Performance characteristics of seven automated CA 15-3 assays.

Measurements of serum cancer antigen (CA) 15-3 are used to monitor tumor recurrence and treatment of advanced disease. We evaluated the performance characteristics, including limit of detection, linearity, method comparison, and reference intervals, of 7 automated methods for CA 15-3, including the Access 2 (Beckman Coulter, Brea, CA), ADVIA Centaur (Bayer Diagnostics, Tarrytown, NY), ARCHITECT i2000 and AxSYM (Abbott Diagnostics, Abbott Park, IL), Elecsys 2010 (Roche Diagnostics, Indianapolis, IN), IMMULITE 2000 (Diagnostic Products, Los Angeles, CA), and VITROS ECi (Ortho Clinical Diagnostics, Raritan, NJ) assays. The limit of detection for each assay was less than 1.0 kU/L. The maximum deviation for the target values for linearity samples was less than 10% for all methods. Method comparison studies revealed large differences for some individual samples. Overall slopes ranged from 0.50 to 1.48, and correlation coefficients were 0.90 to 0.96 when the ADVIA Centaur was the comparison method. The 97.5 percentile upper reference limit ranged from 23.3 to 51.7 kU/L. Additional standardization efforts are needed, and the availability of reference material is required. Substantial intermethod differences exist for some patient samples, indicating that redetermining the baseline is required when changing methods.