SensReg: A Regulon-Based Predictor of Neoadjuvant Chemotherapy Response in ER- Positive/HER2-Negative Breast Cancer
Year: 2026
Type: preprint
Abstract: Pretreatment predictors of neoadjuvant chemotherapy (NAC) response are lacking, especially in ER-positive/HER2-negative breast cancer, where cytotoxic benefit is most uncertain. From 1,158 transcription-factor regulons across five discovery cohorts, we derived SensReg, a locked 13-regulon logistic signature of chemotherapy sensitivity applied without refitting. SensReg was evaluated across 40 chemotherapy-only NAC datasets (n = 3,852, mean AUROC = 0.72), of which 39 (n = 3,840) were meta-analyzable (pooled AUROC 0.72, 95% CI 0.69–0.75), including 35 validation datasets independent of discovery (n = 3,092), of which 34 (n = 3,080, mean AUROC = 0.71) contributed to the pooled validation meta-analysis (pooled AUROC 0.70, 95% CI 0.67–0.73), and 20 HER2-antibody treatment arms from 17 series (pooled AUROC 0.66, 95% CI 0.60–0.72), outperformed 17 established signatures, and showed the strongest independent association with response after multivariable adjustment, the only signature significant at BH-FDR < 0.001 (adjusted OR 1.81, 95% CI 1.57–2.09; BH-FDR = 6.3×10⁻¹⁵), although this mutually adjusted comparison is exploratory because of limited events relative to the number of predictors. It provided positive estimated net benefit in ER-positive/HER2-negative disease (22 cohorts, n = 1,228) and retained discrimination within the recurrence-score intermediate stratum (AUROC 0.735). SensReg shows no detectable prognostic signal in a treatment-naive cohort but was associated with endocrine resistance and was carried mainly by malignant epithelium within immune-engaged microenvironments, associating with favorable chemotherapy-treated outcomes. SensReg is an interpretable, transferable predictor of treatment response warranting prospective evaluation.
Source: Research Square
Authors: Bashir Lawal, Xiaosong Wang
Institution: University of Pittsburgh
Language: English
Subfield: Cancer Research
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7224105280