Dr G. Bastida, Gastroenterology Unit, La Fe Hospital, Campanar 21, 46009 Valencia, Spain. E-mail: firstname.lastname@example.org
Background : The incidence of thiopurine-induced hepatotoxicity in patients with inflammatory bowel disease varies in different studies.
Aims : To assess the rate of thiopurine-induced liver toxicity in patients with inflammatory bowel disease; to determine the predictive factors and to characterize its clinical course and management.
Methods : A cohort of 161 patients was prospectively followed for a median of 271 days. Hepatotoxicity was established when alanine transaminase or alkaline phosphatase plasma levels were greater than twice the upper normal limit.
Results : Abnormal liver function was detected in 21 patients (13%; 95% CI: 7–18). Hepatotoxicity occurred in 16 patients (10%; 95% CI: 6–16) after a median of 85 days. In five cases, treatment was withdrawn due to hepatotoxicity. Use of corticosteroids was associated with hepatotoxicity (OR: 4.94; 95% CI: 1.01–23.98) with antitumour necrosis factor concomitant therapy showing a protective role (OR: 0.3; 95% CI: 0.1–3.1). γ-Glutamyl transferase plasma levels at the onset of hepatotoxicity showed the best predictive value for treatment withdrawal (area under the receiver operating characteristic curve: 0.95).
Conclusions : The incidence of hepatotoxicity in inflammatory bowel disease patients receiving thiopurines is relevant, mainly in patients co-treated with corticosteroids. γ-Glutamyl transferase plasma level is a useful biomarker in therapy withdrawal prediction.
Nowadays, thiopurines, azathioprine (AZA) and mercaptopurine (6-mercaptopurine; MP), are the most commonly used immunomodulatory drugs in patients with inflammatory bowel disease (IBD).1 The occurrence of side-effects, however, is a major drawback in the use of AZA or MP and concerns regarding both short and long-term toxicity have restricted their use.2
The side-effects of thiopurines can be divided into dose-independent or ‘allergic/idiosincrasic’ and pharmacologically explainable dose-dependent events. The most frequently observed dose-independent events are rash, fever and artralgia, pancreatitis and hepatitis.3, 4 The most important dose-related side-effect is myelosuppression. Hepatotoxicity may also be dose-dependent and can be reversible upon reduction of dose or discontinuation of treatment. Both acute hepatocelular hepatitis and colestatic hepatitis have been described during thiopurines treatment.5 The incidence of hepatotoxicity was initially described as low; Present et al.3 with 1800 patients-year of follow-up reported an incidence of drug hepatitis of 0.3%, despite this data, incidence varies among different series6–8 and its real impact and management in clinical practice has not been fully characterized.
Recently, a variation on incidence of hepatotoxicity among different diseases has been reported,9 probably reflecting the influence of different genetic and acquired factors on individual susceptibility; the role of these acquired factors, as concomitant medications have been poorly documented.
The aims of this study were first to assess the rate of hepatotoxicity in IBD patients under thiopurine treatment; secondly, to determine the potential predictive factors for the occurrence of hepatotoxicity and thirdly, to characterize its clinical course and management.
Patients and methods
All patients starting treatment with AZA or MP since May 2000 were prospectively included. Diagnoses of IBD was established according to the criteria of Lennard-Jones.10 In patients with Crohn's disease (CD), lesion location (ileum, colon, ileum and colon or upper gastrointestinal tract) and behaviour of disease (stricturing, penetrating or non-stricturing and non-penetrating) were classified in accordance with the Vienna criteria.11 Patients with ulcerative colitis (UC) were classified in accordance with the extent of the disease. Demographic information, gender, age at diagnosis, concomitant medication, extraintestinal manifestations and indication of immunosuppressive therapy were obtained before starting on treatment. Thiopurine methyl transferase (TPMT) activity was recorded since May 2002.
The AZA was introduced at 50 mg daily and increased gradually to doses of 2.5 mg/kg daily.12 Those patients who had gastrointestinal intolerance to AZA were treated with MP initially at 0.5 mg/kg daily, increasing to 1.5 mg/kg daily.
Subjects were seen in out-patient clinic at 2, 4 and 8 weeks and then every 3 months. Complete blood count and differential, alanine transaminase (ALT), aspartate transaminase (AST), alkaline phosphatase (AP), γ-glutamyl transferase (GGT), bilirubin, sedimentation rate, C-reactive protein, fibrinogen, albumin and iron were monitored at every visit. Amylase and lipase were not determined routinely. In each visit the patient was interrogated about any drug side-effect using a specific questionnaire. If treatment was discontinued, reasons were recorded.
