Carcinoembryonic Antigen

Carcinoembryonic antigen (CEA) was first reported in colorectal cancer in 1965.¹⁶⁷ It was detected in embryonic and fetal gut, pancreas, and liver.¹⁶⁸ CEA is a glycoprotein present in the cell membrane.¹⁶⁹ It weighs approximately 200 kilodaltons (kDa).¹⁷⁰ CEA is a member of the immunoglobulin superfamily.¹⁷⁰ Levels of CEA or CEA-like molecules have been detected up to 60 times higher in tumor tissue than in normal healthy tissue.¹⁷¹ The functions of CEA are not well defined. It is postulated that CEA is involved in cell adhesion and in immune surveillance.^170,172

Serum CEA levels can be elevated in ovarian, breast, lung, pancreas, and other gastrointestinal tract cancers.¹⁷³ Not only malignant conditions cause an elevation of CEA; benign conditions, such as liver dysfunction, hydronephrosis, peptic ulcer disease, pancreatitis, biliary obstruction, bowel obstruction, post-5-fluorouracil (5-FU)/-levamisole chemotherapy, and lung cancer, also elevate this marker.¹⁷⁴

In spite of its lack of sensitivity and specificity, CEA is a useful tumor marker in colorectal cancer. It is important to recognize that tumor burden may not correlate with CEA levels.^173,175 However, in general, CEA levels correlate with stages of colorectal cancer.¹⁷⁶ Well-differentiated colorectal adenocarcinomas produce more CEA than do poorly differentiated adenocarcinomas.¹⁷⁷ Similarly, aneuploid colorectal tumors produce more CEA than near-diploid tumors.^178,179

CEA is not recommended for screening purposes in colorectal cancer. It is useful as a prognostic indicator after curative resection and during follow up.^176,180,181 The College of American Pathologists Consensus Statement recommended that preoperative CEA levels > 5 ng/mL be considered elevated and treated independently in multivariate analyses.¹⁸² After curative resection, an elevated CEA should drop to normal levels within 4 to 8 weeks.¹⁸³ If this decrease to normal does not occur, persistent or recurrent disease is likely. A rising CEA at any time during followup should alert the clinician about a possible recurrence.¹⁷³ The sensitivity of CEA reportedly is 78% and 75% for hepatic metastases and retroperitoneal recurrences, respectively, whereas for the detection of pulmonary metastases and peritoneal surface recurrences it is 42% and 46%, respectively.¹⁸⁴ Unfortunately, the lead time in the detection of recurrent disease by CEA monitoring has not translated into improved survival. A prospective randomized trial evaluating intensive followup with CEA levels versus no followup has not been reported.¹⁸⁵ In a meta-analysis based on the literature from 1972 to 1996, at a mean 2 years followup, Rosen and colleagues reported twice as many resections for recurrent disease and slightly higher cumulative 5-year survival in series with patients who underwent resection for recurrent disease diagnosed by intensive followup, which included history, physical examination, and CEA levels every 3 months.¹⁸⁶ Approximately 30% of colorectal cancers do not produce CEA, and approximately 40% of patients with a normal preoperative CEA will have an elevated level at recurrence.^187,188 The American Society of Clinical Oncology recommends that serum CEA levels be performed every 2 to 3 months in patients with stage II or III for > 2 years after diagnosis.¹⁸¹ CEA levels have also been used to monitor response to chemotherapy in metastatic disease. The reader is cautioned that this is not an accepted objective criterion for tumor response.

Monoclonal antibodies against CEA were developed in an attempt to improve tumor localization in patients with elevated CEA whose conventional testing had not localized the recurrence site. The sensitivity in some of the studies that used this technique is reported to be between 60% and 90%.^189,190 Because of disappointing results, the use of these scans has decreased. 2-(18F)-fluorodeoxy-d-glucose-positron emission tomography (FDG-PET) scans appear to be more useful than monoclonal antibody scans in detecting recurrences. Libutti and colleagues reported a prospective blind study evaluating FDG-PET scans, ^99mtechnetium (Tc)-labeled arcitumomab (CEA scan), and second-look laparotomy.¹⁹¹ Twenty-eight patients were explored. Disease was noted in 26 patients. FDG-PET scans predicted unresectable disease in 90% of patients, whereas CEA scans failed to predict unresectable disease in any patient.¹⁹¹ In patients who were resected, FDG-PET predicted resectability in 81% of the patients, whereas CEA scan predicted resectability in 13%.¹⁹¹ FDG-PET scan alone has been reported to have an 89% positive-predictive value and 100% negative predictive value in patients who had a rising CEA and normal conventional radiographic studies.¹⁹² Diagnosing and treating recurrences detected by CEA monitoring is not cheap. The cost per cure of recurrent colorectal cancer using CEA monitoring is estimated to be approximately $500,000 (US).¹⁹³

CEA has also been used to detect micrometastases in histologically negative lymph nodes. Cutait and colleagues reported on the detection of micrometastases by using immunoperoxidase staining of CEA and cyto-keratins.¹⁹⁴ At 5-year followup, there was no difference in survival between those patients who were upstaged and those whose lymph nodes were free of micrometastases.¹⁹⁴ However, Liefers and colleagues analyzed 192 lymph nodes from 26 consecutive stage II CRCs by using a CEA-specific reverse transcriptase (RT)-PCR technique.¹⁹⁵ Fourteen of 26 patients were upstaged. The adjusted 5-year survival was 91%, versus 50% in those whose lymph nodes were devoid of micrometastases.¹⁹⁵ Consequently, the latter investigators concluded that the detection of micrometastases using a CEA-specific RT-PCR was a prognostic tool in stage II colorectal cancer.¹⁹⁵