Drug-induced liver injury in humans: the case of ximelagatran.

Ximelagatran was the first orally available direct thrombin inhibitor under clinical development that also reached the market. Ximelagatran was tested in an extensive clinical programme. Short-term use (<12 days) in humans including the phase III clinical trials did not indicate any hepatotoxic potential. Increased hepatic enzyme levels were first observed at a higher frequency when evaluating the long-term (>35 days) use of ximelagatran (incidence of >3x upper limit of normal (ULN) plasma ALT was 7.9%). The frequency of elevated total bilirubin levels was similar in the ximelagatran and the comparator groups. However, the combination of ALT > 3x ULN and total bilirubin > 2xULN was 0.5% among patients treated with ximelagatran and 0.1% among patients in the comparator group. Symptoms such as fever and rash potentially indicating hypersensitivity (immunologic type of reaction) were low and did not differ between ximelagatran and the comparators. The withdrawal of ximelagatran from the market and termination of the ximelagatran development program was triggered by safety data from a 35-day study, indicating that severe hepatic injury in a patient could develop after exposure to the drug has been completed and that regular liver function monitoring may not mitigate the possible risk of severe hepatic injury. As for many drugs causing liver injury, the standard preclinical toxicological studies provided no indication that ximelagatran affected hepatic functions. In addition, extensive investigations using human-based in vitro models have not been able to define mechanisms explaining the pattern of hepatic injury observed in long-term clinical trials. A pharmacogenomic study provided evidence that the ALT increases were associated with major histocompatibility complex (MHC) alleles DRB1'07 and DQA1*02 suggesting a possible immunogenic pathogenesis. This example provides important clues to the mechanism of idiosyncratic drug-induced liver toxicity.

Utility of the Roussel Uclaf Causality Assessment Method (RUCAM) to analyze the hepatic findings in a clinical trial program: evaluation of the direct thrombin inhibitor ximelagatran.

AIMS: Causality assessment in drug-induced liver injury is often based on circumstantial evidence rather than a formal, systematic review. The Roussel Uclaf Causality Assessment Method (RUCAM) provides a more objective means of assessing causality of a suspected hepatotoxin but, to our knowledge, has never been used in the assessment of a single drug with unknown hepatotoxic potential in a clinical trial setting. METHODS: We studied the utility of RUCAM in assessing the hepatic events during the long-term clinical trials of the oral direct thrombin inhibitor ximelagatran, which has been associated with an increased incidence of alanine aminotransferase (ALT) elevations. A total of 233 subjects with elevated ALT values signalling possibly severe hepatic injury were eligible for RUCAM analysis (198 ximelagatran and 35 comparator anticoagulants). RESULTS: RUCAM scores, calculated independently by the assessors, using the existing numerical criteria provided in its methodology, suggested a possible or probable causal relationship between ALT and ximelagatran in 37 and 27% of cases, respectively. Causality was excluded or unlikely in the remaining 36% of cases. However, in the course of utilizing RUCAM, several limitations to the methodology came to light, including awarding additional points for age > 55 years, an unspecified use of alcohol, and a latency period of < 90 days, which may have had the unintentional effect of raising the overall score. Moreover, rechallenge is highly rewarded by RUCAM but is seldom done in clinical practice or in clinical trials. We also found ambiguities in the extent to which other causes of liver injury were excluded, what constitutes a significant hepatotoxic concomitant medication, and whether a clinical trial drug should be considered as having an unknown hepatotoxic potential for purposes of RUCAM scoring. Increasing familiarity with the RUCAM over the course of the study allowed for only a slight improvement in concordance between and among the assessors regarding the scoring. CONCLUSIONS: While the results indicate that RUCAM can provide for an objective assessment of causality of the hepatotoxicity of a drug under development in the clinical trial setting, this study highlights a number of problems with the current scoring system that should be addressed by future enhancements of the methodology.

Influence of food intake on electrocardiograms of healthy male volunteers.

OBJECTIVE: To study the influence of food intake on electrocardiogram (ECG) variables (heart rate, QT-, QTc-, PR-intervals, QRS-time) and morphological alterations of the T-waves in 12 healthy male volunteers. METHODS: The study was of open, three-period crossover design. On each occasion, all subjects fasted from midnight. During two of the study periods, the subjects were given a standardised meal at 1.5 h and 5.5 h after the baseline assessments, respectively, whereas, during the third period, they remained fasting for the entire study period of about 9 h. ECG and blood pressure were recorded at baseline and thereafter every hour for 8 h. RESULTS: No ECG changes were observed following the fasting condition, whereas a clear change in ECG and an increased heart rate were recorded in response to the meal intake during the other two periods. The most prominent ECG effect was the change in the size and shape of the T-waves, which were described as flattened to biphasic and, occasionally, negative. These alterations were most pronounced in the precordial leads V4 to V6 in the ECG recording immediately following the meal intake, with a gradual return to baseline conditions over 4-5 h. Moreover, a transient increase of supine systolic blood pressure was also recorded in response to the meal intake. CONCLUSIONS: The intake of a meal can cause clear and consistent ECG changes in healthy male subjects, comprising increases in heart rate as well as alterations in the size and shape of the T-waves (flattened to biphasic and, occasionally, negative). Also, a post-meal increase in the supine systolic blood pressure was recorded.

