CEA, CA 242, CA 19-9, CA 72-4 and hCGbeta in the diagnosis of recurrent colorectal cancer.

Objective: The purpose of this study was to compare the utility of serum CEA, CA 19-9, CA 242, CA 72-4 and human chorionic gonadotropin (hCG)beta in the follow-up of 102 surgically treated colorectal cancer patients, out of which 40 patients developed clinical recurrence.

Methods: In patients with recurrent disease, serum samples were obtained at the time of clinical recurrence, and in the disease-free group, they were obtained postoperatively. The combined use of the markers was evaluated with logistic regression analysis. The sensitivities of the different tumour markers at various specificity levels were compared by receiver operating characteristic (ROC) curve analysis.

Results: When the five tumour markers, Dukes stage and location of the primary tumour were evaluated together in the same model, only CEA provided significant diagnostic information (p < 0.0005) in addition to the location of the primary tumour (p = 0.003). The diagnostic information provided by the other serum tumour markers was insignificant, although CA 72-4 approached borderline significance (p = 0.053). ROC curves were constructed and the difference in the values of the area under the curve (AUC) between the different serum tumour markers was determined at the time of clinical recurrence. Of the individual markers, the highest AUC was observed for CEA (AUC = 0.931). The difference in AUC values between CEA and the other tumour markers was highly significant (p < or = 0.001).

Conclusions: CEA had the highest diagnostic accuracy in detecting recurrent colorectal cancer. Inclusion of CA 19-9, CA 242, CA 72-4 or hCGbeta in the model did not improve the accuracy, although CA 72-4 approached borderline significance (p = 0.053). Thus, CEA seems to retain its position as the surveillance marker of choice for patients surgically treated for colorectal cancer.