On a mammogram, fat appears radiolucent or dark, whereas stromal and epithelial tissue appears radio-dense or white. The amount of mammographic density is a strong independent predictor of breast cancer risk [
1,
2]. The biological basis for this increased risk is poorly understood. A critical question is whether densities are directly related to risk or are simply a marker of risk. We addressed this question recently by studying the location of small ductal carcinoma
in situ (DCIS) lesions as revealed by microcalcifications, and showed that such DCIS occurs overwhelmingly in the mammographically dense areas of the breast [
3]. Most DCIS lesions in our study occurred in the lateral-superior quadrant, as has been found in previous studies [
4], and 'correlated strongly with the average percentage density in the different mammographic quadrants' [
3]. Pre-DCIS mammograms that were taken on average about two years previously showed that the areas subsequently exhibiting DCIS were clearly dense at the time of the earlier mammogram, and this suggests that this relationship was not brought about by the presence of the DCIS. The reasons for these findings are not clear; two obvious possibilities are increased epithelial cell proliferation in mammographically dense areas of the breast and increased breast epithelium in women with mammographically dense breasts. Two groups have investigated the relationship between the amount of mammographic density of a woman and the amount of her breast epithelial tissue [
5,
6]. Alowami and colleagues [
5] used tissue obtained from biopsies investigating breast lesions that were subsequently diagnosed as benign or pre-invasive breast disease; they studied tissue 'distant from the diagnostic lesion' without reference to its location as regards mammographic density (that is, 'random' tissue). They found that the median density of duct lobular units was 28% higher in breasts whose overall mammographic density was 50% or more (
n = 27) than in breasts whose overall mammographic density was less than 25% (
n = 35); this result was not statistically significant and the result was described as showing 'no difference in the density of epithelial components' [
5]. Li and colleagues [
6] also found in their much larger study (
n = 236) of 'random' breast tissue collected from normal women by Bartow and colleagues [
7] in their autopsy study of accidental deaths in New Mexico that women with high mammographic density had greater amounts of epithelial tissue (as measured by area of epithelial nuclear staining) and the result was highly statistically significant. Breast epithelial proliferation rates as they relate to mammographic densities in healthy women have not been well studied [
8]. We have addressed these questions by studying the number of epithelial cells in terminal duct lobular units (TDLUs) and in breast ducts, and their respective proliferation rates as they relate to local histological breast densities within individual women.