Long-term results of tacrolimus in cyclosporine-and prednisone-dependent myasthenia gravis

摘要

The article by Ponseti et al. presents an unmasked single group study purporting to show that tacrolimus might replace prednisone and cyclosporine "as a sole immunosuppressiveagent for the treatment of MG." We do not believe the authors present adequate evidence in support of this conclusion, and have a number of specific concerns about the study.1. No protocol was used to establish eligibility and the patient population does not appear to be homogeneous. For example, it is unclear how long patients were on prednisone and cyclosporine, and how much the duration of this therapy varied. Also, the time since thymectomy varied considerably, some patients being within 3 months of surgery, and thymomatous and non-thymomatous myasthenia gravis (MG) are mixed, in contradiction to their reference 9.2. A third of the patients were in "MM-0" status at baseline (taking no MG medications), which contradicts the stated entrance criteria.3. The steps taken to reduce or withdraw prednisone and cyclosporine are not defined.4. Improvement was noted within 10 days, yet other studies do not mention such an early response to tacrolimus. Although this could represent an early therapeutic effect, it is suspect without blinded observations.5. The endpoint was drug dosage reduction, but there was no formal drug withdrawal protocol. Also, one does not know what would have happened if prednisone had been reduced without tacrolimus.6. Unequal follow-up times and use of the last observation for each patient ignore the time course of the disease. Also, only 74 of 79 patients are represented at the final visit, the others presumably not having at least 1 year of follow-up. Two standard study design issues are illustrated here: 1) "intention to treat"-results from all patients given the drug must be analyzed; 2) when calculating event rates for unequal treatment times, life table or adjusted event rates should be used.7. The adverse event rate appears to drop from 6 per patient at baseline to 0.8 at follow-up, but this cannot be determined withoutknowing the time periods of reference. Further, it is not clear how the 0.8 was calculated, since eight patients were withdrawn from therapy.This article illustrates the disadvantages of single group cohort studies in determining therapy for MG and indicates the need for well-designed-controlled prospective studies, as outlined in the MGFA recommendations for clinical research standards in MG and the accompanying, editorial.