Improvement of breast cancer cell detection by immunomagnetic enrichment

Abstract
Contaminating breast cancer cells in leukapheresis harvested for reinfusion to rebuild hemopoiesis after high-dose therapy have been described by several investigators. Methods for tumor cell detection are conventional immunocytochemistry, culture techniques and reverse transcriptase PCR. The percentage of tumor cell positive leukaphereses shows a wide variation. An approach to clarify if these cells can induce systemic relapse is to characterize them molecular-genetically and immunologically, but these techniques require a sufficient cell count. We compared conventional immunocytochemistry with immunocytochemistry after immunomagnetic enrichment of cancer cells by HEA-125 magnetic microbeads for the detection of micrometastatic tumor cells. A total of 25 samples, consisting of 16 samples from G-CSF-mobilized peripheral stem cell harvests, eight BM aspirations and one peripheral blood sample were investigated without [median 2, range 1–3×106 MNCs] and after [median 5, range 1–10×107, MNCs] HEA-microbead selection. Additionally 10 buffy coat samples from healthy subjects were investigated. Using conventional immunocytochemistry, tumor cells could be detected in nine stem cell samples. Two BM samples and the blood sample (48%) were positive, with a median tumor cell load of 0 (0–12) cells per sample (mean: 2.4). By HEA-bead selection the rate of positivity could be increased to 88% (13 stem cell samples, eight marrow samples and one blood sample) with a median load of 6 (0–47) (mean 10.6) suspected cells (p < 0.007). However, calculation of recovery revealed tumor cell losses by immunobead selection. False positive results were not seen. We conclude first that immunomagnetic selection is an excellent and highly sensitive tool to enrich contaminating cancer cells from marrow and stem cell samples; second that the existence of real tumor cell negative stem cell harvests is doubtful; and third that immunobead selection delivers sufficient tumor cell counts for their further characterization by molecular and immunological methods.

This publication has 14 references indexed in Scilit: