Meta-analysis should not have included one of the studies

EDITOR—Several points relating to Knoll and Bell's meta-analysis of randomised trials comparing cyclosporin (Sandimmun) and tacrolimus may have a bearing on its conclusions.1 Unfortunately, the first study cited—that by Shapiro et al—does not have a minimum of one year's follow up; the “median follow-up is 1.12 years, with a range of 0.17 to 2.25 years.”2 This trial by Shapiro et al should also be excluded on the grounds that it is not a randomised trial but compares a group of patients treated with tacrolimus with “a nearly concurrent group of patients treated with cyclosporine.”

Drawing conclusions is also difficult because two of the studies routinely used antibody induction therapy and a third study did not detail whether it did or not. This is a practice not commonly adopted in the United Kingdom. Furthermore, the drug doses, routes of administration, drug monitoring, and treatment of rejection episodes differed considerably between studies. The fact that three of the studies were multicentre trials with allowances for local protocols leads to further potential variation.

All four studies evaluated the old cyclosporin preparation Sandimmun, which is no longer available in the United Kingdom. Its successor, Neoral, has improved absorption and a lower acute rejection rate than Sandimmun, so that any direct comparisons between the formulations are impossible.

In an attempt to compare cyclosporin (Neoral) and tacrolimus (Prograf) we are conducting a single centre randomised controlled trial for patients undergoing cadaveric renal transplantation. We have so far recruited 200 patients, 179 of whom (90 treated with tacrolimus and 89 treated with cyclosporin) have been followed up for a minimum of six months (median follow up 25 months). The acute rejection rate was 32% for patients treated with tacrolimus and 39% for patients treated with cyclosporin (NS).

Serum …