Francesca Saffioti, Kurinchi Selvan Gurusamy, Neil Hawkins, Clare D Toon, Emmanuel Tsochatzis, Brian R Davidson, Douglas Thorburn

Primary sclerosing cholangitis is a chronic cholestatic liver disease that is associated with both hepatobiliary and colorectal malignancies, which can result in liver cirrhosis and its complications. The optimal pharmacological treatment for patients with primary sclerosing cholangitis remains controversial. To assess the comparative benefits and harms of different pharmacological interventions in people with primary sclerosing cholangitis by performing a network meta-analysis, and to generate rankings of available pharmacological interventions according to their safety and efficacy. Given that it was not possible to assess whether potential effect modifiers were similar across comparisons, we did not perform the network meta-analysis but instead used standard Cochrane methods.When trials begin to provide an adequate description of potential effect modifiers, we will attempt to conduct network meta-analysis. We searched CENTRAL, MEDLINE, Embase, Science Citation Index - Expanded, the WHO International Clinical Trials Registry Platform, and randomised controlled trials registers until February 2017 to identify randomised clinical trials (RCT) on pharmacological interventions for primary sclerosing cholangitis. We included only RCTs, irrespective of language, blinding, or publication status, in which participants were given a diagnosis of primary sclerosing cholangitis. We excluded trials that included previously liver-transplanted participants. We considered any of various pharmacological interventions compared with one other or with placebo. We excluded trials that compared different doses of various pharmacological interventions or that reported different treatment durations, except for ursodeoxycholic acid (UDCA). As UDCA is the drug most commonly investigated for primary sclerosing cholangitis, we performed a second analysis in which we stratified the dose of UDCA. We calculated the odds ratio and the rate ratio with 95% confidence intervals (CIs) using both fixed-effect and random-effects models based on available-participant analysis with Review Manager. We assessed risk of bias according to Cochrane, controlled risk of random errors with Trial Sequential Analysis, and assessed the quality of the evidence using GRADE. We identified 22 RCTs in which 1211 participants were randomised to 13 different interventions. Most were placebo-controlled trials. Trials had few restrictions apart from an established diagnosis of primary sclerosing cholangitis, evidence of cholestasis, absence of decompensated liver disease, and absence of malignancy. However, some trials included symptomatic participants only, and others included both symptomatic and asymptomatic participants. A total of 11 RCTs (706 participants) provided data for one or more outcomes. The period of follow-up ranged from three months to three years in most trials. Only three trials reported follow-up longer than three years. Investigators found no evidence of differences in important clinical benefits such as reduction in mortality at maximal follow-up and improvement in health-related quality of life. Primary outcomes Mortality: Effect estimates: colchicine versus placebo: odds ratio 0.44, 95% CI 0.04 to 5.07, participants = 84, one trial; penicillamine versus placebo: odds ratio 1.18, 95% CI 0.39 to 3.58, participants = 70, one trial; steroids versus placebo: odds ratio 3.00, 95% CI 0.10 to 90.96, participants = 11, one trial; ursodeoxycholic acid versus placebo: odds ratio 1.51, 95% CI 0.63 to 3.63, participants = 348, two trials, I(2) = 0%; vancomycin versus placebo: not estimable because no events in either group, participants = 29, one trial. Serious adverse events (proportion): Effect estimates: infliximab versus placebo: odds ratio not estimable (because of zero events in both arms), participants = 7, one trial; steroids versus placebo: odds ratio 20.00, 95% CI 0.93 to 429.90, participants = 11, one trial; vancomycin versus placebo: not estimable because no events in either group, participants = 29, one trial. Serious adverse events (number): Effect estimates: infliximab versus placebo: rate ratio 0.80, 95% CI 0.02 to 40.44, participants = 7, one trial; penicillamine versus placebo: rate ratio 13.60, 95% CI 0.78 to 237.83, participants = 70, one trial; steroids versus placebo: rate ratio 3.32, 95% CI 0.71 to 15.62, participants = 11, one trial. Adverse events (proportion): Effect estimates: steroids versus placebo: odds ratio 20.00, 95% CI 0.93 to 429.90, participants = 11, one trial; ursodeoxycholic acid versus placebo: odds ratio 1.22, 95% CI 0.68 to 2.17, participants = 198, one trial; vancomycin versus placebo: not estimable because no events in either group, participants = 29, one trial. Adverse events (number): Effect estimates: cyclosporin versus placebo: rate ratio 2.64, 95% CI 0.99 to 7.03, participants = 26, one trial; steroids versus placebo: rate ratio 3.32, 95% CI 0.71 to 15.62, participants = 11, one trial; ursodeoxycholic acid plus metronidazole versus ursodeoxycholic acid: rate ratio 2.36, 95% CI 0.98 to 5.71, participants = 71, one trial. Health-related quality of life: ursodeoxycholic acid versus placebo: mean difference 1.30, 95% CI -5.61 to 8.21, participants = 198, one trial (Short Form (SF)-36 General Health Scale). Secondary outcomes Studies provided no evidence of differences in clinical benefits such as a reduction in the requirement for liver transplantation or a reduction in the incidence proportion of cholangiocarcinoma. One small trial (29 participants) comparing vancomycin versus placebo reported no malignancies, no liver decompensation, and no liver transplantation in either group after a very short follow-up period of 12 weeks after treatment. None of the remaining trials clearly reported other clinical benefits such as decreased development of all malignancies, colorectal cancer, liver decompensation, time to liver decompensation, time to liver transplantation, or requirement for cholecystectomy to allow comparisons between different interventions. Fifteen trials reported the source of funding; three were funded by parties without vested interest in results of the trial, and 12 were funded in part or in full by drug companies. Evidence is currently insufficient to show differences in effectiveness measures such as mortality, health-related quality of life, cirrhosis, or liver transplantation between any active pharmacological intervention and no intervention. However, trials were at high risk of bias and included small numbers of participants, had short follow-up periods, and reported few clinical outcomes. An urgent need exists to identify an effective medical treatment for primary sclerosing cholangitis through well-designed RCTs with adequate follow-up that aim to identify differences in outcomes important to people with primary sclerosing cholangitis.