Recurrence remains a major clinical challenge following curative-intent surgery for biliary tract cancer. Although postoperative surveillance commonly includes cross-sectional imaging and serum tumour markers such as carbohydrate antigen 19-9 and carcinoembryonic antigen, these methods may not reliably identify residual or recurrent disease at its earliest stage.
A multicentre real-world analysis has shown that circulating tumour DNA positivity after surgery was strongly associated with inferior relapse-free and overall survival in patients with resected stage I–III biliary tract cancer. During postdefinitive treatment surveillance, ctDNA also demonstrated greater prognostic value than CA19-9 and CEA.
Among 39 patients whose clinical recurrence was preceded by a positive ctDNA result, molecular recurrence was detected a median of 3.7 months before radiographic relapse.
The article, titled “Real-world analysis of ctDNA and other biomarkers in patients with curatively resected stage I–III biliary tract cancer,” is listed in the September 2026 issue of ESMO Gastrointestinal Oncology, Volume 13, article 100344.
Authors: M. Malla, A. Esmail, A. Tin, M.M. Khayat, L. Bathla, C. Bridges, C.M. Brewer, J. Ferguson, J.B. Ortiz, J. Ensor, K. Magee, J. Yu, A.R. He, T. Cannon, A. Winer, G.G. King, A. Mahipal, A. Nagarajan, J.W. Franses, A. Fungtammasan, A. Jurdi, M.C. Liu, S. Kodali, A.A. Connor, R.M. Ghobrial, R.D. Kim, and M. Abdelrahim.
Evaluating ctDNA After Curative-Intent Surgery
Biliary tract cancer includes intrahepatic and extrahepatic cholangiocarcinoma, gallbladder cancer, and ampullary carcinoma. Despite surgery and the use of adjuvant chemotherapy, recurrence rates remain high, and outcomes after radiologically detected relapse are poor.
ctDNA offers a tumour-specific, noninvasive approach to detecting molecular residual disease and monitoring for recurrence. Although postoperative ctDNA has demonstrated prognostic value in several gastrointestinal malignancies, evidence in resected biliary tract cancer has remained limited.
The investigators therefore assessed whether ctDNA status during the postoperative molecular residual disease window and subsequent surveillance was associated with recurrence and survival.
Study Design
The retrospective study analysed real-world data from prospectively collected longitudinal plasma samples in a multi-institutional cohort of patients with stage I–III resectable biliary tract cancer treated between July 2020 and February 2024. A total of 223 patients were initially identified. Fifty-six were excluded because of stage IV disease at diagnosis, unresectable disease, less than two months of follow-up, or missing ctDNA data. The final analysis included 167 assessable patients.
The cohort included:
- 74 patients with intrahepatic cholangiocarcinoma
- 35 with extrahepatic cholangiocarcinoma
- 30 with gallbladder cancer
- 28 with ampullary carcinoma
The median age was 68 years, and 92 patients, or 55.1%, were men. Stage I disease was present in 35 patients, stage II disease in 64, and stage III disease in 67; disease stage was unspecified in one patient. A total of 115 patients, or 68.9%, received adjuvant therapy. During follow-up, 70 patients experienced recurrence and 18 died. Median follow-up was 21 months, with a range of 2–97 months.
Tumour-Informed ctDNA Testing
ctDNA was analysed using a personalised, tumour-informed 16-plex multiplex polymerase chain reaction assay coupled with next-generation sequencing. Whole-exome sequencing was performed on tumour tissue and matched normal blood samples. Based on these results, primers were designed to track 16 patient-specific somatic single-nucleotide variants in plasma. Samples were classified as ctDNA-positive when at least two variants were detected above the assay’s predefined confidence threshold.
A total of 751 plasma samples were analysed, with an average of 4.5 samples per patient. The molecular residual disease window was defined as two to 12 weeks after surgery. The postdefinitive treatment surveillance window began after the MRD period in patients who did not receive adjuvant therapy or two weeks after completion of adjuvant chemotherapy.
