Prediction of PD-L1 expression and prognosis of mass-forming intrahepatic cholangiocarcinoma based on preoperative magnetic resonance imaging

Introduction

Intrahepatic cholangiocarcinoma is the second most common primary malignant liver tumor (1). There has been a concerning global rise in the incidence and mortality of its predominant variant, mass-forming intrahepatic cholangiocarcinoma (MICC) (2). Surgical resection remains a viable option for a minority of patients with MICC, as the majority are diagnosed at an advanced stage (3). The prognosis for patients with MICC is unfavorable, necessitating the development of novel and effective treatment strategies (4).

In recent years, there has been a growing focus on advancements in immunotherapy strategies. Notably, immune checkpoint blockades, such as for programmed cell death 1 (PD-1) and programmed cell death ligand 1 (PD-L1), have demonstrated significant potential in treating various malignant tumors (5-7). These agents offer optimism in achieving optimal treatment outcomes in patients with MICC. For example, a pivotal clinical trial evaluating the combination of durvalumab with chemotherapy revealed a substantial improvement in clinical outcomes. The estimated 24-month overall survival (OS) rate was 24.9% [95% confidence interval (CI): 17.9–32.5%] in the durvalumab plus chemotherapy cohort, as compared to 10.4% (95% CI: 4.7–18.8%) in the placebo group. Furthermore, the hazard ratio (HR) for progression-free survival (PFS) was 0.75 (95% CI: 0.63–0.89; P=0.001). The objective response rate (ORR) was higher in the experimental arm (26.7%) compared to the control group (18.7%) (8). In a separate investigation, the combination of toripalimab with a lenvatinib, gemcitabine, and oxaliplatin chemotherapy regimen demonstrated remarkable efficacy in advanced patients with intrahepatic cholangiocarcinoma, achieving an ORR of 80%. The median OS and PFS were 22.5 and 10.2 months, respectively (9).

However, only a subset of patients with tumor respond to blockade of the PD-1/PD-L1 pathway (10), and thus the primary challenge lies in identifying the subgroup of candidates who will derive the greatest benefit from this strategy (11) to avoid ineffective treatments and potential side effects associated with autoimmune reactions (12). Therefore, it is crucial to discover biomarkers that can predict immunotherapy response in patients with MICC (13). Previous studies have demonstrated that PD-L1-positive expression in tumors is linked to treatment response when the PD-1/PD-L1 pathway is inhibited (14). Therefore, accurately predicting the preoperative status of PD-L1 expression can inform the clinical practice related to immune checkpoint blockade in patients with MICC.

The gold standard for evaluating the expression status of PD-L1 in tumors is immunohistochemical staining after needle biopsy, but the heterogeneous expression of PD-L1 often confounds patient outcomes, limiting its clinical application. Molecular imaging is able to visualize the real-time expression of target molecules and cells, even in context of immune checkpoint blockade (15,16). However, the lack of validation for patients with MICC has restricted the clinical application of this strategy. Radiomics, which transforms medical images into quantitative data, can be used to predict the response to immune checkpoint inhibitors (17). However, it is limited in its robustness and reproducibility and may generate unreliable results (18), thus restricting its application.

Magnetic resonance imaging (MRI) offers the benefit of superior soft tissue contrast and the ability to evaluate various parameters, thereby offering insights into the underlying pathophysiological processes (18). The relationship between MRI features and PD-L1 expression has been explored in hepatocellular carcinoma (HCC) (18). Nevertheless, there is a lack of research on MRI features for predicting PD-L1 expression and survival in patients with MICC. Therefore, the purpose of this study was to investigate the potential of preoperative MRI in predicting PD-L1 expression and prognosis in patients with MICC. We present this article in accordance with the TRIPOD+AI reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-2130/rc).

Methods

Patients

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Clinical Research Ethics Committee of The First Affiliated Hospital, Zhejiang University School of Medicine (No. 2022-251). The requirement of individual consent for this study was waived due to the retrospective nature of the analysis. During the period from January 2017 to December 2018, The First Affiliated Hospital, Zhejiang University School of Medicine, identified cases of MICC that were pathologically proven and documented in medical records. The inclusion criteria for the study were the pathological confirmation of MICC, the availability of tumor specimens, presurgical MRI scans, absence of antitumor therapy before MRI scanning, complete follow-up data, and completion of routine preoperative laboratory examinations. The exclusion criteria were a lack of tumor specimens, administration of antitumor treatment before MRI scanning, missing or poor-quality MR imaging, loss to follow-up, and incomplete clinical data. A total of 92 patients with MICC were included in the study cohort during the specified timeframe. The flowchart of participant inclusion is provided in Figure 1.

Figure 1 Flowchart of patient enrollment.

