Limitations in the Application of Clinicopathologic Factors Alone in Predicting Radiation Benefit for Women With Low-Risk DCIS After Breast Conserving Surgery: The Impact of a 7-Gene Biosignature Based on 10-Year Ipsilateral Breast Recurrence (IBR) Rates

Document Type

Conference Proceeding

Publication Date


Publication Title

International Journal of Radiation Oncology, Biology, Physics


Purpose/Objective(s): Most women diagnosed with ductal carcinoma in situ (DCIS) receive radiotherapy (RT) after breast conserving surgery (BCS); however, clinical trials show that over 70% of women with BCS alone will not have a recurrence and therefore not benefit from RT. Traditionally, clinicopathologic (CP) factors have been used to select for whom to de-escalate treatment, but prospective trials have failed to identify a low risk CP group that did not benefit from RT with respect to local control. This study assessed the re-classification of patients with low-risk CP into Risk groups defined by the 7-gene biosignature and compared to 10-yr IBR rates. Materials/Methods: Women (n = 926) from four international DCIS cohorts treated with BCS had formalin-fixed paraffin embedded tissue samples analyzed at a CLIA lab (Laguna Hills, CA). CP low-risk patients were identified using a) RTOG-9804-like criteria [Nuclear Grade 1 or 2 & Size ≤2.5 cm & non-Palpable & Screen Detected & margin negative (no-ink on tumor)] and b) MSKCC-like criteria [low-risk score<220, determined using nomogram weighted factors (excluding: number of re-excisions and RT treatment), and using no-ink-on-tumor instead of close margin]. The 7- gene DCIS biosignature combined biomarkers with CP factors (age, size, palpability, and margin status) using an algorithm reporting a Decision Score (DS) and Residual Risk subtype (RRt). Women with low-risk CP were classified into biosignature Low Risk (DS≤2.8, no RRt) or High Risk (DS>2.8 +/- RRt) groups. 10yr in-breast event (IBR) rates with and without RT were assessed by Kaplan-Meier rates and Cox proportional hazard analyses. Results: Overall, 37% of all women were classified into the biosignature Low Risk group, while 51% and 34% were classified into CP low-risk groups (RTOG-9804-like, MSKCC-like, respectively). The biosignature Low Risk group (n = 338) had a 10-yr IBR risk of 5.6% after BCS and no significant RT benefit (absolute RT benefit = 0.8%, p = 0.70), 99% negative predictive value (NPV) for RT benefit. CP low-risk groups had 10-yr IBR rates of 12% and 8% after BCS without RT with absolute 6% (p = 0.04) and 4% (p = 0.1) IBR rate reductions with RT. The biosignature reclassified 51% and 63% of CP low-risk patients into the biosignature High Risk group. Importantly, these patients had higher IBR rates without RT (20% and 12%) and significant 13% (p = 0.005) and 8% (p = 0.01) absolute IBR rate reductions from RT. CP low-risk patients with concordant biosignature Low Risk demonstrated no significant RT benefit. Conclusion: The 7-gene predictive biosignature more reliably identified patients with low 10-yr IBR rates and no significant RT benefit than the traditional CP low-risk criteria (RTOG-9804-like, MSKCC-like). Importantly, those CP low-risk patients who were re-classified as biosignature High Risk had increased 10-year IBR rates and significant RT benefit.




2 Suppl.

First Page



American Society for Radiation Oncology 65th Annual Meeting ASTRO 2023, October 1-4, 2023, San Diego, CA