Mammographic density and breast cancer

© 2017 Dr. Kavitha KrishnanBackground: Mammographic density (MD) refers to the white or bright areas on a mammogram. MD adjusted for age and body mass index (BMI), whether considered as the absolute dense area (DA) and/or the percent dense area of the breast (PDA), has been found to be a risk factor for breast cancer and a risk factor for masking of tumours. Adjustment for age and BMI is due to negative confounding, given that both factors are positively associated with risk but negatively associated with MD. Before MD can be applied in the clinical setting, several issues need to be clarified. The role of non-dense area (NDA) of the breast on a mammogram is controversial and not well studied. It is also not known how well MD risk measures are maintained over time within an individual, especially from early adulthood. BMI is also an important risk factor of breast cancer but no study of MD and breast cancer risk have allowed for the BMI association to vary by age. Studies on prognosis of breast cancer and MD are also few. The aims of this thesis are, therefore, to study all three mammographic measures, DA, PDA and NDA and: 1) investigate how stable these measures are over time, before and after adjusting for age and BMI; 2) investigate risk estimates associated with both risk of developing a tumour in the breast, and risk of masking of an existing tumour on a mammogram, and to find out if these associations vary by tumour characteristics; and 3) for women with breast cancer, investigate the associations with risk of recurrence and risk of death from breast cancer and to find out if these associations vary by the tumour characteristics at first diagnosis. Methods: DA, PDA and NDA were measured by a semi-automated method using the Cumulus software. Data from the Melbourne Collaborative Cohort Study (MCCS) and the Australian Breast Cancer Family Registry were combined for the longitudinal study. There were 970 women who had an average of 4.5 mammograms (range, 1-14) and the age range of all the mammograms in the data was 24 to 83 years. To quantify stability of measures over time, within-woman correlations for DA, PDA and NDA measures over different time intervals were estimated using mixed-effects models and fitting Gompertz curves. A case-control study nested within the MCCS was used to investigate the MD associations with inherent risk and risk of masking. We assumed that the risk of screen-detected breast cancer mostly reflects inherent risk, while the risk of whether a breast cancer is interval versus screen-detected mostly reflects risk of masking. There were 244 screen-detected cases matched to 700 controls and 148 interval cases matched to 446 controls. The following tumour characteristics were included: size, grade, lymph node involvement, and ER, PR, and HER2 status. Conditional and unconditional logistic regression were applied as appropriate to estimate the Odds per Adjusted Standard Deviation (OPERA) adjusted for age and BMI, while allowing the association with BMI to be a function of age at diagnosis. For the study investigating the risks of recurrence and death due to breast cancer, 643 women with breast cancer from the MCCS were followed up and 74 recurrence cases were identified after an average of 3 years, and 89 deaths due to breast cancer were identified after an average of 5 years. Cox regression was applied, with duration as the time axis, to estimate the hazard ratios (HR). Results: For all three measures, DA, PDA and NDA adjusted for age and BMI, the within-woman correlation estimates were 0.9 and above for mammograms taken 10 years apart. Both DA and PDA were positively associated with risk of screen-detected cancer with similar risk estimates (OPERA ~ 1.2). NDA was not associated with risk of screen-detected cancer. PDA was strongly associated with risk of interval cancer (OPERA = 2.24; 95% Confidence Interval (CI): 1.75, 2.86) and risk of interval versus screen-detected cancer (OPERA = 1.76; 1.39, 2.22). NDA was negatively associated with risk of interval cancer (OPERA = 0.63; 0.51, 0.78) and risk of interval versus screen-detected cancer (OPERA = 0.67; 0.54, 0.84). Similar results were obtained when analyses were restricted to small tumours to reduce the influence of increased tumour growth rate. From the analyses by tumour characteristics, both DA and PDA were associated with increased risk of large tumours (OPERA ~ 1.6) and positive lymph node involvement (OPERA ~ 1.8) with similar risk estimates. No association was observed for NDA. Associations between all three mammographic measures and risk of interval cancer and risk of interval versus screen-detected cancer did not vary by tumour characteristics. Only DA was positively associated with risk of recurrence (HR = 1.01; 1.00, 1.02) while no association was found for PDA and NDA. None of the mammographic measures were associated with risk of death from breast cancer. Associations between all three mammographic measures and risk of recurrence and risk of death from breast cancer did not vary by tumour characteristics. Conclusion: All three mammographic measures are highly stable over time from an age younger than 40 years, which is before the target age when screening is recommended. After adjusting for age and BMI, risk of developing breast cancer is predicted by both DA and PDA but not NDA. Risk of masking is best predicted by PDA, and not by BMI. The negative association found between NDA and risk of masking might be due to DA and NDA being “both sides of the same coin”. Both DA and PDA were predictors of inherent risk of larger breast tumours and positive nodal status. The associations with risk of masking did not vary by tumour characteristics. While mammograms have mainly been used for detection of tumours, MD measures might also be useful for predicting inherent risk of breast cancer, of masking of tumours, and of recurrence. Population-based screening could start at a younger age than is current practice, and be tailored based on a woman’s MD measurements. These findings on MD measures could also aid in understanding the etiology of breast cancer