Decision support tool to individualize cyclosporine dose in stable, long-term heart transplant recipients receiving metabolic inhibitors: overcoming limitations of cyclosporine C2 monitoring.

BACKGROUND

Monitoring of the 2-hour post-dose sample (C(2)) for cyclosporine (CsA) has gained favor; however, choosing a single-point surrogate marker of therapeutic effect for a drug with extensive pharmacokinetic variability is problematic and has limitations.

METHODS

A Bayesian decision support tool was developed using published pharmacokinetic information implemented using ABBOTTBASE pharmacokinetic software. The model was evaluated in 47 stable heart transplant recipients who received concomitant administration of drugs known to inhibit CsA metabolism: diltiazem; ketoconazole; and a combination of diltiazem and ketoconazole.

RESULTS

A 3-point feedback strategy with samples collected at 0, 1 and 2 hours after an oral CsA dose was used to predict area under the concentration-time curve in the first 12 hours post-dose (AUC(0-12)). In Group A, patients who received CsA alone showed a good correlation between observed and model-predicted CsA AUC(0-12) (r(2) = 0.871, p < 0.001, precision of 12%, accuracy of 13%). Furthermore, the Bayesian model provided acceptable predictions in patients who received CsA with metabolic inhibitors: Group B (diltiazem), r(2) = 0.791, p < 0.001, precision of 19%, accuracy of 22%; Group C (ketoconazole), r(2) = 0.761, p < 0.001, precision of 4%, accuracy of 12%; and Group D (diltiazem plus ketoconazole), r(2) = 0.818, p < 0.001, precision of 14%, accuracy of 17%.

CONCLUSIONS

A Bayesian decision support tool is described that can predict CsA AUC(0-12) in a cohort of patients with variable CsA absorption who received metabolic inhibitors. Bayesian modeling offers a number of advantages over single point metrics that are used to adjust CsA dose and may provide another refinement to optimize CsA therapy.