Validation of an automated method for the liquid chromatographic determination of atenolol in plasma: application of a new validation protocol

In order to test the applicability of a new strategy by a Commission of the Societe Francaise des Sciences et Techniques Pharmaceutiques (SFSTP) for the validation of bioanalytical methods, an automated method was developed for the determination of atenolol in human plasma. This method was based on the use of dialysis as sample purification step, followed by enrichment of the dialysate on a precolumn and liquid chromatographic analysis. All sample handling operations were carried out automatically by means of an ASTED system. Atenolol and its internal standard (sotalol) were separated on a C-8 column with a mixture of pH 7.0 phosphate buffer containing 1-octanesulphonate and methanol (81/19; v:v) as mobile phase and monitored photometrically at 225 nm. The validation strategy comprises two steps. The experiments performed during the first step, the so called pre-validation step, have permitted the selection of the most appropriate model for the calibration curve by means of a decision tree, i.e. a least squares regression model obtained after transformation of data (square root in this case), the estimation of the limit of quantitation at 25 ng/ml by means of an accuracy profile, the determination of the calibration range (from 25 to 1000 ng/ml), the estimation of the limit of detection at 9 ng/ml and the calculation of the mean extraction efficiency (about 65%). The second step is the validation itself, comprising the evaluation of method selectivity towards endogenous components, the confirmation of the limit of quantitation, the verification of linearity and the assessment of method precision (repeatability and intermediate precision) as well as accuracy using quality control samples at different concentration levels over the range investigated. The relative standard deviation values for repeatability and intermediate precision were between 2.7% and 9.0%. Moreover, the method was found to be accurate. Indeed, the 95% one-sided confidence limits of the mean recovery did not exceed the acceptance limits of 80% and 120%. (C) 1999 Elsevier Science B.V. All rights reserved.