The comparison of effects of biologic agents on rheumatoid arthritis damage progression is biased by period of enrollment: data from a systematic review and meta-analysis.

OBJECTIVES

To indirectly compare the 12-month effects of available biologic agents in slowing RA radiographic progression.

METHODS

A systematic review of literature of randomised, double-blind, controlled trials (RCTs) evaluating RA radiographic progression as end point was conducted using a PubMed searching of MEDLINE from January 1995 to May 2012. For each trial, the mean change from baseline of the standardised annual radiographic progression score (weighted for estimated annual progression rate) was estimated, and the effect size was calculated as the difference between biologic and non-biologic-treated groups. In order to optimise data homogeneity and improve RCTs comparison, a mixed-effect model was applied including previous responsiveness to methotrexate (MTX-experienced or MTX-naïve populations) and period of study enrollment as moderators.

RESULTS

The PubMed search resulted in 183 references, and 14 were eligible for the meta-analysis. The analysis of study distribution in forest plots showed a high correlation between the study period of enrollment and the impact of biological therapy in both MTX-naïve and MTX-experienced subgroups. In particular, effect size was the highest for older trials and progressively decreased in the most recent ones, suggesting a highest propensity to radiographic progression in populations enrolled in older trials. Some statistically significant differences among RCTs were found in both subgroups but were significantly biased by the different propensity to radiographic progression due to period of enrollment.

CONCLUSIONS

Our meta-analysis demonstrated that period of enrollment deeply influence study population propensity to radiographic progression in each trial. This finding does not allow the indirect comparison of various biologic agents, despite our mixed-model significantly reducing heterogeneity among RCTs.