Abnormal liver function tests (LFTs) were established when ALT or AP levels were >50% the upper normal limit. Hepatotoxicity was defined by ALT or AP levels greater than twice the upper normal limit in patients with previous normal values.13 Lower elevations of LFTs were considered as ‘mild’ elevations. To assess the causality we used the Council for International Organizations of Medical Sciences (CIOMS) criteria.14 When abnormal LFTs was detected the patient was specifically interrogated about drugs or alcohol intake. All patients underwent an abdominal ultrasound, serological test to rule out hepatitis B and C infection and laboratory test including autoantibodies (antinuclear autoantibodies, antismooth muscle autoantibodies, antiliver/kidney autoantibodies and antimitochondrial autoantibodies), ceruloplasmin, iron metabolism and α-1-antitrypsin. Hepatotoxicity was classified in three groups using biochemical criteria based on serum activity of ALT and AP and their ratio:14 (i) acute hepatocelular hepatitis defined by ALT above twice the upper normal limit or by a ratio ALT/AP >5; (ii) acute cholestatic hepatitis defined by an isolated increase of AP or by a ratio ALT/AP <2 and (iii) mixed pattern defined by an ALT/AP ratio between 2 and 5. When abnormal LFTs were detected the dose of thiopurines was reduced until LFTs normalization, and then the dosage of immunosuppressant was increased to the higher tolerated dose. If LFTs did not return to normal values with tapering of thiopurines, therapy was withdrawn.
Results are expressed as median and range. Data were tested for normality distribution, the natural logarithms of variables were used to reduce the heterogeneity of variance if necessary. Categorical data are expressed as counts (%). Chi-square with continuity correction or Fisher's exact tests were used to assess differences between categorical variables. Independent samples Student's t-test was used to assess differences between continuous variables. A two-sided P-value of <0.05 was considered statistically significant. To identify independent predictors during out-patient follow-up, variables that achieved a P-value of <0.1 and those we considered clinically relevant were included in a multivariate analysis by using a logistic regression procedure analysis. Receiver operating characteristic curves (ROC) were generated by plotting sensitivity (y-axis) vs. 1 − specificity (x-axis) or true-positive rates vs. false-positive rates.
During the period of study 172 patients started on immunosuppressive treatment; 11 refused the participation in the study, thus a cohort of 161 patients were included and followed.
Patient characteristics, indications of treatment and concomitant medications are listed in Table 1, and baseline laboratory findings, TPMT activity and dose of thiopurines are listed in Table 2.
Table 1. Comparison of patient characteristics, indications of treatment and concomitant medications between patients with and without liver toxicity
No hepatotoxicity, n (%; n = 145)
Hepatotoxicity, n (%; n = 16)
Age (years; range)
Ileum and colon
Upper gastrointestinal tract
Disease behaviour in patients with CD
Indication for treatment
Induction of remission
Prevention of recurrence
Maintenance after severe attack
Table 2. Comparison of baseline laboratory findings, thiopurine methyl transferase (TPMT) activity and dose of thiopurines between patients with and without liver toxicity
No hepatotoxicity (n = 145)
Hepatotoxicity (n = 16)
Leucocyte count (×103/μL)
Neutrophil count (× 103/μL)
Platelet count (×103/μL)
Sedimentation rate (mm/h)
C-reactive protein (mg/dL)
Aspartate aminotransferase (IU/L)
Alanine aminotransferase (IU/L)
Alkaline phosphatase (IU/L)
γ-Glutamyl transferase (IU/L)
TPMT in (IU/mL) red blood cells
Dose of AZA (mg/kg)
Globally occurred 95 side-effects in 70 (44%) patients: 16 patients had two side-effects, three patients had three side-effects and one patient had four side-effects. The first side-effect occurred after a median time of 41 days (range: 1–951).
Treatment was withdrawn in 46 (29%) patients after a median time of 58 days (range: 2–1134). The major reason for stopping the medication was the appearance of side-effects (32 patients, 20%). Eleven patients switch from AZA to MP because of gastrointestinal upset, symptoms recurred in five patients and thiopurine treatment was definitely removed. Other reasons for discontinuation of medication were that surgery become necessary in five (3%), the treatment was considered to be ineffective in four cases (2.5%), the patients requested stopping the medication because the potential side-effects in four (2.5%) or the patient wished to become pregnant while off the therapy in one (0.6%) patient.
Abnormal LFTs and hepatotoxicity
Elevation of LFTs was detected in 21 (13%; 95% CI: 7–18) patients. Five (3%) patients (95% CI: 1–7) had a ‘mild’ elevation of LFTs, none of them stopped the medication.