Imaging acquisition

All MRI scans were performed with the same state-of-the-art system (MAGNETOM Trio 3.0-T, Siemens Healthineers, Erlangen, Germany). The standard sequences included T1-weighted imaging, in-and-out-of-phase, fat-suppressed T2-weighted imaging, diffusion-weighted imaging (DWI), and dynamic multiphase-enhanced imaging (volume interpolated breath-hold examination sequence: repetition time, 4.03 ms; echo time, 1.43 ms; flip angle, 9°; slice thickness, 2.3 mm; and field of view, 42×25 cm²). The contrast agent (Omniscan, GE HealthCare, Chicago, IL, USA) was administered at a dosage of 0.2 mL/kg at a rate of 3 mL/s, followed by a 30-mL saline flush. Arterial phase imaging was delayed by 20–35 s, whereas portal venous phase imaging occurred at 60–70 s’ postinjection. The detailed parameters for MRI scanning are available in Table S1.

Imaging evaluation

The magnetic resonance (MR) images of all the liver tumors were evaluated by two senior radiologists, each with extensive expertise in the field and a record of examining more than 5,000 liver MR images. Before the assessment, they underwent a comprehensive hands-on instructional session elucidating the intricacies of liver MRI evaluation. This included individual and joint analyses of 20 randomly selected MR images of MICC cases from the picture archiving and communication system database (not included in this study cohort). Within 1 week following the conclusion of the aforementioned session, both senior radiologists independently reviewed all the MR images of MICC case included in this cohort before conducting a joint reassessment after 2 weeks. Any discrepancies between the two senior radiologists were deliberated until a consensus was achieved. Individual assessment was employed to determine interobserver agreement, and concordant assessments were used for performance classification. In cases of multiple lesions, the largest lesion was reevaluated.

In the absence of established imaging guidelines specific to MICC, therefore, this study incorporated the most widely validated MRI features. The MRI evaluation system comprised primarily 10 imaging features, including shapes [regular (round or oval) or irregular (lobulated)], margins [well-defined (smooth) or ill-defined (infiltrative)], peritumoral bile duct dilatation (dilation of bile ducts adjacent to the tumor), hepatic lobe atrophy (shrinkage of a portion of the liver), satellite nodules (small tumor nodules near the primary tumor), lymph node metastasis (LNM) (enlarged or abnormal lymph nodes near the tumor), capsular retraction (a locally flattened or depressed contour of the extrahepatic region), the target sign on DWI (a hyperintense tumor in the periphery and a hypointense tumor in the central region), enhancement patterns in the arterial phase (categorized as hypo-/mild enhancement and hyperenhancement), and intratumor vascularity (presence of blood vessels within the tumor).

Histopathology

For the immunohistochemical analysis, 4-µm-thick tissue slides obtained from paraffin-embedded tumor specimens were processed. The tissue slides were stained with PD-L1 (14-5983-82; Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Immunohistochemistry of the paraffin slides was conducted with a Dako REAL EnVision Detection System in accordance with the manufacturer’s instructions (Agilent Technologies, Santa Clara, CA, USA). First, the paraffin-embedded tissue sections were dewaxed and hydrated. Following microwave antigen retrieval, ethylenediaminetetraacetic acid-based antigen retrieval buffer (pH 9.0) was prepared and preheated, after which the deparaffinized and rehydrated tissue slides were fully immersed in the buffer within a heat-resistant container. The slides were then heated in a microwave oven maintained at 95–100 ℃ for 20 minutes, with care taken to prevent excessive evaporation through periodic replenishment with distilled water. Following heating, the container was removed and allowed to cool at room temperature (20–25 ℃) for 30 minutes while the slides were kept submerged in the retrieval buffer. Finally, the slides were rinsed three times with phosphate-buffered saline (pH 7.4) for 5 minutes each immunohistochemical staining was conducted. Endogenous peroxidase activity was blocked by incubating the tissue slides in 0.3% H₂O₂ for 15 minutes, while nonspecific binding sites were blocked via a protein block (x0909; Agilent Technologies).

The tissue slides were subsequently incubated with mouse anti-human PDL-1 primary antibodies at a concentration of 5 µg/mL overnight at 4 ℃. The secondary antibodies, coupled with horseradish peroxidase, were then visualized via a detection kit (K5007; Agilent Technologies) according to the manufacturer’s instructions. The sections were visualized with 3,3'-diaminobenzidine under a microscope and delicately counterstained with hematoxylin.

The tissue sections were digitized via a state-of-the-art Zoomer digital pathology C9600-01 virtual slide light microscope scanner (Hamamatsu Photonics, Hamamatsu, Japan) following staining. For each immunization variable, two expert pathologists who were blinded to the clinical data independently selected five nonoverlapping and discontinuous regions for calculating the mean in the statistical analysis. The number of PD-L1-expressing cells was quantified at 400× (0.0484 mm²). PD-L1 positivity was determined on a per sample basis via a 5% expression cutoff value (PD-L1-positive tumor cells/total tumor cells). Patients whose PD-L1 expression was greater than 5% were classified as PD-L1 positive.

