Detection of circulating tumor cells in cholangiocarcinoma via epithelial-mesenchymal transition markers and their application.

Conceptualization, Y.L., X.Y., T.H.S., W.L., and Y.Q.Z.; methodology, Y.L., X.Y., T.H.S., W.L., and Y.Q.Z.; investigation, Y.L., X.Y., T.H.S., Q.Y.L., T.H.J., W.L., and Y.Q.Z.; writing – original draft, Y.L., X.Y., T.H.S., Q.Y.L., T.H.J., W.L., and Y.Q.Z.; writing – review and editing, Y.L., X.Y., W.L., and Y.Q.Z.; funding acquisition, Y.L., W.L., and Y.Q.Z.; resources, Y.L., W.L., and Y.Q.Z.; supervision, Y.L., W.L., and Y.Q.Z.

The average particle size of Ep-LMB was 241.62 ± 6.62 nm with a polydispersity index (PDI) of 0.141 ( Figure 2 A); the average particle size of Vi-LMB was 238.73 ± 7.82 nm, with a PDI of 0.196 ( Figure 2 B). TEM observations showed that both Ep-LMB and Vi-LMB were of varying sizes and spheroidal shapes, with diameters distributed around 240 nm ( Figures 2 A and 2B). In the PBS system, the capture efficiency reached more than 90% when 10 μL of Ep-LMB and Vi-LMB were added sequentially for capture ( Figure 2 C). In the blood system, when 10 μL of Ep-LMB and Vi-LMB were added sequentially for capture, the capture efficiency reached more than 90% ( Figure 2 D). Therefore, we determined to use the capture protocol of sequentially adding 10 μL of Ep-LMB and Vi-LMB for capturing CTCs in the blood of patients with CCA. In addition, we validated capture efficiency in HUCCT1, RBE, and CCLP-1 cells, achieving an average capture efficiency of 94.66% in the PBS system ( Figure 2 E) and 91.06% in the blood system ( Figure 2 F). Figure 2 Characterization testing and capture efficiency detection (A) Particle size distribution and TEM morphology of Ep-LMB (scale bars, 50 nm); (B) particle size distribution and TEM morphology of Vi-LMB (scale bars, 50 nm); (C) in the PBS system, different dosages of Ep-LMB and Vi-LMB were tested for HUCCT1 cell capture efficiency; (D) in the blood system, different dosages of Ep-LMB and Vi-LMB were tested for HUCCT1 cell capture efficiency; (E) capture efficiency tests of Ep-LMB and Vi-LMB on HUCCT1, RBE, and CCLP-1 cells in the PBS system, respectively; and (F) capture efficiency testing of Ep-LMB and Vi-LMB on HUCCT1, RBE, and CCLP-1 cells, respectively, in the blood system. Data are represented as mean ± SD. Characterization testing and capture efficiency detection (A) Particle size distribution and TEM morphology of Ep-LMB (scale bars, 50 nm); (B) particle size distribution and TEM morphology of Vi-LMB (scale bars, 50 nm); (C) in the PBS system, different dosages of Ep-LMB and Vi-LMB were tested for HUCCT1 cell capture efficiency; (D) in the blood system, different dosages of Ep-LMB and Vi-LMB were tested for HUCCT1 cell capture efficiency; (E) capture efficiency tests of Ep-LMB and Vi-LMB on HUCCT1, RBE, and CCLP-1 cells in the PBS system, respectively; and (F) capture efficiency testing of Ep-LMB and Vi-LMB on HUCCT1, RBE, and CCLP-1 cells, respectively, in the blood system. Data are represented as mean ± SD. CTCs were identified in the blood of patients with CCA by immunofluorescence microscopy ( Figures S1 A and S1C). CTCs were identified when green fluorescence-labeled CK19-FITC was positive, blue fluorescence-labeled DAPI was positive, and red fluorescence-labeled CD45-PE was negative, and there was a clear cellular morphology under white light. In addition, we found CTC clusters in the blood of patients with CCA ( Figures S1 B and S1D). The definitive criteria for identifying epithelial-type CTCs (ECTC) were EpCAM+CK19+DAPI+CD45 − and mesenchymal-type CTCs (VCTC) were Vimentin+CK19+DAPI+CD45 − . Correlation analysis between the number of CTCs and clinicopathological parameters of patients with CCA showed ( Tables S1–S3 ) that the number of total CTCs had no significant correlation with gender and age, and had significant differences with tumor size, number of tumors, bile duct invasion, biliary hyperplasia, clinical stage, distal metastasis, and lymph node metastasis ( p < 0.01). As can be seen from the CTC distribution graph ( Figures S2 A−S2F, S3 A−S3F, and 3 A–3F), the numbers of