KRAS mutations were more frequently detected in eCCA, with G12D (37.0%), G12V (24.0%), and Q61H (8.2%) being the main sites. KRAS G12/13 mutation is an important prognostic marker for CCA and is closely associated with shortened OS and recurrence-free survival (RFS) (sensitivity 80%, specificity 93%). Meanwhile, ddPCR analysis of plasma ctDNA showed that patients with higher KRAS G12/G13 mutant allele frequency (MAF) and CA19-9 levels (MAF >0.174% and CA19-9 >49.99 U/mL) had a significantly worse prognosis (15.8 vs. 39.0 months; p = 0.046) (Thongyoo et al., 2025). It can be found that ctDNA detection found that KRAS is associated with poor survival in CCA. This approach offers a less invasive diagnostic alternative to traditional biopsy methods and provides critical insights into the potential of cfDNA analysis to act as a predictive tool for patient survival. Although targeted therapies for the KRAS G12C mutation, such as sotolacizumab, have been approved by the FDA, the development of inhibitors targeting other KRAS mutation subtypes is still in its early stages. In addition, studies have shown that inhibition of casein kinase 2 (CK2) may affect metabolic pathways in KRAS mutant CCA, providing a new research direction for targeting KRAS-driven tumor metabolism and is expected to provide a potential new approach for the treatment of CCA (Lee D. S. et al., 2024). In addition, KRAS mutations induce high expression of PD-L1/CTLA-4 by activating the MAPK pathway, forming an immunosuppressive microenvironment, thereby leading to resistance to immunotherapy (Job et al., 2020).

Homologous recombination repair (HR) pathway gene mutations (such as TP53, BRCA2 and RAD51D) have also been shown to be associated with a poor prognosis in CCA. TP53 mutation is the most common mutation type in CCA, with a detection rate of up to 38.1% (Silverman et al., 2019). Sanger sequencing of bile cfDNA revealed that patients with HR pathway mutations had significantly shorter survival than those without mutations (P = 0.0049), which may be related to the accelerated tumor progression caused by genomic instability caused by DNA damage repair defects (Yin et al., 2024). In addition, the study also found that patients with HR pathway mutations may be more sensitive to PARP inhibitors (such as olaparib), suggesting that it can be used as a biomarker for targeted therapy to improve patient prognosis. However, there are still few clinical trials of PARP inhibitors for CCA, and their efficacy needs to be further verified.

It has been reported that approximately 13%–25% of iCCA patients have isocitrate dehydrogenase 1 (IDH1) mutations (Rizzo et al., 2021). Berchuck et al. performed NGS analysis on ctDNA in the blood of 1,671 patients with advanced BTC, and the results showed that 9.1% of patients had IDH1 mutations. Among patients with IDH1 mutations detected in tissues, 87% (41/47) of patients also had these mutations detected in ctDNA. This shows that IDH1 mutations have high tissue consistency and can be used for targeted therapy (Berchuck et al., 2022). Ivosidenib, an oral, selective mutant IDH1 inhibitor, has shown promising results in a Phase III trial, significantly improving progression-free survival (PFS) and OS in advanced CCA (Abou-Alfa et al., 2020a). Resistance is particularly evident in IDH1-targeted therapies and may be related to acquired resistance mutations in IDH1 or IDH2 that prevent ivosidenib from binding to its target site (Cleary et al., 2022).

The incidence of rearrangement or fusion changes (3.5%) was higher than that of amplification (2.6%), and mutation events were rare (0.9%) (Helsten et al., 2016). Alberto et al. conducted a retrospective study of 18 iCCA patients who were found to have FGFR2 fusions. Plasma samples were analyzed using a custom hybrid capture gene panel with NGS (VHIO-iCCA panel). The results showed that patients in the high ctDNA group had a worse prognosis than those in the low ctDNA group (median PFS 6.53 months vs. 13.3 months, P = 0.0018; median OS 10.6 months vs. 21.2 months, P = 0.0198). In addition, 16 patients (88.9%) were found to have positive FGFR2 fusion events in plasma ctDNA, proving that VHIO-iCCA can accurately detect FGFR2 fusion in plasma ctDNA, thereby quickly screening patients who benefit better from targeted therapy. (González-Medina et al., 2024). A variety of FGFR receptor inhibitors have been developed, such as infigratinib, futibatinib, pemigatinib, erdafitinib and derazantinib, etc., and phase I and phase II trials of these drugs have shown good results (Javle et al., 2018; Goyal et al., 2019; Abou-Alfa et al., 2020b; Bahleda et al., 2019; Mazzaferro et al., 2019). Goyal et al. reported evidence of acquired resistance to FGFR inhibitors in CCA. PCR and Sanger sequencing of plasma ctDNA were performed in three patients receiving FGFR inhibitors, and the results showed that the gatekeeper mutation FGFR2 V564F was found in all patients, suggesting that secondary mutations in the FGFR2 kinase domain are a potential resistance mechanism (Goyal et al., 2017).

Imaging examinations (ultrasound/CT/MRI) are still the first-line tools for CCA screening. Common manifestations include bile duct obstruction, dilatation, and masses. They can evaluate tumor morphology (such as local invasion, vascular encapsulation) and metastasis status (Khan et al., 2012; Rushbrook et al., 2024). However, imaging methods have low sensitivity for small tumors and early lesions, and can only provide structural information of the tumor, but cannot reveal the molecular biological characteristics of the tumor and pathological changes at the cellular level (European Association for the Study of the Liver, 2023). Endoscopic examination techniques such as endoscopic ultrasound (EUS), magnetic resonance cholangiopancreatography (MRCP) and endoscopic retrograde cholangiopancreatography (ERCP) can directly observe bile duct lesions, detect early lesions and tiny tumors, and can obtain pathological information when combined with tissue biopsy. For example, ERCP combined with brush cytology can obtain tissue or cells, but its sensitivity is low (45%), so negative cytology does not exclude the diagnosis of CCA (Navaneethan et al., 2015). However, endoscopic examination is greatly affected by the quality of tissue samples and the sampling location, and has low sensitivity. In addition, the entry of the endoscope into the bile duct may lead to related complications, such as pancreatitis and bleeding (Tamada et al., 2011). Pathological biopsy is still the gold standard for diagnosing bile duct cancer. Pathological diagnosis can be performed directly through brush cytology, fine needle puncture or percutaneous methods. However, tissue specimens are difficult to obtain and can only reflect information about the sampling site, which is greatly affected by sample heterogeneity. Moreover, as an invasive examination, it is easy to cause harm to patients and is not convenient for continuous monitoring of disease progression (Vaidyanathan et al., 2018).

Compared with the above traditional methods, liquid biopsy is a minimally invasive sample collection method, especially suitable for patients or high-risk groups who cannot obtain tissue samples (Heitzer et al., 2019). Carbohydrate antigen 19–9 (CA19-9) is currently the most commonly used serum tumor marker for CCA diagnosis. Its false positive rate in patients with biliary obstruction exceeds 50%, and Lewis antigen-negative individuals (about 10% of the population) cannot secrete CA19-9, which seriously limits its universality (Ballehaninna and Chamberlain, 2012; Liang et al., 2015; Luo et al., 2016). The American Association for the Study of Liver Diseases and the American College of Gastroenterology both pointed out that the clinical application of CA19-9 faces limitations in CCA (European Association for the Study of the Liver, 2022; Boberg et al., 2006; Levy et al., 2005; Sinakos et al., 2011; Lindor et al., 2015; Bowlus et al., 2023).

In recent years, ctDNA has demonstrated significant advantages in the early diagnosis, disease monitoring and prognosis assessment of CCA due to its non-invasive, real-time monitoring, and dynamic change monitoring. Genetic and epigenetic changes in ctDNA are closely related to changes in tumor tissue, which lay the foundation for the application of ctDNA in CCA (Ettrich et al., 2019). Currently, cfDNA detection of CCA has been achieved through multiple sample sources such as blood, bile, and duodenal fluid (Table 2 Comparison of liquid biopsy specimens for CCA).

