Answers are grounded in this paper and cite their sources.
BACKGROUND: ctDNA is increasingly used to detect minimal residual disease, however, its sensitivity varies by tumor type. Its utility in CCA remains incompletely defined. METHODS: In this single-institution retrospective study, we evaluated ctDNA using a commercial tumor-informed assay (Signatera), compared with CA19-9, for predicting relapse after surgical resection during postoperative (<90 days) and surveillance (≥90 days) windows. We assessed associations between ctDNA status and clinical features, including resection pathology and sites of radiographic relapse. RESULTS: Of 48 patients with CCA who underwent resection between 2020-2025, 26 (54.2%) experienced relapse (median follow-up 18.3 months). ctDNA positivity during both postoperative and surveillance windows was associated with worse median disease-free survival, whereas CA19-9 was not. During the postoperative window, ctDNA sensitivity was 41.7% and specificity was 85.7% for predicting disease relapse any time post-resection. During the surveillance window, sensitivity was 60.0% and specificity was 90.9%. Landmark analysis for relapse at 1-year for ctDNA collected prior to 90 days had a sensitivity of 83.3% and specificity of 90.0%. Among patients with detectable ctDNA who subsequently developed radiographic relapse, median lead time from first ctDNA positivity to relapse was 34 days (IQR 26-132 days), with a maximum lead time of 292 days. ctDNA performance was higher among patients without perineural or lymphovascular invasion and was most frequently positive in those who developed extrahepatic metastases (n = 10/13, 76.9%). CONCLUSIONS: ctDNA is a specific prognostic biomarker that outperforms CA19-9 for predicting relapse after CCA. However, negative ctDNA results should be interpreted cautiously given modest assay sensitivity.
No takes yet. Share an insight, caveat, or question.
Guo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: