Mammographic density and breast cancer phenotypes

Abstract: Mammographic density is one of the strongest risk factors for breast cancer and has been thoroughly studied as such. Extensive mammographic density also decreases screening sensitivity, thereby increasing the risk of interval cancers. Whether density acts as fertile ground for all types of breast cancer, or whether it influences tumor growth in a specific direction, was not known when we embarked upon the studies of this thesis. We therefore aimed to investigate the association between density, tumor characteristics, molecular subtypes, recurrence, and survival, focusing on interval cancers in the last study. For studies I, III, and IV, we used the cases included in a population-based case-control study, in which cases were all Swedish women, aged 50-74, with incident breast cancer, diagnosed 1993-1995 (n=3345). We only included postmenopausal women with no prior history of cancer other than non-melanoma skin cancer and cervical cancer in situ (n=2720). Of these women, 1774 women had eligible mammograms. For study II, in which we investigated the relationship between density and molecular subtypes, the study population was based on all women with breast cancer operated at a large university hospital in Stockholm 1994-1996 (n=524). Women with available gene expression profiling and mammograms were included in the study (n=110). Pre-diagnostic/diagnostic density of the unaffected breast was assessed using a semi-automated, computer-assisted thresholding technique, Cumulus. Density was either measured as the dense area in cm2 (absolute density=AD) or percentage density (PD) (the absolute dense area/the total breast area). We did not find an association between density and tumor characteristics (lymph node metastasis, hormone-receptor status, grade, and histopathological classification) except for tumor size. However, this association seemed at least in part to be due to masking delaying diagnosis. In accordance with the lack of association between PD and most tumor characteristics, we did not find an association between density and molecular subtypes, nor between density, distant recurrence, and survival. We did, however, see a relatively strong association between PD and both local and locoregional recurrence, independent of established risk factors. In the last study, we investigated the differences in survival between interval cancers and screening-detected cancers, taking mammographic density into account. We could show that interval cancers in both dense and non-dense breasts were associated with poorer prognosis compared to screening-detected cancers. However, the poorer prognosis seen in interval cancers in dense breasts seemed mainly attributable to delayed detection, whereas the group of interval cancers in non-dense breasts primarily seemed composed of truly aggressive tumors which we believe need further study.