The estimated lifetime risk of breast cancer in American women is approximately 13%, however, some individuals are at higher risk to develop a primary breast malignancy than others. Current breast cancer risk assessment is based on clinical features associated with increased risk in large cohort populations. The most familiar factors are a woman's age at menarche and menopause, first-degree family history of breast cancer, and breast biopsy details. Multiple models exist to collate these factors into an estimated breast cancer risk, the oldest and best known being the Gail Model. Although based on an individual's specific health features, these calculators generate a gross estimate, rather than a quantifiable, individual, tissue-based, molecularly- driven risk assessment. Therefore, identification of women at risk for malignant transformation based on tissue from a benign breast biopsy has great potential application in clinical practice. Our goal is to develop a robust multi-gene clinical assay to quantify risk of breast cancer occurrence in unaffected individuals who have undergone a benign breast biopsy. The greatest impact of a validated MR is the estimation of malignant risk in an unaffected individual; the diverse and growing population of women undergoing biopsy for undiagnosed (but ultimately benign) lesions is of particular interest, and further information regarding personal risk will significantly alter future care of these patients. Because diagnostic breast biopsies are fixed in formalin and embedded in paraffin blocks, we will focus on formalin-fixed paraffin-embedded (FFPE)-specific malignancy risk-signature (MR) assay development, as FFPE represents the majority of archived breast tissue samples. We propose three specific aims, building upon our prior experience, to accomplish our goal. First, we intend to perform technical validation of the MR signature by testing the FFPE-based NanoString platform in breast cancer cases with frozen and fixed specimens. Second, we will clinically validate the NanoString MR for prediction of breast cancer risk in a cohort of women who have received a benign breast biopsy at our institution. And third, we will compare the predictive value of our NanoString MR to the Gail Model in prediction of breast cancer risk. We will analyze the MR score in two types of data: a dichotomized MR score (low/high MR) and a continuous MR score. Fisher exact test for the dichotomized MR score and two-sample t-test for the continuous MR score will be used to validate the MR signature for its detecting power of cancer cases versus no cancer cases for each risk group. In addition to median cutoff, various approaches, such as ROC curve analysis and tree analysis, will be used to determine the optimal cutoff of the MR score with the goal to identify who is susceptible for breast cancer development. Subset analyses stratified by ER/PR/Her2neu status, use of TAM, and years of follow-up will also be performed by logistic regression.
Benign breast biopsies after abnormal imaging are very common; this application seeks to develop personalized gene signature testing of tissue obtained at such biopsies to predict an individual's subsequent risk of breast cancer.