The primary objective was relapse-free survival, measured from surgery until radiological or clinical progression or death from any cause. Overall survival and the comparative prognostic value of ctDNA, CA19-9, and CEA were also evaluated.
ctDNA Positivity During the MRD Window
Among the 82 patients with ctDNA results available during the postoperative MRD window, 19, or 23%, tested positive. ctDNA positivity during this period was significantly associated with inferior relapse-free survival compared with ctDNA-negative status:
- Hazard ratio: 9.33
- 95% confidence interval: 3.40–25.64
- P<0.0001
Estimated 12-month relapse-free survival was 12.9% among ctDNA-positive patients and 75.3% among ctDNA-negative patients. Postoperative ctDNA positivity was also associated with inferior overall survival, with a hazard ratio of 5.29. Estimated 12-month overall survival was 55.7% among ctDNA-positive patients, compared with 95.8% among those who were ctDNA-negative.
In multivariable analysis incorporating ctDNA status, biliary tract cancer subtype, disease stage, tumour grade, and resection margin status, ctDNA positivity during the MRD window remained the most statistically significant factor associated with relapse-free survival:
- Adjusted hazard ratio: 15.86
- 95% confidence interval: 4.69–53.60
- P<0.001
ctDNA During Postdefinitive Treatment Surveillance
During postdefinitive treatment surveillance, 31 of 82 evaluable patients, or 38%, tested ctDNA-positive. Compared with patients who remained serially ctDNA-negative, those who tested positive during surveillance had significantly shorter relapse-free survival:
- Hazard ratio: 18.12
- 95% confidence interval: 8.21–40.01
- P<0.0001
Estimated 12-month relapse-free survival was 32.3% among ctDNA-positive patients, compared with 91.8% among serially ctDNA-negative patients. ctDNA positivity during surveillance was also associated with inferior overall survival. The reported hazard ratio was 47.54, although the confidence interval was wide, ranging from 2.31 to 976.40. Estimated 12-month overall survival was 85.1% in the ctDNA-positive group and 98.0% in the serially ctDNA-negative group. In multivariable analysis, surveillance ctDNA positivity remained the most statistically significant prognostic factor associated with relapse-free survival:
- Adjusted hazard ratio: 14.93
- 95% confidence interval: 5.33–41.90
- P<0.001
For detecting recurrence during postdefinitive treatment surveillance, ctDNA demonstrated a sensitivity of 80.6% and specificity of 95.7%. Among 39 patients who experienced clinical recurrence and had a positive ctDNA result before recurrence, the median lead time between molecular and radiographic relapse was 3.7 months, with a range of 0.164–18.3 months.
Findings According to Adjuvant Therapy
The prognostic association between surveillance ctDNA positivity and recurrence was observed both in patients who received adjuvant chemotherapy and in those who did not. Among 46 patients who received adjuvant chemotherapy and had surveillance ctDNA results, positivity was associated with inferior relapse-free survival:
- Hazard ratio: 16.40
- 95% confidence interval: 5.46–49.28
- P<0.0001
Fifteen of 17 ctDNA-positive patients in this group experienced clinical recurrence. Among 36 patients who did not receive adjuvant chemotherapy, surveillance ctDNA positivity was also associated with inferior relapse-free survival:
- Hazard ratio: 28.72
- 95% confidence interval: 6.16–133.98
- P<0.0001
All 14 ctDNA-positive patients in this group experienced clinical recurrence. ctDNA dynamics during adjuvant treatment were also prognostic. Patients with persistent ctDNA positivity had substantially inferior relapse-free survival compared with those who remained persistently ctDNA-negative. Patients whose last available on-treatment result was positive also had poorer outcomes. These findings support the potential value of longitudinal ctDNA monitoring during and after adjuvant treatment. However, the study did not test whether changing treatment according to ctDNA dynamics improves clinical outcomes.
