Statistical validation of the mammaglobin-nested RT-PCR assay for tumor cell detection in blood of breast cancer patients.

A stochastic model was developed to validate the results obtained with the mammaglobin-nested RT-PCR assay for tumor cell detection in peripheral blood of breast cancer patients. Since the assay consists of four PCR setups per peripheral blood sample, the probabilities for receiving 0, 1, 2, 3, or 4 positive setups were calculated. In this model, samples with just 500 mammaglobin mRNA molecules are highly probable to result in at least three positive setups, whereas lower quantities shift the probabilities towards one or two positive setups. In the clinical trial, samples with one or two mammaglobin positive setups were detected in 6/143 (4%) patients with benign lesions of the breast, in 41/310 (13%) breast cancer patients with no evidence of disease and in 39/157 (25%) breast cancer patients with metastatic disease. On the contrary, no sample from patients with benign lesions of the breast resulted in three or four positive setups, but 5/310 (2%) breast cancer patients with no evidence of disease and 46/157 (29%) with metastatic disease. These results correspond with the model: an increased number of tumor cells in peripheral blood lead to a higher amount of mammaglobin mRNA molecules, and these samples may result in at least three positive setups. Samples with three orfour positive setups were mainly derived from breast cancer patients with metastatic disease and only occasionally from patients with no evidence of disease. On account of these results, samples with at least three positive setups are of prognostic value and regarded as tumor cell positive.