ECTC, VCTC, and ECTC+VCTC were all significantly different from tumor size, bile duct invasion, bile duct hyperplasia, clinical stage, distal metastasis, and lymph node metastasis ( p < 0.05). In particular, the mean value of VCTC was significantly higher in patients with distal metastasis (3 ± 0.45 vs. 7 ± 0.32) and lymph node metastasis (3 ± 0.47 vs. 6 ± 0.32) compared with patients with CCA who did not have distal metastasis and those who did not have lymph node metastasis ( p < 0.001). This suggests that the number of VCTC is elevated when the tumor metastasizes. Figure 3 Distribution of CTCs in the blood of patients with CCA (A) Number of ECTC, VCTC, and ECTC+VCTC in the blood of patients with different stages; (B) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of patients with different stages; (C) distribution of ECTC, VCTC, ECTC+VCTC in the blood of metastatic patients; (D) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of metastatic patients; (E) distribution of ECTC, VCTC, ECTC+VCTC in the blood of patients with lymph node metastases; and (F) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of patients with lymph node metastases. CCA, cholangiocarcinoma; ECTC, EpCAM-CTC; VCTC, Vimentin-CTC. Data are represented as mean ± SD. ∗ p < 0.05 and ∗∗∗ p < 0.001. Distribution of CTCs in the blood of patients with CCA (A) Number of ECTC, VCTC, and ECTC+VCTC in the blood of patients with different stages; (B) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of patients with different stages; (C) distribution of ECTC, VCTC, ECTC+VCTC in the blood of metastatic patients; (D) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of metastatic patients; (E) distribution of ECTC, VCTC, ECTC+VCTC in the blood of patients with lymph node metastases; and (F) mean levels of ECTC, VCTC, and ECTC+VCTC in the blood of patients with lymph node metastases. CCA, cholangiocarcinoma; ECTC, EpCAM-CTC; VCTC, Vimentin-CTC. Data are represented as mean ± SD. ∗ p < 0.05 and ∗∗∗ p < 0.001. The statistical results of CTCs in the blood of patients with CCA showed that the mean values of ECTC, VCTC, and ECTC+VCTC were 4 ± 0.25, 4 ± 0.37, and 8 ± 0.51, respectively. The statistical results of CTCs in the blood of healthy individuals, patients with HRP and CCA showed that the mean values of ECTC, VCTC and ECTC+VCTC in the blood of patients with CCA were significantly higher than those of healthy individuals and HRP ( p < 0.001); the levels of ECTC, VCTC and ECTC+VCTC in the blood of HRP were significantly higher than those of healthy individuals ( p < 0.001); and the blood of patients with CCA with III+IV had significantly higher ECTC, VCTC and ECTC+VCTC levels were significantly higher in III+IV than in I + II ( p < 0.001) ( Figures S4 A−S4F and S5 ). CTC tests were performed on 49 patients with CCA 1 day before and 7 days after treatment ( Figures 4 A–4D). The results showed that the numbers of ECTC, VCTC, and ECTC+VCTC were significantly reduced at 7 days after treatment compared with 1 day before treatment ( p < 0.001). Correlation analysis of pre- and post-treatment VCTC levels with clinical information showed ( Table S4 ) that both pre- and post-treatment VCTC levels were significantly correlated with the number of tumors, bile duct invasion, biliary hyperplasia, lymph node metastasis, and distal tumor metastasis ( p < 0.05). These results suggest that VCTC may better reflect the metastatic status of CCA tumors. Figure 4 Distribution of the number of CTCs before and after treatment (A) Distribution of ECTC before and after treatment; (B) distribution of VCTC before and after treatment; (C) ECTC+VCTC distribution before and after treatment; and (D) ECTC+VCTC distribution before and after treatment. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC. Data are represented as mean ± SD. ∗∗∗ p < 0.001. Distribution of the number of CTCs before and after treatment (A) Distribution of ECTC before and after treatment; (B) distribution of VCTC before and after treatment; (C) ECTC+VCTC distribution before and after treatment; and (D) ECTC+VCTC distribution before and after treatment. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC. Data are represented as mean ± SD. ∗∗∗ p < 0.001. A total of 149 patients with CCA were followed up, and at the end of the follow-up, 73 patients died, and 76 patients survived, 40 males and 33 females out of the 73 dead patients, aged 34–87 years. Among the surviving patients, there were 37 males and 39 females, ranging in age from 30 to 86 years. Survival analysis of the patients showed ( Table S5 ) statistically significant median survival for tumor size, number of tumors, bile duct invasion, biliary hyperplasia, clinical stage, distal metastasis, and lymph node metastasis ( p < 0.05). Among the 73 patients who died, there were 47 cases of III+IV and 26 cases of I + II, with median survival periods of 23 and 47 months, respectively, which were significantly different ( p < 0.01) ( Figure 5 A). The median survival of patients who developed distal metastasis and lymph node metastasis was 25 and 23 months, which was much lower than the median survival of patients who did not develop metastasis (58 and 43 months), with a significant difference ( p < 0.001) ( Figures 5 B and 5C). When blood ECTC <4, the median survival was 36 months, which was not significantly different from the median survival when ECTC ≥4 ( Figure 5 D). Median survival was 46 months when VCTC ≤4 (average) and 23 months when VCTC >4, a significant difference between the two groups ( p < 0.001) ( Figure 5 E). The median survival for total CTCs at < 8 and ≥8 was 47 and 33 months, respectively, with a significant difference ( Figure 5 F). This suggests that when a tumor develops metastasis, the VCTC count is elevated, the survival time becomes shorter, and the prognosis worsens. Figure 5 Patient DFS curves (A) Relationship between clinical stage and DFS; (B) relationship between distal metastasis and DFS; (C) relationship between lymph node metastasis and DFS; (D) relationship between ECTC and DFS; (E) the relationship between VCTC and DFS; and (F) relationship of ECTC + VCTC to DFS. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC; DFS, disease-free survival. Patient DFS curves (A) Relationship between clinical stage and DFS; (B) relationship between distal metastasis and DFS; (C) relationship between lymph node metastasis and DFS; (D) relationship between ECTC and DFS; (E) the relationship between VCTC and DFS; and (F) relationship of ECTC + VCTC to DFS. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC; DFS, disease-free survival. A total of 149 patients with CCA were followed up, and the number of cases still alive after 5 years of follow-up was 70, with a 5-year DFS of 46.9%. There was no significant correlation between DFS and patients’ gender ( p = 0.350) and age ( p = 0.560), and there were significant differences with tumor size ( p < 0.001), number of tumors ( p < 0.001), bile duct invasion ( p = 0.007), and bile duct proliferation ( p < 0.001) ( Table S5 ). The 5-year DFS was 28.1% for stage I + II and 18.1% for stage III+IV, which was significantly different ( p < 0.001) ( Figure S6 A). When patients developed distal metastasis and lymph node metastasis, their 5-year DFS was 9.4% and 12.0%, respectively, and the 5-year DFS of patients without metastasis was 37.5% and 34.2%, respectively, with significant differences ( p < 0.001) ( Figures S6 B and S6C). When ECTC <4, VCTC ≤4, and ECTC+VCTC 4 (9.3%), and ECTC+VCTC ≥8 (12.0%) ( p < 0.05) ( Figures S6 D−S6F). It indicated that patients with tumor metastasis had an elevated risk of recurrence, in which an increase in the number of VCTC resulted in a significant decrease in DFS, suggesting that an increase in the number of VCTC was associated with a decrease in DFS. Additionally, a higher number of CTC clusters were detected in the blood of metastatic patients, with the majority of these clusters concentrated in VCTC ( Figures 6 A and 6B). Subsequently, the survival rates of patients exhibiting CTC clusters were analyzed. Patients with CTC clusters demonstrated a median overall survival of 41 months and a 5-year DFS of 2.01%, whereas those without CTC clusters had a median overall survival