Model study detecting breast cancer cells in peripheral blood mononuclear cells at frequencies as low as 10(-7).

A flow cytometric assay was developed to detect rare cancer cells in blood and bone marrow. Multiple markers; each identified by a separate color of immunofluorescence (yellow and two shades of red), are used to reliably identify the cancer cells. Blood or bone marrow cells, which are not of interest but interfere in detecting the cancer cells, are identified by a panel of immunofluorescence markers, each of which has the same color (green). Thus, the rare cancer cells of interest are yellow and two different shades of red but not green. The requirement that the rare cancer cell be simultaneously positive for three separate colors (the specific markers) and negative for a fourth color (the exclusion color) allowed detection of as few as one cancer cell in 10(7) nucleated blood cells (a frequency of 10(-7). To test this rare-event assay prior to clinical studies, a model study was performed in which the clinical sample was simulated by mixing small numbers of cells from the breast carcinoma line BT-20 with peripheral blood mononuclear cells. We detected statistically significant numbers of BT-20 cells at mixing frequencies of 10(-5), 10(-6), and 10(-7). In control samples, no target events (BT-20) were observed when more than 10(8) cells were analyzed. For additional confirmation that the BT-20 cells in the model study were correctly identified and counted, the BT-20 cells (and only BT-20 cells) were covalently stained with a fifth fluorescence dye, 7-amino-4-chloromethylcoumarin (CMAC). CMAC fluorescence data were not used in the assay for detecting BT-20 cells. Only after the analysis using data from the specific and exclusion colors had been completed were the events identified as BT-20 cells checked for CMAC fluorescence. The putative BT-20 events were always found to be positive for CMAC fluorescence, which further increases confidence in the assay. Manual data analysis and an automated computer program were compared. Results were comparable with the manual and automated methods, but the automated "genetic algorithm" always found more BT-20 events. Cell sorting of BT-20 cells from samples that contained BT-20 at frequencies of 10(-5), 10(-6), and 10(-7) provided further evidence that these rare cells could be reliably detected. The good performance of the assay with the model system will encourage further studies on clinical samples.