A Novel Breast Cancer Index for Prediction of Distant Recurrence in HR⁺ Early-Stage Breast Cancer with One to Three Positive Nodes.

Abstract

<b>Purpose:</b> The study objective was to characterize the prognostic performance of a novel Breast Cancer Index model (BCIN⁺), an integration of BCI gene expression, tumor size, and grade, specifically developed for assessment of distant recurrence (DR) risk in HR⁺ breast cancer patients with one to three positive lymph nodes (pN1).<b>Experimental Design:</b> Analysis was conducted in a well-annotated retrospective series of pN1 patients (<i>N</i> = 402) treated with adjuvant endocrine therapy with or without chemotherapy using a prespecified model. The primary endpoint was time-to-DR. Results were determined blinded to clinical outcome. Kaplan-Meier estimates of overall (0-15 years) and late (≥5 years) DR, HRs, and 95% confidence interval (CIs) were estimated. Likelihood ratio statistics assessed relative contributions of prognostic information.<b>Results:</b> BCIN⁺ classified 81 patients (20%) as low risk with a 15-year DR rate of 1.3% (95% CI, 0.0%-3.7%) versus 321 patients as high risk with a DR rate of 29.0% (95% CI, 23.2%-34.4%). In patients DR-free for ≥5 years (<i>n</i> = 349), the late DR rate was 1.3% (95% CI, 0.0%-3.7%) and 16.1% (95% CI, 10.6%-21.3%) in low- and high-risk groups, respectively. BCI gene expression alone was significantly prognostic (ΔLR-χ² = 20.12; <i>P</i> < 0.0001). Addition of tumor size (ΔLR-χ² = 13.29, <i>P</i> = 0.0003) and grade (ΔLR-χ² = 12.72; <i>P</i> = 0.0004) significantly improved prognostic performance. BCI added significant prognostic information to tumor size (ΔLR-χ² = 17.55; <i>P</i> < 0.0001); addition to tumor grade was incremental (ΔLR-χ² = 2.38; <i>P</i> = 0.1) with considerable overlap between prognostic values (ΔLR-χ² = 17.74).<b>Conclusions:</b> The integrated BCIN⁺ identified 20% of pN1 patients with limited risk of recurrence over 15 years, in whom extended endocrine treatment may be spared. Ongoing studies will characterize combined clinical-genomic risk assessment in node-positive patients. <i>Clin Cancer Res; 23(23); 7217-24. ©2017 AACR</i>.