KEYNOTE-361: Quantitative ctDNA Analysis in Patients With Advanced UC

At the 2024 American Society of Clinical Oncology Annual Meeting, Thomas Powles, MBBS, MRCP, MD, and colleagues at the Barts Cancer Institute, Queen Mary University of London, are presenting a retrospective analysis of pretreatment and on-treatment circulating tumor DNA (ctDNA) data by clinical outcomes with pembrolizumab monotherapy versus chemotherapy in patients with advanced urothelial carcinoma (aUC) from the phase 3 KEYNOTE-361 trial.

ctDNA is regarded as a potential biomarker of disease in early-stage UC, but little is known about the utility of ctDNA in aUC.

In the KEYNOTE-361 trial, patients with previously untreated aUC were randomly assigned (1:1:1) to receive pembrolizumab plus chemotherapy, pembrolizumab alone, or chemotherapy alone. In this analysis, Dr. Powles and colleagues evaluated tumor tissue mutations with whole exome sequencing of the tumor and matched normal DNA. ctDNA changes from pretreatment cycle 1 (C1) to on-treatment cycle 2 (C2) were quantified by maximum variant allele frequency (maxVAF) of tumor tissue-specific mutations or Guardant Health molecular response score.

Researchers evaluated the association between C2/C1 ratios or baseline tumor-informed maxVAF and clinical outcomes, including overall response rate (ORR), progression-free survival (PFS), and overall survival (OS).

They analyzed ctDNA samples from 263 patients, of whom 131 received chemotherapy and 132 received pembrolizumab. Researchers noted that clinical characteristics and baseline ctDNA levels within the treatment arms were comparable.

Lower C1 maxVAF was found to be associated with improved ORR, PFS, and OS in the pembrolizumab arm (P<.01) and was robust to adjustment for tumor mutation burden and PD-L1, but not in the chemotherapy arm (P>.05). Larger C2 reductions in ctDNA levels were observed in the chemotherapy arm versus the pembrolizumab arm (median ratio of C2/C1 tumor-informed maxVAF, 0.03 vs 0.71, respectively). Similar observations were made for molecular response score.

Additionally, Dr. Powles and colleagues reported that ctDNA reductions were associated with improved ORR, PFS (P<.001), and OS (P<.01) for tumor-informed maxVAF in the chemotherapy arm. The molecular response score was significantly associated with ORR and PFS (P<.01). However, associations were stronger in the pembrolizumab arm (P<.001) and were robust to adjustment for tumor mutation burden and PD-L1.

ctDNA changes from C1 to C2 did not show independent explanatory value for OS when Response Evaluation Criteria In Solid Tumors v1.1 response status was added as a variable, they noted.

Researchers concluded that ctDNA levels at baseline may be prognostic for pembrolizumab, and reduction in ctDNA during a patient’s first treatment cycle may be associated with outcomes. “Distinct patterns of ctDNA response were observed with the different treatments, and stronger associations with long-term clinical outcome were observed with pembrolizumab,” they wrote, adding that “early ctDNA dynamics did not offer additional benefit in predicting outcome beyond radiographic assessment of tumor size change.”