Background: This study was designed to investigate the clinical performance of the Access GI Monitor (Beckman Coulter) on the UniCel DxI 800, a method for CA19-9 antigen determination, and to compare with CA19-9 assay on the AxSYM system (Abbott).
Methods: 1,063 serum samples from unselected patients with different underlying diagnoses were tested with both methods. Passing-Bablok regression analysis and Bland Altman analysis was performed. In addition, using ROC analysis, the distribution of Access GI Monitor and AxSYM CA19-9 antigen levels was tested in patients with pancreatic cancer (n = 50), acute inflammatory disease (n = 20), and with chronic inflammation of the pancreatic gland (n = 18). Furthermore, four patients with pancreatic cancer were monitored individually in their courses of the disease (before, during, and after therapeutic procedures) to compare their CA19-9 values with regard to inter-method concordance.
Results: Passing-Bablok analysis showed a systematic difference with R = 0.93, slope 0.75, and intercept -1.0. Bland Altman analysis showed a wide scatter of relative differences between both methods, especially in the low end measuring range. In the selected group of patients with pancreatic diseases the analysis of concordance revealed 95.5 % agreement between both methods with a comparable area under the ROC curves (0.73 vs. 0.76). A clear concordance was found for all four selected patients.
Conclusions: Although we found significant systematic measuring variations in the global analysis, the two different automated methods for the quantitative determination of CA19-9 antigen were comparable with respect to their clinical accuracy and applicability to support decision making in the management of pancreatic cancer.