Reliability of quantitative reverse-transcriptase-PCR-based detection of tumour cells in the blood between different laboratories using a standardised protocol

Differences in methods of reverse-transcriptase (RT)-polymerase chain reaction (PCR)-based detection of tumour cells in the blood gives rise to conflicting results, and standardisation is urgently needed. This pilot study aimed to assess the variation of RTPCR-based detection of tumour cells in blood between four different laboratories using a commercially available kit with a standardised protocol. This kit allows comparison of results from different laboratories and facilitates the investigation of the influence of pre-analytical parameters. All laboratories analysed identical sets of blood samples spiked with tumour cells in a concentration range of 1-100 tumour cells/ml. To study at which level variation was introduced, three kinds of sample sets were generated in which (i) tumour cell RNA was spiked in the RNA of mononuclear cells (MNC), (ii) tumour cells were spiked in isolated MNC, and (iii) tumour cells were spiked in blood. Real-time quantitative RT-PCR was used to detect and quantify cytokeratin 20 (CK20) expression, which is indicative for the presence of epithelial tumour cells. All laboratories were able to detect CK20 expression in all spiked-RNA samples with limited variation in expression levels between laboratories. There was a positive correlation between the amount of spiked tumour cell RNA and CK20 expression level. RT-PCR analysis of spiked-MNC samples resulted in more variation in the CK20 expression levels between laboratories, however again all spiked samples were reported to be positive by all of the laboratories. The evaluation of spiked-blood samples gave rise to considerable quantitative and qualitative variation between the laboratories. Our results underline the importance and need for standardisation and extended quality control studies in the field of pre-analytics. (C) 2003 Elsevier Science Ltd. All rights reserved.