Abstract PS3-07-21: Validation of the CTS5 as a predictor of late distant recurrence (DR) for HR+ HER2-negative Invasive Lobular Carcinoma (ILC)

Abstract Background: ILC is a distinct subtype of breast cancer characterized by unique histopathological features and a higher risk of late distant recurrence (DR) compared to invasive ductal carcinoma (IDC). Accurate risk stratification is critical for optimizing adjuvant management. While CTS5 is a widely used and readily available tool for predicting late DR (after 5 years) in HR+/HER2- breast cancer, it has not been evaluated and validated in ILC. We report the first validation of CTS5 in ILC and compare its performance to the recently developed ILC-specific prognostic model (MDA-iLobulaRX). Methods: We retrospectively analyzed data from patients with stage I-III HR+/HER2- ILC treated at MD Anderson Cancer Center, between 1980-2020, using a prospectively curated database. Two models were evaluated for predicting late DR: (1) MDA-iLobulaRX, incorporating age, ER%, tumor grade, presence of LCIS, ILC histology, lymph node status, tumor size, and adjuvant endocrine therapy (tamoxifen or aromatase inhibitor [AI]); and (2) CTS5, including age, tumor size and grade, and nodal status. Patients with less than five years of follow-up were excluded to ensure adequate time at risk for late DR. Additionally, patients who did not receive endocrine therapy were excluded. Model performance was assessed using: (1) Akaike Information Criterion (AIC) to compare model fit; (2) Harrell’s concordance index (c-index) to assess discrimination; and (3) Receiver operating characteristic (ROC) curves with area under the curve (AUC), based on a binary classification of recurrence (yes/no) after 5 years for clinical interpretability. Results: The cohort included 2,253 women with a median follow up time of 10.3 years with a range of 5 to 40 years. The median age was 57 years, median tumor size was 23 mm, 52% were node-positive and 67% were postmenopausal. Classical ILC was the predominant subtype (90%) compared to 10% non-classical subtypes. Endocrine therapy included tamoxifen (43%), non-steroidal AIs (55%) and steroidal AI (2%). Model 1 (MDA-iLobulaRX) had a slightly better fit (AIC: 9115 vs 9157) while both models demonstrated comparable discrimination (c-index 0.698 for model 1 vs 0.704 for model 2) and AUC (0.74 for model 1 and 0.75 for model 2). Conclusion: CTS5 is a valid, readily accessible tool for predicting late DR in HR+/HER2- ILC, with performance comparable to ILC-specific models. While CTS5 does not guide treatment or surveillance decisions, it may support risk stratification discussions in clinical practice and offers a pragmatic tool for estimating late recurrence risk in this population. Citation Format: J. A. Mouabbi, A. S. Raghavendra, S. Pasyar, B. L. Roland, R. A. Mukhtar, M. Giancarlo, M. D. Wright, A. K. Bisen, A. N. Iheme, C. H. Barcenas, V. Valero, A. Nasrazadani, B. Lim, D. L. Ramirez, A. Hassan, H. M. Kuerer, T. Adesoye, P. R. Paula, F. Meric-Bernstam, S. H. Giordano, J. K. Litton, R. M. Layman. Validation of the CTS5 as a predictor of late distant recurrence (DR) for HR+ HER2-negative Invasive Lobular Carcinoma (ILC) [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 PS3-07-21.