Blood ctDNA: Multiple studies have confirmed that blood ctDNA is as effective as traditional tissue genomic analysis and can be used as a marker for the diagnosis of CCA (Wasenang et al., 2019; Yang J. D. et al., 2021; Awosika et al., 2024). Hwang S et al. showed that the ctDNA in the blood of patients with BTC was consistent with the gene profile of tumor tissue, with a sensitivity of 84.8% and a positive predictive value (PPV) of 79.4% (Hwang et al., 2024).

Bile cfDNA: As the environment for the growth of bile duct tumor cells, bile is rich in tumor markers released by tumor cells through paracrine or autocrine pathways, making the detection of bile tumor markers an important means for early diagnosis of bile duct cancer and monitoring of disease progression (Shu et al., 2024; Liu et al., 2023). Multiple studies have reported that the mutation spectrum of bile cfDNA (such as TP53, KRAS, IDH1) is highly consistent with that in tumor tissue (P < 0.001) (Shen et al., 2019; Arechederra et al., 2022; Yin et al., 2024). By combining machine learning to optimize variant screening methods, the sensitivity of bile cfDNA detection can be increased by 2 times (Ito et al., 2024). Moreover, the content of bile cfDNA is much higher than that of plasma ctDNA, reaching 68.2 times that of plasma. The detection rate of driver mutations (bile 54% vs. plasma 17%) and sensitivity (96.2% vs. 31.6%) are significantly higher than those in plasma (Driescher et al., 2020; Arrichiello et al., 2022; Ito et al., 2024; Yin et al., 2024). Although bile sample acquisition is more technically demanding than blood sampling, biliary drainage constitutes an integral part of the initial therapeutic protocol for CCA patients with obstructive jaundice. Consequently, bile cfDNA analysis is particularly suitable for patients with biliary tract obstruction (Kearney et al., 2023; Shu et al., 2024).

Duodenal fluid ctDNA: Duodenal fluid (DF) contains components from the bile duct, pancreatic duct, gastric juice, and intestinal juice, and is mainly affected by pancreatic juice and bile. A study on DF analysis found that the cfDNA concentration of DF was significantly higher than that of plasma cfDNA, and it was more advantageous than plasma cfDNA in detecting low-abundance variants, and could reflect the overall microenvironment of the pancreatic and biliary system (González-Medina et al., 2024; Tavano et al., 2024). Compared with the precise detection of CCA using bile cfDNA, DF is more suitable for the combined detection of biliary and pancreatic system tumors.

In addition to detecting specific mutations in ctDNA, cancer type-specific methyl groups can also be found (Khosla et al., 2024). Wasenang et al. used a qPCR-based methylation-sensitive high-resolution melting (MS-HRM) method and found that hypermethylation of OPCML, HOXA9, and HOXD9 genes in serum ctDNA can effectively distinguish CCA from benign biliary diseases (such as gallstones), with a sensitivity and specificity of 62.5% and 100%, respectively (Wasenang et al., 2019). The NGS-based targeted methylation detection system developed by the Circulating Cell-free Genome Atlas (CCGA; NCT02889978) Consortium has validated the diagnostic value of cfDNA methylation patterns in a multi-cancer cohort study covering CCA (Liu et al., 2020; 2018). This technology can simultaneously analyze hundreds of methylation sites, significantly improving the detection throughput. Yang et al. found that the sensitivity of reduced-representation bisulfite sequencing (RRBS) technology in the early detection of ctDNA methylation combination markers can reach 76% and the specificity is 94% (Yang J. D. et al., 2021). Moreover, among patients with low levels of the traditional marker serum CA19-9 (≤100 U/mL), the test was able to successfully identify 64% of patients who were suitable for transplantation or surgical resection. The above research results show that ctDNA methylation markers can be effectively used as an early screening and identification tool for CCA. In the future, its clinical application can be further broadened through combined detection with other tumor biopsy markers.

In summary, ctDNA and bile cfDNA have great clinical potential for the accurate detection of CCA-related gene mutations. They can not only make up for the shortcomings of tissue biopsy, but provide a more sensitive and accurate detection method for early diagnosis, disease progression monitoring and targeted therapy of CCA.

