Low risk of hepatotoxicity from rifampicin when used for cholestatic pruritus: a cross‐disease cohort study

Summary Background The use of rifampicin for cholestatic pruritus is accompanied by concerns over safety, but the availability of real‐world prescribing data is relatively limited. Aim We sought to describe the rate and characteristics of rifampicin‐induced hepatitis in a mixed aetiology cohort of patients with established liver disease and cholestatic pruritus. Methods Retrospective review of records for out‐patients commenced on rifampicin for pruritus 2012‐2016 inclusive. Rifampicin‐induced hepatitis was recorded where alanine aminotransferase activity (ALT) increased to both ≥5 × baseline and ≥5 × upper limit of normal (ULN), or to both ≥3 × baseline and ≥3 × ULN with concurrent elevation in serum bilirubin to ≥2 × baseline and ≥2 × ULN, in addition to a Roussel‐Uclaf Causality Assessment Method score of “probable” or “highly probable” for rifampicin causality. Results After exclusions, we reviewed 105 patients who took rifampicin for a median of 131 days. Most had primary biliary cholangitis or primary sclerosing cholangitis; 40 (38.1%) were men and median age was 44 years (IQR: 32‐57). 44 (41.9%) patients had baseline serum bilirubin ≥2 × ULN and 28 (26.7%) ALT ≥3 × ULN. 5 (4.8%) developed rifampicin‐induced hepatitis at a median of 70(range 27‐130) days after drug initiation. No individual or laboratory baseline characteristics were significantly associated with subsequent development of hepatitis. All cases of hepatitis recovered after drug cessation, although one patient was hospitalised and received corticosteroids. Conclusions Given the efficacy of rifampicin for an important sub‐group of those with cholestatic pruritus, adult patients, including those with jaundice, can be counselled that 95% of prescriptions are safe, and where hepatitis occurs, including at long latency, drug cessation appears effective.

Comparative statistics were used to compare groups that did, and that did not, develop hepatitis whilst taking rifampicin: the chi-squared test was used to compare categorical variables whilst the Mann-Whitney U-test was used to compare non-normally distributed numeric variables with normality assessed using the Shapiro-Wilk test; a P value of < 0.05 was considered significant.

| RESULTS
We identified 116 out-patients prescribed rifampicin without concurrent isoniazid by the department of liver medicine between 2012 and 2016 inclusive ( Figure 1). Of these, four (2.4%) prescriptions were not made for the treatment pruritus and seven (6.0%) prescriptions were never commenced. 105 patients were therefore included in the final analysis.
One thousand three hundred and eighteen patients were prescribed rifampicin without isoniazid by departments other than liver medicine over the same time period. To assess whether these patients had received rifampicin for pruritus, 116 records were ran-   Table 2). The median time to diagnosis of rifampicin-induced hepatitis was 70 days (range 27-130).
There were no significant differences in baseline characteristics between patients who were diagnosed with rifampicin-induced hepatitis and those that were not (Table 1). One further patient was diagnosed with rifampicin-associated acute kidney injury without associated hepatitis.
During our median follow-up of 809 days, we identified 29 patients who met biochemical criteria for potential DILI. Of these, 21 had undergone liver transplantation immediately prior to the derangement in liver biochemistry and were not further analysed. Of the remaining 8, 5 represented rifampicin-related DILI as assessed by RUCAM and are described in Table 2. Three patients' derangements in liver biochemistry were scored as not related to rifampicin.
One patient reached biochemical criteria at 451 days following rifampicin initiation with a RUCAM score of À2 and a clinical course involving excess alcohol consumption; one at 545 days with a RUCAM score of À2 and a clinical diagnosis of deteriorating PSC; and one at 1239 days with a RUCAM score of À1 and with a clinical course that involved an undiagnosed febrile illness associated with cerebrospinal fluid lymphocytosis.
During the first 120 days of follow-up, we assessed for milder fluctuations in liver biochemistry. Serum ALT activity rose to   continued to take rifampicin at the point of data collection. Figure 3 details the proportion of patients who remained free of major adverse events, minor adverse events and who continued to take rifampicin over the time period studied.

| DISCUSSION
Here, we present the largest, real-world, cohort of patients treated with rifampicin for pruritus in liver disease published to date. Within our mixed aetiology cohort, we demonstrate that 95% of patients did not have any concern for hepatotoxicity, but that in 5% a rifampicininduced hepatitis was diagnosed. This finding is in contrast with the literature from controlled trials of rifampicin for pruritus where no cases of hepatitis were reported, but is consistent with isolated case reports and the literature derived from tuberculosis therapy. Our hepatitis rate of 4.8% is higher than the median of 1.1% reported for tuberculosis treatment regimens not containing isoniazid by one meta-analysis, but is within the range of rates reported for tuberculosis patients treated with both isoniazid and rifampicin, 11 and less than reported rates of 10%-20% for any increase in serum transaminase activity reported with rifampicin therapy. 19 Attributing liver injury to a given potential causative agent is challenging. In this study we used the established RUCAM, which has been widely used in other studies and has the benefit of being points-based rather than explicitly relying on individual opinion.
However, it is important to note that recent commentary and guidelines have also emphasised the value of expert opinion in diagnosing DILI. 20,21 Furthermore, the RUCAM has been suggested to potentially underestimate the rate of drug-induced liver injury and to demonstrate more inter-assessor variability than other methodology. 22 In this study, we used a variation on the original RUCAM to account for deranged baseline liver biochemistry. Although such an adjustment is consistent with that promoted by a body expert opinion, it represents a variation from the initial RUCAM specification and is therefore likely to have differing sensitivity and sensitivity for diagnosing DILI. 17 Although the number of patients who developed rifampicininduced hepatitis in our cohort is small and our analysis is retrospective, there were no statistically significant distinguishing features that predicted subsequent development of hepatitis; a much larger cohort would, however, be necessary to investigate risk factors more conclusively. This is in contrast with the work on those treated for mycobacterial infection with rifampicin, where alcohol excess, low BMI, low serum albumin, age and gender have been suggested as predisposing factors. 10,11,23,24 Multi-centre studies will be needed to further investigate potential risk factors for rifampicin-induced hepatitis in the context of liver disease.   Table 2 Our cohort included some 23 patients with a baseline serum bilirubin of over 100 lmol/L ( ≥ 5.8 mg/dL). The UK package insert of rifampicin states that its use is contra-indicated in jaundice, although anecdotally many liver clinicians will consider using rifampicin for the treatment of pruritus despite jaundice. 25 We note, however, that it is our practice to co-prescribe oral vitamin K for icteric patients receiving rifampicin to reduce the risk of coagulopathy. 26 None of our markedly jaundiced patients developed hepatitis and we note that a separately reported group of markedly jaundiced patients with hepatocellular secretory failure predominantly attributed to biliary transporters dysfunction are reported as having safely received rifampicin for up to 10 weeks. 27  F I G U R E 3 Proportion of patients event-free after initial rifampicin prescription. Those who stopped taking rifampicin because of a major adverse event (hepatitis or renal failure) are shown in green; those who stopped because of an adverse event or intolerance are shown in red; those who stopped taking rifampicin because of an adverse event, intolerance or reported ineffectiveness are shown in blue. In all instances if the drug was stopped because of resolution of pruritus, liver transplantation, death or reasons other than an adverse event, intolerance or ineffectiveness, this was not recorded as a positive event at censoring