Association of tumor-informed circulating tumor DNA dynamics with radiologic disease assessment in a single-center real-world sarcoma cohort

BACKGROUND: Sarcomas are rare and heterogeneous malignant tumors for which reliable blood-based biomarkers for longitudinal disease monitoring remain limited. Circulating tumor DNA (ctDNA) has emerged as a minimally invasive biomarker in several solid tumors, but its clinical utility in sarcoma remains incompletely defined because of biological heterogeneity, variable ctDNA shedding, and the limited availability of recurrent, panel-coverable genomic alterations across different subtypes. METHODS: We conducted a retrospective real-world study of patients with bone and soft tissue sarcomas treated at a certified high-volume sarcoma center. Patients underwent tumor-normal molecular profiling to identify tumor-informed ctDNA targets. Longitudinal plasma monitoring was performed using a unique molecular identifier (UMI)-based targeted sequencing approach in patients with an identifiable target. ctDNA findings were compared with radiologic assessment, treatment response, and clinical outcomes. In two selected patients with extended longitudinal follow-up, whole-genome-based molecular residual disease (MRD) analysis was performed as an exploratory analysis. RESULTS: Twenty patients were screened, of whom 14 (70%) had an identifiable tumor-informed ctDNA target and were included in the longitudinal analysis. A total of 67 longitudinal plasma samples were analyzed, with a median of 4 samples per patient (range: 2-10). A total of 67 paired liquid biopsy-radiologic assessments were included in the concordance analysis. Concordant assessments were observed in 59 of 67 cases (88.1%), corresponding to substantial agreement between ctDNA and imaging (Cohen's κ = 0.755; 95% CI, 0.598-0.913). Using radiologic assessment as the reference standard, ctDNA demonstrated a sensitivity of 80.0% and specificity of 94.6% for detecting progressive disease. These assessment-level estimates were descriptive and did not account for repeated measurements within patients. Descriptive review of the longitudinal ctDNA identified patterns consistent with radiologic and clinical disease course in individual patients. In a 3-month landmark analysis, patients with ctDNA-defined progression had numerically shorter progression-free survival than patients without ctDNA-defined progression (9.0 vs. 16.0 months), although the difference was not statistically significant. Exploratory whole-genome MRD analysis was concordant with radiologic disease course in two selected patients. CONCLUSIONS: Tumor-informed ctDNA monitoring was feasible in 70% of screened patients and showed high concordance with radiologic disease assessment in this real-world sarcoma cohort. These findings support prospective evaluation of ctDNA as an adjunct to imaging for longitudinal disease monitoring in patients with a suitable tumor-informed target, while broader tumor-informed approaches may be required for patients whose tumors are not adequately covered by predefined targeted panels.