Diminishing impact of preoperative carcinoembryonic antigen (CEA) in prognosis of Dukes' C colorectal cancer.

BACKGROUND In curable colorectal cancer (CRC), preoperative serum carcinoembryonic antigen (CEA) (preCEA) has been reported to have predictive prognostic value. However, data remains insufficient to support its clinical use. The aim of the current study was to validate the prognostic impact of preCEA in Dukes' C CRC. PATIENTS AND METHODS The prognostic significance of preCEA for 237 Dukes' C CRC patients assessed retrospectively (between 1990 and 2000: previous cases) and the prospective relevance for 197 counterparts (between 2001 and 2004: recent cases) according to preCEA, were both examined. RESULTS The previous cases showed the most potent impact of preCEA as an independent prognostic factor (hazard ratio=2.0, p=0.003) among the clinicopathological factors using a multivariate proportional hazard model, while the recent cases did not even show a univariate prognostic impact. A significant difference in the prognosis between the two periods was only found in the patients with elevated preCEA administered adjuvant chemotherapy (ADT) (p=0.03). Between the two terms, a dramatic change of ADT regimens from 5FU alone (p<0.001) to 5FU in combination with leucovorin (p<0.001) and/or irinotecan (p-0.0009/0.005) was introduced, and N2 patient survival was dramatically improved. However, a significant prognostic difference for the elevated preCEA patients with ADT could not be demonstrated by sub-analysis of N1 and N2 disease due to diminished correlation of preCEA and the N factor (p=0.02 to 0.5), indicating that preCEA did not predict chemosensitivity. CONCLUSION The preCEA is no longer useful in predicting prognosis with Dukes' C CRC patients, because of the loss of preCEA association with the N factor, putatively through undefined diagnostic or therapeutic advancement.