Blood-Based Biomarkers for Personalized Risk Assessment of Breast and Ovarian Cancer

Project Summary There remains a need to develop biomarker tests for personalized risk assessment of breast and ovarian cancers. Such tests would not replace screening programs but would instead be a basic tool that can be integrated with other risk models based on a subject’s characteristics to personalize the risk of harboring cancer and inform on the need for screening and surveillance for earlier detection. The primary translational objective of this proposal is to develop a multi-analyte blood-based biomarker panel based on circulating proteins and autoantibodies against tumor antigens that inform about a subject’s probability of harboring a breast or ovarian cancer. Studies by the applicant team have led to the identification of a panel of cancer-relevant circulating proteins as well as autoantibodies against tumor proteins, including TP53 and novel citrullinated antigens, for detection of breast and ovarian cancers. The PLCO cohort is an excellent resource to further advance testing of candidate biomarkers and to also establish combination rules together with subject characteristics for individualized risk assessment of breast and ovarian cancers to optimized screening and surveillance for earlier detection of these diseases. In Specific Aim 1, leveraging pre-diagnostic plasmas from 969 breast cancer cases and 106 ovarian cancer cases as well as four times the number of non-case plasmas from female PLCO participants that did not develop cancer during study follow-up, we will assess the time-dependent (e.g. 0-1 year, 1-2 years, etc) predictive performance (AUC, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV)) of candidate biomarkers for detection of breast and ovarian cancers. Priority biomarker candidates will be advanced to establish models together with pertinent patient characteristics (e.g. Gail Model for breast cancer) for 1-year risk prediction of BrCa and OvCa. For modeling, we will adhere to the Predictability, Computability and Stability framework. The entire PLCO specimen set will be divided into a Development Set and a Set-Aside Test Set. Modeling and tuning of hyperparameters as well as initial validation will be performed in the Development Set. The model with the best predictive performance (AUC) in the Development Set will be selected for subsequent testing in the Set-Aside Test Set. In Specific Aim 2, we will leverage serial samples procured from cases preceding a diagnosis of a BrCa or OvCa and serial samples for non-cases, and we will test whether longitudinal trajectories of biomarker panel scores improve risk prediction. The proposed study represents a validation of cancer-associated protein and autoantibody biomarkers and has high probability to develop a blood test that can be implemented in the clinical setting for individualized risk assessment of breast and ovarian cancers.