Servicio de Farmacología Clínica, Grupo de Estudio para las Hepatopatías Asociadas a Medicamentos, Coordinating Centre, Hospital Universitario Virgen de la Victoria, Facultad de Medicina, Campus Universitario de Teatinos s/n, Málaga, Spain
Servicio de Farmacología Clínica, Departamento de Farmacología Facultad de Medicina, Boulevard Louis Pasteur 32, Campus de Teatinos s/n, 29071 Málaga, Spain
Unidad de Hepatología, Grupo de Estudio para las Hepatopatías Asociadas a Medicamentos, Coordinating Centre, Hospital Universitario Virgen de la Victoria, Facultad de Medicina, Campus Universitario de Teatinos s/n, Málaga, Spain
Servicio de Farmacología Clínica, Grupo de Estudio para las Hepatopatías Asociadas a Medicamentos, Coordinating Centre, Hospital Universitario Virgen de la Victoria, Facultad de Medicina, Campus Universitario de Teatinos s/n, Málaga, Spain
Potential conflict of interest: No co-author has any involvement that could be construed as a conflict of interest.
The funding source had no involvement in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.
Amoxicillin-clavulanate (AC) hepatotoxicity has been reported to exhibit a higher predominance of cholestatic types of damage, especially in males. However, the determinants of its clinical expression are unknown. This study prospectively evaluated the profile of AC hepatotoxicity. Data on all cases of hepatotoxicity reported to the Spanish Registry attributed to AC and assessed as definite or probable on the Council for International Organizations of Medical Sciences (CIOMS) scale were collated and compared to published case series. Hepatotoxicity related to amoxicillin-clavulanate was identified in 69 patients (36 males; mean age 56 years) representing 14% of all cases of hepatotoxicity submitted to the Registry. There was an overall sex distribution and the predominant pattern of lesion was hepatocellular (36%) which occurred at a shorter duration of treatment (P < .03). Mean time lapse between therapy initiation and jaundice onset was 16 days. Late onset of symptoms following end of treatment occurred in half the cases. Multiple logistic regression analysis identified advancing age as the factor associated with the development of cholestatic/mixed type of injury (odds ratio for an age interval for 1 year: 1.045 [95% CI = 1.013–1.078; P = .005). An unfavorable outcome was seen in 7% of patients. In conclusion, age is the most important determinant in the biochemical expression of AC hepatotoxicity; younger age is associated with cytolytic damage and shorter treatment duration, whereas cholestatic/mixed type of damage is related to older age and prolonged AC therapy. (HEPATOLOGY 2006;44:850–856.)
Amoxicillin-clavulanate (AC) has become one of the most widely prescribed antibiotics1 and, although the penicillin variants are considered relatively safe, adverse reactions such as those in the gastrointestinal tract can occur more frequently with this combination.2 Several reports incriminate AC in the development of cholestatic hepatitis.3–7 Yet despite the potentially serious adverse reactions, the sales of this antibiotic combination have increased steadily to become the leading antibiotics prescribed in Spain and other European countries.8 A recent multicentered Spanish study9 showed that AC was the most common drug involved in drug-induced liver injury and was the single most frequently prescribed drug leading to hospitalization for drug-induced liver disease. Other studies, using different methodologies, also found a strong association between the AC combination and acute liver injury.10, 11 In 1996, using the UK-based General Practice Research Database, the incidence of acute hepatitis related to AC was estimated as 1.7 per 10,000 prescriptions, a 6-fold higher incidence than for amoxicillin alone (0.3 per 10,000 prescriptions) despite its more widespread use.12 Similarly, in a recent retrospective analysis in the United Kingdom of 800 patients presenting with jaundice, AC and flucloxacillin (not registered for use in Spain) ranked as the main drugs responsible for pharmacologically induced cases of liver disease.13 More than 200 cases of liver damage related to AC use have been collated in a recent review of the literature by Gresser.14
The type of liver damage most frequently associated with AC has been cholestasis with signs of hypersensitivity.5–7, 12 The hepatocellular and mixed patterns of injury have been described less frequently,14 but the determinants of these clinical expressions are unknown. Some studies5, 12, 14 suggested that older age, longer duration of therapy and male gender are predisposing risk factors, whereas other studies10, 11 contradict these findings. The general consensus is that complete clinical recovery is the norm and that hepatotoxicity associated with AC use is mild but which, in some instances, could follow a more severe, or protracted, course.15, 16
In this study, we analyzed the clinical profile of the largest cohort to-date of well-characterized patients of AC hepatotoxicity and sought factors that could explain its clinical presentation. We performed a comparative analysis to provide a perspective for our Spanish data in relation to other large published studies from different countries.
