The sensitive and specific detection of micrometastasis holds great promise for earlier staging of cancer patients. By amplification of tissue-specific gene expression, the reverse transcriptase polymerase chain reaction (rtPCR) readily detects single tumour cells in different tissues. An increasing number of rtPCR assays with possible relevance for routine laboratory diagnostic procedures is currently being reported in the literature. Interestingly, when used in the clinical setting, assays for the same target mRNA perform very differently, despite comparable sensitivities and specificities in-vitro. Using rtPCRs specific for carcinoembryonic antigen (CEA), prostate-specific antigen (PSA) and cytokeratin 18 (CK 18), we have started to systematically investigate, both experimentally and in clinical specimens, a number of factors that contribute to the varying and seemingly implausible test results. Here we have concentrated on sample collection modalities, assay stability and test reproducibility at the sensitivity limit. Our results demonstrate in detail that, at the maximum sensitivity required for micrometastasis detection, preanalytical and statistical influences increasingly become important for the consistency of the assay results. We conclude that the prerequisite for translating the results from highly sensitive and specific rtPCR assays into clinically relevant data is the thorough definition of assay procedures and the number of tests performed on a sample. Addressing questions of standardization and quality control management is a central aspect yet to be emphasized in assay development and application of routine laboratory rtPCR tests in oncology.
