COMPLEX ANALYTE-DEPENDENT AND ANALYTE-INDEPENDENT INTERFERENCES WITH CONJUGATED BILIRUBIN IN THE ENZYMATIC PHENOL-AMINOPHENAZONE PEROXIDASE(PAP) METHOD FOR CREATININE DETERMINATION

Although bilirubin interferes with the enzymatic assays for creatinine , neither a consensus of the degree of interference nor the mechanism has been established. Using multiple regression analysis, we demonstra te that the interference is negative and caused by both analyte-depend ent and analyte-independent mechanisms. Furthermore, the correlative m odel includes terms non-linear with respect to creatinine. In the kine tic creatinine phenol-aminophenazone peroxidase method, there are anal yte-dependent and analyte independent mechanisms at work. The multivar iate equation is: Crea' = 0.9879 Crea - 0.4524 Bili - 0.000828 Crea x Bili + 2.094 x 10(-7) Crea(2) x Bili + 5.0 (Crea' = measured creatinin e (mu mol/l), Crea = true creatinine (mu mol/l), and Bili = conjugated bilirubin (mu mol/l)). The endpoint mode was affected less than the k inetic mode and exhibited different relationships in which two models describe the interference equally well. One is strictly analyte-depend ent: Crea' = 0.9991 Crea - 0.00203 Crea x Bili + 2.390 x 10(-6) Crea(2 ) x Bili - 1.464 x 10(-9) Crea(3) x Bili + 3.261 x 10(-13) Crea(4) x B ili - 9.9. The other is a complex combined analyte-dependent and analy te-independent: Crea' = 0.9834 Crea - 0.00680 Crea x Bili + 2.477 x 10 (-7) Crea(2) x Bili - 3.233 x 10(-7) Crea x Bili(2) + 0.4652 Bili - 0. 000458 Bili(2) + 12.2. These models are valid for creatinine concentra tions up to 2200 mu mol/l (24.9 mg/dl) and bilirubin up to 660 mu mol/ l (38.6 mg/dl). The interference increases with increments of either b ilirubin or creatinine. In addition, we found that unconjugated biliru bin interferes differently from conjugated bilirubin in degree and mec hanism. Model building, contour plots, surface plots, and possible mec hanisms are discussed. We propose multiple regression analysis as the proper way to evaluate interferences because analyte-dependence can be easily missed by simple regression analysis. True creatinine concentr ations can be estimated despite the interference from conjugated bilir ubin. Other phenol-aminophenazone peroxidase methods may be similarly affected.