AC, amoxicillin-clavulanate; GEHAM, Grupo de Estudio para las Hepatopatías Asociadas a Medicamentos; CIOMS, Council for International Organizations of Medical Sciences; NHS, National Health System; ALT, alanine aminotransferase; AP, alkaline phosphatase; ULN, upper limit of normal; SPSS, statistical package for social sciences; HLA, human leukocyte antigen; FIS, fondo investigaciones sanitarias.
Patients and Methods
The study protocol was approved by the local ethics committee of the Coordinating Centre at the Virgen de la Victoria University Hospital of Malaga, Spain. With the collaboration of several Liver Units from hospitals within the National Health System (NHS) in various parts of Spain, 69 cases of acute liver injury associated with the use of AC were collated between November 1995 and October 2005. The same structured assessment protocol was used in all cases.9 A prospective registry was set up and coordinated by two authors of this manuscript (MIL and RJA). We assessed all the cases of AC in the registry to review their clinical features and outcomes. Some of these cases have been reported elsewhere.17, 18
A clinical history was obtained from each patient. Details included antecedents of liver or biliary tract disease, drug addiction and/or alcohol abuse, transfusion of blood products, surgery within the 6 months preceding the onset of hepatitis, prescription drugs, and over-the-counter medications. Duration of therapy with AC, time to symptoms onset, time lapse between cessation of therapy and jaundice onset, and clinical features with special emphasis on evidence of hypersensitivity were noted, as was the clinical and laboratory course of the reaction once the drug was withdrawn.
Serological markers of acute hepatitis viral infection were evaluated in all patients, as was serum ceruloplasmin where appropriate. Screening for autoimmune liver disorders was conducted systematically. Ultrasonography of the liver and billiary tract was performed in all cases. Additional imaging tests (magnetic resonance and/or endoscopic cholangiography) were performed in dubious cases to exclude biliary obstruction.
The patterns of liver injury and the chronological relationship between the drug and the onset of hepatitis were defined according to the criteria of the International Consensus Meeting for Drug-induced Liver Injury.19 The pattern of liver damage was defined as hepatocellular when R > 5 [R is the ratio of ALT/AP activity, expressed as a multiple of the upper limit of normality (ULN)], cholestatic when R < 2, and mixed when R > 2 but < 5. The liver function tests used in the classification of liver damage were those from the first blood test available following liver injury. Liver damage was established according to liver biopsy results, when available.
The direct involvement of other drugs in hepatic injury was assessed on the basis of clinical features or time course of disease. Case ascertainment was initially that of the attending physician and, subsequently, evaluated at the Coordinating Centre where causality was determined by applying the CIOMS scale.20 Only the cases assigned as “definite” or “highly probable” have been included in this report.
An age ≥55 years was taken as a cutoff for analysis in keeping with the CIOMS scale stratification of age as a risk factor. Cases were classified as hypersensitivity in nature if they presented with any of the classical clinical or laboratory features of allergy (fever, rash, serum eosinophilia, cytopenia, or detectable titers of autoantibodies) and/or there was accompanying suggestive pathological findings (eosinophil-rich inflammatory infiltrate and/or granuloma formation). In the remaining cases, the mechanism was presumed to be metabolic idiosyncrasy.
