Abstract
Background
Combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a rare primary liver cancer characterized by features of both hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA). As cHCC-CCA prognosis often falls between HCC and CCA or resembles CCA, we compared the outcomes of durvalumab plus gemcitabine–cisplatin (GCD) in patients with cHCC-CCA and CCA in a real-world setting.
Materials and methods
This retrospective multicenter study included patients with histologically confirmed unresectable or metastatic CCA or cHCC-CCA who underwent first-line GCD at four centers in Taiwan between 2021 and 2025. Outcomes included overall survival (OS), progression-free survival (PFS), tumor response, and safety. Propensity score matching (PSM) was performed to balance baseline characteristics.
Results
Among the 175 patients, 154 had CCA and 21 had cHCC-CCA. Before matching, patients with cHCC-CCA more frequently had elevated α-fetoprotein levels and poor biliary drainage; however, these imbalances were resolved after PSM. Before PSM, patients with CCA and cHCC-CCA had median OS of 13.3 months [95% confidence interval (CI) 9.5-17.1] and 13.8 months (95% CI 2.1–25.6, P = 0.60) and median PFS of 5.1 months (95% CI 4.7-5.5) and 5.3 months (95% CI 3.1-7.5, P = 0.40), respectively. After PSM, both OS and PFS rates were similar between the groups. The objective response rate was higher in the cHCC-CCA group (38.1% versus 23.4% before PSM, P = 0.15; 38.1% versus 23.8% after PSM, P = 0.24), whereas the disease control rates were identical (66.7% after PSM). The most common toxicities were hematologic events, with both groups demonstrating similar rates of anemia, thrombocytopenia, leukopenia, and neutropenia.
Conclusions
This study demonstrates that patients with cHCC-CCA derive comparable clinical benefit from GCD as those with CCA in terms of efficacy and safety. These findings support the incorporation of immunochemotherapy into the treatment framework for cHCC-CCA and provide a rationale for its use in clinical practice.
Key words: cholangiocarcinoma, hepatocellular carcinoma, immune checkpoint inhibitor, mixed hepatocellular-cholangiocarcinoma, propensity score matching
Highlights
-
•
Real-world multicenter study comparing durvalumab plus GCD outcomes between cHCC-CCA and pure CCA.
-
•
cHCC-CCA and CCA showed comparable survival and response, before and after matching.
-
•
Safety profiles were similar in both groups, with hematologic toxicities most common.
-
•
Our findings support GCD as a feasible first-line option for cHCC-CCA, consistent with CCA care.
Introduction
Cholangiocarcinoma (CCA) is an aggressive malignancy with limited treatment options and a poor prognosis, with a 5-year survival rate of <20%.1, 2, 3 For over a decade, chemotherapy with gemcitabine plus cisplatin (GC) was the standard first-line regimen, providing only modest survival benefits with a median overall survival (OS) of ∼11 months.4 Recent advances have transformed the systemic treatment landscape for CCA by adding immune checkpoint inhibitors to GC chemotherapy.5, 6, 7 The phase III TOPAZ-1 trial demonstrated that the addition of durvalumab to GC substantially improved OS, progression-free survival (PFS), and objective response rate (ORR), establishing durvalumab plus gemcitabine–cisplatin (GCD) as a new standard of care.5,6 Likewise, the phase III KEYNOTE-966 trial revealed that pembrolizumab combined with GC significantly improved OS compared with GC alone [median OS 12.7 versus 10.9 months; hazard ratio (HR) 0.83], further confirming immunochemotherapy as the first-line backbone for advanced CCA.7
Within the spectrum of primary liver cancers, combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a rare entity characterized by histological features of both hepatocellular carcinoma (HCC) and CCA.8 Incidence rates vary widely, ranging from 1% to 14%, depending on diagnostic criteria and geographic region.9, 10, 11, 12, 13, 14 Clinically, patients with cHCC-CCA frequently present with overlapping characteristics of both tumor types, with some features resembling HCC [e.g. hepatitis virus infection and elevated α-fetoprotein (AFP)] and others similar to CCA (e.g. incomplete tumor capsules, elevated carbohydrate antigen 19-9, and frequent lymph node involvement).15, 16, 17 Nationwide cancer registry analyses from Japan and Taiwan demonstrated that cHCC-CCA shares demographic features with HCC but displays malignant behavior more akin to CCA, with long-term survival outcomes generally intermediate between the two entities.14,18 Consistently, most studies have reported that the prognosis of cHCC-CCA is generally intermediate between HCC and CCA, although some suggest outcomes that more closely resemble CCA when adjusted for stage and liver function.19, 20, 21
Despite advances in systemic therapies for CCA and HCC, a notable knowledge gap remains regarding cHCC-CCA, primarily because patients with cHCC-CCA have been largely excluded from pivotal clinical trials. For example, the TOPAZ-1 trial failed to enroll any patients with cHCC-CCA,5 while the KEYNOTE-966 trial included only a very small proportion (1%-2%) without reporting subgroup analyses.7 This exclusion has highlighted a clinical challenge in determining whether cHCC-CCA should follow the standard CCA treatment paradigm or necessitates a tailored approach reflecting its distinct biological characteristics.10 Accumulated evidence is limited to small retrospective studies and case series with highly variable outcomes, highlighting the biological heterogeneity of the disease and the absence of a standardized therapeutic approach.21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 To address this crucial question, we conducted a multicenter, real-world study to compare the efficacy and safety of the standard-of-care GCD regimen in patients with advanced cHCC-CCA versus those with pure CCA.
