Background: CEA, CA 19-9, CA 242 and CA 72-4 are commonly used tumour markers for gastrointestinal malignancies. The advantage of the concomitant use of these markers is under debate.
Materials and methods: Serum concentrations of the markers were measured at the time of diagnosis in 161 patients with benign and 125 with malignant gastrointestinal diseases. Concomitant use of the markers was evaluated in a logistic regression model.
Results: CA 19-9, CA 242 or CA 72-4 showed similar sensitivity of 44% for gastric cancer, whereas CEA was elevated in 25% of the cases. In patients with colorectal cancer, CEA was most frequently elevated (54%), followed by CA 242 (46%), CA 19-9 (36%) and CA 72-4 (25%). High CA 19-9 and CA 242 serum levels were frequent in patients with cholangiocarcinoma (86% and 68%, respectively) and pancreatic cancer (80% and 63%, respectively). In the benign disease group, serum CA 19-9 was most frequently elevated, i.e. in 24%, 25% and 38% of patients with pancreatic, biliary and liver disorders, respectively. The overall accuracy of CEA, CA 19-9, CA 242 and CA 72-4 was 66%, 71%, 71% and 66%, respectively (p > 0.18). When combined in a logistic regression model, CA 72-4, CA 19-9 and CEA provided independent diagnostic information, whereas CA 242 contributed with independent diagnostic information only on excluding CA 19-9. The probability of cancer for each patient, calculated with the model, was applied as a diagnostic test and was compared with the single markers by ROC-curve analysis. The AUC value of the probability index was significantly higher than the values of the different tumour markers.
Conclusion: An algorithm based on the combination of CEA, CA 19-9 and CA 72-4 improved the diagnostic accuracy in gastrointestinal tract malignancies compared with these markers alone.