Although carcinomas had traditionally been viewed as an “epithelial disease”, recent breast cancer studies have highlighted the importance of the tumor microenvironment, that is, the interactions between cancer epithelium and tumor stroma. In order to explore the genetic background of the tumor microenvironment, a total genome loss-of-heterozygosity/allelic imbalance (LOH/AI) scan was performed on epithelial and stromal DNA from 134 sporadic invasive breast carcinomas. We obtained tumor epithelium and tumor stroma with laser capture microdissection (LCM) technique from 134 samples of clinically sporadic invasive breast cancers, and subsequently extracted genomic DNA from LCM-captured tissues. We performed multiplex-PCR genome-wide LOH/AI mapping by using 381 microsatellite markers on the extracted DNA. Although, the average LOH/AI frequencies at each maker were higher in epithelium than in stroma for the majority of markers, we detected more frequent LOH/AI at 3 loci in stroma than in epithelium (ATA79C10 at 1p35.1, D6S942 at 6p25.3 and D8S1136 at 8q13.1). Moreover, we found strong evidence that LOH/AI frequencies were significantly elevated in specific regions of each chromosome. We detected 57 markers which preferentially showed LOH/AI either in stroma (38 markers from 19 chromosomes) or in epithelium (19 markers from 15 chromosomes), relative to the background LOH/AI frequencies on their respective chromosomes (we named those markers “hotspot markers”). Especially, chromosome arm 11q contained 6 hotspot markers (D11S2365, D11S4459, D11S2006 and D11S4464 in stroma, and D11S2002 and D11S1986 in epithelium), suggesting the existence of 3 distinct LOH/AI regions (2 in stroma and 1 in epithelium). The “multiplicity” of stromal cell LOH/AI targets and hence alterations of genetic material provide a possible mechanism for inter-patient variation in host-stromal response to invading adenocarcinoma cells: after transformation by alterations at a smaller number of loci in the epithelium, inter-patient diversity may result from the multiplicity and larger variety of stromal hotspot markers. This is consistent with a model in which initial, random LOH/AI occurs equally amongst epithelium and stroma, but subsequent clonal selection is driven by factors which appear to be distinctly different between malignant epithelial and surrounding stromal cells. Genetic alterations in stroma did not mimic those of epithelium, but they could play a different and parallel role in carcinogenesis and tumor progression, probably by modifying some features specific to breast cancer. In the future, such data may also be useful in designing novel drugs which target the stroma for both prevention and treatment.

[Proc Amer Assoc Cancer Res, Volume 46, 2005]