Leave-one-out procedure in the validation of elimination rate constant analysis.

Many registration agencies and other organizations define how to calculate the elimination rate constant (kel) value. No validation procedures have been introduced to verify the correct selection of the concentration-time (C-T) points used for the kel calculation. The purpose of this paper is to discover whether kel analysis can be subjected to the condensed validation procedure and what acceptance criteria should be adopted for such a procedure. For the analysis, data collected during bioequivalence studies of 4 drugs were selected, including 2 highly lipophilic drugs (itraconazole, atorvastatin) and 2 weakly lipophilic drugs (trimetazidine, perindopril). Pharmacokinetic calculations were performed with the use of WinNonlin Professional v 5.3. Internal validation of the kel analysis using leave-one-out cross-validation was performed. The present analysis proves that the C-T selection process for the kel calculations cannot be automated. In each of the analysed data series there were such C-T sequences that did not meet even one of the validation criteria. This paper proposes 3 validation criteria which need to be met in order to confirm the optimal selection of C-T data to calculate kel: Q 2≥0.6, R2≥ 0.85, Q 2-R2<0.3, were Q 2 - squared cross-validated correlation coefficient, R2 - coefficient of determination). Application of the validation procedure for the kel analysis under discussion proves the accuracy of the calculations, even if repeated kel analysis is based on a different sequence of points in the elimination phase.