Multicomponent analysis for alkaline phosphatase isoenzyme determination by multiple linear regression.

Alkaline phosphatase (EC 3.1.3.1) isoenzymes in serum may be determined by multicomponent analysis of the enzyme activities in the presence of multiple inhibitors. To determine inhibition coefficients of the isoenzymes, we used multiple linear regression analysis to compare alkaline phosphatase activities in the presence of known inhibitors with electrophoretically determined isoenzyme activities in plasma and serum samples. All possible combinations of exactly determined and overdetermined linear systems of inhibitors were ranked according to their prediction error to select an optimum set. The best multicomponent system for prediction included the use of levamisole, phenylalanine, and heat inhibition at 56 degrees C and 65 degrees C to determine bone, hepatic, intestinal, and placental isoenzymes. Consideration of the hepatic isoenzyme as liver and macromolecular fractions resulted in significantly worse predictions. Error analysis involving repeat determinations and a simplex optimization of the inhibition coefficients indicated that the inaccuracy of the comparison electrophoretic method may have been a major factor affecting poor isoenzyme prediction in some samples.