Eight trials (n=2,059) were included in the review, comprising four open-label trials (n=1,045) and four RCTs (n=1,014); the RCTs were included in the meta-analysis. Sample sizes ranged from 12 to 896 patients in the open-label trials and 135 to 464 patients in the RCTs. Follow-up durations of RCTs ranged from 16 to 28 weeks. The results of the quality assessment were only reported for the four RCTs. The concealment of allocation was unclear in all four RCTs; the random allocation of patients and patient blinding were judged to be adequate in all four RCTs.
Statistically significant benefits were observed with treatment with 3mg tacrolimus compared with control treatments for response rates classified as ACR50 (RR 2.583, 95% CI 1.095 to 6.092; I2=66.6%; four RCTs), ACR20 (RR 3.141, 95% CI 2.327 to 4.240; I2=17.3%; four RCTs) and ACR70 (RR 5.618, 95% CI 1.440 to 21.92; I2=26.3%; two RCTs,). Withdrawals due to lack of efficacy were also significantly lower in the tacrolimus group (RR 0.391, 95% CI 0.217 to 0.705, I2=72.7%).
The use of tacrolimus given at doses of 1.5 to 2.0 mg was associated with significant benefits compared with placebo in response rates of ACR50 (RR 2.143, 95% CI 1.097 to 4.187; I2=0%; two RCTs), ACR20 (RR 1.849, 95% CI 1.187 to 2.880; I2=0%; two RCTs) and ACR70 (RR 8.209, 95% CI 1.039 to 64.85; I2=0%; one RCT). There was also a non-significant beneficial difference in withdrawals due to lack of efficacy in the tacrolimus group.
At both dose levels, there were significant improvements for the patients who received tacrolimus for all clinical outcomes related to rheumatoid arthritis disease activity, but there were no differences between the tacrolimus and control groups in physical function. The results did not vary significantly from the main analysis of tacrolimus for results stratified by methodological quality criteria.
In the four open-label trials, tacrolimus appeared to be well-tolerated, safe and was associated with clinical benefits.
The incidence of adverse events was higher for the tacrolimus-treated groups than the control groups (RR1.214, 95% CI 1.066 to 1.384).
There was no evidence of publication bias shown in the Egger's test; however, as the authors acknowledged, the number of studies included in the analysis was insufficient for these analyses.