Background: Although tumor markers are frequently used in the follow-up of patients with breast cancer, two points are still being debated: 1) their cost/effectiveness has been neither demonstrated nor disproved; 2) the reliability of the currently used dichotomous division into a positive/negative cut-off should be definitely validated. Dynamic criteria of interpretation based on serial serum samples would probably be more effective for early detection of relapse.
Patients and methods: The aim of the present study was to compare the dichotomous cut-off based decision criteria to a dynamic serial sample based assessment of tumor markers. Since 1989, 794 patients have been followed in 11 institutions. CEA and CA15.3 were measured once a month for three months before every clinical examination. The present paper concerns the evaluation variability in 405 patients without evidence of disease in the first three institutions joining the study.
Results: In patients without evidence of disease, the coefficient of variation of all samples for every patient showed a median value of 19 for CEA and 21 for CA15.3. Variability was negatively associated with the antigen level and was most likely due to the analytical component. This was also confirmed by the significant difference in variability among the three institutions evaluated. The median value of the critical difference was 53% for CEA and 57% for CA15.3.
Conclusions: 1) Individually tailored dynamic decision criteria are applicable in about 50% of the cases. 2) The problem of improving the precision of tumor marker assays in the low dose range must be urgently addressed to the manufacturers of tumor markers by the scientific community in order to apply individually tailored decision criteria for patients in whom the serum level of biological markers is low.