Estimation of lithium clearance from routine clinical data in Egyptian bipolar patients. A population pharmacokinetic approach.

Population pharmacokinetics (PK) of lithium as a mood stabilizer was investigated in Egyptian patients with bipolar affective disorders (n = 50) of whom 31 were suffering from lithium toxicity. The mean (+/- SD) age and body weight of patients were 33 +/- 10 years and 67 +/- 3.6 kg, respectively. Patients selected were maintained on lithium carbonate controlled release tablets at doses of 400 mg/12 hours (n = 43) or 200 mg/12 hours (n = 7) respectively. In 19 patients who continued lithium therapy, 1 blood sample/patient was withdrawn for lithium level determination before the morning dose of the drug was given while for 31 patients who suffered from lithium-related toxicity and cessation of drug intake was therapeutically decided, a single blood was drawn at variable time (36, 48 or 72 h) following the last administered dose of the drug. The data was subjected to population PK analysis using NONMEM and a two-compartment model was used. Due to single point sparse data, not all parameters and their between subject variability (BSV) could be determined. Therefore, lithium clearance (CL) and BSV were estimated while other PK parameters were fixed using available literature information. First order (FO) estimation method was used in the analysis. Covariates were evaluated by univariate analysis using likelihood ratio test. The most significant covariate on lithium CL was found to be creatinine clearance (CrCL). The population CL of lithium in the final model was expressed as CLpop = 0.51 x (CrCL/105.3)0.44. The final population PK parameters estimates of lithium were: CL = 0.51 l/h with 12.7% BSV, V1 (Fixed) = 15.2 l, Q (Fixed) = 7.44 l/h, and V2 (Fixed) = 6.7 l. The mean value of lithium concentration at 12 hours as predicted by the final model in the patients with drug toxicity was 1.3 +/- 0.1 mmol/l versus 0.8 +/- 0.14 mmol/l in patients without toxic signs. External validation of the final model on another group of adult bipolar patients (n = 12) maintained on lithium therapy showed a predictive ability of -35 to 65% as represented by% error for the predictions.