Outcomes were assessed by clinical analytical imaging tests and histological methods, when available. Cases were defined as “resolved” when liver test values returned to within the laboratory reference range within 3 months for hepatocellular pattern of damage, or 6 months for cholestatic/mixed pattern of injury; or “chronic” when liver test values remained altered.
For statistical comparisons, we retrieved individual data (when available) from the larger studies published on AC-induced hepatotoxicity. We compared our data with that of the American study by Reddy et al.,4 the French study by Larrey et al.,5 and the Belgian study by Hautekeete et al.6 In the UK study performed by O'Donohue et al.,7 individual data had not been provided and therefore was not suitable for the present comparisons. However, mean values and ranges were included when available.
Data were analyzed with the SPSS software for Windows (version 12.0). Bivariate associations were measured using the Student t test for continuous variables and the χ2 test for categorical items. Analysis of variance (ANOVA) was used for comparisons of groups. Nonparametric analyses were performed where variables did not follow a normal distribution. All P values of less than .05 were considered statistically significant. Variables relevant to the development of cholestatic/mixed type of injury were included in a multiple logistic regression model. Calibration of the model was assessed using the Hosmer and Lemeshow χ2 statistics (P < .05).
Between November 1995 and October 2005, of the 486 cases of idiopathic liver injury submitted to the Coordinating Centre, 69 cases (14%) were classified as having hepatotoxicity attributable to AC use. Using the CIOMS diagnostic scale, these patients were classified as “definite” in 55 (80%) cases and “probable” in 14. There were 36 men (52%), the overall mean age was 56 years, and hepatocellular damage predominated (36%) (Table 1).
Table 1. Demographic and Clinical Parameters of Amoxicillin-Clavulanate Hepatotoxicity Cases in the Spanish Registry Compared to Other Published Large Patient Series in Other Countries
1 case of 18 was not categorized according to pattern of liver injury;
5 cases of 22 were not categorized according to pattern of liver injury;
No individual data of patients were provided, therefore this study was not used for comparison purposes. However, mean values and range when available and are included in the table; Kruskal-Wallis test
Overall, there was a similar gender distribution. The indication for AC treatment was respiratory infection (35 cases; 51%), followed by urinary infection (9 cases; 13%), and ear-nose-throat infections (7 cases; 10%). The mean daily dose administered was 2017 mg (range: 500 mg to 3000 mg) orally, and 2 cases intravenously (daily dose 3000 and 4500 mg). The AC formulation ratio was 4:1 (500/125 mg) in 42 cases and 7:1 (875/125 mg) in 12 cases. The mean duration of treatment was 12 days (ranging from 1 to 37 days). Symptoms of liver disease occurred within a mean of 16 days (ranging from 1 to 71 days) after the initiation of treatment. In 31 patients, there was a clear time-course relationship between AC administration and the onset of liver injury, as well as between withdrawal of treatment and the resolution of liver dysfunction. The exception was 1 patient who died and 5 patients who developed chronic disease. In 34 patients, hepatotoxicity was late onset and the mean time lapse from cessation of therapy to onset of jaundice was 15 days (Table 1). Other drugs taken concurrently by 6 patients were diclofenac (2 cases, the pattern of damage was cholestatic), ibuprofen (2 cases, one hepatocellular and the other cholestatic), and erythromycin and ketorolac trometamol (1 case of mixed damage each).
Jaundice was the most frequent manifestation at the time the case was first identified (56 cases; 81%), and 46 patients (67%) required hospitalization. All patients were negative for viral hepatitis (cytomegalovirus; Epstein-Barr virus; hepatitis A, B, and C viruses). Ultrasonographic examination revealed no abnormalities of the liver or the intrahepatic bile ducts in any of the patients. None of the patients had had a previous history of hepatitis, blood transfusions, or recent acute hypotension. A history of moderate alcohol consumption (mean 47 g/day; range: 6–177 g/day) was noted in 15 patients.