An epidemiological study of aplastic anaemia: relationship of drug exposures to clinical features and outcome.

Two hypotheses were examined in the combined data from 3 case-control studies of aplastic anaemia, conducted in Thailand, Europe/Israel and the US: 1. Cases exposed to drugs associated with a significantly increased risk of aplastic anaemia are more likely to present with thrombocytopenia (e.g. petechiae, easy bruising); and 2. cases exposed to these drugs are more likely to recover quickly than non-exposed cases. After excluding all cases who lacked information on timing of symptoms and those whose symptoms began > or = 180 d before hospital admission, 392 cases remained for analysis. A total of 51 (13%) had been exposed to one of the significantly associated drugs; the remaining 341 (87%) had not. Among the former, 31% reported thrombocytopenia either before or at the same time as non-bleeding symptoms (e.g. pallor, fatigue); the corresponding proportion among the non-exposed was 53%. Data on time to recovery (return of the 3 blood cell lines to normal levels) were not available for the Thai cases; among the others, the median time to recovery for the non-fatal cases was 7 and 6 months in the 29 exposed and the 83 non-exposed cases, respectively. The data do not support either hypothesis: the two groups of aplastic anaemia cases appeared to be similar in both the presenting symptoms and the recovery time.

Comparing risk estimates of sulphonamide-induced agranulocytosis from the Swedish Drug Monitoring System and a case-control study.

A comparison has been made of risk estimates for agranulocytosis connected with sulphasalazine and trimethoprim-sulphamethoxazole (T-SM) therapy calculated from data in the Swedish Drug Monitoring System ("spontaneous" reports, sales and prescription information) and a population based case-control study (the IAAAS). The relative risk for agranulocytosis during sulphasalazine treatment was calculated to be 107 and 123 by the spontaneous reporting system and the case-control study, respectively. The corresponding excess risk in both systems was 1.8. For T-SM the relative risk was 17 in the spontaneous reporting system and 12 in the case-control study, while the excess risk calculated for 3 days of treatment was 0.9 in the spontaneous reporting system, and 1.6 for 3 or more days of treatment in the case-control study. It is concluded that the Swedish Drug Monitoring System gives an appropriate estimate of the risk of developing agranulocytosis in association with the two drugs studied.

Sulfasalazine associated agranulocytosis in Sweden 1972-1989. Clinical features, and estimation of its incidence.

During the 18 year period 1972-1989 a total of 62 cases of agranulocytosis associated with the use of sulphasalazine were reported to the Swedish Adverse Drug Reactions Advisory Committee (SADRAC). The median age of the patients was 52 y and median duration of sulphasalazine treatment was 43 days. The fatality rate was 6.5%, and among patients who recovered the median recovery time was 10 days. Twelve patients were treated concomitantly with other drugs generally suspected to induce agranulocytosis. From sales and prescription data the average incidence of agranulocytosis during sulphasalazine therapy was estimated to be 1/1750 patient years of exposure. From an ongoing Prescription Monitoring Project in a Swedish county it was possible to calculate the proportion of patients receiving sulphasalazine for different periods of time. The incidence of agranulocytosis during the first 30 days was estimated to be 1/2400 patients, while it was 1/700 between Days 31-90 and 1/11200 during Days 91-365. The risk of developing agranulocytosis during sulphasalazine treatment is considerable during the first three months of treatment, and the traditional way of expressing the risk (1/1750 patient years) underestimates the risk for the individual patient.

How safe is proguanil? A post-marketing investigation of side-effects.

Side-effects of proguanil reported to the Swedish Adverse Drug Reaction (ADR) register from 1981 to 1988 are described and related to sales figures of the drug in Sweden during the same period. One serious reaction, thrombocytopenia, and 7 minor reactions, mainly urticaria and exanthema were believed to be causally related to proguanil intake in an estimated 60,000 users of the drug. Proguanil can be considered a very safe drug but rare hematological side-effects may possibly occur.

Acetazolamide-associated aplastic anaemia.

Eleven cases of acetazolamide-associated aplastic anaemia were reported in Sweden during a 17-year period. There were six women and five men with a median age of 71 years (range 63-85 years). The median dose of acetazolamide was 500 mg, and the median duration of treatment was 3 months (range 2-71 months). Ten of the eleven patients died, all within 8 weeks after detection of their aplastic anaemia. The relative risk of developing aplastic anaemia when taking acetazolamide was 13.3 (95% confidence limits (CL); 6.8-25.3). The estimated incidence of reported acetazolamide-associated aplastic anaemia is approximately one in 18,000 patient years. The results strongly indicate that acetazolamide treatment is associated with a substantial increase in the risk of developing aplastic anaemia.

