Abstract PD1-08: Development and Validation of a Combined Residual Risk Score to Predict Breast Cancer Risk in Unaffected Women Negative for Mutations on a Multi-Gene Hereditary Cancer Panel

Abstract Background: Unaffected women with a strong family history of breast cancer (BC) are often referred for hereditary cancer testing with multi-gene panels; however, typically <10% test positive for clinically actionable mutations. Large-scale genotyping studies have identified common variants (primarily single-nucleotide polymorphisms) that individually confer modest BC risk, but together partially explain BC genetic susceptibility in many women without monogenic mutations. In addition, a number of factors relating to reproductive and medical history modify risk for BC. Here, we describe the development and validation of a combined polygenic residual risk score (cRRS) which takes into account non-genetic factors, in a large, consecutive cohort of women who tested negative for mutations in known BC susceptibility genes. Methods: This IRB-approved study includes women of European ancestry tested with a multi-gene hereditary cancer panel who were negative for mutations in 11 genes associated with BC (BRCA1, BRCA2, TP53, PTEN, STK11, CDH1, PALB2, CHEK2, ATM, NBN, BARD1). Clinical information was collected from provider-completed test request forms. The dataset was divided into a training (July 2016– February 2017) and validation cohort (February 2017 – May 2017). 94 previously published variants (Mavaddat et al 2015; Michailidou et al 2015) were genotyped using NGS. Multivariable logistic regression models were used to evaluate the 94 variants, to develop a residual risk score (RRS) as a predictor of personal BC history in the training cohort, and to assess the performance of the RRS in the validation cohort. Independent variables included age, personal/family cancer history, and ancestry. In an additional cohort, reproductive and medical history variables will be recorded and used to calculate BC risk according to the Tyrer-Cuzick model. Results: Accurate genotyping results were produced for 92 out of 94 variants. The training (validation) cohort included 24,259 (10,575) women, 18% (15%) of whom reported a personal history of BC. In the validation cohort RRS was strongly associated with personal history of BC (p<10-31) with odds ratio per unit standard deviation of the RRS being 1.41 (95% CI = 1.33-1.49). The RRS outperformed a published polygenic risk score (PRS) based on 77 SNPs (Mavaddat et al 2015): in a model with both scores included, the RRS score was significantly associated with BC (p=2x10-6) while the PRS was not (p=0.29). The RRS score also outperformed a PRS score based on the 92 variants with literature-derived odds ratios for association with BC: in a model with both scores included, the RRS score was significantly associated with BC (p=0.022) while the PRS was not (p=0.28). The RRS will be combined with the Tyrer-Cuzick model to deliver an optimized combined residual risk score (cRRS) and compared to risk predicted by Tyrer-Cuzick model alone in a separate cohort of women testing negative for BC mutations. Conclusions: The validation and clinical implementation of a combined residual risk score for women at risk for hereditary BC may offer significant potential for the management of high-risk, unaffected women who test negative for monogenic BC mutations. Citation Format: Hughes E, Judkins T, Wagner S, Rosenthal E, Wenstrup R, Lanchbury JS, Gutin A. Development and validation of a combined residual risk score to predict breast cancer risk in unaffected women negative for mutations on a multi-gene hereditary cancer panel [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr PD1-08.