Follow-up

Patients were consistently followed up after surgery and were prospectively monitored through chest X-ray, computed tomography, and/or MR imaging at intervals of 3–6 months. OS was defined as the duration between surgery and either death or the last follow-up. Data were censored at the time of the last follow-up for surviving patients, while deaths were duly recorded.

Development of the OS model

The clinical OS model was developed on the basis of clinal-radiologic factors via multivariate Cox regression analysis. According to the median cutoff value in this model, patients with MICC were stratified into high- and low-risk groups. A log-rank test was then used to compare the different Kaplan-Meier (KM) survival curves. The performance of the model was assessed with the concordance index (C-index). Additionally, a nomogram was constructed for the clinical OS model to predict the probabilities of 1-, 3-, and 5-year OS, and a calibration curve was generated to evaluate its prognostic accuracy.

Statistical analysis

The t-test or the Mann-Whitney test was employed for numerical variables, whereas the Chi-squared test or Fisher’s exact test was used for categorical variables. Interobserver agreement for assessing the reliability of MRI evaluation was examined in accordance with a previous report (19). Logistic regression analysis was used to identify the significant MRI and clinicopathological factors associated with PD-L1 expression. The efficacy of the model in predicting PD-L1 expression was assessed through receiver operating characteristic (ROC) curve analysis and the area under the curve (AUC). OS was evaluated on the basis of KM survival curves and the Cox proportional hazards model. Statistical analysis was conducted via R software (version 3.5.2). A P value less than 0.05 was deemed significant in two-tailed analyses.

Results

Clinicopathological characteristics and MRI features

Among the 92 patients with MICC included, 32 had positive PD-L1 expression, and 60 had negative PD-L1 expression. The kappa values for the MR images ranged from 0.85 to 0.92. Specifically, kappa values were as follows: shapes, 0.92; margins, 0.88; peritumoral bile duct dilatation, 0.91; hepatic lobe atrophy, 0.87; satellite nodules, 0.89; LNM, 0.86; capsular retraction, 0.87; target sign on DWI, 0.87; enhancement patterns in the arterial phase, 0.85; and intratumor vascularity, 0.91. PD-L1 expression was found to be significantly associated with enhancement patterns in the arterial phase of enhanced MRI [odds ratio (OR): 0.12; 95% CI: 0.04–0.37] on the basis of multivariate analysis. The findings revealed that patients with PD-L1-positive MICC were predominantly associated with hypo-/mild enhancement, whereas those with PD-L1-negative MICC tended to exhibit greater hyperenhancement. Nevertheless, no statistically significant differences in clinicopathological factors were identified between the PD-L1-positive and PD-L1-negative groups. The results of the univariate analysis can be found in Tables 1,2.

Correlations of PD-L1 expression with patient prognosis and the prediction of PD-L1 expression

Compared with PD-L1-negative patients, patients with MICC with PD-L1-positive expression were significantly associated with poorer outcomes, as evidenced by a median survival time of 18.5 months [interquartile range (IQR), 10.8–29.2 months] for PD-L1-positive patients, in contrast to 34.5 months (IQR, 13.8–63 months) for PD-L1-negative patients. The 1-, 3-, and 5-year survival rates were also notably lower in the PD-L1-positive expression group (65.6%, 21.9%, and 21.9%, respectively) than in the PD-L1-negative expression group (76.7%, 46.7%, and 40%, respectively). A predictive model for PD-L1 expression was established on the basis of the enhancement patterns in the arterial phase of enhanced MR images. The model achieved an impressive AUC of 0.722 (95% CI: 0.632–0.811), with an accuracy of 0.685, a specificity of 0.6, and a sensitivity of 0.844 (Figure 2).

Figure 2 Prognostic and predictive analysis of PD-L1 expression in patients with MICC. (A) KM survival curves showing significant associations between PD-L1 expression status and OS. (B) ROC curve of the model for predicting PD-L1 expression status. AUC, area under the curve; CI, confidence interval; KM, Kaplan-Meier; MICC, mass-forming intrahepatic cholangiocarcinoma; OS, overall survival; PD-L1, programmed cell death ligand 1; ROC, receiver operating characteristic.

Performance of the clinical OS model

Hepatic lobe atrophy (HR: 2.036; 95% CI: 1.136–3.650), enhancement pattern (HR: 0.509; 95% CI: 0.295–0.877), LNM (HR: 1.715; 95% CI: 1.005–2.926), and carcinoembryonic antigen (CEA) (HR: 1.664; 95% CI: 1.007–2.750) were identified as prognostic factors for patients with MICC based on multivariate Cox regression analysis (Table 3). Consequently, these variables were integrated into the clinical OS model (C-index: 0.875; 95% CI: 0.812–0.905).

The formula for the clinical OS model was as follows: OS = 0.711 × hepatic lobe atrophy − 0.676 × enhancement patterns + 0.539 × LNM + 0.509 × CEA.