Indicators of hypersensitivity were present in 38% of the patients. Underlying liver disease was evident in 3 patients (4%). A positive (inadvertent) rechallenge occurred in 3 (4%) patients, 1 of whom developed liver cirrhosis and underwent liver transplantation.18 This was a young patient (44 years) that did not have any other risk factors for chronic liver disease. Presumably, re-exposure to the offending drug was critical for the evolution to liver cirrhosis.
In the 55 patients in whom we were able to continue a follow-up for an extended period, liver test values returned to within the laboratory reference range within a mean of 77 days (range: 10–300 days). The overall outcome was death in 1 patient and liver transplantation in another (2.9%). The only fatal case had compensated liver cirrhosis. In a mean follow-up of 8.5 months, 4 patients (5.8%) fulfilled the criteria of chronic outcome with persistent alterations in liver test values.
Liver histology was available in 16 patients. Cholestasis with hepatitis was the most common histopathologic diagnosis in 11 (69 %) patients. Centrilobular cholestasis was the diagnosis in 3 cases; 1 patient presented with bile duct proliferation and intense fibrosis. This patient eventually developed liver cirrhosis. In patients with histological diagnosis of hepatocellular necrosis, 2 had focal necrosis and 1 had chronic hepatitis. Eosinophils were found in biopsy specimens from 3 patients.
Comparison of Demographic Characteristics and Clinical and Laboratory Features According to Type of Damage.
The findings are summarized in Table 2. Patients with cholestatic pattern were significantly older than patients with other patterns of liver damage (P < .0001). Conversely, patients with the hepatocellular type damage had the highest mean serum alanine aminotransferase (ALT) values (P < .001). It is noteworthy that chronicity was more common with cholestatic reactions than hepatic ones.
Table 2. Demographic, Clinical, and Laboratory Parameters of the 69 Cases of Liver Injury Due to Amoxicillin-Clavulanate Use, and Segregated According to Type of Liver Damage
Type of damage
Hepatocellular (n = 25)
Cholestatic (n = 21)
Mixed (n = 23)
Abbreviations: ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase Values are expressed as multiples of the upper limit of normal (ULN). The ALT and AST values are those at presentation, whereas bilirubin values are at peak expression. Hypersensitivity refers to the presence of fever, rash, and/or eosinophilia;
Demographic Characteristics, Clinical and Laboratory Features in Patients Younger or Older Than 55 Years of Age.
Patients aged ≥55 years accounted for 65% of the overall population presenting with jaundice (93%; P < .001). They had the highest mean plasma bilirubin values, in keeping with the largest number of cases with a cholestatic/mixed damage presentation (P < .003). Overall, cholestatic presentation was only 40% of those patients classified as over 55 years, but when considering patients aged ≥70 years, 10 of 16 patients (63%) had cholestatic reactions (P < .002).
Duration of treatment was shorter in patients younger than 55 years of age (8.8 days; P < .03), exhibiting, predominantly, hepatocellular damage (P < .003) (Table 3). In a multiple logistic regression model, older age was the variable associated with the development of cholestatic/mixed type of injury induced by AC [OR for an age interval for 1 year: 1.045 (95% CI: 1.013–1.078); P = .005]. Patients with hepatotoxicity during treatment did not differ in laboratory findings and outcome from those with a late onset.
Table 3. Comparison of Demographic Characteristics and Clinical and Laboratory Findings in Patients Above and Below the Age of 55 Years and Segregated With Respect to Amoxicillin-Clavulanate Hepatotoxicity
<55 years (n = 22)
≥55 years (n = 45)
Men, n (%)
Daily dose; mg (±SD)
2075 ± 595
2000 ± 485
Treatment duration; days (±SD)
8.8 ± 5
12.7 ± 9
Onset from treatment cessation; days (±SD)
18 ± 14
17 ± 16
Symptom resolution; days (±SD)
56 ± 31
86 ± 70
Hypersensitivity features; n (%)
Total billirubin (mg/dL)
ALT (× ULN)
AP (× ULN)
Chronic damage, n (%)
Influence of Time-to-Hepatic Reaction on the Pattern of Liver Damage.
