Circulating tumor DNA (ctDNA) from a phase II study of adjuvant dostarlimab with pelvic radiation in locally advanced, mismatch repair–deficient (MMR-D) endometrial cancer (D-RT study).

Abstract

5613 Background: Although ctDNA is a promising, non-invasive tool for cancer evaluation and monitoring, it’s use in endometrial cancer (EC) is currently limited. We sought to evaluate ctDNA dynamics and associations with clinical outcomes as part of a novel study of adjuvant immune checkpoint blockade (ICB) with radiation (RT) in locally advanced MMR-D EC. Methods: The D-RT study was an investigator-initiated, single-arm, phase II trial evaluating the safety and efficacy of 5 cycles of dostarlimab given concurrently with standard pelvic intensity modulated RT (IMRT), NCT04774419. The study included patients with newly diagnosed clinical stage III/IVA MMR-D EC. Primary endpoint was progression-free survival (PFS) at 2 years defined by RECIST V1.1 compared with historical control. All patients consented to MSK-IMPACT tumor-normal genetic testing. We performed MSK-ACCESS targeted ctDNA sequencing in a subset of patients with at least 3 timepoints (baseline post-surgery, on-treatment, post-treatment/follow-up) and correlated with baseline tumor genomics and clinical outcomes. Results: We previously reported the primary endpoint in 31 evaluable patients with 2-year PFS of 80.6% (95% 1-sided CI 65.3-100%), meeting pre-specified threshold for a positive study (data cutoff 8/1/2025). Here, we report ctDNA results from a subset of 15 patients with 58 total samples. Median age was 62 years (range 51-89 years) with 1 stage IB (i+), 3 stage IIIA, 9 stage IIIC1, and 2 stage IIIC2 MMR-D ECs (FIGO 2009). Histology included 8 endometrioid grade I/II, 5 endometrioid grade III, and 2 dedifferentiated carcinomas. Six patients (40%) had detectable baseline ctDNA post-surgery, with 5/6 (84%) having visible/residual disease at baseline. Of these, 3/6 (50%) had eventual disease recurrence post adjuvant treatment. In 8 patients with visible disease post-surgery, only 2 did not have detectable baseline ctDNA. ctDNA cleared post-treatment in all but 2 patients who had aggressive, dedifferentiated carcinomas. Among the 5 patients with recurrence (4 distant, 1 local), ctDNA detection preceded or coincided with radiographic detection of disease in the 4 patients (80%) with distant disease. One patient had a biopsy-proven, vaginal cuff recurrence only and never had detectable ctDNA. In the 10 patients without disease recurrence, none had detectable ctDNA post-treatment. Conclusions: In patients with locally advanced MMR-D EC, ctDNA detection after surgery correlated with visible disease. ctDNA levels declined during dostarlimab and RT treatment for most patients, and ctDNA detection was correlated with disease recurrence in distant but not local disease, suggesting its potential as a non-invasive marker of disease. Future studies should explore the utility of ctDNA as a disease monitoring tool to guide therapy decisions. Clinical trial information: NCT04774419 .