In the future, the integration of liquid biopsy with imaging genomics and artificial intelligence is expected to achieve full-cycle management of CCA: “Non-invasive early-stage cancer detection, precise classification, and dynamic monitoring”.

Early-stage CCA lacks specific clinical and diagnostic methods and has a poor prognosis, with a 5-year OS rate of only 3%–24% (Tawarungruang et al., 2021). After receiving the standard treatment of surgery combined with adjuvant therapy, 60%–70% of patients will still experience recurrence, and the 5-year survival rate is usually less than 25% (Brandi et al., 2012; Banales et al., 2016; Blechacz, 2017). Monitoring tools related to CCA prognosis are relatively lacking, and expanding prognostic monitoring methods is crucial to improving the clinical management of CCA.

Although CA19-9 and carcinoembryonic antigen (CEA) are widely used in the diagnosis and disease monitoring of CCA in clinical practice, their sensitivity and specificity are still low, and their role as independent prognostic markers is limited. In recent years, ctDNA has shown potential as a powerful prognostic biomarker for postoperative MRD in multiple cancer types (Huffman et al., 2022; Kotani et al., 2023; Hou et al., 2022). ctDNA can be used to detect tiny residual lesions early after surgery, thereby providing more accurate disease monitoring and prognosis assessment, and providing a basis for personalized medicine.

ctDNA positivity can serve as a strong predictor of decreased disease-free survival (DFS). Many studies in the CRC field (such as the GALAXY, BESPOKE, and DYNAMIC trials) have demonstrated that the level of ctDNA is closely related to the patient’s prognosis. Patients with persistently positive ctDNA or ctDNA conversion from negative to positive after surgery have a poor prognosis, while patients with ctDNA conversion from positive to negative have a significant survival benefits (Yu et al., 2025). The ctDNA mutation levels of 60 plasma samples were evaluated using NGS, and the results showed that patients with low ctDNA levels had a progression-free survival (PFS) of 12.3 months and an overall survival (OS) of 22.6 months, which were significantly longer than those of patients with high ctDNA levels (PFS = 5.8 months, OS = 9.4 months), suggesting that ctDNA can serve as a predictor of disease burden and treatment response (González-Medina et al., 2024).

The variant allele frequency (VAF) can reflect the degree of tumor clonal advantage. Hwang S et al. confirmed that the median OS of patients with VAF>3.9% in the NGS results of blood ctDNA was only 4.9 months, which was significantly lower than that of patients with VAF≤0.9% (16.4 months, p = 3.8 × 10⁻⁷) (Hwang et al., 2024). Allele frequency variance (AFV), as a measure of the uniformity of VAF distribution, can reflect the genomic instability (GI) of tumors. High AFV values are usually associated with greater tumor clonal heterogeneity, high chromosomal instability, and subclonal evolution, features that usually predict a poor prognosis. Liu R et al. proposed that AFV can be used as a new method to evaluate the prognosis of BTC. Patients with higher preoperative AFV levels had poorer OS (p = 0.004), and preoperative AFV showed better predictive value compared with postoperative AFV (Liu et al., 2024).

NGS is a commonly used ctNDA detection method in clinical practice, which can detect low-frequency mutations below 0.1%. At the same time, different panels of NGS can analyze the full spectrum of mutations such as point mutations, insertions and deletions (Indel), copy number variations (CNV), and fusion genes, reducing technical costs. Currently, NGS is widely used in MRD monitoring, prognosis assessment, and drug resistance mechanism research of CCA. However, recent studies have also showed that liver function status may affect NGS plasma ctDNA concentration determination (González-Medina et al., 2024). Studies have found that the NGS plasma ctDNA level in patients with liver dysfunction is 2.1 times higher than that in patients with normal liver function, which may be due to decreased liver metabolic capacity, resulting in reduced ctDNA clearance. Therefore, future studies should consider the impact of liver function on ctDNA load to avoid misjudging disease progression due to liver dysfunction.

