New procedure for analysing drug binding data, exemplified by warfarin binding to human serum albumin.

Determination of binding constants for multiple binding of a ligand usually results in highly variable figures. We have found that the variations depend mainly upon cooperativity of ligand binding, and that cooperativity is generally absent on binding to human serum albumin. When this is taken into account it becomes possible to obtain binding constants with only slight variation. A computerized curve fitting procedure for analysing binding data has been established consisting of the following steps. (1) Fitting of Scatchard's equation to observed binding equilibrium data to obtain a best-fit set of Scatchard binding constants. (2) Repetition of the fitting procedure, not to obtain a best fit but to generate 30 acceptable sets of Scatchard binding constants. (3) Fitting of Adair's equation to the observed points to obtain a best fit. If the sum of weighted and squared deviations is significantly smaller than the fitting of Scatchard's equation, cooperativity should be considered. If not, cooperativity cannot be demonstrated and the binding constants obtained by fitting Scatchard's equation can be accepted, with the variations found. (4) Final transformation of all Scatchard constants to Adair's. To illustrate the method warfarin data obtained by equilibrium dialysis was used.