Background: Estrogens are responsible for the development of the secondary female gender characteristics. Serum estradiol concentration measurements depend widely on automated immunoassay systems such as the Architect i2000SR, Cobas e601, and UniCel DxI 800. Here, we compare the performance of three automated immunoassay systems and investigate the correlation of serum estradiol levels.
Methods: The precision of the three automated immunoassay systems were evaluated by the Bio-Rad Laboratories Quality Control products according to the EP15-A2 document. Measurements of 81 serum samples from routine clinical specimens were completed on the Architect i2000SR, Cobas e601, and UniCel DxI 800 on the same day in our laboratory. Method comparison was performed using Bland-Altman and Passing-Bablok analyses. The correlation between the three detection systems was analyzed using the concordance correlation coefficient r.
Results: With the Bio-Rad Laboratories Quality Control products, we clearly demonstrated that the total inaccuracy of the Architect i2000SR detection systems is in the range of 1.61 - 5.64, the Roche Cobas e601 inaccuracy is in the range of 1.88 - 6.56, and the inaccuracy of the UniCel DxI 800 is in the range of 4.69 - 8.33. A Bland-Altman plot showed that serum estradiol concentrations determined by the Architect i2000SR were about 0.69 times those of the Roche Cobas e601, the Architect i2000SR concentrations were about 0.65 times those of the UniCel DxI 800, and the Roche Cobas e601 concentrations were about 0.96 times those of the UniCel DxI 800. The correlation coefficient r was 0.9936, 0.9857, and 0.9774 for the Architect i2000SR versus the Roche Cobas e601, the Architect i2000SR versus the UniCel DxI 800, and the Roche Cobas e601 versus the UniCel DxI 800, respectively.
Conclusions: The Architect i2000SR system had better precision. The three detection systems have good correlation with each other, but there is a large gap between the results of their detection of estradiol.