of 18 months and a 5-year DFS of 44.97%. These findings revealed significant differences ( p < 0.001) ( Figures 6 C and 6D). Subsequently, in the ROC curve analysis simulating prognostic diagnosis ( Figure 6 E), although the AUC values for VCTC and ECTC+VCTC (0.676 and 0.694) did not meet ideal standards, they remained higher than that of CEA (0.586). This indicates that CTC markers hold an advantage over the traditional tumor marker CEA in predicting prognosis for patients with CCA. Figure 6 Relationship between CTC clusters, metastasis, and survival rates, and the application of CTCs in prognostic diagnosis (A) CTC cluster levels in different distant metastases; (B) CTC cluster levels in different lymph node metastases; (C) relationship between CTC cluster levels and survival time; (D) relationship between CTC cluster levels and disease-free survival; and (E) ROC curve for simulated prognostic diagnosis. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC; ROC, receiver operating characteristic. Data are represented as mean ± SD. Relationship between CTC clusters, metastasis, and survival rates, and the application of CTCs in prognostic diagnosis (A) CTC cluster levels in different distant metastases; (B) CTC cluster levels in different lymph node metastases; (C) relationship between CTC cluster levels and survival time; (D) relationship between CTC cluster levels and disease-free survival; and (E) ROC curve for simulated prognostic diagnosis. ECTC, EpCAM-CTC; VCTC, Vimentin-CTC; ROC, receiver operating characteristic. Data are represented as mean ± SD.

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Yongqiang Zhu ( [email protected] ). This study did not generate new unique reagents. • Data: All data reported in this paper will be shared by the lead contact upon request. • Code: This paper does not report original code. • Additional information: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Data: All data reported in this paper will be shared by the lead contact upon request. Code: This paper does not report original code. Additional information: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

CCA is a malignant tumor with a high degree of malignancy, and its incidence, morbidity, and mortality have been increasing year by year in recent years. So far, there is a lack of effective tumor markers for biliary tract tumors. 23 , 24 With the development of testing technology, it has become possible to detect CTCs in peripheral blood. Studies have shown that CTCs and AFP were used to diagnose patients with hepatocellular carcinoma, and the diagnostic efficacy of CTCs for hepatocellular carcinoma was considered to be better than that of AFP, and the blood CTC level of patients with hepatocellular carcinoma was significantly higher than that of patients with benign liver disease. 25 , 26 In this study, patients with CCA were found to have significantly higher levels than healthy people and HRP, which is of great significance for the adjuvant progression and early screening and warning of patients with CCA. CTCs originates from solid tumors, and through gene replication and transcription, as well as protein modification and cellular phenotypic changes, the invasive power increases, and enters into the circulatory system through neovascularization or peripheral blood vessels, and in this change, EMT plays an important role. 27 , 28 Some studies have shown that EMT is widely present in malignant tumors such as lung cancer, colorectal cancer, breast cancer, and pancreatic cancer. 27 , 28 , 29 , 30 Hypoxia-induced signaling promotes EMT, inducing alterations in multiple cytoskeletal elements that contribute to hallmarks of invasive tumors, such as enhanced cell motility, weakened adherens junctions, metastatic dissemination, and downregulation of E-cadherin, which diminishes intercellular adhesion in epithelial tissues. Cyclophilin β relocates from the plasma membrane to the nucleus, wavelet protein expression increases, and matrix metalloproteinase production rises, contributing to the acquisition of mesenchymal traits. During EMT, CTCs undergo structural and molecular alterations: epithelial cells lose their characteristic polygonal, cobblestone morphology and acquire mesenchymal features, presenting a spindle-shaped mesenchymal appearance that enhances their invasive and migratory capabilities. 