Khoshnoud R, He Q, Sylván M, Khoshnoud A, Ivarsson M, Fornander T, Bergh J, Frisell J, Rutqvist L, Skog S
Int. J. Mol. Med. 25 (5) 735-741 [2010-05-00; online 2010-04-08]
cDNA microarray is an established technique. However, difficulties such as handling tissue samples under RNase-free conditions, the heterogeneous tumor composition, i.e. non-malignant versus malignant cells and different pathologic types of malignant cells, and lack of appropriate reference may limit the potentially benefit of this method in clinical use. In this study, we examined how standardization of gene expression to total mg RNA or mg tissue and tumor heterogeneity affect the final results. We found that the gene expression of human breast tumors was approximately 9 times higher in malignant tissue as compared to the non-malignant tissue when expressed per total mg RNA, but approximately 40 times higher when expressed per mg tissue. Genes that were expected to act as housekeeping genes (PUC18, RPL and beta-actin) varied between different parts of the tumor and also between non-malignant and malignant tissues, excluding them as reference genes. We also found that the gene expression differed in various parts of the breast tumor, probably due to a mixture of different types of cells, i.e. non-malignant and malignant cells. To find out if the variations in the gene expression were due to cell heterogeneity we used microdissection to collect malignant cells separately. We found that the gene expression was markedly different in the isolated malignant cells as compared to the gene expression of the bulk tumor tissue. Thus, to be able to evaluate results from cDNA array gene expression experiments it is, to our opinion, necessary to work with pure tumor cell populations, until solid information is available on the impact of stromal component. Housekeeping genes should be handling with care and mg tissue may be preferred instead of microg RNA for standardization.