Glucocorticosteroid-free versus glucocorticosteroid-containing immunosuppression for liver transplanted patients.Cochrane Database Syst Rev. 2018 04 09; 4:CD007606.CD
Liver transplantation is an established treatment option for end-stage liver failure. Now that newer, more potent immunosuppressants have been developed, glucocorticosteroids may no longer be needed and their removal may prevent adverse effects.
To assess the benefits and harms of glucocorticosteroid avoidance (excluding intra-operative use or treatment of acute rejection) or withdrawal versus glucocorticosteroid-containing immunosuppression following liver transplantation.
We searched the Cochrane Hepato-Biliary Group Controlled Trials Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, Science Citation Index Expanded and Conference Proceedings Citation Index - Science, Literatura Americano e do Caribe em Ciencias da Saude (LILACS), World Health Organization International Clinical Trials Registry Platform, ClinicalTrials.gov, and The Transplant Library until May 2017.
Randomised clinical trials assessing glucocorticosteroid avoidance or withdrawal versus glucocorticosteroid-containing immunosuppression for liver transplanted people. Our inclusion criteria stated that participants should have received the same co-interventions. We included trials that assessed complete glucocorticosteroid avoidance (excluding intra-operative use or treatment of acute rejection) versus short-term glucocorticosteroids, as well as trials that assessed short-term glucocorticosteroids versus long-term glucocorticosteroids.
DATA COLLECTION AND ANALYSIS
We used RevMan to conduct meta-analyses, calculating risk ratio (RR) for dichotomous variables and mean difference (MD) for continuous variables, both with 95% confidence intervals (CIs). We used a random-effects model and a fixed-effect model and reported both results where a discrepancy existed; otherwise we reported only the results from the fixed-effect model. We assessed the risk of systematic errors using 'Risk of bias' domains. We controlled for random errors by performing Trial Sequential Analysis. We presented our results in a 'Summary of findings' table.
We included 17 completed randomised clinical trials, but only 16 studies with 1347 participants provided data for the meta-analyses. Ten of the 16 trials assessed complete postoperative glucocorticosteroid avoidance (excluding intra-operative use or treatment of acute rejection) versus short-term glucocorticosteroids (782 participants) and six trials assessed short-term glucocorticosteroids versus long-term glucocorticosteroids (565 participants). One additional study assessed complete post-operative glucocorticosteroid avoidance but could only be incorporated into qualitative analysis of the results due to limited data published in an abstract. All trials were at high risk of bias. Only eight trials reported on the type of donor used. Overall, we found no statistically significant difference for mortality (RR 1.15, 95% CI 0.93 to 1.44; low-quality evidence), graft loss including death (RR 1.15, 95% CI 0.90 to 1.46; low-quality evidence), or infection (RR 0.88, 95% CI 0.73 to 1.05; very low-quality evidence) when glucocorticosteroid avoidance or withdrawal was compared with glucocorticosteroid-containing immunosuppression. Acute rejection and glucocorticosteroid-resistant rejection were statistically significantly more frequent when glucocorticosteroid avoidance or withdrawal was compared with glucocorticosteroid-containing immunosuppression (RR 1.33, 95% CI 1.08 to 1.64; low-quality evidence; and RR 2.14, 95% CI 1.13 to 4.02; very low-quality evidence). Diabetes mellitus and hypertension were statistically significantly less frequent when glucocorticosteroid avoidance or withdrawal was compared with glucocorticosteroid-containing immunosuppression (RR 0.81, 95% CI 0.66 to 0.99; low-quality evidence; and RR 0.76, 95% CI 0.65 to 0.90; low-quality evidence). We performed Trial Sequential Analysis for all outcomes. None of the outcomes crossed the monitoring boundaries or reached the required information size. Hence, we cannot exclude random errors from the results of the conventional meta-analyses.