Evaluation of the human serum albumin column as a discovery screening tool for plasma protein binding.

A total of 69 compounds with a variety of chemical structures were assayed using a human serum albumin column in combination with UV and mass spectrometric detection. A moderate correlation, R(2)=0.661, between the plasma protein binding, determined by traditional techniques of equilibrium dialysis or ultrafiltration, and chromatographic retention factor (k'/k'+1) was observed. Disparity between the regression line and numerous samples was observed across the entire range of plasma protein binding. Attempts to discriminate between compounds from the data set to achieve better correlation based physico-chemical properties were unsuccessful. Good agreement was observed for retention times obtained with UV detection with mobile phase containing phosphate buffer and mass spectrometric detection with mobile phase containing acetate buffer. Essentially identical data were obtained for compounds analyzed in singlet or cassette for minimally or highly bound (>90% bound) compounds. Analysis of cassettes containing compounds with plasma protein binding greater than 90% did not cause column overload, even at analyte concentrations up to 100 microg/ml. Diverse results were obtained when chromatographic retention was used to rank order various classes of compounds. Better correlation with ordering from known binding was obtained when a compound class contained a wide range of protein binding, in contrast to when compounds within a given class were all highly bound.