The time lapse from start of AC therapy to the onset of hepatitis, and segregated according to type of liver damage, is represented in Fig. 1. Cases of cholestatic and hepatocellular damage peaked at 1–2 weeks from start of treatment. Appearance of a second peak at around 4–5 weeks was less meaningful whereas the numbers of patients with mixed liver injury became more evident after 3–4 weeks.
Demographic and Clinical Parameters of Amoxicillin-Clavulanate Hepatotoxicity in the Spanish Case Registry Compared to Published Data Frrom Other Countries.
The segregation with respect to gender did not differ significantly in relation to other case series (χ2 = .06) (Table 1). Patients in the Belgian study were the oldest (66 years; P < .024) and treatment duration was the longest in the French study (18 days; P < .018). The Spanish patient population had the shortest mean time interval between initiation of therapy and onset of jaundice (16 days; P < .0001). The mean time lapse between cessation of therapy and onset of jaundice was similar between the studies. In the Spanish and Belgian studies, approximately half the patients had late-onset disease, while in the French and English series the percentages were much higher (87% and 95%, respectively). Liver injury was predominantly cholestatic in the French (74%), Belgian (66%), and U.S. (41%; P < .05) studies, as classified according to the International Consensus Criteria.19
This study contains clinical and epidemiological features of hepatotoxicity induced by AC derived from one of the largest cohorts of patients ever collated prospectively, and follows the same structured report format from several hospitals. It suggests that AC might be the most common cause of unpredictable hepatotoxicity in adults in Spain, accounting for 14% of all cases recorded in the registry. Our results show that age is the most important determinant in the biochemical expression of AC hepatotoxicity, cholestatic/mixed injury being related to older age, whereas younger age is associated with a cytolytic damage. The pattern of age-associated damage has been highlighted recently in the analysis of a cohort of patients with hepatotoxicity which considered all drugs collectively.9 Thomson et al.10 also suggested increasing age as a risk factor for jaundice resulting from AC treatment.
Indeed, patients younger than 55 years of age exhibited predominantly hepatocellular damage which occurred at a significantly shorter AC exposure. Interestingly, our data also indicate that the appearance of hepatic injury may follow a distinct periodicity from onset of AC therapy. Hepatocellular damage appears to predominate at 1 week, cholestatic liver injury occurs mostly at 2–3 weeks, and the mixed liver injury proportionally predominates after 3 weeks. In the French and Belgian series there was a predominance of the cholestatic/mixed type of damage. The French case series5 had a longer treatment period, the Belgian6 had the oldest set of patients, and all the other countries4–7 had a longer interval between commencement of treatment and onset of jaundice, compared to our Spanish study. All together, these factors may explain the higher proportion of patients with cholestatic/mixed presentation associated with AC combination therapy. Furthermore, in a retrospective cohort study conducted in the United Kingdom, Garcia Rodriguez et al.12 found that the greatest absolute risk of acute liver injury associated with the AC combination was among elderly patients who were receiving long-term therapy, and of note was that the type of liver injury in this cohort was cholestatic in three quarters of the cases.
Explanations for the age-related pattern of liver damage induced by AC remain speculative. We may hypothesize that a slower drug elimination related to advanced age and its concomitant retention in the body would, in turn, allow a prolonged exposure of the bile duct cells to the drug metabolite through canalicular excretion, which might trigger an immune response against haptenized duct cell proteins and a periductular inflammatory reaction.21 Recent data show an association between cholestatic/mixed injury and human leukocyte antigen (HLA) class II antigens (DRB1*15 and DQB1*06 alleles).22 This supports the concept that most cases of hepatotoxicity with cholestatic/mixed expression of damage are probably allergy based.