The model effectively stratified patients with MICC into high- and low-risk categories (median, 0.035, 95% CI: −0.167 to 0.373 vs. −0.843, 95% CI: −1.352 to 0.676) with a cutoff value of −0.66, with 46 patients in each group. The median survival times for the high- and low-risk groups were 15.5 months (IQR, 9.3–27 months) and 60.5 months (IQR, 20.5–64.0 months), respectively; furthermore, the 1-, 3-, and 5-year survival rates for the two groups were 56.5%, 17.4%, and 15.2%, respectively, and 89.1%, 58.7%, and 52.2%, respectively (P<0.001). Notably, the calibration curve of the clinical OS model exhibited exceptional concordance between the predictions and actual observations (Figure 3). An illustrative example is shown in Figure 4.

Figure 3 Evaluation of the performance and clinical application of the OS model. (A) A multiparameter nomogram. CEA is located on the CEA axis, and a vertical line is drawn to determine the corresponding points for each variable. The points obtained for each risk factor are summed, and the total sum is located on the total points axis. A vertical line is drawn downward to ascertain the patient’s OS probability. (B) The results of the KM survival analysis-based OS prediction model for patients. (C) Calibration curves for the prediction model for 1-, 3-, and 5-year OS. CEA, carcinoembryonic antigen; KM, Kaplan-Meier; OS, overall survival.

Figure 4 Representative imaging and immunohistochemical characteristics of PD-L1 expression in patients with MICC. (A1-A3) PD-L1-negative case. (A1) Arterial phase contrast-enhanced T1-weighted image. (A2) Portal venous phase contrast-enhanced T1-weighted image. (A3) Immunohistochemical staining showing negative PD-L1 expression [paraffin-embedded sections (4 µm) were stained with mouse anti-human monoclonal antibody clone 14-5983-82 (Invitrogen; 5 µg/mL) and visualized with the Dako REAL EnVision Detection System with 3,3'-diaminobenzidine chromogen and hematoxylin counterstain; original magnification 400×]. (B1-B3) PD-L1-positive case. (B1) Arterial phase contrast-enhanced T1-weighted image. (B2) Portal venous phase contrast-enhanced T1-weighted image. (B3) Immunohistochemical staining (400× magnification) demonstrating positive PD-L1 expression [≥5% tumor cell membrane staining; same protocol as that for (A3)]. MICC, mass-forming intrahepatic cholangiocarcinoma; PD-L1, programmed cell death ligand 1.

Discussion

In this investigation, we found that patients with MICC with positive PD-L1 expression demonstrated a more unfavorable prognosis than did those with negative PD-L1 expression. Multivariate logistic regression analysis revealed a correlation between the tumor enhancement patterns in the arterial phase of enhanced MR and the expression of PD-L1. Therefore, we developed a model for predicting PD-L1 expression, achieving an AUC of 0.722. Furthermore, our findings indicated that three MRI features and one clinical parameter were associated with the prognosis of patients with MICC. Leveraging these parameters, we established a MICC prognostic model with a C-index of 0.875, effectively stratifying patients with MICC into high- and low-risk groups. These results suggest that preoperative MRI may offer valuable guidance for immunotherapy and prognosis in patients with MICC.

We found a substantial correlation between positive PD-L1 expression and unfavorable outcomes in patients with MICC. The PD-1/PD-L1 pathway plays a role in promoting malignant potential and immune tolerance in MICC (20). However, previous studies are conflicting in terms of the impact of positive PD-L1 expression on the prognosis of patients with MICC (21-29). These discrepant data may be attributed to variations in evaluation methods, antibodies, and cutoff values. Furthermore, a meta-analysis indicated that positive PD-L1 expression does not correlate with outcomes in patients with cholangiocarcinoma (30). However, it is important to note that this conclusion was drawn cautiously because both MICC and extrahepatic cholangiocarcinoma were included and therefore cannot be directly applied to MICC.

Nonetheless, our findings are consistent with the established role of PD-L1 in HCC. Patients with PD-L1-positive HCC demonstrate significantly reduced OS and disease-free survival (31,32). As a result, future efforts should focus on elucidating the significance of PD-L1 expression in patients with MICC.

We observed a correlation between PD-L1 expression and enhancement patterns in the arterial phase of enhanced MR images, as radiographic images contain nonvisible information regarding the differences in protein expression in tumors (33). Our findings indicated that patients with PD-L1–positive MICC were predominantly associated with hypo-/mild enhancement in the arterial phase, whereas patients with PD-L1-negative MICC tended to exhibit hyperenhancement in the arterial phase. These findings align with other emerging evidence, including recent studies establishing arterial phase hypo-enhancement on MRI as a predictor of PD-L1 expression in HCC (34). Considering the enhancement patterns in the arterial phase, we devised a predictive model for PD-L1 expression, which yielded an impressive AUC of 0.722 and a sensitivity of 0.844. These findings are consistent with a recent high-impact study, which reported an AUC value of 0.714 for a predictive model assessing the efficacy of checkpoint inhibitor immunotherapy (35). This alignment suggests that our model, achieving an AUC of 0.722, retains meaningful predictive value despite its moderate performance, particularly in the context of patients with MICC, for whom robust and reliable predictive biomarkers are currently lacking. Although preoperative MRI demonstrates potential in predicting PD-L1 expression, the irreplaceable role of histopathology as the gold standard for biomarker assessment must be emphasized. Our findings represent a preliminary yet significant step toward integrating imaging and histopathological data, offering a foundation for future research to explore the complementary roles of noninvasive imaging and tissue-based analysis in guiding immunotherapy strategies.

