Clemens, Keelia

3052 Background: Minimal residual disease (MRD) detected using circulating tumor DNA (ctDNA) is associated with increased breast cancer (BC) relapse risk in patients (pts) with early BC after curative-intent surgery. However, the clinical utility of ctDNA-based MRD testing for post-surgical surveillance and risk stratification remains incompletely defined. We used real-world data to evaluate the performance of Guardant Reveal, a tissue-free epigenomic assay for MRD detection, and the clinical significance of a positive MRD test (+MRD) in pts with resected early BC. Methods: Data of pts with resected stage I-IIIA BC in the Guardant InfinityAI Data Library who underwent Guardant Reveal testing between February 2023 and September 2025 was analyzed. Pts with ≥1 post-surgical MRD were included if they had either no distant metastatic BC recurrence with ≥1 year of follow up after the last MRD test or a documented distant metastatic BC postoperatively; pts with a new non-breast primary malignancy were excluded. Sensitivity was assessed by site of distant metastasis and interval from MRD test to recurrence; nodal recurrences were excluded due to claims data limitations. Specificity was assessed using all tests from pts without recurrence and ≥1 year of follow up after the last post-treatment MRD test. Claims data were used to infer treatment changes following +MRD. Results: A total of 822 pts was analyzed. 114 pts with real-world clinico-genomics data were evaluable for sensitivity and 708 for specificity. One-year sensitivity for distant metastatic recurrence was 71%, and was highest for bone-only (97%, 35/36 cases) or lung (89%, 8/9 cases) recurrences and lowest for brain-only recurrences (33%, 1/4 cases). Sensitivity within 12 months of recurrence was higher in hormone receptor-positive (HR+)/HER2-negative BC (79%) and stage IIIA BC (81%) and for multiple-organ versus single-organ metastasis (83% versus 69%). Sample-level specificity was 94%. Following +MRD, 41 pts did not initiate or switch systemic therapy; 29 had distant BC recurrence, including 24 within 90 days after +MRD. Twenty pts switched therapy within 90 days of +MRD; 19 recurred, including 17 within 90 days after +MRD. Among 10 pts with post-recurrence MRD testing, only 3 had +MRD. Conclusions: In a real-world cohort of pts with resected early-stage BC, a tissue-free epigenomic ctDNA MRD assay demonstrated high specificity and clinically meaningful sensitivity for distant metastatic recurrence, particularly for bone-only and lung metastases and in pts with HR+ BC. These findings support ctDNA-based MRD testing as a noninvasive tool for postsurgical surveillance and risk stratification in early BC.

2561 Background: Some cancers, including breast, ovarian, pancreatic, and prostate cancers, are traditionally considered “cold” tumors, less likely to trigger an immune response and/or respond to immune checkpoint inhibitors (ICIs). However, some patients with these cancer types may benefit from ICIs. Here we investigate if a high blood tumor mutational burden (bTMB-H) may predict benefit to ICIs in such tumor types. Methods: We retrospectively queried the Guardant Health database for patients with advanced breast (n = 8531), ovarian (n = 1106), upper GI/pancreatic (n = 5110), and prostate (n = 7153) cancer who had ctDNA next-generation sequencing (Guardant360, Redwood City, CA) as part of routine clinical care from 2020-2022. Mutation profiles were evaluated for patients with bTMB-H (based on previously reported 80th percentiles for each tumor type, breast ≥15.31mut/Mb, ovarian ≥14.98mut/Mb, pancreatic ≥11.36mut/Mb, prostate ≥13.4mut/Mb) vs those with bTMB below this threshold (bTMB-L). A subset of cases were analyzed in a multi-institutional clinical cohort with a bTMB of ≥10mut/Mb (n = 35). Results: bTMB-H was observed in 9.5% of patients with breast cancer, 5.1% ovarian, 2.0% pancreatic, and 4.4% prostate. Genes more often mutated in patients with pTMB-H vs pTMB-L in each cancer type included TP53 and PIK3CA in breast cancer; TP53 and CHEK2 in ovarian; ATM and CHEK2 in pancreatic; TP53 and AR in prostate (all p values < 0.01). Patients in the clinical cohort (n = 35) had a median age of 74 (range: 40-85) 29/35 White / 4/35 Black, 27 prostate, 1 breast, 3 upper GI. Patients underwent first ctDNA analysis after a median of 3 (range 0-11) lines of therapy. The median bTMB detected was 16.59 mut/Mb (range: 10.02-152.85); six patients (6/35) were MSI-H. Twenty-three patients received ICI after a median of 6 (range: 0-8) lines of therapy, either alone (9, 39%) or in combination with hormonal therapy (5, 22%), chemotherapy (3, 13%) or targeted therapy (6, 26%). The median time to progression on ICI was 5.0 (95% CI, 3.6-6.4) months and median duration of response was 19.8 months (12.1-20.2). The ORR for evaluable patients was 13% and 74% of patients had progressed on therapy at time of data cut-off. Conclusions: "Cold” tumors represent a meaningful portion of patients with advanced cancer in which ICIs are not typically leveraged, and ctDNA was a feasible tool for identifying patients with bTMB-H who achieved durable responses from ICI-based therapy. The efficacy in this clinical cohort of traditionally “cold” tumors approaches other pan-cancer cohorts that include more traditionally immunoresponsive tumor types; further research is warranted to explore predictive value of bTMB, including potential tumor-specific thresholds, for optimizing patient selection for ICIs.