Carcinoembryonic antigen in metastatic colorectal cancer

Objective: To determine the reliability and validity of the carcinoembryonic antigen (CEA) level in following the response of tumours to chemotherapy in patients with metastatic colorectal cancer. At present, CEA has an accepted role in detecting recurrence of colorectal cancer following complete resection. Design: Retrospective case series. Patients: Eighty-one patients with metastatic colorectal cancer seen at the Ottawa Regional Cancer Centre in 1992 whose CEA levels were above the upper limit of normal. Main outcome measures: Change in CEA levels as compared with the tumour response to chemotherapy (as assessed by radiologic studies, physical examination and surgery) and with survival. Results: Over one-half of the CEA levels measured were not consistent with the tumour responses to chemotherapy. The sensitivity of CEA in following the tumour response was 0.54 (95% confidence interval [CI] 0.37 to 0.75) and the specificity was 0.53 (95% CI 0.42 to 0.63). A sample size of 2258 was calculated as necessary to determine the minimum number for establishing the reliability of CEA in measuring tumour response to chemotherapy. Survival was analyzed using the Kaplan-Meier method. CEA was found to have no statistical value in determining survival in this group of patients (p = 0.09). Sample size was calculated for the survival statistics; an adequate sample size was used in this study (the minimum necessary being 20). Conclusion: Neither this study nor a review of the literature supports the use of the CEA level in the setting of metastatic colorectal cancer. Sample size calculations discourage the implementation of a large prospective trial to investigate this topic further.