31 , 32 , 33 Furthermore, Vimentin was found to be highly expressed in a variety of tumor cells, especially in tumor cells with EMT. 21 , 34 Therefore, in this study, CTCs were captured in the blood of patients with CCA using tumor cell-specific markers before and after the occurrence of EMT, i.e., EpCAM and Vimentin phenotypes, and the results suggested that increased CTCs of Vimentin phenotype may indicate poor prognosis and high risk of tumor recurrence and metastasis. Previous studies have shown that the content of CTCs in peripheral blood is closely related to the pathological characteristics and stage of tumor. 35 , 36 In patients with CCA, the higher the peripheral blood CTC content, the worse the patient’s prognosis 37 ; a higher number of CTCs is positively correlated with the metastatic rate of regional lymph nodes in patients with CCA. 37 Some studies have shown that the presence of CTCs is positively associated with poor prognosis and that Vimentin-positive CTCs predict poorer survival, suggesting that Vimentin-positive CTCs detected by sensitive platforms could be used to improve the prognostic value of patients with advanced colorectal cancer. 38 In addition, it has also been demonstrated that the EMT process occurs when hepatocellular carcinoma CTCs are shed from the primary lesion into the peripheral vein, and that the spatiotemporal heterogeneity of CTC molecular phenotypes is closely related to postoperative intrahepatic recurrence and lung metastasis in patients with hepatocellular carcinoma. 39 Fründt et al. 1 used the CellSearch system to measure CTCs in the peripheral blood of patients with CCA, achieving a preoperative detection rate exceeding 50%. This finding was significantly correlated with survival duration in metastatic patients. Jiang et al. 40 discovered that 92.2% of patients with advanced intrahepatic CCA harbored more than one CTC, with higher CTC counts associated with poor prognosis. In this study, we found significant differences between the number of CTCs and tumor size, number of tumors, bile duct invasion, biliary hyperplasia, clinical stage, distal metastasis, and lymph node metastasis. Among them, the number of VCTC detected in patients with CCA with metastasis was significantly higher than that in patients with CCA without metastasis, indicating that the number of VCTC in the peripheral blood of patients with CCA at different stages was different, and that the VCTC gradually increased with the deterioration of the cancer, which suggests that the monitoring of the VCTC in patients with CCA is able to determine the progression of the disease and metastasis in patients. In addition, we found that when metastasis occurs in CCA, the higher the number of VCTC, the shorter the survival time, the lower the DFS, and the worse the prognosis. Several clinical studies have shown that mesenchymal CTCs are of great value in evaluating the efficacy and prognosis of breast cancer, prostate cancer, lung cancer, and other tumors. 41 , 42 , 43 , 44 , 45 , 46 In this study, we examined the CTCs of patients with CCA before and after treatment, and found that the degree of change in CTCs was different in different patients, with an overall trend of decreasing and reducing, and there was a significant difference between the number of CTCs before treatment and the number of CTCs after treatment, and patients with CCA with a higher-than-average VCTC before treatment were more likely to recur after treatment, and their tumors recurred for a shorter period of time. Therefore, changes in the number of VCTC can help to dynamically monitor the disease progression of patients with CCA in real time, which may provide clinical guidance for the evaluation of the efficacy of patients with CCA and the monitoring of recurrence and metastasis. This study also found that tumor cells meeting established CTC criteria could be detected in both healthy individuals and those at elevated cancer risk. We hypothesize that these tumor cells represent CTCs. This suggests such tumor cells may represent a highly promising new sample type for cancer health screening and early/ultra-early detection. 