Abstract PS2-09-04: Predicting outcomes for patients with mixed ductal/lobular carcinoma of the breast based on circulating tumor DNA positivity patterns

Abstract Background: Invasive lobular carcinoma (ILC) exhibits distinct histopathologic characteristics and prognoses when compared to breast carcinomas of no special type (NST). However, mixed NST/ILC, an uncommon and less studied subtype, poses unique clinical challenges due to its mixed histology and heterogeneous expression of hormone receptors and HER2. In this study, we developed a quantitative model to classify these tumors based on their “NST-like” or “ILC-like” molecular relapse patterns as defined by circulating tumor DNA (ctDNA) results. Methods: In this study, patients with early-stage breast cancer (eBC) were identified using Natera’s proprietary real-world database, which is linked to Komodo's Healthcare Map® claims database and available results from tumor-informed ctDNA testing (SignateraTM, Natera, Inc.). Receptor status and the date of relapse were inferred from the claims codes associated with interventions and secondary neoplasms. The quantitative model developed to classify tumors involved comparing mixed NST/ILC ctDNA profiles against cumulative ctDNA positivity curves of canonical NST and ILC using squared error minimization technique. To validate the model, we performed a separate analysis based on clinical recurrences. We assigned each patient with mixed NST/ILC who recurred to the “NST-like” or “ILC-like” category by comparing their clinical recurrence behavior against the subtype-specific cumulative molecular relapse distributions derived from known NST and ILC cases. Thus, the concordance between ctDNA-based and recurrence-based classifications was assessed. Results: A total of 10,122 patients with eBC (NST, N=7,073; ILC, N=1,147; mixed NST/ILC, N=271) were identified with inferred clinical outcomes and longitudinal ctDNA results for analysis. ctDNA-positivity at any time point after surgery was observed in 47 patients with mixed NST/ILC, of whom 16 (34%) were classified as “NST-like” and 31 (66%) as “ILC-like” by the ctDNA-based classification model. A secondary neoplasm was reported in 21/47 (44.6%) of patients. A 100% (21/21) concordance was observed between the ctDNA-based and recurrence-based classifications. We next analyzed the mutational landscape of newly classified tumors using whole-exome sequencing data. Interestingly, for “NST-like” tumors, TP53 and CDH1 were mutated in 38% (6/16) of cases, followed by PIK3CA and USP40 (31%, 5/16). For “ILC-like tumors”, the most frequently altered genes were: TP53 (45%, 13/29), PIK3CA (34%, 10/29), UNC13C (28%, 8/29), and CDH1 (21%, 6/29). Conclusion: In phenotypically diverse disease types, such as mixed NST/ILC, ctDNA-based relapse patterns can help predict the clinical trajectory of the disease and discriminate between “NST-like” vs “ILC-like” phenotype. Upon further refinement and validation, the model can potentially inform prognosis among patients with mixed NST/ILC. Citation Format: J. Foldi, S. Satta, C. Palsuledesai, S. Oesterreich, A. Lee, J. McKenzie, A. Rodriguez, E. Kalashnikova, M. Liu, M. Balic. Predicting outcomes for patients with mixed ductal/lobular carcinoma of the breast based on circulating tumor DNA positivity patterns 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 PS2-09-04.