5%)	0.164
Mixed	1/9 (11%)	3/8 (37.5%)
Hepatocellular	8/9 (89%)	4/8 (50.0%)
Peak laboratory values, Median (Q1, Q3)
ALT (U/L)	1046 (517, 1432)	696 (303, 1517)	0.386
Alk P (U/L)	453 (326, 522)	273 (229, 505)	0.29
Total bilirubin (mg/dL)	12.6 (3, 28)	10.7 (6.3, 23.9)	0.847
INR	2.4 (1.5, 3.1)	1.2 (1.1, 1.3)	0.02
Days from onset to peak values, Median (Q1,Q3)
ALT	0 (0, 0)	7.5 (0.5, 17.0)	0.048
Alk P	11 (3.0, 16.0)	33.0 (4.0, 34.5)	0.247
Total bilirubin	9.0 (0.0, 17.0)	18.5 (9.5, 30.0)	0.188
Days from peak to normalization, Median (Q1, Q3)
ALT	82	76.5	0.489
Alk P	19	53.5	0.334
Total bilirubin	100	40	0.897
Severity score (%)
Mild	0/9	1/8 (12.5%)	0.075
Moderate¶	1/9 (11%)	5/8 (62.5%
Severe	2/9 (22%)	1/8 (12.5%)
Fatalδ	6/9 (67%)	1/8 (12.5%)
Hospitalized (%)	9/9 (100%)	4/8 (50%)	0.029
Causality (%)
Definite/	9/9 (100%)	1/8 (12.5%)	<0.001
Highly likely/	0/9	5/8 (62.5%)
Probable	0/9	2/8 (25%)
Steroid therapy (%)	9/9 (100%)	1/8 (12.5%)	<0.001
All deaths (%)	6/9 (67%)	0/8	0.009
Liver-related deaths (n)	0	0	NC
Liver transplant (n)	0	1 (12.5%)	0.471
Chronic DILI (n)	0	3/6	0.044

^¶Moderate & Moderate-hospitalized combined

^δFatal includes liver transplantation.

Abbreviations: Alk P: Alkaline phosphatase; ALT: Alanine aminotransferase; BMI: Body mass index; CS: Cholestatic; DILI: Drug induced liver injury; DRESS: Drug reaction with eosinophilia and systemic symptoms; HC: Hepatocellular; INR: International normalized ratio; SD: Standard deviation; SJS: Stevens-Johnson syndrome; TEN: Toxic epidermal necrolysis

Notably, all 9 cases from India were associated with SCAR, 67% with DRESS syndrome (among whom 3 had eosinophilia), and 33% with SJS alone and 2 with TEN. All patients with DRESS syndrome exhibited >50% skin involvement with features varying from morbilliform rashes to exfoliative dermatitis and erythema multiforme. Of the 3 patients with SJS, two had extensive oral mucosal ulcers and while one had corneal ulcers that was treated with amniotic membrane graft. In contrast, none of the patients in the US DILIN series [n= 8] exhibited either SCAR or eosinophilia. All but two patients exhibited clinically significant jaundice. The predominant pattern of injury was hepatocellular.

Hepatocellular injury, which was more common than mixed or cholestatic particularly in women (92% vs. 20%), was associated with younger mean age (41 vs. 59 years), higher INR (2.3 vs. 1.2) and higher mortality (58% vs. 0%), respectively.

While 3 of the 9 patients who survived in the Indian series all normalized serum aminotransferases at 3–4 months of follow-up, a DILIN patient continued to have serum aminotransferase elevations without symptoms 6 months after onset, suggesting chronic injury. However, elevated liver enzymes normalized within 12 months and a liver biopsy demonstrated expansion of portal tracts due to edema. Information on autoantibodies across both registries are stated in Supplement Table 1.

While no patient in the Indian series underwent liver biopsy, findings of liver biopsy was available in 5 patients in the DILIN series (See Supplement Table 1). These were performed in the course of clinical care by providers and not per DILIN study protocol. The overall features ranged from mild portal expansion, chronic hepatitis, cholestatic hepatitis and in one case hepatocellular necrosis.

Non-hepatic Complications

Three Indian female patients developed myocarditis manifesting as breathlessness, shock, and rhythm disturbances. One patient with carditis presented with flash pulmonary edema (pro-BNP 35,000 ng/L) needing transfer to an intensive care unit and escalation of steroids. The second patient developed myocarditis leading to scar tissue on cardiac MRI and pericardial effusion which progressed rapidly despite escalation of intravenous steroids. Coronary angiogram was normal. The third patient was a 14 year old girl who developed rapid onset diabetes mellitus with severe ketoacidosis needing intensive care that predated the onset of carditis by 2 months. This girl was on tapering doses of oral prednisolone when these complications occurred. Another patient who discontinued steroids 10 days after resolution of DRESS developed bone marrow suppression during the recovery phase of DILI and then presented with febrile neutropenia and pneumatosis intestinalis. She died from these complications. Thyroiditis with hypothyroidism occurred during follow-up in another patient 2 months after onset of DILI.

