A comprehensive review of the literature (1978–2008) was performed, using search term [Carcinoembryonic antigen] AND [Colorectal neoplasm], limited to the subheadings [analysis], [diagnostic use], [blood], [standards] and limited to ‘human’ and ‘English language’. Additional relevant references found in articles were included. Review articles were used as a reference but not included in the analysis. Selection of the abstracts was made on the available information concerning clinical use of CEA in follow-up after curative resection of CRC.
From 1980 until 2008, 26 original clinical trials evaluating the outcome of follow-up after curative treatment of colorectal carcinoma which included serum CEA measurements were published () [6
]. One publication was excluded from analysis because it concerned a double publication. These 25 clinical trials, and a few follow-up studies that did not include serum CEA measurements, have resulted in five meta-analyses on follow-up in CRC [1
]. The 25 original articles were analyzed on actual testing of the diagnostic capacity of serum-CEA, the clinically applied threshold value of serum-CEA, timing of follow-up and measurement frequency.
Clinical studies on follow-up including CEA measurements (n = 25).
Most studies evaluated follow-up regimes consisting of various diagnostic tests. In only five of 25 clinical trials, different
serum CEA regimes were compared [6
], from which three studies were randomized controlled trials [8
]. The results of these studies show a consistent beneficial effect of serum CEA measurements on both the rate of curative resection of recurrent disease and survival (). In the meta-analyses the majority of included trials did not compare different
CEA regimes (). These meta-analyses therefore, do not reflect the specific diagnostic capacity of serum CEA in follow-up.
Effect of CEA measurements in FU on resectability of recurrent disease and 5 year survival.
Summary of meta-analyses on follow-up for colorectal cancer and inclusion of trials comparing different CEA regimes.
The static normal value of serum-CEA as advised by manufacturers is 2.0–2.5 ng/mL, depending on the actual test. To increase specificity and prevent the situation of inability to localize recurrent disease with imaging techniques or relaparotomy in the past, the clinically applied static threshold value is often increased to 5 ng/mL or higher. From the 25 studies, all except one study used a static threshold value. A threshold value of ≤2.5 ng/mL was applied in three studies (13%), of ≥5 ng/mL in eight studies (32%) and in 14 studies the applied threshold value was not mentioned. Applying a higher threshold value, results in a decrease in sensitivity together with a possible delay in diagnosis of recurrent disease.
Diagnostic tests for recurrent disease are most sensitive when the timing is aimed at the expected time to recurrence through optimizing the pre-test probability. Data from the 25 clinical studies revealed that local recurrences were found on average between six months and two years, liver metastases between six and 18 months and pulmonary metastases between 24 and 36 months. In the clinical trials (calculation based op 22 trials), a measurement frequency of ≤3 months was attained in 91% of trials in the first year, 53% in the second year and 25% in the third year. The measurement frequency actually
carried out is only 46% to 62% of the reported
measurement frequency, as was shown in six independent studies [7
]. This implies that the timing and measurement frequency in most studies and in clinical practice has been insufficient to expect any beneficial effect of serum-CEA in follow-up for CRC.