Comparison of the diagnostic accuracy of CA27.29 and CA15.3 in primary breast cancer

Background: A new, fully automated method that measures the breast cancer-associated glycoprotein CA27.29 has become commercially available. The aim of the present study was to compare this CA27.29 assay with the assay that measures CA15.3 in primary breast cancer. Methods: The study was performed retrospectively on preoperative serum samples collected from 275 patients with untreated primary breast cancer (154 node positive and 121 node negative). Eighty-three healthy control subjects were also evaluated. CA27.29 was measured using the fully automated Chiron Diagnostics immunochemiluminescent system (ACS:180 BR). CA15.3 was measured with a manual immunoradiometric method (Centocor CA15.3 RIA). Results: In healthy subjects, CA15.3 was significantly higher than CA27.29 (P <0.0001). On the other hand, in breast cancer patients CA27.29 was higher than CA15.3 (P = 0.013). The mean value found in the control group plus 2 SD was chosen as the positive/negative cutoff point. The overall positivity rates were 34.9% for CA27.29 and 22.5% for CA15.3. The area under the ROC curve was greater (P <0.001) for CA27.29 (0.72) than for CA15.3 (0.61). Both markers showed a statistically significant, direct relationship, with pathological stage being higher in node-positive than in node-negative cases and in larger than in smaller tumors. Neither CA27.29 nor CA15.3 showed significant associations with age, menopausal status, or tumor receptor status. Conclusions: CA27.29 discriminates primary breast cancer from healthy subjects better than CA15.3, especially in patients with limited disease. Prospective studies are necessary to confirm this conclusion. (C) 1999 American Association for Clinical Chemistry.