We found that prominent enhancement in the arterial phase was associated with a favorable prognosis in patients with MICC, which is in line with previous studies (36,37) since MICC hypervascularity is associated with an abundance of tumor cells and minimal interstitial fibrosis (38). We also discovered liver lobe atrophy was an independent prognostic factor in patients with MICC. Nevertheless, the etiology remains elusive, and we postulate that it may be associated with the tumor infiltration of the bile ducts or portal vein. Further investigations are warranted to elucidate the specific mechanism involved. The aforementioned MRI features can predict the prognosis of MICC, which is in line with previous reports. For instance, the prognostic value of MRI has been shown to be associated with early progression in unresectable intrahepatic cholangiocarcinoma treated with combined targeted immunotherapy (39). A recent study highlighted the utility of DWI as an imaging biomarker for MICC survival, stratifying tumors based on restriction area (≥1/3 vs. <1/3 of the tumor) (40). In contrast, our study categorized DWI findings according to the presence or absence of the target sign, which showed no statistical significance. This discrepancy may stem from the differences in study populations, including variations in epidemiological characteristics, tumor biology, and imaging protocols, such as patient demographics, tumor stage, or DWI acquisition parameters. Our findings are consistent with other studies in HCC, in which MRI features were validated as prognostic predictors (41). Although tumor-specific biological differences may result in distinct prognostic signatures, MRI characteristics retain their utility as robust imaging biomarkers for outcome prediction across liver malignancies (41,42). LNM is a widely recognized malignant characteristic of tumors and is linked to unfavorable outcomes in MICC (43). In our study, we further found that LNM served as an independent prognostic risk indicator for patients with MICC, which is consistent with previous research (44). CEA, a serum biomarker reflecting tumor burden in MICC, was identified as an independent factor impacting the OS of patients with MICC, which aligns with findings from prior studies (45).

Our findings indicated that enhancement patterns in the arterial phase, including hepatic lobe atrophy, LNM, and CEA, were independent prognostic risk factors for patients with MICC. A clinical OS model incorporating these factors was developed, with a C-index of 0.875 and exceptional ability to stratify risk in patients with MICC. The calibration curve of the clinical OS model displayed remarkable concordance between the predictions and actual observations. This sophisticated clinical OS model can provide valuable support for clinical decision-making.

This study was subject to several limitations. First, many patients with MICC who did not undergo surgery or MRI scanning were excluded due to the retrospective design of the study, potentially introducing selection bias. Second, the study was conducted at a single center with a small sample size, and we will thus further expand the patient cohort and incorporate multicenter data in future studies to enhance statistical power and generalizability. Finally, we evaluated the prognostic significance of imaging biomarkers integrated with clinical variables. Although pairwise correlations between individual parameters (e.g., PD-L1 expression vs. CEA levels or specific MRI metrics) were not explicitly examined in this study, we plan to address this in the future.

Conclusions

This study demonstrated that MRI features could predict PD-L1 expression in patients with MICC. The integration of MRI features and clinical factors in the clinical OS model showed outstanding performance in stratifying patients with MICC into high- and low-risk groups. These findings suggest that the preoperative MRI could offer promising and noninvasive means to indicating the response to immunotherapy and the prognosis of patients with MICC.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD+AI reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-2130/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-24-2130/dss

