Integrated tumor-informed and tumor-agnostic ctDNA molecular residual disease (MRD) assessment with imaging in early breast cancer.

Abstract

11130 Background: Accurate ctDNA-based MRD detection in early breast cancer is limited by tumor heterogeneity, clonal evolution and methodological issues, limiting tumor-informed (TI) assays reliant on predefined variants. Tumor-agnostic (TA) approaches, while broader, are susceptible to false positives from clonal hematopoiesis and reduced sensitivity in low–tumor burden settings. Methods: We analyzed a single-center cohort of 74 female patients with stage I–III breast cancer undergoing post-treatment surveillance after completion of curative-intent therapy (surgery, radiation, and adjuvant systemic therapy as appropriate). Plasma samples collected post-treatment were analyzed using an integrated MRD assay combining TI ddPCR and TA NGS ctDNA detection strategies. CHIP filtering was incorporated to improve analytical robustness. MRD status was correlated with clinicopathologic features and contemporaneous imaging. MRD positivity was defined as detection by either TI or TA component. Results: At the time of ctDNA assessment, which was performed during standard post-treatment surveillance following completion of curative-intent therapy, 94.6% (70/74) of patients were imaging-negative. MRD was detected in 31.1% (23/74) of patients overall. All imaging-positive patients (4/4) were MRD-positive. Notably, among imaging-negative patients, 27.1% (19/70) were MRD-positive, indicating molecular residual disease not captured by conventional imaging. No imaging-positive cases were MRD-negative. MRD positivity was observed across all stages, including early-stage disease (I–II: 26%, 14/54), with the highest prevalence in stage III (47.4%, 9/19); one case had unknown staging. MRD was detected in 33.3% (18/54) of HR + /HER2 - , 25% (2/8) of TNBC, and 25% (3/12) of HER2 + tumors, demonstrating applicability across biological subtypes. Importantly, the hybrid approach increased detection sensitivity significantly. Among MRD-positive patients (n=23), 47.8% were detected exclusively by TA component, 43.5% by the TI component alone, and only 8.7% were detected by both. Reliance on a single MRD strategy would have failed to identify 91.3% (21/23) of MRD-positive cases, highlighting substantial biological complementarity between the two approaches. At early follow-up (median 113.5 days), MRD was detected in 23 patients, including 19 who were imaging-negative; among those with subsequent imaging-confirmed recurrence (n=4), the median MRD–imaging interval was 9.75 months. Conclusions: A hybrid TI and TA ctDNA MRD platform detects clinically occult molecular disease beyond imaging in stage I–III breast cancer. The complementary performance of TI and TA approaches substantially improves MRD detection and supports the use of hybrid MRD testing for early risk stratification and MRD-guided clinical trials.