Re-Thinking Clinicopathologic Risk Assessment in DCIS: Pooled Data from Validation Studies Comparing a 7-gene DCIS Assay to Clinicopathologic Features Alone

Abstract

<h3>Purpose/Objective(s)</h3> There is a need for a tool to predict recurrence risk and RT benefit to enable personalized management of DCIS after breast conserving surgery (BCS). Clinicopathologic (CP) factors alone have failed to identify a low-risk group with little benefit from RT, or a high-risk group with high recurrence risk after RT. The purpose of this study was to prospectively validate a novel biosignature, Residual Risk Subtype (RRt), integrated with the 7-gene DCIS assay to assess 10yr ipsilateral breast recurrence (IBR) rates and RT benefit compared to CP criteria alone. <h3>Materials/Methods</h3> Women (n=926) from four international cohorts treated with BCS with negative margins had FFPE tissue samples analyzed at a CLIA lab. The 7-gene DCIS assay combines biomarkers with CP factors (age, size, palpability, and margin status) using an algorithm and reports a decision score (DS). Women were classified into 3 risk groups, using DS (0-10) with integrated RRt (yes/no): 1) Low Risk DS ≤2.8 without RRt), 2) Elevated Risk (DS >2.8 without RRt), 3) Residual Risk (DS >2.8 with RRt). Within these 3 biosignature risk groups, IBR Kaplan-Meier rates by RT treatment and Cox proportional hazard ratios for RT-effect were assessed and reported by CP criteria, including RTOG 9804 study ‘good-risk' groupings. <h3>Results</h3> The biosignature classified 37% of the women as Low Risk who had low 10yr IBR rates with or without RT (4.8% vs 5.6%) and 43% as Elevated Risk who had an 20.6% IBR rate without RT and 4.9% after RT, which was an 80% IBR benefit from RT (p<0.001). The significant benefit persisted even in 'good risk' CP patients (∼17% IBR without RT and ∼6% with RT, p<0.03). 20% of women were classified as Residual Risk and had a 42.1% IBR rate without RT and a significantly elevated IBR rate of 14.7% after RT (p<0.001). The distribution of patients in the 3 biosignature risk groups differed by CP criteria (RTOG 9804 'good-risk', or young age or high grade). However, once classified into these 3 biosignature risk groups, the IBR rates and RT benefit were independent of CP criteria. About 30% of women with a high-risk CP profile were re-classified as Low Risk by the biosignature with almost no benefit of RT while ∼30% were re-classified as Residual Risk with ∼14% IBR risk after RT. In women with favorable CP criteria, the biosignature reclassified > 50% of patients as Elevated Risk or Residual Risk, where Residual Risk patients had high IBR (27.1%) risk remaining after BCS. <h3>Conclusion</h3> The 7-gene predictive DCIS assay integrated with the novel Residual Risk Subtype (RRt) biosignature classified women into 3 risk groups with distinct 10yr IBR rates and RT-benefit profiles. Within biosignature risk groups, CP criteria were not prognostic or predictive for 10yr IBR rates or RT benefit while a clear delineation was seen for the 3 risk groups. The incorporation of genomic information yields superior risk and RT benefit prediction as compared to CP features alone.