Development of retention prediction models for adrenoreceptor agonists and antagonists on a polyvinyl alcohol-bonded...

Retention prediction models based on multiple linear regression (MLR) and artificial neural network (ANN) for adrenoreceptor agonists and antagonists chromatographed on a polyvinyl alcohol-bonded stationary phase under hydrophilic interaction chromatography were described. The models showed the combined effects of solute structure and mobile phase composition on the retention behavior of the analytes. Using stepwise MLR, the retentions of the studied compounds were satisfactorily described by a five-predictor model; the predictors being the %ACN, the logarithm of the partition coefficient (log D), the number of hydrogen bond donors (HBD), the desolvation energy for octanol (FOct), and the total absolute atomic charge (TAAC). The inclusion of the solute-related descriptors suggested that hydrophilic interactions such as hydrogen bonding and also ionic interactions are possible mechanisms by which analytes are retained on the studied system. ANN prediction models were also derived using the predictors derived from MLR as inputs and log k as outputs. The best network architectures were found to be 5-3-1 for the datasets at pH 3.0 and 4.0, and 5-4-1 for the dataset at pH 5.0. The optimized ANNs showed better predictive properties than the MLR models for both training and test sets under all pH conditions.