Hepatotoxicity occurred in 16 (10%; 95% CI: 6–16) patients. Patient demographics, baseline values of LFTs, TPMT activity and indications of treatment of all the patients with liver injury are listed in Table 3. Hepatotoxicity was detected after a median of 85 days (range: 2–951; Figure 1), an acute hepatocelular hepatitis in 14 cases and an acute cholestatic hepatitis in two. All patients were treated with AZA when hepatotoxicity occurred. In all cases abdominal ultrasound, antinuclear autoantibodies, antismooth muscle autoantibodies, antiliver/kidney autoantibodies, antimitochondrial autoantibodies, ceruloplasmin, iron metabolism, α-1-antitrypsin and serological markers of the hepatitis B virus (HBV) were normal. One patient had antihepatitis C virus (HCV) antibodies and was not considered to have hepatotoxicity. This patient was tested for viral RNA and underwent a liver biopsy. The patient had a HCV viral load of 7 692 310 copies/mL with HCV genotype 1b and the liver biopsy revealed a histological activity index (HAI) of seven of 14 with and a fibrosis stage of two of four. In this patient the LFTs returned to normal values, continuing on full dose of AZA.
Table 3. Patient characteristics, baseline values of liver function test and concomitant medication of patients with liver injury
Univariate analyses revealed that concomitant treatment with systemic steroids, haemoglobin level, leucocyte and neutrophil count at the onset of treatment were significant predictive factors for the appearance of liver toxicity.
Multivariate analyses revealed that only the use of corticosteroids at the onset of treatment was found to be a significant positive predictive factor for the appearance of hepatoxicity (OR: 4.94; 95% CI: 1.01–23.98). Concomitant anti-tumour necrosis factor (TNF) therapy showed a protective effect (OR: 0.3; 95% CI: 0.1–3.1) although with no statistical significance.
Clinical course of liver injury, therapy withdrawal and predictive factors for treatment discontinuation
Full details of the 16 patients with liver injury are listed in Table 4. Treatment was abandoned due to hepatotoxicity in five patients after a median of 51 days (range: 2–919): four cases with hepatocelular hepatitis and one with cholestatic hepatitis. In all patients abnormal LFTs normalized after treatment was discontinued. In two of these five patients in whom the treatment was withdrawn, rechallenge was attempted with lower doses of AZA with recurrence of elevated LFTs within 2 weeks. In the remaining 11 patients, seven were able to tolerate full doses of thiopurines without further elevations of LFTs and in four patients the treatment should be withdrawn because of other causes different from liver toxicity during the follow-up before they reached the full doses of thiopurines.
Table 4. Time, dose of azathioprine, laboratory findings at the onset and outcome of liver injury
Among the patients with liver injury, those patients in whom the treatment was withdrawn due to hepatotoxicity showed higher levels of AST, ALT, AP, GGT and total bilirubin at the onset of hepatotoxicity, although with non-statistical significance. Plasma levels of GGT at the onset of liver toxicity, had, in the multivariate analysis, the best predictive value for treatment withdrawal with an area under the ROC curve of 0.95 (95% CI: 0.85–1.0). The best positive and negative predictive values were achieved in this model using a cut-off level of four times the upper normal limit: sensitivity of 1.0, specificity of 0.87, predictive value of a positive test of 0.57, predictive value of a negative test of 1.0, likelihood ratio of a positive test of 0.76 and likelihood ratio of a negative test of 0.
To our knowledge, this is the first prospective epidemiological study that specifically addresses the incidence of hepatotoxicity in IBD patients treated with thiopurinic immunomodulators. The incidence of abnormal LFTs or liver toxicity is a relevant finding during the follow-up of patients treated with thiopurinic immunomodulators as we showed in this study conducted with a cohort of 161 IBD patients. There is a lack of recent published series that specifically assess thiopurine-induced hepatotoxocity, but some recent studies with IBD patients do not describe any case of liver injury during the follow-up.15, 16 This is important because systematic biochemical analysis is not performed routinely in recent studies that analyse the side-effects of thiopurinic immunomodulators.17 Thus, the first clinical application that derives directly from our study is that LFTs monitoring is mandatory in these patients and should be performed for the duration of therapy as it has been proposed previously.18
Probably the main limitation of studies evaluating drug-induced liver toxicity is related with diagnosis, due to the absence of specific markers or tests. Therefore, the diagnosis relied entirely on circumstantial evidence and only in the cases of relapse after rechallenge we had the certainty that AZA or MP were the offending drug.19 Liver biopsy is an invasive procedure with significant morbidity and was not performed routinely to all patient presented with abnormal LFTs. Although liver biopsy is not required to establish the diagnosis of drug-induced liver toxicity,19 it is impossible to know in our study whether there was a correlation between histological damage and elevated enzymes. Based on the absence of histological confirmation and in the fact that an important percentage of patients were able to tolerate full-dose therapy, we cannot assert that we are handling with a true hepatotoxicity which implies hepatocyte damage rather than a form of tolerance. It is important to remark that we ruled out other causes that might have explained the liver injury as well as alcohol or hepatotoxic drugs intake, but we should be aware that the patient could be hiding the consumption of illegal drugs or herbal remedies.20
