Background: Low frequency epithelial cells in bone marrow from colorectal carcinoma patients are associated with an increased risk of recurrence and decreased survival. Current immunohistochemical approaches to detect epithelial cells in bone marrow aspirates rely on antibodies against cytokeratin 18 (CK18). The predictive value of CK18-based detection strategies is limited by false-positives that occur in approximately 30% of cases. Cross-reactivity of anti-CK18 antibodies with nontumor cells may contribute to the false-positive rate. Cytokeratin 20 (CK20) shows more restricted expression than CK18 and labels cells in colorectal tumors.
Methods: Immunofluorescence assays were used to quantify CK20-labeled cells in bone marrow aspirates and tumors from 18 Dukes stage C and D colorectal carcinoma patients to determine whether CK20 is useful in detecting micrometastases. Fluorescent in situ hybridization was used to determine whether CK-labeled subpopulations carried genomic aberrations associated with colorectal carcinoma.
Results: CK20-labeled cells occurred at frequencies approximately 5 x 10(-5) in control bone marrow aspirates from patients without colorectal carcinomas. Approximately 10(-4) CK20-labeled cells were present in 4 of 11 bone marrow aspirates (45%) from patients with Dukes stage D colon carcinoma. The mean frequency (5 x 10(-5) of CK20-labeled cells in Dukes stage C and D rectal carcinoma patients was statistically similar to control values. CONCLUSIONS. A subset of CK20-labeled cells in primary tumors and hepatic metastases are aneusomic. CK20-labeled cells in bone marrow aspirates are cytogenetically normal. These data demonstrate that CK20 cells in solid tumors may be cytogenetically aberrant, but suggest caution in the use of CK20 to detect micrometastases in bone marrow aspirates.