OBJECTIVE We aimed identification of cell surface molecules,which might serve as diagnostic biomarkers or useful targets for therapies, in breast cancer.METHODS We developed unique DNA microarray coupled with spherical self-organizing map (sSOM) analysis to characterize cells and tissues by the cell surface markers. In the microarray 1,797 probes for human genes coding membrane bound proteins were spotted. With this microarray the gene expression profiles of eight breast carcinoma cell lines were compared to identify the genes that were commonly expressed in breast carcinomas but not in normal cells.RESULTS The gene expression profiles of sSOM from the eight breast carcinoma cell lines were successfully distinguished from that of normal breast tissue derived cells suggesting the presence of genes of interest, sSOM on the data extensively filtered revealed several candidate genes, of which expression was significant in carcinoma cells but low in normal cells. Finally, TM9SF2 was nominated through validations of PCR procedures together with CD24 and ErbB3, which are known breast carcinoma markers. TM9SF2 expression was further confirmed by immunological staining. Interestingly, TM9SF2 was found to be expressed in all the cell lines evaluated while CD24 and ErbB3 were not in all of the carcinoma cells, supporting their relationship in sSOM. Although physiological significance of TM9SF2 is unknown yet, siRNA treatment significantly inhibited the growth of MDA-MB-231 cells.CONCLUSION We propose TM9SF2 as a novel and useful diagnostic marker as well as a potential molecular target specific to breast carcinoma cells covering wide range of breast cancer.