Clinical Management

After suspicion of leflunomide or teriflunomide as the implicated agent causing DILI, these drugs were stopped immediately. For patients with SCAR, treatment measures included administration of intravenous fluids, local emollients (creams applied on the skin to reduce pain or make it softer), topical antimicrobials, and/or betamethasone as appropriate. Furthermore, patients with SCAR received treatment with systemic corticosteroids, either 60 mg prednisolone or its equivalent dose of dexamethasone, and tapered over 1 to 2 months at the discretion of the treating clinician. When recurrence of skin lesions was encountered two patients, oral steroids were escalated, while in one patient with resolved liver tests and reappearance of skin rashes, oral cyclosporine was initiated at a dose of 2 mg/kg and continued for 2 months until complete resolution of skin rashes had occurred.

All patients in the Indian series (all with skin rashes) and three in the US series underwent “cholestyramine washout”, i.e. 4 grams of cholestyramine three times a day orally for 11 days.⁹

Outcome

Six of 9 patients from India died in hospital compared to one of 8 from the US who underwent liver transplantation but died thereafter. Liver injury was the primary cause of death in the one DILIN patient, whereas the relationship to death was variable in the Indian cohort (See Table 2). In particular, two patients from India died of myocarditis, one of whom developed acute fulminant diabetes mellitus, and another patient died from pancytopenia leading to sepsis and shock. Of interest, despite the fatal outcomes, both patients with myocarditis had almost complete resolution in liver biochemical tests and skin reactions prior to deaths that were ascribed to causes other than liver disease. The woman who had been treated with teriflunomide developed gradually increasing breathlessness and giddiness and died five and one-half months after diagnosis of DILI. Cardiac ECHO and MRI showed global dyskinesia. In the other patient, carditis was preceded by development of fulminant diabetic ketoacidosis (which was initially ascribed to treatment with steroids given for management of DRESS and DILI). She died four and a half months after diagnosis of DRESS. Three patients from Bangalore died due to SJS/TEN and related complications (sepsis in all 3, along with associated intracerebral bleed in one). All 3 with SJS and DILI had persisting liver abnormalities at the time of death. One other patient developed bone marrow suppression with pancytopenia, which was complicated by infection, sepsis and pneumatosis intestinalis. The lone patient from the DILIN series underwent liver transplantation after developing acute liver failure 9 months after starting leflunomide. The 3 Indian patients are well and free of liver and skin injury over a follow up period of 1, 4, and 14 years respectively.

Table 2:

Summary of clinical and laboratory characteristics of the six patients who died and the one patient that underwent liver transplantation

Age/sex	Indication for leflunomide	Nationality	Extrahepatic manifestation	Time to death from initial presentation	Clinical complication	Bilirubin at time of death (mg/dl)	Peak ALT	Peak INR	Role of DILI in fatality
21/F	Seronegative arthritis	Indian	SJS/TEN	6 weeks	SJS Intracerebral bleed	29	1046	2.5	Contributory role
22/F	Seronegative arthritis	Indian	DRESS	7 weeks	Bone marrow suppression septicemia	17	1900	6.3	Contributory role
14/F	Seronegative arthritis	Indian	DRESS	18 weeks	Carditis, shock DM	0.6	435	3	No role
46/F	Multiple sclerosis	Indian	DRESS	23 weeks	Carditis Cardiogenic shock	2	2531	1.6	No role
33/F	Seronegative arthritis	Indian	SJS	9 weeks	SJS Sepsis	9.3	1151	2.2	Contributory role
40/F	Rheumatoid arthritis	Indian	SJS	8 weeks	SJS SIRS	10.7	1008	1.1	Insufficient data
*58/F	Rheumatoid arthritis	African American	None	4 weeks	Progressive jaundice/coagulopathy	39	1714	2.8	Primary

^*Underwent liver transplantation

Abbreviations: DILI: Drug induced liver injury; DM: Diabetes mellitus; DRESS: Drug reaction with eosinophilia and systemic symptoms; SIRS: Systemic inflammatory response syndrome; SJS: Stevens-Johnson syndrome; TEN: Toxic epidermal necrolysis

Discussion

Fortunately, despite its wide use clinically, severe leflunomide -induced liver injury is uncommon.¹⁰ Although mild elevations of serum aminotransferases have been reported in 2–13% of patients treated with leflunomide, serious drug-induced hepatotoxicity leading to hospitalization is rare (0.02%). However, anecdotal reports of fatalities from liver failure have been described.¹⁰