Funding: This research was supported by the Zhejiang Provincial Natural Science Foundation of China (No. LTGY24H180001).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-2130/coif). X.Z. reports being a full-time employee of GE HealthCare during the conduct of the study. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Clinical Research Ethics Committee of The First Affiliated Hospital, Zhejiang University School of Medicine (No. 2022-251). The requirement of individual consent for this study was waived due to the retrospective nature of the analysis.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. El Homsi M, Alkhasawneh A, Arif-Tiwari H, Czeyda-Pommersheim F, Khasawneh H, Kierans AS, Paspulati RM, Singh C. Classification of intrahepatic cholangiocarcinoma. Abdom Radiol (NY) 2025;50:2522-32. [Crossref] [PubMed]
2. Qurashi M, Vithayathil M, Khan SA. Epidemiology of cholangiocarcinoma. Eur J Surg Oncol 2025;51:107064. [Crossref] [PubMed]
3. Zeng D, Wang Y, Wen N, Li B, Cheng N, Lu J. Short-term and Long-term Clinical Outcomes of Combined Caudate Lobectomy for Intrahepatic Cholangiocarcinoma Involving the Hepatic Hilus: A Propensity Score Analysis. Gut Liver 2025;19:438-53. [Crossref] [PubMed]
4. Song M, Li J, Li Y, Zhang C, Sigdel M, Hou R, Jiao D, Zhou X. Efficacy of microwave ablation for intrahepatic cholangiocarcinoma: a systematic review and meta-analysis. Quant Imaging Med Surg 2025;15:760-9. [Crossref] [PubMed]
5. Kaufman B, Abu-Ahmad M, Radinsky O, Gharra E, Manko T, Bhattacharya B, Gologan D, Erlichman N, Meshel T, Nuta Y, Cooks T, Elkabets M, Ben-Baruch A, Porgador A. N-glycosylation of PD-L1 modulates the efficacy of immune checkpoint blockades targeting PD-L1 and PD-1. Mol Cancer 2025;24:140. [Crossref] [PubMed]
6. Cox C, Brown S, Walpole E, Roy E, Dousset L, Ladwa R, Khosrotehrani K. Immune Checkpoint Inhibitors in Field Cancerization and Keratinocyte Cancer Prevention. JAMA Dermatol 2025;161:383-90. [Crossref] [PubMed]
7. Idris OA, Westgate D, Saadaie Jahromi B, Shebrain A, Zhang T, Ashour HM. PD-L1 Inhibitor Cosibelimab for Cutaneous Squamous Cell Carcinoma: Comprehensive Evaluation of Efficacy, Mechanism, and Clinical Trial Insights. Biomedicines 2025.
8. Oh DY, Ruth He A, Qin S, Chen LT, Okusaka T, Vogel A, et al. Durvalumab plus Gemcitabine and Cisplatin in Advanced Biliary Tract Cancer. NEJM Evid 2022;1:EVIDoa2200015.
9. Shi GM, Huang XY, Wu D, Sun HC, Liang F, Ji Y, et al. Toripalimab combined with lenvatinib and GEMOX is a promising regimen as first-line treatment for advanced intrahepatic cholangiocarcinoma: a single-center, single-arm, phase 2 study. Signal Transduct Target Ther 2023;8:106. [Crossref] [PubMed]
10. Zabeti Touchaei A, Vahidi S. MicroRNAs as regulators of immune checkpoints in cancer immunotherapy: targeting PD-1/PD-L1 and CTLA-4 pathways. Cancer Cell Int 2024;24:102. [Crossref] [PubMed]
11. Moniot A, Schneider C, Chardin L, Yaniz-Galende E, Genestie C, Etiennot M, et al. The CD47/TSP-1 axis: a promising avenue for ovarian cancer treatment and biomarker research. Mol Cancer 2024;23:166. [Crossref] [PubMed]
12. Dong J, Liu M, Li S, Zhang S, Fu P, Liu M, Jin S, Fan C, Fang M, Wu L, Li Z. Adverse reactions associated with SSRIs and PD-1/PD-L1 inhibitors: a disproportionality analysis of the FDA Adverse Event Reporting System. Expert Opin Drug Saf 2025;24:453-60. [Crossref] [PubMed]
13. Yan H, Deng Q, Meng Y, Zhang Y, Wu J, Liu W. IL-21 and IL-33 May Be Effective Biomarkers to Predict the Efficacy of PD-1 Monoclonal Antibody for Advanced Cholangiocarcinoma. Cancer Biother Radiopharm 2025;40:78-88. [Crossref] [PubMed]
14. Liu F, Zhang X, Lu M, Liu C, Zhang X, Chu Q, Chen Y, Zhang P. The association of genomic alterations with PD-L1 expression in Chinese patients with EGFR/ALK wild-type lung adenocarcinoma and potential predictive value of Hippo pathway mutations to immunotherapy. Cancer Med 2024;13:e7038. [Crossref] [PubMed]