A characteristic of the hepatotoxicity induced by AC is the variation in time lapse between the cessation of therapy and the onset of jaundice. Late onset after treatment was observed in our study in half the cases, and the time lapse ranged from a few days up to 7.8 weeks. This late onset has been recognized for AC and is a feature shared by very few drugs.23, 24 Contrary to previous studies,4, 5 our findings do not support the view that male gender acts as risk factor in the development of AC hepatitis.
Assigning causality when more than one drug could be the culprit, and when treatments are started simultaneously, remains a major challenge. The complicity of a single agent can rarely be determined unequivocally. In all but 6 of the cases we report, AC was the only hepatotoxin identified. Of these 6 patients, 2 received diclofenac and developed cholestatic damage. Diclofenac-associated hepatitis is a rare complication that occurs mainly as hepatocellular damage in females.25 In 2 cases, an involvement of ibuprofen, which has been implicated in hepatitis of various types, could not be discarded unequivocally, whereas in another case, the potential for erythromycin to be implicated in causing hepatitis (usually cholestatic) was acknowledged.26
Although AC is widely prescribed and is administered to all age groups, the clinical and biological characteristics of the associated hepatotoxicities have been described in relation mainly to adult populations. A potential limitation on the interpretation of the data is that pediatric populations have been clearly underrepresented. The case-control study11 that showed AC to have had the strongest association of acute and clinically relevant liver injury (adjusted OR: 94.8 [95% CI: 27.8–323]), was conducted using the UK-based General Practice Research Database as the source of information. This database includes children of all ages but the youngest patient in the study was 17 years old, similar to the population ages of our data. As such, the speculation could be that a pediatric patient population might be less susceptible to developing hepatotoxicity related to the AC combination. The need for, and a proposal to establish, a multidisciplinary network to monitor drug-induced pediatric liver disease has been highlighted.27
Our data indicate that the probability of an unfavorable outcome (death, liver transplant, and persistent liver damage) is 7%. A severe outcome (death/transplant) occurred in 2.9% of the patients, numbers that are similar in other series.14, 28 This brings into question the generally-held opinion that the clinical outcome of hepatotoxicity induced by AC is invariably toward recovery.
In summary, age is the most important determinant in the biochemical expression of AC hepatotoxicity; younger age is associated with cytolitic damage and shorter treatment duration while cholestatic/mixed type of damage is related to older age and a prolonged AC therapy. These features could be helpful in the initial clinical evaluation of a patient exposed to AC and presenting with hepatitis.
We are grateful to D. Ramón Hidalgo of the Servicio Central de Informática de la Universidad de Málaga for his invaluable help in the statistical analyses.
Table . GEHAM Principal Investigators and Their Affiliations
R. J. Andrade, M. I. Lucena, B. García-Ruiz, R. Camargo, E. García-Ruiz, M. García-Cortés, K. Pachkoria, Y. Borraz.
Hospital Universitario Virgen de la Victoria, Málaga (Co-ordinating Centre)
M. C. Fernández, G. Peláez, M. Casado, J. L. Vega, M. González-Sánchez, F. Suárez.
Hospital Torrecárdenas, Almería
J. A. Durán, M. Jiménez-Sáez, M. Fernández, A. Ruiz, J. Alanis-López, M. Villar
Hospital Virgen Macarena, Sevilla
M. Romero, C. Trabadela
Hospital Universitario Virgen de Valme, Sevilla
Hospital Central de Asturias, Oviedo
J. M. Pérez-Moreno, M. Puertas
Hospital de Puerto Real, Cádiz
J. Salmerón, M. A. López-Garrido, A. Caballero, A. Gila
Hospital Universitario San Cecilio, Granada
R. Planas, A. Soler, J. Costa, A. Barriocanal
Hospital Germans Trias i Puyol, Barcelona
R. Martín-Vivaldi, F. Nogueras
Hospital Universitario Virgen de las Nieves, Granada