Some limitations of using equilibrium dialysis to study human serum albumin-testosterone interaction.

The apparent association constant of testosterone binding to pure human serum albumin was found to be significantly less at 5 g/dl (1.79 x 10(4) M-1) than at 1 g/dl (2.29 x 10(4) M-1). This demonstrates that the common assumption that binding site characteristics are constant with serum protein dilution is not strictly valid. Determination of the percent free testosterone at normal serum protein concentrations by equilibrium dialysis may also be in error by 20% or more due to fluid shifts within the test system.