A statistical evaluation of baseline and follow-up carcinoembryonic antigen in patients with resectable colorectal carcinoma.

A long-term CEA follow-up was evaluated statistically for a series of 74 patients with primary colorectal carcinoma who underwent resections for cure. Thirty-three recurrences and 29 deaths were reported among this population after a median follow-up interval of 55 months. Preoperative CEA levels correlated with the Dukes' classification. However, the preoperative level adds significant information to Dukes' classification in the prediction of recurrence. Postoperative CEA assays taken later in the clinical evaluation process carry the most prognostic information for subsequent recurrence. Using matched-pairs techniques, the authors noted that CEA often rose in anticipation of recurrence, sometimes as early as one year before recurrence. The following four follow-up CEA events were evaluated: 1) three consecutive rising CEA's 2) CEA rises 5.0 ng/ml; 3) CEA exceeds 5.0 ng/ml; 4) CEA exceeds 5.0 ng/ml two consecutive times. The follow-up CEA events examined were characterized by false positive rates ranging from 17% to 66% and true positive rates ranging from 43% to 56% in the prediction of recurrence. From follow-up CEA evaluations, further clinical investigation appeared most warranted when CEA exceeds 5.0 ng/ml at two consecutive follow-up evaluations, but the chances of finding any subsequent recurrent disease was near 50%. Among patients experiencing recurrences, although forewarned by CEA elevations, there was no subsequent time when patients were most likely to have recurrence of disease.