Transient pancytopenia. A report from the International Agranulocytosis and Aplastic Study.

From a population-based study on the incidence of potentially drug-associated blood dyscrasias 28 cases were identified with pancytopenia. Who recovered within 90 days after diagnosis. Early recovery occurred more frequently in patients showing normal or increased cellularity of the bone marrow than in patients with bone marrow hypoplasia. Median recovery times of leukocytes were 14 and 10 days and of platelets 21 and 9 days in patients with and without bone marrow hypoplasia, respectively. Age and sex distribution were similar in both groups. Of 28 patients, 11 reported a period of fever before onset of pancytopenia. Sixteen patients in whom information on drug use was available had taken a median of 4 drugs before the onset of symptoms that were related to pancytopenia. From these results we present the hypothesis that transient pancytopenia with or without marrow hypoplasia can be the expression of the same type of bone marrow injury and that drugs or viral infections should be considered as etiological factors.

Trimethoprim-sulphamethoxazole-associated blood dyscrasias. Ten years' experience of the Swedish spontaneous reporting system.

During the 10-year period 1976-1985, a total of 154 cases of blood dyscrasia were reported in Sweden which were evaluated as having a probable or possible causal relationship with trimethoprim-sulphamethoxazole (T-SM). There were 61 cases of leucopenia (of which 16 had agranulocytosis), 28 cases of thrombocytopenia, and two of non-haemolytic anaemia. There were also 32 cases of bicytopenia and 31 cases of tricytopenia. The median age varied from 38 years in the leucopenia group to 81 years in those with tricytopenia. The overall fatality rate was 17%, ranging from 2% in the group with mild leucopenia to 52% in the group with tricytopenia. In relation to sales and prescription data, the overall incidence of reported T-SM blood dyscrasias was 5.3 per million defined daily doses, and among out-patients the incidence was one case per 18,000 prescriptions. Thus the overall incidence of any blood reaction to T-SM appears to be low. In relation to prescription data, elderly people were overrepresented among the serious reactions.

Diagnoses in patients with severe pancytopenia suspected of having aplastic anemia.

There were 100 patients with pancytopenia notified to the Swedish part of an international study of aplastic anemia (Aa) (1983-1986). Aa was the cause in 16 patients and 50 patients had different conditions explaining their pancytopenia, whereas in 34 patients no obvious explanation was found at the time of discovery of their pancytopenia. A follow-up study in 1987 comprised the patients with Aa (n = 16) and the patients with uncharacterized pancytopenia (n = 34.3 additional patients had then been diagnosed as having Aa whereas 3 patients had a diagnosis of a low-grade NHL and 8 patients a diagnosis of MDS. No (or an inadequate) bone marrow sample had been taken in 9 of the 34 patients. This study shows that pancytopenia may be found in many serious conditions. Patients with pancytopenia should be carefully followed as Aa can develop slowly and as the MDS diagnosis can be difficult to establish at an early stage.

The incidence of agranulocytosis during treatment of dermatitis herpetiformis with dapsone as reported in Sweden, 1972 through 1988.

During the 17-year period 1972 through 1988, a total of seven cases of agranulocytosis associated with the use of dapsone for the treatment of dermatitis herpetiformis were reported in Sweden. The median age of the patients involved was 61 years; three of them were male. The median duration of dapsone treatment was 7 weeks and the daily prescribed dose was 100 mg. Based on sales and prescription data, the crude relative risk of agranulocytosis during dapsone treatment of dermatitis herpetiformis was 50, and the total risk was one case per 3000 patient years of exposure to dapsone. In relation to the number of new cases of dermatitis herpetiformis, agranulocytosis was estimated to develop in 1 of 240 to 425 patients receiving dapsone therapy. Patients should be instructed to seek medical care immediately in case of fever.

Could Swedish 'yellow cards' be substituted by E-coded summaries?

Case summaries of 2490 patients treated at the Department of Infectious Diseases, Danderyd Hospital, in 1986 were reviewed for discharge diagnoses where the International Classification of Disease (ICD) code indicated an adverse drug reaction (ADR) [E 939,9 in ICD 8]. Of 48 patients whose case summaries indicated an ADR, only 10 (21%) had been reported on 'yellow cards' to the Swedish Adverse Drug Reactions Advisory Committee. The Committee had also received from the same department 3 reports where the case summary lacked an ADR code. The information on drug treatment with respect to dosage, duration and treatment with concomitant drugs was not complete in the case summaries but sufficient for a preliminary evaluation of the suspected reactions in all but 4 of the cases. On the basis of these findings the authors do not reach an outright conclusion in favour of replacing the 'yellow card' system but the possible benefits of a system in which 'yellow cards' are supplemented by automatic referral of all case summaries containing an ICD code indicating an adverse drug reaction was judged sufficient to recommend and initiate a large field study.