BAYESIAN MIXED TREATMENT COMPARISON META-ANALYSIS OF OUTCOMES OF TREATMENT FOR CHRONIC HEPATITIS B

Purpose: Chronic hepatitis B (CHB) accounts for 50% of case of cirrhosis, liver failure and hepatocellular carcinoma worldwide.  Many treatments are available; but, comparative efficacy evidence from direct, head-to-head, trials is not. We set out to estimate the relative effects of treatments for CHB in HBeAg positive patients, and if feasible, provide a ranking of treatment effects.

Methods:  The mixed treatment comparisons (MTC) method for meta-analysis of randomized controlled trial data was used to infer direct and indirect comparisons of CHB treatments. Data from 14 RCTs reporting dichotomous outcomes of undetectable HBV DNA, ALT normalization and HBeAg seroconversion; and evaluating 10 CHB treatment strategies was included.  The Bayesian random-effects MTC logistic regression model was implemented in the Bayesian software WinBUGS. The model was structured so that (A) Lamivudine (LAM) was considered the reference treatment; (B) pooled response rates of LAM were estimates across treatment arms, (C) direct effects derived and (D) indirect effects were estimated using data from (A-C). Assuming the relative efficacy was unknown; uninformative prior distributions for all efficacy parameters were used. Probability estimates of the clinical outcomes were then derived from the fitted MTC model. These were used in the final ranking.

Results:  Tenofovir (TDV) monotherapy was estimated with the highest probability of undetectable HBV DNA, at 84% (95%CrI:27%-99%) among the 10 competing treatments.  LAM and adefovir (ADV) combination therapy was estimated with the second highest probability at 65% (95%CrI:10-97%) for undetectable HBV DNA.   LAM+ADV combination therapy was estimated with the highest probability for normalization of ALT, at 80% (95%CrI:74%-98%) followed by monotherapy with TDV at 71% (95%CrI:59%-97%).  Combination therapy with LAM+ADV was estimated with the highest probability of HBeAg serconversion, at 20% (95%CrI:2%-55%) followed closely by entecavir monotherapy with an estimated probability of 16% (95%CrI:6%-33%).

Conclusion: Given the lack of direct evidence comparing these alternative CHB treatments, the approach we used provides an evidence-based solution to discern the relative efficacy of the multitude of treatment options. No treatment strategy was superior for all three outcomes however LAM+ADV combination therapy was ranked first or second for all three clinical outcomes.  Monotherapy with TDV was ranked in the top four in each of the outcomes.  The analysis suggests that LAM+ADV is estimated as the most efficacious treatment option for these three clinical outcomes.