Validating a custom multiplex ELISA against individual commercial immunoassays using clinical samples

• Published in: BioTechniques Vol. 42; no. 3; pp. 327 - 333
• Main Authors: Liew, Michael, Groll, Matthew C, Thompson, James E, Call, Sara L, Moser, Joann E, Hoopes, Justin D, Voelkerding, Karl, Wittwer, Carl, Spendlove, Rex S
• Format: Journal Article
• Language: English
• Published: England Taylor & Francis Group 01-03-2007
• Subjects: alpha-Fetoproteins - metabolism; Carcinoembryonic Antigen - biosynthesis; Enzyme-Linked Immunosorbent Assay - methods; Humans; Immunoassay - methods; Immunoenzyme Techniques - methods; Models, Chemical; Models, Statistical; Prostate-Specific Antigen - biosynthesis; Regression Analysis; Reproducibility of Results; Sensitivity and Specificity
• ISSN: 0736-6205; 1940-9818
• DOI: 10.2144/000112332

The measurement of multiple antigens in a single sample poses clinical and methodological challenges. Here we describe the validation of a multiplexed sandwich enzyme-linked immunosorbent assay (ELISA) array (microELISA) of nine antigens. The antigens tested simultaneously were: alpha-fetoprotein (AFP), prostate specific antigen (PSA), carcinoembryonic antigen (CEA), cancer antigen 125 (CA 125), CA 15-3, CA 19-9, beta-human chorionic gonadotropin (beta-hCG), luteinizing hormone (LH), and follicle stimulating hormone (FSH). At least 44 clinical samples were tested for each antigen. microELISA results for the nine antigens were then compared with clinical laboratory results obtained for the same antigens in individual chemiluminescent immunoassays. The microELISA had a coefficient of variation (cv) of 7.3% within an assay and 12.6% for assays run at different times. A statistical comparison of results from the microELISA with results from the clinical laboratory showed that the assays had correlation coefficients ranging from 0.99 to 0.76, and Deming regression demonstrated that four of the nine assays were high-quality assays and not statistically different to the individual assays. To determine if the differences in the assays were due to methodology, the microELISA was also compared with conventional ELISAs using identical antibodies and reagents. Deming regression demonstrated that five of the eight assays were high-quality, indicating that a poor correlation between a microELISA and an individual immunoassay are partly due to antibody differences.