Evaluation of five immunoturbidimetric assays for urinary albumin quantification and their impact on albuminuria categorization

Abstract

ObjectivesThe study was designed to evaluate the performance of five automated immunoturbidimetric assays to quantify urinary albumin, each corresponding to a combination of a reagent and an analyzer (Olympus on AU640®, Roche on Cobas Integra, Abbott on Architect, Ortho-Clinical Diagnostics Vitros on Fusion and Siemens on Dimension). Design and methodsTo assess imprecision, albumin was measured in three urinary pools with a mean value of 25, 66 and 131mg/L. One hundred and eight patient urine samples were then used to compare each turbidimetric method using the Passing–Bablok regression and Bland–Altman analyses. Concordance of the albumin/creatinine ratio (ACR), according to the albuminuria classifications proposed by the KDIGO, was calculated to test the agreement between the different assays. ResultsAll immunoturbidimetric methods evaluated in this study exhibited acceptable imprecision (CV<6%). Mean values for 108 urine samples varied from 0.5 to 762.2mg/L. Significant differences were found (p<0.05) between all methods except between Olympus and Ortho (p=1.0) and between Abbott and Roche (p=0.12). Regarding the albuminuria categories based on the ACR proposed by the KDIGO, only the classification obtained with the Roche method was significantly different from the four other methods (p<0.001). ConclusionsWe demonstrated that all assays were not strictly equivalent which could affect ACR categories in clinical practice, suggesting the need for harmonization of commercial methods.