Differences of prostate-specific antigen assays: a small light at the end of the tunnel?
| Autor: | Davide Giavarina, Paolo Rizzotti, Sandra Meneghelli, Romolo M. Dorizzi |
|---|---|
| Rok vydání: | 2006 |
| Předmět: | |
| Zdroj: | Annals of Clinical Biochemistry: International Journal of Laboratory Medicine. 43:420-421 |
| ISSN: | 1758-1001 0004-5632 |
| Popis: | Roddam et al. convincingly and elegantly investigated the diierences of 11 analytical methods for prostatespeci¢c antigen (PSA). The authors concluded, on the basis of the results obtained in 15 samples sent to 223 laboratories in England, that analytical methods for PSA show unacceptable inter-method diierences and some methods have unacceptable estimated biases. Even if the industry rapidly adopted the Standford University calibrators eliminating the most extreme cases of bias, Roddam et al. found an estimated positive bias of Beckman Hybritech and Immulite 2000 Third generation PSA, respectively, around 20% and10%. Notwithstanding the claim of the manufacturers that their PSA assays are calibrated against the PSA reference material, the inter-assay variability has not been eliminated, the interchangeability of the results is inadequate and the goal of interchangeable PSA has not been reached. In a preliminary evaluation of the recently marketed automated chemiluminescent analyser Unicel DxI 800 Immunoassay System (Beckman Coulter, Brea, CA, USA), we measured PSA in 64 randomly selected samples (concentration: 0.01--33.0 mg/L) using both the Hybritech PSA-Hyb reagents with Unicel DxI 800 analyser andThird-generation PSA reagents, and Immulite 2000 analyser (DPC, Los Angeles, CA, USA); calculations were carried out using MedCalc Version 8.1 (Mariakerke, Belgium). DxI results ranged from 0.01 to 35.4 mg/L, with mean 5.3778.50 mg/L (95% con¢dence interval [CI]: 3.25--7.49) and median 1.86 mg/L (95% CI: 0.64--4.67). Immulite results ranged from 0.01 to 34.2 mg/L, mean 5.4078.71 mg/L (95% CI: 3.23--7.58), median 1.89 mg/L (95% CI: 0.55--4.47). The Passing Bablok regression showed an intercept of 0.0001 (95% CI: 0.0004--0.0003) and a slope of 0.9879 (95% CI: 0.9738--1.0171). The Bland Altman plot (Figure 1) eiectively demonstrates the good consistency and the lack of bias between the two methods along the entire spectrum of concentrations.The ¢gure shows that 95% of the diierence of the results were between 26.6% and 26.9% with a mean diierence of 0.2% (95% CI: 3.1--3.5). In conclusion, while the Hybritech reagents processed with the previously available platforms yielded results with relevant bias as recently pointed out also by Stephan et al., our ¢ndings suggest that the industry is taking in account the profession’s recommendations and is improving the performance of PSA assays. It seems that, in some cases, the manufacturers are re-calibrating their assays; therefore, it is mandatory that the laboratories compare very carefully the performance of the reagents used with diierent analysers and indicate the method used on the report. The goal of interchangeable PSA results has not yet been achieved but the ‘joint venture’of industry and laboratory is improving things. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|