47 , 48 , 49 These findings align with the clinical phenomenon that primary tumors can metastasize via the bloodstream many years before symptoms appear and become diagnosable through medical imaging. 50 The CTC sorting system developed in this study demonstrates significant advantages in cost control. First, the system does not rely on expensive, large-scale instruments, substantially reducing initial investment and operational costs. Second, the immunolipid magnetic beads used in the system are independently developed and designed, ensuring controllable product quality and performance while further reducing dependence on external suppliers, thereby lowering detection costs. Furthermore, these proprietary immunolipid magnetic beads can be optimized and adjusted according to specific application requirements to meet tailored detection needs. Overall, the CTC sorting system developed in this study demonstrates excellent cost control, offering not only high economic efficiency but also broad application prospects and market potential. In conclusion, we successfully detected CTCs in the blood of patients with CCA using a CTC sorting technique and revealed the importance of CTCs as a tool for clinical diagnosis and prognostic assessment of CCA. This study has several limitations. First, this study involved only a small number of samples for CTC detection, and larger randomized clinical trials and multicenter joint studies are needed in the future. Second, we found the presence of CTC cell clusters in the blood of some patients with CCA during the detection process, but due to the sample size limitation, there was no statistical analysis of the joint patient prognosis, and the correlation between the CTC clusters and the patient’s prognosis needs to be further analyzed in future studies. Finally, we only detected the changes in CTC counts in patients with CCA before and 7 days after treatment, and whether the increase of VCTC counts after treatment implies the occurrence of tumor metastasis needs to be analyzed by further and longer monitoring of CTC dynamic changes in blood.

Cholangiocarcinoma (CCA) is a tumor originating from bile duct epithelial cells with a high degree of malignancy, and its diagnosis mainly relies on enhanced CT, MRI, and laboratory tests. 1 , 2 In recent years, the incidence of CCA has continued to rise worldwide. However, CCA has a high degree of malignancy, is prone to recurrence after surgery, and has no typical symptoms in the early stage, so most of the patients are already in the advanced stage at the time of diagnosis, and even if they are operated on in time, the 5-year survival rate is less than 10%. 3 , 4 , 5 Recurrence and metastasis are still the first cause of patient’s death. 3 , 4 , 5 , 6 , 7 Circulating tumor cells (CTCs), which are shed from primary or metastatic foci and enter the peripheral blood circulation, have tumor-initiating properties, especially when they survive in the peripheral blood and aggregate with each other to form tiny cancer emboli. They are more likely to promote the formation of metastatic foci and thus lead to cancer recurrence. 8 , 9 Therefore, the presence and number of CTCs in the peripheral blood represent both the ability of the primary tumor to infiltrate into the vasculature and its potential to form metastatic foci. Currently, many studies have evaluated the role and significance of CTCs as a biomarker in predicting recurrence, prognostic assessment, or therapeutic response in different tumors. 10 , 11 , 12 CellSearch system is the only method approved by the FDA for clinical CTC detection, and its basic principle is based on the capture of epithelial phenotypic CTCs (ECTC) by epithelial cell adhesion molecule (EpCAM) antibody, but this method has great limitations for CCA CTC detection in clinical application, and its sensitivity is low. 13 , 14 , 15 It has been shown that CTCs shed from the primary tumor foci into the blood circulation undergo epithelial-mesenchymal transition (EMT) and are transformed into mesenchymal-type CTCs (VCTC), which possess stem cell-like properties with stronger invasiveness and migration ability while losing the EpCAM antigen. 