2986 - The PREDICT II Registry: A Prospective Study to Evaluate the Clinical Utility of a 7-Gene Predictive Biosignature on Treatment Decisions in Patients with Ductal Carcinoma In Situ

Purpose/Objective(s):

For women with ductal carcinoma in situ (DCIS) treated with breast conserving surgery (BCS), determination of which patients can be safely treated without radiation therapy (RT) remains controversial. Although evidence from randomized clinical trials supports the role of RT in reducing the risk of local recurrence for all patients, 70-90% of women with DCIS do not recur after BCS alone. Current guidelines recommend the use of clinicopathologic (CP) features to determine which patients may be considered for RT-omission. To overcome limitations of CP risk assessment, a biosignature has been developed and validated that integrates the protein expression of seven genes and four CP features. The 7-gene biosignature has been clinically validated to be both prognostic for in-breast recurrence (IBR) risk, predictive for response to RT, and the molecular residual risk subtype which identifies patients at a high risk of IBR after BCS and RT. The primary objective of the PREDICT registries is to understand the decision impact such a tool has on shared decision making and outcomes.

Materials/Methods:

This prospective observational registry is open to females age 30-85 who are candidates for BCS and eligible for RT and/or systemic treatment, with 10 year follow-up. Primary endpoints are changes in treatment recommendations for surgical, radiation, and hormonal therapy, as well as 10-year IBR and overall survival (OS) rates. Secondary endpoints are identi?cation of key drivers for treatment recommendations, such as age, size, grade, patient preference, biosignature status and recurrence. Subjects must not have been previously treated for DCIS or invasive breast cancer. Physicians complete surveys before and after receiving biosignature test results. Changes in treatment recommendations will be analyzed using McNemar's test (alpha level = 0.05). Multivariate logistic regression will be used to determine odds ratios of CP factors affecting pre- and post-test treatment recommendations. Covariates include hormone receptor status, nuclear grade, tumor necrosis, family history of breast cancer, race, ethnicity, patient preference, physician specialty, and the biosignature score. Differences in recurrence-free and OS will be assessed by Kaplan-Meier survival analysis using the log-rank test and/or the Cox Proportional Hazards model. Enrollment is open to 3,000 women from 30 sites.

Results: TBD

Conclusion: TBD. The study has been approved by WCG IRB (Tracking #20172841) and/or local IRBs at each site. ClinicalTrials.gov: NCT03448926.