Department of Nephrology, University of Queensland at Princess Alexandra Hospital, Brisbane, Australia.
Ther Drug Monit. 2010 Dec;32(6):723-33. doi: 10.1097/FTD.0b013e3181fc8fbb.
Multiple limited sampling strategies (LSSs) have been proposed for estimation of mycophenolic acid (MPA) area under the concentration-time curve from 0 to 12 hours postdose (AUC 0-12) after mycophenolate mofetil intake. The aim of this study was to provide summary information on all published LSSs for MPA and to evaluate their predictive performance in an independent population of kidney transplant recipients. Seventy-eight LSSs for MPA were identified. Sixty-nine full AUC profiles were collected from 45 subjects (25 cotreated with cyclosporine and 20 with tacrolimus). Predicted MPA AUC 0-12, calculated by applying the relevant concentration measurements within the LSS equations, was compared with full AUC calculated by using all concentration measurements in the linear trapezoidal rule. Four error indices (median prediction error, median percentage prediction error [MPPE], root median squared prediction error, and median absolute percentage prediction error [MAPE]) were used to evaluate bias and imprecision. Twelve of the 25 LSSs for cyclosporine-cotreated recipients and one of the 53 LSSs for tacrolimus-cotreated recipients displayed acceptable (less than 15%) bias and imprecision. In the cyclosporine group, two equations demonstrated the highest predictive power, one that used four time points in the first 6 hours postdose (r2 = 0.84, MPPE 1.6%, MAPE 7.8%) and one that used four time points in the first 4 hours postdose (r2 = 0.76, MPPE -0.8%, MAPE 10.2%). In the tacrolimus group, an equation that used two time points in the first 4 hours postdose was superior (r2 = 0.80, MPPE -3.0%, MAPE 13.6%). Application of the LSSs most appropriate for cyclosporine-cotreated patients to the tacrolimus-cotreated group resulted in clinically unacceptable bias and imprecision and vice versa. High variability in performance of LSSs highlights the importance of validating any LSS before applying it to an alternative population. Attention to comedication use is of particular relevance when choosing a LSS.
已有多种有限采样策略（LSS）被提出，用于估算霉酚酸（MPA）在霉酚酸酯给药后 12 小时内的浓度-时间曲线下面积（AUC 0-12）。本研究旨在提供所有已发表的 MPA LSS 的总结信息，并在独立的肾移植受者群体中评估其预测性能。共鉴定出 78 种 MPA LSS。从 45 名受试者（25 名环孢素合用，20 名他克莫司合用）中采集了 69 个完整的 AUC 谱。通过应用 LSS 方程中的相关浓度测量值来计算预测的 MPA AUC 0-12，并与使用线性梯形规则的所有浓度测量值计算的全 AUC 进行比较。使用四个误差指数（中位数预测误差、中位数百分比预测误差[MPPE]、根均方预测误差和中位数绝对百分比预测误差[MAPE]）来评估偏差和不精密度。25 名环孢素合用受者的 25 个 LSS 中有 12 个和 53 个他克莫司合用受者的 53 个 LSS 中有 1 个显示出可接受的（小于 15%）偏差和不精密度。在环孢素组中，两个方程显示出最高的预测能力，一个是在给药后 6 小时内使用 4 个时间点（r2=0.84，MPPE 1.6%，MAPE 7.8%），另一个是在给药后 4 小时内使用 4 个时间点（r2=0.76，MPPE-0.8%，MAPE 10.2%）。在他克莫司组中，使用前 4 小时内 2 个时间点的方程更优（r2=0.80，MPPE-3.0%，MAPE 13.6%）。将最适合环孢素合用患者的 LSS 应用于他克莫司合用患者组会导致不可接受的临床偏差和不精密度，反之亦然。LSS 性能的高度变异性突出表明在将其应用于替代人群之前验证任何 LSS 的重要性。在选择 LSS 时，需要特别注意合并用药的使用。
