We will develop and validate new breast cancer risk prediction models that include familial factors related to genetic risk, personal risk factors unrelated to genetic risk, qualitative breast density measures, novel quantitative dense tissue measures and types of dense tissue structure, endogenous sex hormone levels and 18 candidate single-nucleotide polymorphisms (SNPs) identified by genome-wide association studies (GWAS). In addition, we will assess the comparative effectiveness of risk-based versus age-based mammography screening on age to start and stop screening and how often to screen. We will build on the existing Breast Cancer Surveillance Consortium (BCSC) infrastructure, expanding data collection and constructing two cohorts that will be the basis for developing and validating new breast cancer risk prediction models (Specific Aim 1). Additional data collected will be detailed breast and ovarian cancer family history;an automated measure of dense breast volume;development and validation of new quantitative density structure measures based on image texture features;endogenous sex hormone levels; and, genetic profiles. Using the existing and expanded BCSC infrastructure, we will develop and validate new breast cancer risk prediction models (Specific Aim 2) using traditional and novel risk factors for up to 1,000,000 women and 20,000 breast cancers, generating separate 5-year breast cancer risk estimates for: (1) all-type breast cancers (2) specific invasive cancer sub-types including estrogen receptor positive (ER+), ER-negative (ER-), HER2/neu-oncoprotein positive (HER2+), triple-negative (ER-/PR-/HER2-) and HER2+/ER- and ductal carcinoma in situ, and (3) women aged 35-49 and 50-79 years and those white, black, Hispanic and Asian, We will determine the accuracy of the new risk prediction models to classify women into Low (<1%), Average (1-1.66%), Intermediate (1.67%-2.49%), High (2.5%-3,99%) and Very high (>4%) 5-year risk groups and compare our new models to the BCSC Breast Density 5-year risk model using reclassification methods. We will also explore whether genetic variants improve the discriminatory accuracy of prediction models for women with intermediate-to-high breast cancer risk and if endogenous sex hormone levels improve prediction in postmenopausal women, Lastly, we will apply the risk estimates from prediction models in Specific Aim 2 to three established breast cancer simulation models to compare risk-targeted mammography screening at various starting and stopping ages and intervals, with current age-based guidelines to assess the benefits (e.g., percent mortality reductions and life years gained), harms (e,g., false-positive tests and unnecessary biopsies), and costs (Specific Aim 3) of each strategy.
These studies will develop and validate new breast cancer risk prediction models and show whether predicted risk levels are useful for determining the appropriate ages to start and stop screening mammography, and how frequently to screen to give the best outcomes. Our findings will inform national guidelines by determining how screening strategies can be further personalized based on breast cancer risk.
Showing the most recent 10 out of 190 publications
