The significance of CEA, CA19-9 and CA72-4 was evaluated the for early detection of disease recurrence, on the basis of retrospective evaluation of routine data in patients with colorectal carcinoma. They also considered the dependence of the results of these data analyses on the definition of groups of patients, both with no evidence of disease (NED) and with recurrence of disease (RD).
Patients and methods: From January 1994 to March 1999 serum levels of CEA, CA19-9 and CA72-4 were determined in the follow-up of 517 patients with colorectal cancer and compared with the retrospectively confirmed clinical status of the patients.
Results: CEA and CA19-9 showed comparable sensitivities in the detection of locoregional recurrence of colorectal carcinoma, whilst the sensitivity of CA72-4 was considerably lower. CEA is an optimal marker for detecting distant metastases, in particular liver metastases, since its sensitivity considerably exceeds the sensitivities of the other two monitored markers.
Conclusion: Using routine data required detailed analysis and clear definitions of groups of patients with NED and RD. The following conclusions for the evaluation of data were drawn from this analysis: a) Tumor marker cut-off values and sensitivities related to 95% specificity of remission values depended strongly on the given definition of the groups of patients with NED and RD. b) The patient group with NED is best characterized as the group of patients who never developed progression and where all the values which were assessed within a period shorter than six months from the end of therapy and follow-up, or less than six months before progression, death, or before the last marker assessment in the patient, were excluded. c) For the optimal characterisation of the group of patients with RD it is recommended only to consider values obtained during the first progression, after the period of complete post-operative or post-therapeutic remission. d) These conclusions refer not only to routine data, where this correction represents a condition for reliable evaluation, but also to any research done, since they ensure complete homogeneity of the group and mutual comparability of the results.