In our series striking differences in the phenotypic features at presentation were noted in leflunomide-induced DILI: in India, all 9 cases of hepatotoxicity were hospitalized and associated with SCAR; 6 were associated with DRESS and 3 with SJS/TEN. In contrast in the patients from the US, DRESS or SJS/TEN did not occur, despite a comparable severity of liver injury in both the groups. It appears from published reports that DRESS or hypersensitivity reactions are reported more commonly in Indians from the sub-continent than European-Americans.^11–14 In the US DILIN experience, risks of SJS/TEN are higher among African-Americans than among other ethnic and racial groups.¹⁵ The phenotypic differences suggest that genetic, pharmacokinetic, and environmental factors are important factors in the pathophysiology of lefluonomide-induced DILI.¹⁰ The influence of genetic factors is illustrated by reduced metabolism of leflunomide in the presence of the CYP2C9*3 homozygous state resulting in increased toxic products.¹⁴ While this is only a speculative comment, CYP2C9*3 is reported to have a high frequency (~13%) in South Asian populations compared to other populations (0.4–1%),¹⁶ although there is wide inter- and intra-ethnic variability.¹⁷ In the Indian series, regrettably, we did not have DNA saved from a majority of affected patients, and we were unable to perform genetic testing. However, this will be important to perform in future cases whenever possible.

Reports of acute liver failure were reported to the US FDA and European Medicine Evaluation Agency (EMEA), leading to black-box warnings in the package inserts of leflunomide and terflunomide.¹⁸ Nineteen of 49 patients with severe liver injury reviewed by FDA either died or received liver transplantation.¹⁸ Among 129 cases of serious hepatic injury reported to EMEA, 15 cases developed liver failure and 9 died.¹⁹ The limited information available from FDA and EMEA precludes detailed characterization of liver injury and, as in some of our cases, other factors if any were responsible for the severe injury and fatality.

Reports suggest that elimination of leflunomide and its metabolite can be dramatically increased by using oral charcoal or cholestyramine.²⁰ However, despite cessation of leflunomide and treatment with cholestyramine in all cases of DILI in the Indian series, death occurred in 6 of 9 patients. Deaths in the Indian series were not primarily related to liver injury although it did contribute to fatality in 50% of deaths. In the DILIN series, leflunomide did lead to one case of acute liver failure needing liver transplantation. While in the Indian cohort, bone marrow failure leading to neutropenia and secondary infection was a cause of death in one patient, at least 2 deaths occurred 2–3 months after both liver and skin injury had resolved. The two patients died of myocarditis, when both liver and skin lesions had abated. One of these also developed “fulminant” diabetes mellitus with a previous admission of ketosis. We think is likely that myocarditis and pancreatic islet cell insufficiency are due an immune-mediated disease, triggered by prior exposure to leflunomide/teriflunomide. Myocarditis following drug-induced hypersensitivity reaction is rare.²¹Although a host of drugs including immune checkpoint inhibitors have been linked to hypersensitivity myocarditis, to the best of our knowledge, leflunomide has not been implicated in hypersensitivity myocarditis.²² Steroid administration soon after the diagnosis of myocarditis has been associated with better outcome,²¹ but not in the acute necrotizing variety.²³ In one patient, accelerated diabetes mellitus with ketosis preceded the onset of carditis by 2 months. Accelerated development of diabetes is a rare occurrence in patients with DRESS syndrome.²⁴ Finally, deaths in 3 other patients in the Indian series were associated with SJS/TEN highlighting the poor outcome in patients with concurrent severe DILI (with clinical jaundice) and SJS/TEN.²⁵ None of our patients had hepatitis B virus infection. Published reports suggests reactivation of quiescent hepatitis B although rare could occur when treated with leflunomide or when supplemented by biologicals.^{26, 27}

Teriflunomide was approved in 2012 and appears to be associated with lower frequency of liver injury than leflunomide. In one report, teriflunomide-induced DILI was associated with complete recovery after early identification and cholestyramine ‘washout’,²⁸ unlike the lone patient in the Indian series who died due to myocarditis well after recovery from liver and skin injury.

In conclusion, leflunomide-induced liver injury is predominantly hepatocellular, which is the more common injury pattern in women. In India, it is predominantly observed in females, is frequently accompanied by severe skin injury and has a high mortality. The marked differences in phenotypic presentations and outcomes among subjects from India vs. the US suggest the role of as yet unknown genetic or environmental factors in the pathogenesis of leflunomide-induced liver injury.

Supplementary Material

supinfo

Supplementary Table 1: Clinical, demographic, laboratory data

Abbreviations

ALT

alanine aminotransferase

AST

aspartate aminotransferase

DILI

drug-induced liver injury

DILIN

Drug-Induced Liver Injury Network

DM

diabetes mellitus

DRESS

drug-reaction with eosinophilia and systemic symptoms

INR

international normalized ratio

RegiSCAR

International Registry of Severe Cutaneous Adverse Reactions

SCAR

severe cutaneous adverse reaction

SJS/TEN

Stevens-Johnson syndrome-Toxic Epidermal Necrolysis

SIRS

systemic inflammatory response syndrome

Data Availability Statement:

The data that supports the findings of this study are available in the supplementary material of this article