Performance evaluation of nine commonly used urine albumin assay systems / 中华检验医学杂志

ABSTRACT

Objective By investigating precision,linearity and accuracy of 9 commonly urine albumin assay systems (8 of immuno-turbidimetric assays and 1 of immuno-nephlometric assay),and comparing the concordance of measurement results,to elucidate the quality of the existing analytical systems.Methods Referring to Clinical and Laboratory Standards Institute (CLSI) EP15-A2,two mixed urines with U-Alb levels of 20 mg/L and 200 mg/L were made to validate precision; Referring to CLSI EP14-A2,fourty fresh urines were selected to evaluate matrix effect of saline diluted European Reference Materials (ERM) DA 470 and saline diluted urine,also to reflect the variation of measurement results among systems; Referring to EP6-A,saline diluted urines (10 levels) were made to validate linearity; Taking the theoretical concentration of precisely saline-diluted ERM-DA 470 as the target value,accuracy of each assay system was evaluated.Maximal allowable coefficient variation (CV) of ≤ 15％ was taken as the acceptable precision for each assay system,as rccommcnded by International Federation of Clinical Chemistry (IFCC)-and National Kidney Disease Education Program (NKDEP) ; maximum allowable bias of ≤25％ was taken as criteria for accuracy evaluation as used in Proficiency Test (PT) sponsored by College of American Pathologists (CAP).Results At level of micro-albuminuria(20-200 mg/L),all 9 systems total CVs were ≤ 15％ ; No matrix effect or interference were found in saline diluted ERM DA 470 and saline diluted urine.For A,B,E,F,G and I systems,validated linear regions were close to those stated in kit instruction;For C,D and H systems,the lower limits of validated linear region (18.7,3.6 and 12.0 mg/L,respectively) were higher than those stated in kits instruction (0,0.9 and 5.0 mg/L,respectively) ;For B and C systems,the lower limits of validated linear region were close to the upper limits of reference interval stated in kit instruction.When urine albumin was ≤ 12.6 mg/L,A,E,F,G and I systems showed good accuracy,absolute biases at all dilution were below 3 mg/L,D system showed higher positive bias (5.0-14.4 mg/L),B,C and H systems' biases were not evaluated because of high in-batch CV (the CV of B system≥ 18.1％,of C system ≥ 14.5％,of H system ≥ 39.1％); when U-Alb ranged in 25.2-201.0 mg/L,all 8 systems' relative biases were ≤25％,except D systems,which showed an un-acceptable positive bias (15.9％-44.3％).Good concordance among systems' results was present at level of microalbuminuria(20-200 mg/L),with CV among systems ＜ 15％ ;when urine albumin was ＜ 20 mg/L,CV among systems increased as allumin concentration decreased.The main contribution of variation came from B,C and H systems,which lower limits of linearity were relatively high.Conclusions At level of microalbuminuria(20-200 mg/L),except D system,the other 8 systems show good precision and accuracy;at low level of urine albumin(＜20 mg/L,especially ＜ 10 mg/L),precision and accuracy of some systems(B,C and H system) needs to be improved.