It is worth noting some considerations related to the clinical course of thiopurine-induced liver injury. At first it is important to point out that a small percentage of patients, <5%, presented with a slight elevation of LFTs that did not have clinical implications: the abnormalities in liver chemical tests returned to normal values during the follow-up and was not necessary to adjust dose of immunomodulator. As with other drugs,21 thiopurine-induced liver injury occurred more frequently within the first months of treatment, 50% of cases within the first 3 months. Moreover, treatment withdrawal because of hepatotoxicity occurred in most of cases, 75%, during this period of time. Despite this, in some cases the liver injury was only detected after a long period of follow-up leading to therapy withdrawal. This finding is surprising because the long delay makes the role of the suspected drug unlikely; the explanation of this event could be related to a cumulative effect of the metabolites on liver,22 or to the confluence of multiple factors that could be triggers of an autoimmune liver injury. We found acute hepatocelular hepatitis in 87.5% of patients, in contrast with previous descriptions that considered pure cholestasis as the typical pattern.23 In all cases LFTs returned to normal values and no chronic disease was detected. When hepatotoxicity occurred, the treatment was withdrawn in 31% of patients, but an important percentage, 44%, was able to continue on full dose of thiopurine once the dose was temporarily adjusted. This group of patients had a dose-dependent hepatotoxicity rather than an immunoallergic hepatitis. The rationale about how these patients were able to return to full doses of thiopurinic immunomodulators may be theoretically explained by the confluence of multiple factors in the onset of hepatotoxicity: dose of immunomodulator, concomitant treatment, quality of nutrition, drug interaction, etc.
If we consider the high percentage of patients able to continue with proper doses of AZA or MP after temporal adjustment of the dose, it would be of particular usefulness to predict those patients prone to tolerate immunomodulators. We noted that the values of LFTs when toxicity is detected (first determination) were higher in the patients in whom the treatment was finally withdrawn, and the plasma levels of GGT had the best sensibility and specificity to discriminate those patients in whom the treatment was removed. This is an additional important observation and could be very useful in clinical practice because many patients could continue on immunomodulator treatment and thus, maximize the clinical effectiveness of thiopurines, cornerstone of maintenance treatment in IBD. One important limitation of this study is the number of patients – only 16 – used to calculate the sensitivity and specificity, so we should be cautious when analysing the conclusions; this approach should be tested prospectively in a larger series to corroborate our findings.
Previous reports have shown an increase risk of hepatotoxicity in men,23 in the present study men had higher risk of liver injury as well, although this difference did not reach statistical significance. Now it is well recognized the role of TNF as a key mediator in hepatototoxicity24 and could explain the protective effect observed in our study in those patients with anti-TNF (infliximab) concomitant therapy. One interesting finding of our study is that the use of steroids was associated with an increased risk to develop liver injury. Animal studies suggest that AZA effects correlate to oxidative stress, mitochondrial dysfunction and activation of stress-activated protein kinase pathways leading to necrotic cell death.25–27 Thus, steroids may precipitate hepatotoxicity by exacerbating insulin resistance, recognized as a proinflammatory condition with increased oxidative stress and activation of cellular signalling pathways,28 or by exacerbating risk factors for fatty liver as central obesity, diabetes or hypertriglyceridaemia, which predisposes to oxidative stress by amplifying the capacity for free radical chain reactions.29In vitro studies25, 27 support the role of intracellular reduced glutathione (GSH) depletion in AZA-induced hepatocyte injury and the protective role of N-acetyl-l-cysteine (NAC), however, these findings must be completed.27 Based on these considerations, the use of NAC or other antioxidants could represent a novel strategy for: hepatotoxicity prevention, mainly in those receiving AZA and steroids concomitantly.
In conclusion, this study provides an approach to the real incidence of thiopurine-related hepatotoxicty in IBD patients in clinical practice setting. We have identified the corticosteroids usage at the onset of treatment as a positive predictive factor to hepatotoxicity and, we have proposed the use of GGT plasma levels to monitor liver toxicity clinical course, in order to optimize effectiveness of thiopurinic immunomodulators.
This study was supported in part by a grant from the ‘Instituto de Salud Carlos III’ (C03/03). Guillermo Bastida MD was supported by a grant from the ‘Fundacion para la investigacion Hospital Universitario La Fe’. The authors would like to thank Dr Carlos Cara (UCB Pharma, Spain) for his help in writing the manuscript.