Although radical resection can remove most of the tumor tissue, MRD may still exist, which cannot be detected by conventional imaging examinations. As a sensitive prognostic marker, ctDNA can effectively detect MRD and thus assess the risk of relapse in patients. Postoperative ctDNA detection can be used for early monitoring of CCA recurrence and provide a basis for adjuvant therapy decisions.

Compared with traditional markers and imaging examinations, ctDNA has higher sensitivity in predicting recurrence. In a study of 56 patients with curatively resected stage I-III BTC, blood ctDNA abnormalities were detected in 93.8% of recurrent patients at an average of 3.7 months in advance, and ctDNA had a stronger predictive ability for recurrence than traditional CA19-9 levels (HR = 1.17 [95% CI, 0.24 to 5.71]; P = 0.844) (Yu et al., 2025). A study by Reinert T et al. showed that ctDNA can detect recurrence several months earlier (174–222 days) than imaging examinations in monitoring tumor recurrence (Reinert et al., 2019).

A CCA-based subanalysis of the STAMP Phase II trial showed that ctDNA levels were positively correlated with tumor burden. During postoperative MRD monitoring, plasma ctDNA was evaluated by 16-plex PCR next-generation sequencing, and ctDNA positivity indicated that residual lesions were not completely eliminated and was significantly associated with worse DFS (HR = 1.8; 95% CI 1.06–3.07; p = 0.029) (Yoo Changhoon et al., 2024). During adjuvant chemotherapy, persistently positive ctDNA indicates that the patient will relapse 100% after surgery and their RFS is significantly shortened (Yoo C. et al., 2024).

These results show that ctDNA has a higher sensitivity in predicting recurrence than traditional biomarkers and imaging examinations. With the advancement of larger-scale prospective studies, tumor-specific ctDNA detection is expected to become an important recurrence monitoring biomarker in clinical practice and promote the establishment of a precise postoperative monitoring system.

Most CCA patients are already in the metastatic or locally advanced stage at the time of diagnosis, and chemotherapy (such as gemcitabine/cisplatin regimen) has become the standard treatment for unresectable or recurrent CCA. ctDNA plays an important role in this process. By monitoring the changes in specific gene mutations in ctDNA and the fluctuations in tumor load, it can reflect the patient’s response to treatment in real time. This can help patients optimize follow-up treatment strategies and provide timely personalized and precise treatment. Specifically, a decrease in ctDNA levels after chemotherapy indicates a reduction in tumor burden, indicating that the treatment is effective. If ctDNA turns negative after chemotherapy, it may indicate complete remission (CR) and the patient has a better treatment response. However, patients whose ctDNA levels do not decrease or continue to increase may indicate chemotherapy resistance or disease progression, with a higher risk of recurrence and significantly lower DFS than those whose ctDNA levels decrease. In addition, the study by Hwang S et al. showed that high variant allele frequency in ctDNA was closely associated with poor prognosis after gemcitabine/cisplatin chemotherapy, specifically manifested as a significant shortening of OS and PFS (p = 6.9 × 10⁻⁶ and p = 3.8 × 10⁻⁷) (Hwang et al., 2024). This study further demonstrated the important role of ctDNA in evaluating chemotherapy effects and prognosis.

Molecular targeted therapy and immunotherapy have gradually become the core of personalized treatment for CCA, especially in patients with specific gene mutations (such as FGFR2, IDH1, MSI-H, TMB-H) or ERBB2 gene amplification (Hwang et al., 2024). Currently, targeted therapy is widely used in patients with CCA, and dynamic monitoring of ctDNA can help predict the effect of targeted therapy. If the targeted mutation in ctDNA disappears or decreases, it usually means that the targeted therapy is effective and the patient is sensitive to the drug; if the targeted mutation does not decrease, it may indicate the presence of primary drug resistance, and the targeted therapy is not effective. However, patients may develop acquired resistance mutations during treatment, and ctDNA can be used to monitor the emergence of these resistance mutations, thereby helping to dynamically adjust treatment options (Goyal et al., 2017). Plasma NGS analysis results showed that approximately 34.3% of patients carried mutations that could be targeted for treatment, such as FGFR2, IDH1, MSI-H, and ERBB2, which makes ctDNA play an important role in personalized treatment decisions (Hwang et al., 2024). Individualized precision treatment is particularly critical for patients with different subtypes (Chen et al., 2024).

