Abstract PS3-06-17: Prognostic performance analysis of artificial intelligence test and 21-gene assay in premenopausal node-positive HR+ HER2- breast cancer patients

Abstract Background: The RxPONDER trial was a landmark study demonstrating that the 21-gene Oncotype DX (ODX) assay can help avoid unnecessary adjuvant chemotherapy in postmenopausal women with 1-3 positive nodes (N+) and an ODX recurrence score (RS) ≤25. However, N+ premenopausal women with RS≤25 showed a significant but variable benefit to adjuvant chemotherapy despite low or intermediate ODX RS. Current NCCN guidelines therefore recommend ODX only as a consideration in this subgroup, leaving risk-adapted decision-making unresolved. Ataraxis Breast (ATX) is a multi-modal artificial intelligence test that integrates morphological features extracted from standard H&E-stained slides with baseline clinical data. Here we validate the prognostic accuracy of ATX in four premenopausal N+ cohorts. Methods: 150 premenopausal HR+/HER2- node-positive cases were extracted from Providence Health (n=48), TCGA (n=59), Basel (n=32) and Karmanos (n=11) cohorts. ATX recurrence scores were generated using a locked model. 43 patients (Basel and Karmanos cohorts) also had ODX RS available. Prognostic performance of ATX and ODX RS was assessed for disease-free interval (DFI) endpoint using Harrell’s C-index and hazard ratio (HR per 0.1 increase in ATX). Reported HR and C-index are pooled using random-effects meta-analysis. Results: In a univariate model, ATX was a significant predictor of DFI in premenopausal N+ women with a hazard ratio of 2.77 (1.53-5.01, p=0.02) and C-index of 0.78 (95% CI: 0.64-0.92). ODX did not reach statistical significance with a HR of 1.05 (1.00-1.10, p=0.05) and C-index of 0.46 (0.32-0.59). In multivariate analysis adjusting for age, tumor size and grade, ATX was the only variable significantly associated with DFI (p<0.01). The estimated 5-year recurrence rates for ATX bottom and top quartiles were 9.1% and 25.7%. When divided into exploratory low/high ATX risk groups with threshold at 10% 5-year risk of recurrence (RoR), the ATX <10% RoR group included 39 patients with an observed recurrence rate of 5%, while ATX ≥ 10% RoR included 111 patients with an observed recurrence rate of 16%. Conclusions: ATX demonstrates promising independent prognostic performance in premenopausal node-positive breast cancer across diverse datasets. Larger studies are needed to confirm its efficacy and utility. Citation Format: J. Elayoubi, K. Choucair, K. Kalinsky, K. Pogoda, F. J. Esteva, K. Zeng, J. Cappadona, B. Machura, K. J. Geras, J. Witowski. Prognostic performance analysis of artificial intelligence test and 21-gene assay in premenopausal node-positive HR+ HER2- breast cancer patients [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS3-06-17.