15. Zhao L, Ge J, Zhang R, Wang H, Liu X, Xu K, Liu Y, Zhao W, Zhang W, Ye L, Chen Z, Zeng J, He Y, Gao M. Noninvasive Immunotyping and Immunotherapy Monitoring of Lung Cancers via Nuclear Imaging of LAG-3 and PD-L1. Adv Sci (Weinh) 2025;12:e2404231. [Crossref] [PubMed]
16. Wang Y, He K, Zhang Y, Chen Y, Wang S, Zhao K, Liu Z, Hu M. Peptide-based immuno-PET/CT monitoring of dynamic PD-L1 expression during glioblastoma radiotherapy. J Pharm Anal 2025;15:101082. [Crossref] [PubMed]
17. Liu Z, Yao Y, Zhao M, Zhao Q, Xue J, Huang Y, Qin S. Radiomics Models Derived From Arterial-Phase-Enhanced CT Reliably Predict Both PD-L1 Expression and Immunotherapy Prognosis in Non-small Cell Lung Cancer: A Retrospective, Multicenter Cohort Study. Acad Radiol 2025;32:493-505. [Crossref] [PubMed]
18. Sun L, Mu L, Zhou J, Tang W, Zhang L, Xie S, Chen J, Wang J. Imaging features of gadoxetic acid-enhanced MR imaging for evaluation of tumor-infiltrating CD8 cells and PD-L1 expression in hepatocellular carcinoma. Cancer Immunol Immunother 2022;71:25-38. [Crossref] [PubMed]
19. Zhang J, Wu Z, Zhang X, Liu S, Zhao J, Yuan F, Shi Y, Song B. Machine learning: an approach to preoperatively predict PD-1/PD-L1 expression and outcome in intrahepatic cholangiocarcinoma using MRI biomarkers. ESMO Open 2020;5:e000910. [Crossref] [PubMed]
20. Huang S, Hua C, Ding B, Chen J, Zheng S, Ding C. Advances in improving the efficacy of anti-PD-1/PD-L1 therapy in intrahepatic cholangiocarcinoma. Crit Rev Oncol Hematol 2025; Epub ahead of print. [Crossref]
21. Lu JC, Zeng HY, Sun QM, Meng QN, Huang XY, Zhang PF, et al. Distinct PD-L1/PD1 Profiles and Clinical Implications in Intrahepatic Cholangiocarcinoma Patients with Different Risk Factors. Theranostics 2019;9:4678-87. [Crossref] [PubMed]
22. Gani F, Nagarajan N, Kim Y, Zhu Q, Luan L, Bhaijjee F, Anders RA, Pawlik TM. Program Death 1 Immune Checkpoint and Tumor Microenvironment: Implications for Patients With Intrahepatic Cholangiocarcinoma. Ann Surg Oncol 2016;23:2610-7. [Crossref] [PubMed]
23. Jing CY, Fu YP, Yi Y, Zhang MX, Zheng SS, Huang JL, Gan W, Xu X, Lin JJ, Zhang J, Qiu SJ, Zhang BH. HHLA2 in intrahepatic cholangiocarcinoma: an immune checkpoint with prognostic significance and wider expression compared with PD-L1. J Immunother Cancer 2019;7:77. [Crossref] [PubMed]
24. Asahi Y, Hatanaka KC, Hatanaka Y, Kamiyama T, Orimo T, Shimada S, Nagatsu A, Sakamoto Y, Kamachi H, Kobayashi N, Fukai M, Taketomi A. Prognostic impact of CD8+ T cell distribution and its association with the HLA class I expression in intrahepatic cholangiocarcinoma. Surg Today 2020;50:931-40. [Crossref] [PubMed]
25. Dong Z, Liao B, Shen W, Sui C, Yang J. Expression of Programmed Death Ligand 1 Is Associated with the Prognosis of Intrahepatic Cholangiocarcinoma. Dig Dis Sci 2020;65:480-8. [Crossref] [PubMed]
26. Sabbatino F, Villani V, Yearley JH, Deshpande V, Cai L, Konstantinidis IT, Moon C, Nota S, Wang Y, Al-Sukaini A, Zhu AX, Goyal L, Ting DT, Bardeesy N, Hong TS, Fernandez-del Castillo C, Tanabe KK, Lillemoe KD, Ferrone S, Ferrone CR. PD-L1 and HLA Class I Antigen Expression and Clinical Course of the Disease in Intrahepatic Cholangiocarcinoma. Clin Cancer Res 2016;22:470-8. [Crossref] [PubMed]
27. Wu H, Wei Y, Jian M, Lu H, Song Q, Hao L, Yue Y. Clinicopathological and Prognostic Significance of Immunoscore and PD-L1 in Intrahepatic Cholangiocarcinoma. Onco Targets Ther 2021;14:39-51. [Crossref] [PubMed]
28. Deng M, Li SH, Fu X, Yan XP, Chen J, Qiu YD, Guo RP. Relationship between PD-L1 expression, CD8+ T-cell infiltration and prognosis in intrahepatic cholangiocarcinoma patients. Cancer Cell Int 2021;21:371. [Crossref] [PubMed]
29. Zhu C, Ma J, Zhu K, Yu L, Zheng B, Rao D, Zhang S, Dong L, Gao Q, Zhang X, Xie D. Spatial immunophenotypes predict clinical outcome in intrahepatic cholangiocarcinoma. JHEP Rep 2023;5:100762. [Crossref] [PubMed]
30. Xu G, Sun L, Li Y, Xie F, Zhou X, Yang H, Du S, Xu H, Mao Y. The Clinicopathological and Prognostic Value of PD-L1 Expression in Cholangiocarcinoma: A Meta-Analysis. Front Oncol 2019;9:897. [Crossref] [PubMed]