14 , 15 , 16 , 17 And VCTC will undergo mesenchymal epithelial transformation (MET) to form metastatic foci after passing through the surrounding stroma to enter the bloodstream and become tumor cells within the systemic circulatory system, and can form tumor metastasis in different organs after multiple blood circulations, which is an important cause of recurrent metastasis of tumors. 14 , 18 To address the above limitations of traditional CTC detection, a CTC sorting system with enhanced sensitivity ( Figure 1 ) was developed. Targeting EpCAM alone, this system achieves a CTCs capture efficiency 4.4 ± 1.2 times higher than the Cellsearch system. 19 Building upon this foundation, we combined EpCAM and Vimentin as targets, prepared EpCAM-immunolipid magnetic ball (Ep-LMB) and Vimentin-immunolipid magnetic ball (Vi-LMB), 20 , 21 , 22 respectively, and carried out the combined application to develop a Vi-LMB/Ep -LMB combined sorting system. In this study, we focused on capturing CTCs in the blood of patients with CCA. By monitoring changes in CTC levels in patient blood and integrating these findings with clinical and pathological data, we investigated the relationship with patient prognosis to explore the clinical significance and value of CTCs as an adjunctive marker for disease progression in patients with CCA. Figure 1 Schematic diagram of the preparation process and CTC assay for LMB (A) Flowchart of the preparation of LMB; (B) Schematic diagram of the clinical blood CTC test. Schematic diagram of the preparation process and CTC assay for LMB (A) Flowchart of the preparation of LMB; (B) Schematic diagram of the clinical blood CTC test.

The authors declare they have no competing interests.

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti-epithelial cells adhesion molecule (EpCAM) Abcam ab223582; RRID: AB_2762366 Anti-Vimentin Abcam ab92547; RRID: AB_10562134 Anti-Cytokeratin 19 (CK19) Abcam ab52625; RRID: AB_2281020 Anti-CD45 Abcam ab40763; RRID: AB_726545 Chemicals, peptides, and recombinant proteins DSPE-PEG Merck KGaA 880132P cholesterol Merck KGaA C8667 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC) MedChemExpress (MCE) HY-113424A Fe 3 O 4 Merck KGaA 700312 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) Merck KGaA 39391 N-hydroxysuccinimide (NHS) Merck KGaA 130672 Experimental models: Cell lines HUCCT1 ATCC HTX2115 RBE ATCC EY-X0709 CCLP-1 ATCC AC339759 Software and algorithms SPSS v. 25.0. χ2 IBM / Detailed information on all patients is provided in Tables S1–S3 . A total of 149 patients with CCA who were newly diagnosed and treated at our hospital between October 2019 and October 2025 were enrolled. Among them, 77 were male and 72 were female. Their ages ranged from 30 to 87 years, with a median age of 61 years and a mean age of 60.7 years. All enrolled patients were pathologically diagnosed with CCA during their initial hospitalization, had no prior history of the disease, and were receiving treatment for the first time. Follow-up periods for all patients were within 6 years, with survival and recurrence status determined through inquiries to patients' families via telephone, email, or follow-up examinations. Thirty healthy physical examiners were also recruited as negative controls, and blood was collected from 43 tumors high-risk group (HRP) for testing (including patients with cholecystitis, gallbladder polyps, gallbladder stones, porcelainized gallbladder, gallbladder adenomyosis, obesity, and diabetes mellitus). Among them, 17 were healthy males and 13 were healthy females, ranging in age from 30 to 84 years old, with a median age of 60 years and an average age of 60.9 years. HRP: 20 males and 23 females, aged 30 to 86 years, with a median age of 60 years and an average age of 60.4 years. The study was approved