Currently, FGFR2 fusion mutations (approximately 10%–20%) and IDH1 gene mutations (approximately 15%–20%) are the most common mutations in CCA and have been widely used in clinical practice with FDA-approved targeted therapies (Abou-Alfa et al., 2020a; Uson Junior and Borad, 2022). Various FGFR inhibitors (such as infigratinib (Abou-Alfa et al., 2020a), Pemigatinib (Abou-Alfa et al., 2020b), Futibatinib (Goyal et al., 2023), and the recently observed Tasurgratinib (Morizane et al., 2025), etc.) have shown good therapeutic effects in CCA patients. However, patients with FGFR2 fusion may develop mutations such as FGFR2 V564F and N550K after receiving FGFR inhibitors, leading to drug resistance. Traditional ctDNA detection methods have a low sensitivity for FGFR2 fusion mutations, at only 18% (Vaidyanathan et al., 2018). However, there have been advances in ctDNA profiling techniques in recent years. Hwang S et al.'s study showed that ctDNA profiling can efficiently detect FGFR2 fusion mutations and comprehensively capture tumor genetic changes (Hwang et al., 2024). The optimized ctDNA method specifically targeting FGFR2 fusions (FGFR-Dx) developed by Julie W et al. has for the first time verified the feasibility of FGFR2 fusion detection at low VAF (0.5%) levels in BTC patients (Reeser et al., 2024). This method significantly improves the sensitivity of FGFR2 detection (92.9%, compared with the original 18%), and is lower in cost. It is expected to help more patients adjust their treatment plans in time when drug-resistant mutations appear early. The results of a randomized, double-blind, placebo-controlled phase 3 study (ClarIDHy) showed that ivosidenib can significantly improve the median PFS of patients with advanced IDH1 mutation CCA (2.7 months vs. 1.4 months, P < 0.0001), and has a good benefit in OS, but its objective response rate (tumor size reduction >30%) is only 2% (Abou-Alfa et al., 2020a). Patients receiving ivosidenib may develop new IDH1 R132C mutations or activating IDH2 mutations (R172V), resulting in drug failure (Harding et al., 2018; Lowery et al., 2019). ctDNA testing can identify these mutations in a timely manner, providing evidence for treatment adjustments.

Drug resistance is one of the main reasons for CCA treatment failure. Drug resistance can be divided into primary resistance (no initial response to treatment) and acquired resistance (disease progression during treatment). Its potential mechanisms include secondary mutations of driver genes (such as EGFR, PI3K/Akt, Erk and NF-κB), epigenetic remodeling (such as abnormal DNA methylation) and cell apoptosis escape (Varghese et al., 2021). These resistance mechanisms not only affect the efficacy of chemotherapy and targeted therapy, but may also accelerate tumor metastasis and recurrence. Therefore, real-time tracking of drug resistance is crucial for developing individualized treatment plans and evaluating patient prognosis. ctDNA can break through the limitations of a single tissue biopsy and simultaneously monitor multiple drug resistance mechanisms, providing a basis for dynamically adjusting treatment plans. A study of 42 patients with advanced gastrointestinal tumors showed that among 23 resistant patients, the positive rate of detecting blood ctDNA drug-resistant mutations using NGS and ddPCR reached 87% (20/23), significantly higher than the 48% (11/23) of tissue biopsy. The proportion of ctDNA detection of multiple drug resistance mechanisms (40%) was 4.4 times higher than that of tissue biopsy (9%) (Parikh et al., 2019). Nevertheless, the clinical significance of ctDNA in monitoring drug resistance in CCA has not yet been fully confirmed, and more prospective studies are still needed to verify its application value.

