Characterization of circulating tumor DNA (ctDNA) burden and its association with real-world overall survival among prostate cancer (PC) patients.

69 Background: Circulating tumor DNA (ctDNA) from blood-based liquid biopsies is emerging as a promising biomarker with potential prognostic and predictive value for managing patients across tumor types and therapeutic areas. Yet data and understanding of how ctDNA might be used as a prognostic biomarker in prostate cancer (PC) are limited. This study used real-world data (RWD) to evaluate the association between ctDNA burden and clinical characteristics and outcomes in PC. Methods: Prostate cancer patients were identified from the Guardant INFORM real-world clinical-genomic database, which connects ctDNA results obtained through the plasma-based next-generation sequencing Guardant360 assay (G360) to de-identified claims data. U.S. PC patients who underwent a G360 test between June 2014 and June 2023 were included. A 6-month baseline period prior to the first G360 test (index) was used to gather clinical information on these patients. ctDNA burden for each patient was defined as the maximum variant allele frequency (MVAF) of all somatic variants. The median MVAF across all samples was used to classify patients into high or low ctDNA burden groups. The association between ctDNA and real-world overall survival (rwOS) was assessed using log rank tests and multivariable Cox proportional hazards models adjusted for age, race/ethnicity, year of test, comorbidity indices, metastasis status, PC clinical state, and presence of androgen receptor ligand binding domain (AR LBD) mutations. Results: 16,757 PC patients were identified, of which, 13% had undetectable ctDNA burden. Using a median MVAF cut-off of 1.9% (range: 0.01% to 96.4%), 7,216 had low ctDNA burden and 7,330 had high ctDNA burden. Among patients with detectable ctDNA, higher ctDNA burden was observed in patients with metastatic castration resistant prostate cancer (N=5,996, median=2.4%), smokers (N=2,580, median=2.5%), with higher comorbidity burden (N=3,336, median=3.0%), and those AR LBD mutations (N=2,624, median=8.6%). Patients with high ctDNA burden showed significantly shorter rwOS after the index date (16.4 months [95%CI: 15.8-17.0]) than low and undetectable ctDNA burden groups (34.8 [95%CI: 32.8-37.0] and 53.1 months [95%CI: 48.6-65.9], respectively). Multivariable Cox models showed similar results (high ctDNA vs undetectable ctDNA hazard ratio [HR] = 3.16; low ctDNA vs undetectable ctDNA HR = 1.45; high ctDNA vs low ctDNA HR = 2.18). Conclusions: In this real-world study with large PC population, ctDNA burden varied by severity of disease and was negatively associated with overall survival after adjustment for other prognostic features. Our study adds to the body of evidence suggesting the utility of ctDNA as a potential prognostic biomarker in PC patients to monitor disease progression and patient outcomes in routine clinical practice.