Detection of Cytokeratin-19 in Peripheral Blood of Breast Cancer Patients: Results from a Systematic Review/Meta-Analysis.

Background: The identification of occult micrometastatic disease is potentially important in management of breast cancer (BC) patients. Therefore, detection of cytokeratin 19 (CK-19) and others molecular markers using reverse transcriptase polymerase chain reaction (RT-PCR)-based techniques to detect micrometastases could have significant implications in prevention, early detection and treatment of breast cancer.

Objective: To assess the diagnostic specificity and sensitivity of detection of CK-19 in the peripheral blood (PB) of BC patients as well as to assess its relation to prognostic factors and implication on survival.

Methods: We conducted a systematic review/meta-analysis of all studies that analyzed CK-19 in the PB of BC patients. We searched MEDLINE, LILACS and Cochrane Database of Systematic Reviews (last search June 2004). Studies that had at least 10 BC patients and controls and evaluated the presence of CK-19 in PB using RT-PCR were included in this review. Diagnostic sensitivity was defined as the percentage of detection of CK-19 in patients with histological proven diagnosis of breast cancer. The specificity was defined as the percentage of negative tests for CK-19 in the control group (healthy individuals or non breast cancer patients).

Results: Out of 1842 relevant studies that were identified, 26 met our inclusion criteria, enrolling 2398 patients (1176 treatment and 1222 controls). Results showed that PCR had an overall diagnostic sensitivity of 0.39 (95% CI: 0.36–0.42) and diagnostic specificity of 0.83 (95% CI: 0.76–0.87) regardless of the breast cancer stage. Six studies researched the association between CK-19 and well established prognostic factors (i.e. positive axillary lymph nodes, hormonal receptors, tumor size); however data were not available for quantitative synthesis. Survival was poorly reported making impossible to generate any conclusion about the implication of detection of CK-19 on prognosis. When studies were critically appraised, we also identified many methodological weaknesses in the design, conduct and the analysis of these studies (e.g. use of multiple control arms, use of non-consecutive patients, different timing of sample collection, inclusion of all stages of breast cancer and differential work-up in the study & control groups) hence resulting in biased findings due to selection, verification, detection and spectrum bias.

Conclusion: The existing body of evidence regarding the assessment and prognostic value of CK-19 using RT-PCR techniques is poor and needs ample improvement before the initiation of its clinical application.