by the Ethics Committee of Putuo Hospital Shanghai University of Traditional Chinese Medicine (ID: PTEC-A-2019-18-1), and all participants signed an informed consent form before sample collection. Ep-LMB and Vi-LMB were prepared as follows: first, DSPE-PEG (Merck KGaA, 880132P), cholesterol (Merck KGaA, C8667), 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC, MedChemExpress (MCE), HY-113424A), Hexadecyl-quaternized (carboxymethyl) chitosans (HQCMC), and Fe 3 O 4 (Merck KGaA, 700312) were added to dichloromethane, respectively, and 6 mL of ddH 2 O was added using ultrasonic oscillations for 30 s. Ultrasonic oscillations were continued to 6 min, so as to make emulsification reaction of the mixed solution. Next, the LMB solution was obtained by spin evaporation on dichloromethane (0.09 MPa, 120 rpm) removed by rotary evaporator. Third, GHDC was dissolved in isopropanol, EpCAM (Abcam, ab223582) / Vimentin (Abcam, ab92547) was taken and dissolved in Dimethyloctadecyl[3-(trimethoxysilyl)propyl]ammonium chloride (GHDC) solution, and the coupling agents N-hydroxysuccinimide (NHS) and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) were added, and the solution was left to stand at 4°C overnight. Finally, EpCAM-GHDC/Vimentin-GHDC was dissolved in LMB and vortexed to obtain EpCAM/Vimentin-modified lipid magnetospheres, namely Ep-LMB and Vi-LMB ( Figure 1 A). Measure 10 μL of sample diluted in 1 mL of distilled water, and use BI-90Plus laser particle size meter to test the particle size; Measure 10 μL of sample dilution drops on the copper mesh, and wait for drying to observe the morphology by transmission electron microscope (TEM). We used a protocol that sequentially added Ep-LMB and Vi-LMB to capture CTCs. That is, Ep-LMB was added first to capture CTCs, and then Vi-LMB was added to capture CTCs. The capture efficiency was verified using HUCCT1 (ATCC, HTX2115), RBE (ATCC, EY-X0709), and CCLP-1 (ATCC, AC339759) cell lines in PBS and blood systems, respectively. Cell concentration gradients are 10, 50, 100, 200, 500, and 1000. A 7.5 mL blood sample was collected from CCA patients through the elbow vein, stored in a vacuum blood collection tube containing EDTA anticoagulant, preserved at 4°C, avoiding freezing during storage, handling and transportation, and tested within 72 h. The patients were followed up by telephone, mail, and review to observe the survival and recurrence after treatment, and to analyze the correlation between the post-treatment CTCs test results and the treatment effect, survival, and recurrence. Additionally, progression-free survival (PFS) and overall survival (OS) were calculated. 7.5 mL of peripheral blood was taken, and 10 μL Ep-LMB capture CTCs was added first, followed by 10 μL Vi-LMB capture CTCs. After addition, incubate at room temperature for 15 min respectively, shake every 5 min, and then put into the magnetic separation frame for adsorption for 10 min. Wash with 1 mL of PBS solution twice, then add 20 μL of CK19 (Abcam, ab52625)-FITC, 20 μL of DAPI staining solution, and 20 μL of CD45 (Abcam, ab40763)-PE staining solution, respectively, and avoid light staining for 15 min. After adding 1 mL ddH2O and washing for 2 times, the cells were resuspended with 20 μL ddH 2 O, dropped on slides, observed and counted under immunofluorescence microscope ( Figure 1 B). Data were analyzed using SPSS v. 25.0. χ2 test and Fisher's exact test were used for comparative analysis. PFS and OS were assessed using Kaplan-Meier curves and Log-rank tests were used. Factors associated with PFS or OS were determined by Cox regression analysis, in which the backward stepwise method was used in the multivariate analysis. Data are represented as mean ± SD. P < 0.05 (∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001) was considered a statistically significant difference.

We acknowledge the support of the Young Talents Program of Shanghai Health Commission ( 2022YQ037 ), Shanghai Putuo District Central hospital "clinical talents" program ( 2022-RCJC-07 ), and Shanghai Putuo district talent drainage start-up project ( 2024-YJRC-05 ) for this research.