31. Kong P, Yang H, Tong Q, Dong X, Yi MA, Yan D. Expression of tumor-associated macrophages and PD-L1 in patients with hepatocellular carcinoma and construction of a prognostic model. J Cancer Res Clin Oncol 2023;149:10685-700. [Crossref] [PubMed]
32. Li JH, Ma WJ, Wang GG, Jiang X, Chen X, Wu L, Liu ZS, Zeng XT, Zhou FL, Yuan YF. Clinicopathologic Significance and Prognostic Value of Programmed Cell Death Ligand 1 (PD-L1) in Patients With Hepatocellular Carcinoma: A Meta-Analysis. Front Immunol 2018;9:2077. [Crossref] [PubMed]
33. Chang H, Chen J, Wang D, Li H, Ming L, Li Y, Yu D, Yang YX, Kong P, Jia W, Yan Q, Liu X, Zeng Q. Multimodal apparent diffusion MRI model in noninvasive evaluation of breast cancer and Ki-67 expression. Cancer Imaging 2024;24:137. [Crossref] [PubMed]
34. Kang JG, Han K, Chung T, Rhee H. Prediction of PD-L1 expression in unresectable hepatocellular carcinoma with gadoxetic acid-enhanced MRI. Eur J Radiol 2024;181:111772. [Crossref] [PubMed]
35. Yoo SK, Fitzgerald CW, Cho BA, Fitzgerald BG, Han C, Koh ES, et al. Prediction of checkpoint inhibitor immunotherapy efficacy for cancer using routine blood tests and clinical data. Nat Med 2025;31:869-80. [Crossref] [PubMed]
36. Li X, Yoh T, Shimada K, Hori Y, Koyama Y, Ogiso S, Ishii T, Isoda H, Nakamoto Y, Hatano E. Outcomes and Role of Lymphadenectomy in Hypervascular Intrahepatic Cholangiocarcinoma Based on CT-Vascularity. World J Surg 2025;49:708-16. [Crossref] [PubMed]
37. Nanashima A, Abo T, Murakami G, Matsumoto A, Tou K, Takeshita H, Kunizaki M, Hidaka S, Sakamoto I, Hayashi H, Fukuda T, Kudo T, Nagayasu T. Intrahepatic cholangiocarcinoma: relationship between tumor imaging enhancement by measuring attenuation and clinicopathologic characteristics. Abdom Imaging 2013;38:785-92. [Crossref] [PubMed]
38. Yoshida Y, Imai Y, Murakami T, Nishikawa M, Kurokawa M, Yonezawa T, Tokunaga K, Fukushima Y, Wakasa K, Kim T, Nakamura H, Sakon M, Monden M. Intrahepatic cholangiocarcinoma with marked hypervascularity. Abdom Imaging 1999;24:66-8. [Crossref] [PubMed]
39. Sheng R, Sun W, Huang X, Jin K, Gao S, Zeng M, Wu D, Shi G. Apparent Diffusion Coefficient MRI Shows Association With Early Progression of Unresectable Intrahepatic Cholangiocarcinoma With Combined Targeted-Immunotherapy. J Magn Reson Imaging 2023;57:275-84. [Crossref] [PubMed]
40. Promsorn J, Eurboonyanun K, Chadbunchachai P, Apivatanasiri C, Wirasorn K, Chindaprasirt J, Sookprasert A, Harisinghani M. Diffusion-weighted imaging as an imaging biomarker for assessing survival of patients with intrahepatic mass-forming cholangiocarcinoma. Abdom Radiol (NY) 2022;47:2811-21. [Crossref] [PubMed]
41. Jiang H, Qin Y, Wei H, Zheng T, Yang T, Wu Y, Ding C, Chernyak V, Ronot M, Fowler KJ, Chen W, Bashir MR, Song B. Prognostic MRI features to predict postresection survivals for very early to intermediate stage hepatocellular carcinoma. Eur Radiol 2024;34:3163-82. [Crossref] [PubMed]
42. Jiang H, Wei H, Yang T, Qin Y, Wu Y, Chen W, Shi Y, Ronot M, Bashir MR, Song B. VICT2 Trait: Prognostic Alternative to Peritumoral Hepatobiliary Phase Hypointensity in HCC. Radiology 2023;307:e221835. [Crossref] [PubMed]
43. Chen P, Yang Z, Ning P, Yuan H, Qi Z, Li Q, Meng B, Zhang X, Yu H. To accurately predict lymph node metastasis in patients with mass-forming intrahepatic cholangiocarcinoma by using CT radiomics features of tumor habitat subregions. Cancer Imaging 2025;25:19. [Crossref] [PubMed]
44. Zhu J, Liu C, Li H, Ren H, Cai Y, Lan T, Wu H. Adequate lymph node dissection is essential for accurate nodal staging in intrahepatic cholangiocarcinoma: A population-based study. Cancer Med 2023;12:8184-98. [Crossref] [PubMed]
45. Liguori C, Copparoni C, Felicetti C, Pecci F, Lupi A, Pinterpe G, Berardi R, Giampieri R. Hemoglobin and Neutrophil Count as Prognostic Factors in Cholangiocarcinoma Patients in 2nd Line Treatment Setting: Results from a Small Monocentric Retrospective Study. Curr Oncol 2023;30:1032-45. [Crossref] [PubMed]