Comparison With CA19-9 and CEA
Unlike ctDNA, neither CA19-9 nor CEA was significantly associated with relapse-free survival during the MRD or postdefinitive treatment surveillance periods.
CA19-9 demonstrated:
- Sensitivity for recurrence detection: 42.9%
- Specificity: 61.8%
CEA demonstrated:
- Sensitivity: 33.3%
- Specificity: 55.6%
By comparison, surveillance ctDNA demonstrated a sensitivity of 80.6% and specificity of 95.7%. Serum biomarker data were available only for subsets of the study population. CA19-9 results were available for 50 patients during the MRD window and 62 during surveillance. CEA data were available for only 18 patients during the MRD window and 15 after definitive treatment.
Clinical Interpretation
The study establishes ctDNA as a strong prognostic biomarker in resected biliary tract cancer. A positive result during the early postoperative period identified patients with a substantially increased risk of recurrence and death, while positivity during longitudinal surveillance was strongly associated with subsequent clinical relapse.
The median 3.7-month lead time over radiographic recurrence suggests that ctDNA may complement imaging by identifying molecular relapse before it becomes radiologically apparent. However, the study did not evaluate whether treatment intervention based on ctDNA positivity improves survival. The study was observational and did not evaluate a ctDNA-guided treatment strategy. Prospective interventional studies are therefore needed to determine whether acting on postoperative or surveillance ctDNA positivity can improve outcomes.
A negative ctDNA result should also not be interpreted as confirmation that residual disease is absent. Low disease burden, variation in tumour DNA shedding, and the timing of blood collection may limit molecular detection, particularly when testing is performed at a single postoperative time point.
Study Limitations
The study had several important limitations. Although plasma samples were collected prospectively, the clinical analysis was retrospective. ctDNA testing was performed at the discretion of treating physicians, resulting in variable testing intervals and incomplete data during individual clinical windows. Imaging schedules were also not uniform, which may have influenced the estimated lead time between ctDNA positivity and radiographic recurrence.
The cohort included multiple anatomical subtypes and disease stages with different prognoses and patterns of ctDNA shedding. The inclusion of patients with ampullary carcinoma and stage I disease may have contributed to the relatively low 23% ctDNA positivity rate during the MRD window.
The surveillance analysis was also potentially affected by guarantee-time, or immortal time, bias because patients needed to remain recurrence-free long enough to undergo serial ctDNA testing. Time-dependent statistical modelling was used to reduce this risk, but residual bias could not be excluded. The number of patients with evaluable ctDNA measurements during the MRD and surveillance periods was smaller than the overall cohort, limiting detailed analyses according to disease subtype and molecular characteristics.
Finally, the study demonstrated the prognostic value of ctDNA but did not evaluate the clinical utility of ctDNA-guided treatment decisions. Several authors were employees of Natera, the company associated with the tumour-informed assay used in the analysis, and may own stock and/or stock options. Other investigators reported additional financial relationships.
Conclusion
In this real-world study of patients with curatively resected stage I–III biliary tract cancer, ctDNA positivity during the postoperative MRD window and postdefinitive treatment surveillance was strongly associated with recurrence and inferior survival. ctDNA was the most statistically significant prognostic factor for relapse-free survival in multivariable analyses and demonstrated greater prognostic performance than CA19-9 and CEA.
Among patients whose clinical recurrence was preceded by ctDNA positivity, molecular relapse was identified a median of 3.7 months before radiographic detection. These findings support ctDNA as a potentially valuable tool for postoperative prognostication and risk stratification in biliary tract cancer. Prospective studies are needed to determine whether ctDNA-guided surveillance or treatment strategies can translate earlier molecular detection into improved clinical outcomes.
The full article is available on the ESMO Gastrointestinal Oncology.
Find more videos on OncoDaily YouTube TV.