Our objective was to develop and study the feasibility of a quantitative, nested reverse-transcription polymerase chain reaction (RT-PCR) assay for detection of micrometastatic, epithelial tumor cells using the epithelial glycoprotein EGP-2 gene as a target. Several carcinoma cell lines and peripheral blood samples of 10 healthy volunteers were screened for levels of EGP-2 mRNA. The assay included EGP-2 competitor molecules, carrying an internal deletion, that had been titrated by limiting dilution. Seven carcinoma cell lines showed a wide spectrum of EGP-2 mRNA expression levels, with the highest values (20-100 molecules/cell) seen in 3 breast-cancer cell lines. Unexpectedly, a consistent low level of EGP-2 mRNA expression (0.0004 molecules/cell) was observed in peripheral blood mononuclear cells, probably representing ectopic non-functional expression. Because of this background level, spiking experiments with T47D breast-carcinoma cells added to blood mononuclear cells exhibited a detection limit that was not better than approximately one tumor cell in 2 x 10(4) normal cells. Together with the considerable variation of EGP-2 transcript levels that is observed in different carcinoma cell lines, the extent of expression in normal blood cells would prevent a reliable estimation of low numbers of carcinoma cells in clinical samples. A similar situation might well apply for other target genes. This emphasizes the need for a critical evaluation of the different steps involved in the methods used for RT-PCR detection of micrometastatic tumor cells.