Predictions are circulating

We know that tumour cells disseminate round the body through the blood to form metastases, but measurement of circulating tumour cells has been difficult until recently. Massimo Cristofanilli, Daniel Hayes and colleagues have used a new technique to detect tumour cells in whole blood to show that circulating tumour cells in patients with metastatic breast cancer are associated with short survival.

The technique called CellSearch (developed by Immunicon) enriches a blood sample for epithelial cells using magnetic beads coated with antibodies against epithelial cell-adhesion molecules. Then, fluorescently labelled antibodies against epithelial-lineage-associated cytokeratins and the leukocyte antigen CD45, together with a nuclear stain, further separate out the epithelial cells. The nucleated, CD45-negative, cytokeratin-positive cells are then counted.

The prospective multicentre study included 177 patients with metastatic breast cancer who had had previous adjuvant treatment and/or treatment for metastatic disease. The levels of circulating tumour cells were measured before patients started a new treatment, and then again at the first follow-up visit 3–4 weeks later. The control group consisted of 72 premenopausal healthy women, 73 postmenopausal healthy women, 99 women with benign breast disease and 101 women with non-malignant diseases.

The first 102 patients with metastatic breast cancer enrolled were used as the training set to select the level of circulating tumour cells to be used to stratify patients into favourable and unfavourable prognosis. Circulating epithelial cells were rare in the control group (less than 2 cells per 7.5 ml blood), but 61% of patients with metastatic breast cancer 2 or more cells per 7.5 ml blood. The levels measured in the training set were correlated with progression-free survival (PFS) and overall survival (OS), and a cut-off point of 5 cells per 7.5 ml blood was chosen to distinguish unfavourable and favourable prognosis. When the next 75 patients who were enrolled — the validation set — were evaluated, the ability of the cut-off point chosen to predict outcome was confirmed.

For the 87 patients with more than 5 cells per 7.5 ml of blood, the median PFS was 2.7 months and median OS was 10.1 months — significantly shorter than the PFS of 7 months and OS of over 18 months seen in the 90 patients with less than 5 cells per 7.5 ml of blood. Ten patients who died before the follow-up visit all had high to extremely high levels of circulating tumour cells.

Hormone-receptor status was significantly associated only with OS, but, although time to metastasis, ERBB2 status and type of therapy remained relevant in the multivariate analysis, circulating tumour cells emerged as the strongest predictor of PFS and OS. The association with PFS and OS was not robust in the patients receiving hormonal therapy or immunotherapy.

So, the test reliably estimated disease prognosis and survival much earlier (3–4 weeks) than in usual clinical practice with traditional imaging methods, which are usually done 8–12 weeks after initiation of treatment. Comparison of the CellSearch assay with measurement of serum tumour markers, such as CA15-3, and further investigation of the robustness of the test for prediction of outcome for different types of therapy would be useful next steps.