Evaluation of six erythropoietin kits

Six erythropoietin (EPO) measurement kits from different manufacturers were evaluated for ease of use and performance. Reference values were compared from patient samples with normal and abnormal circulating serum EPO. The kits evaluated were EPO-Trac RIA, EPORIA, DSL RIA, EPO EIA, IBL ELISA and Medac ELISA (see below for manufacturers' details). The radioimmunoassay (RIA) methods were simple to perform with little preliminary preparation of reagents but the non-isotopic methods were of varying degrees of complexity. Imprecision data showed that performance of the RIA methods was acceptable across the range 40-150 mU/mL [within-assay coefficient of variation (CV) = 2-5%, between-assay CV = 3-7%] but poor (CV > 10%) at the lower end of the range (10-15 mU/mL). The IBL enzyme-linked immunosorbent assay (ELISA) kit showed good precision across the range 10-54 mU/mL (within-assay CV = 5-7%, between-assay CV = 8-12%). The performance of the EPO enzyme immunoassay (EIA) and Medac ELISA were acceptable although precision was poor at the lower end of the range (< 10 mU/mL; within-assay CV 10% and 17% and between-assay CV 13% and 18%, respectively). The ELISA and EIA methods had lower limits of detection (0.4-0.8 mU/mL) than the RIA methods (2.3-3.6 mU/mL). Analysis of serum EPO measurement variation between methods showed evidence of significant negative bias with DSL RIA and Medac ELISA when compared with EPO-Trac RIA. Conversely, EPORIA showed significant positive bias and the two remaining methods, EPO EIA and IBL ELISA, showed no evidence of systematic bias when compared with EPO-Trac RIA. Patients with normal circulating concentration of serum EPO gave values that were within the reference range of the kits.