In addition to traditional tumor markers, recent studies have also found that the level of syncytin-1 in bile cfDNA can be used as a new biomarker, especially with potential in the early diagnosis of CCA. The study by He JD et al. showed that syncytin-1 performed better than CA19-9, CEA and AFP in the diagnosis of bile cfDNA, with an AUC of 0.805 (95% CI: 0.719–0.890, p < 0.001), a specificity of 90.2% and a sensitivity of 75%. Combining syncytin-1 with CA19-9 increased the AUC to 0.927 (0.901 95% CI: 0.877–0.978, p < 0.001) (He et al., 2025). The results show that the combination of the two can significantly improve the efficiency and accuracy of early diagnosis, providing an important basis for the early diagnosis of CCA.

In recent years, cell-free miRNA (cf-miRNA), like ctDNA, has shown promise in diagnosing a variety of diseases, including heart disease, infection, and various cancers (Biró et al., 2020; Song et al., 2020; Hadipour et al., 2023). Studies have shown that there is a significant positive correlation between ctDNA and cf-miRNA (Prasopdee et al., 2024). The combination of plasma cf-miRNA and ctDNA can significantly improve the sensitivity and specificity of differential diagnosis of CCA compared with ctDNA detection alone (83.33%, 100% vs. 75.00%, 95.83%). This means that the combination of the two can serve as a potential diagnostic tool to distinguish CCA from other diseases and further improve the accuracy of CCA diagnosis.

Circulating tumor cells (CTCs) spread in the bloodstream through epithelial-mesenchymal transition and are closely related to tumor metastasis. They have been actively studied as prognostic biomarkers (Han et al., 2021). A study conducted by Sung et al. demonstrated that the combined detection of CTC and ctDNA improved the accuracy of predicting recurrence in BTC patients after standard treatment (surgery + adjuvant chemotherapy) (AUC = 0.92, sensitivity 93.8%, specificity 87.5%), which was much higher than CA 19–9 (AUC = 0.75, sensitivity 75.0%, specificity 75.0%) (Park et al., 2024).

The multiplex detection scheme of ctDNA combined with other liquid biopsy markers is still in its infancy, and the data integration standards of different detection platforms have not yet been unified. In the future, with the help of artificial intelligence (AI) algorithms and the development of multicenter cohorts, the clinical operability of combined markers can be further verified.

ctDNA detection can analyze the dynamics of tumor clonal evolution in real time, overcome the limitations of traditional imaging in evaluating immunotherapy response (such as pseudoprogression, delayed response, etc.), and provide a non-invasive means to monitor tumor dynamics and immunotherapy response (Weber et al., 2024).

Li et al. developed a patient-specific ctDNA fingerprint based on NGS (including eight hotspot genes: BRAF, EGFR, ERBB2, KIT, KRAS, MET, NRAS, and PIK3CA), by monitoring ctDNA content fraction (CCF) levels and CCF fold changes. This method has improved the specificity and sensitivity of monitoring the response to immunotherapy in CCA and can identify tumor recurrence earlier than imaging (Li et al., 2020).

Xu et al. constructed a CNV risk scoring model based on copy number variation (CNV) in plasma ctDNA to predict the clinical outcomes of immune checkpoint inhibitor (ICI) treatment in advanced hepatobiliary cancer. The results showed that the low CNV risk group had better clinical outcomes than the high CNV risk group (median PFS: 6.17 months vs. 2.60 months, HR = 0.045; median OS: not reached vs. 6.5 months, HR = 0.39) (Yang X. et al., 2021).

Yu et al. used ctDNA analysis to guide the adjuvant treatment of a patient with stage III CCA after resection and achieved good results. The patient started standard capecitabine adjuvant therapy on the 50th day after surgery. ctDNA was positive for two consecutive times and CA19-9 continued to rise. Before obtaining evidence of radiological recurrence, the patient was given capecitabine combined with pembrolizumab. After 112 days, ctDNA was negative and CA19-9 returned to normal after 155 days (Yu et al., 2024). This case report reveals the guiding significance of ctDNA testing for perioperative adjuvant treatment of CCA, but due to the limited sample size, the results are less credible.

Currently, there are few studies on ctDNA and CCA immunotherapy. In the future, larger-scale prospective studies are needed to demonstrate the role and clinical application value of ctDNA in CCA immunotherapy.