| Objective:1. To assess the external predictive performance of a previous published carbamazepine population pharmacokinetic model through calculating the accurary, precision and consistency of predictive concentration and observed concentration.2. To assess the relationships between predictive concentration and seizure frequency and adverse effects.Methods:1. To assess the external predictive performance of a previous published carbamazepine population pharmacokinetic modelThe patients were involoved this stuty incluing who treated in epilepsy clinics of the department of Neurology of Huashan Hospital, choosen carbamazepine or combined other anti-epileptic drugs according to seizure type. After ethics and informed consent, we collected the data of demographic, medical history, medicine information et al., and then we obtained the predive concentration of carbamazepine based on the prevision population pharmacokinetics model.On the premise of ensuring patients medication adherence (i.e. medication education, Morisky Medication Adherence Scale was used for evaluating medication adherence after 2 to 4 weeks), patients were asked to extract 2ml blood sample on an empty stomach when the concentration of carbamazepine had reached steady state (5 to 7 half-life). And we obtained the observed concentration of carbamazepine using the fluorescence polarization immunoassay method.Mean prediction error (accurary) and mean absolute prediction error (precision) of the predictive concentration and the observed concentration were calculated for evaluating the external predictive performance of a previous population pharmacokinetics model. In the clinical setting, it was considered acceptable when the mean absolute prediction error≤20%.The consistency of the predictive concentration and observed concentration was assessed using the Bland-Altman difference plot. And the 95% consistency limit as measurement was calculated. It was considered had good consistency and could be used interchangeably when data points outside the 95% consistency limit and the differences of two methods within the 95% consistency limit were acceptable in clinical setting.2. The relationships between concentration (observed concentration and predictive concentration) and sizure frequency and adverse effectsThe selected patients with epilepsy were recorded seizure frequency using epilepsy diary. The patients were asked to assess the adverse effects using Liverpool Adverse Events Profile after 2 to 4 weeks. And the relationships between concentration and seizure frequency and adverse effects were examinned using Pearson correlation analysis.Results:1.121 patients were eligible. Final model of the mean prediction error and mean absolute prediction error were -4.76% (-6.18% for mono- and -2.74% for poly-therapy) and 14.93% (15.01% for mono-and 14.80% for poly-therapy), whereas absolute prediction error less or equal to 20% was 74.38% (76.06% for mono- and 72.00% for poly-therapy). Bland-Altman plot showed that 4.13% (7.04% for mono- and 0.00% for poly-therapy) patients were not within 95% confidence interval; and within 95% confidence interval, the maximum bias of the observed and predictive concentration was 2.44μg/mL (2.44μg/mL for mono-and 2.4μg/mL for poly-therapy).2. A positive relation was found between predictive concentration and seizure frequency (r= 0.132, p= 0.001) and adverse effects (r= 0.236, p= 0.009).Conclusion:1. The previous published carbamazepine population pharmacokinetics model had less bias and better precision, regardless of mono-therapy or poly-therapy. Being a valuable tool, it permitted to forecast the carbamazepine concentration before a dosage given in individual patient and ultimately individualize therapy was conducted.2. A relation was found between predictive concentration and seizure frequency and adverse effects, and with the increase of carbamazepine predictive concentration, the seizure frequency and adverse effects increased. The finds could help to provide data for clinical application of this model. |
PDF Full Text Request