Seventeen RCTs (n=1,223) were included.
As the fixed-effect and random-effects models gave similar results, only the effect sizes obtained using the random-effects model were reported.
Effect of IFN versus non-active treatment on total HAI score (12 RCTs).
There was variability across the trials in terms of the IFN schedule. The overall net change was -0.82 (95% CI: -1.25, -0.40, p<0.0001). There was significant heterogeneity among trials (p<0.0001).
Effect of IFN versus non-active treatment on the degree of fibrosis (10 RCTs). The pooled net change was -0.20 (95% CI: -0.35, -0.05, p<0.01); the results were homogeneous across trials.
Effect of IFN versus non-active treatment on periportal and intralobular necrosis and portal inflammation (10 RCTs).
The pooled data showed a statistically-significant decrease of the HAI score in treated patients, compared with controls, for all 3 categories. The overall net change was -0.72 (95% CI: -0.99, -0.46, p<0.0001) for intralobular necrosis, -0.62 (95% CI: -0.97, -0.28, p<0.001) for portal inflammation, and -0.59 (95% CI: -0.91, -0.27, p<0.001) for periportal necrosis. There was significant heterogeneity across the trials for portal inflammation and periportal necrosis (both p<0.05), but not for intralobular necrosis.
There was a correlation between the histological and biochemical responses at the end of IFN therapy.
Histological improvement was confined to the subset of biochemical responders. There was no or very little change in non-responders.