Materials and methods
Study design and participants
This retrospective, multicenter, observational study was conducted at four medical centers in Taiwan. We identified consecutive patients with histologically confirmed, unresectable, recurrent, or metastatic CCA or cHCC-CCA who received first-line GCD between August 2021 and March 2025. cHCC-CCA diagnosis was established according to the revised World Health Organization classification.32 All histological slides were jointly reviewed by two experienced hepatopathologists to ensure diagnostic accuracy. Specifically, biliary differentiation was confirmed by staining for CK7 and CK19, while hepatocellular differentiation was based on CD10 staining with canalicular pattern and cytoplasmic staining for hepatocyte paraffin 1.8 Eligibility criteria were age ≥20 years and receipt of first-line GCD therapy. Patients were excluded if they received durvalumab in the second-line or later settings or if they were lost to follow-up within 4 weeks after receiving the first dose of GCD, resulting in no evaluable clinical, safety, or outcome data. The Institutional Review Board of each participating center approved the study protocol (IRB No. 202501211B0). The requirement for written informed consent was waived owing to the retrospective nature of the study. A flowchart of the study is shown in Figure 1.
Figure 1.
Study flowchart.
CCA, cholangiocarcinoma; cHCC-CCA, combined hepatocellular and cholangiocarcinoma.
Propensity score matching
To reduce potential selection bias and create comparable groups, propensity score matching (PSM) was performed using multivariate logistic regression, adjusting for covariates including age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS), tumor stage, hepatitis status, tumor location, and biliary drainage. Given that intrahepatic (iCCA), extrahepatic (eCCA), and gallbladder CCA differ biologically and clinically, tumor location (iCCA versus eCCA versus gallbladder) was included as a covariate in the propensity score model to reduce allocation bias. To match patients with CCA to those with cHCC-CCA, we applied a 2:1 nearest-neighbor matching algorithm with a caliper width of 0.05 standard deviations of the logit of the PSM.
Treatment protocol
All patients were treated according to the regimen established in the TOPAZ-1 trial.5 Durvalumab was administered a fixed dose of 1500 mg intravenously every 3 weeks, combined with gemcitabine (1000 mg/m2 intravenously on days 1 and 8 of each 21-day cycle) and cisplatin (25 mg/m2 intravenously on days 1 and 8, up to 8 cycles). After completion of GC, maintenance durvalumab (1500 mg intravenously every 4 weeks) was administered until disease progression, unacceptable toxicity, or patient withdrawal. Dose modification and supportive care measures, including hematopoietic growth factor support and transfusions, were provided in accordance with institutional practice.
Assessments and outcome
Baseline demographic, clinical, and laboratory data were collected, including age, sex, ECOG PS, viral hepatitis status, primary tumor site, disease stage, and key laboratory values. Patients were stratified by histology into CCA and cHCC-CCA groups. Radiological tumor response was assessed every 8-12 weeks using computed tomography or magnetic resonance imaging and evaluated according to RECIST version 1.1.33 The best overall response was recorded for each patient. Adverse events (AEs) were monitored throughout the treatment period, graded according to the Common Terminology Criteria for Adverse Events (CTCAE), version 5.0.34
The primary outcome of interest was OS, defined as the time from the initiation of GCD therapy to death from any cause or to the last follow-up (censored on 30 June 2025). Secondary endpoints included PFS, defined as the time from treatment initiation to radiological progression or death; ORR, defined as the proportion of patients achieving complete or partial response (PR); disease control rate (DCR), defined as the proportion of patients achieving complete response (CR), PR, or stable disease (SD); and safety outcomes.
Treatment-exposure metrics, including the number of durvalumab cycles, treatment duration, dose reductions, dose delays, and discontinuation due to AEs, were extracted from medical records. Dose modification was performed based on institutional practice, typically in response to CTCAE grade ≥3 toxicities.
Statistical analysis
The baseline characteristics were summarized using descriptive statistics. Continuous variables are expressed as medians with interquartile ranges (range) and compared between groups using the Mann–Whitney U test. Categorical variables are presented as frequencies with percentages and were compared using the χ2 test or two-tailed test, as appropriate.
Laboratory abnormalities were defined according to institutional reference ranges. Reduced albumin level was defined as a serum albumin level <3.5 g/dL (lower than the lower limit of normal). Elevated bilirubin, aspartate transaminase (AST), alanine transaminase (ALT), and alkaline phosphatase (ALP) were defined as values higher than the upper limit of normal. A neutrophil-to-lymphocyte ratio (NLR) ≥3 was used to define systemic inflammation, consistent with prior biliary tract cancer study.12 Age was analyzed as a continuous variable. Disease stage followed the tumor–node–metastasis (TNM35)-aligned criteria: locally advanced disease was defined as unresectable, nonmetastatic disease (e.g. vascular invasion or nodal involvement precluding curative resection), whereas metastatic disease required the presence of distant metastases (M1).
Survival outcomes were estimated using the Kaplan–Meier method, and differences between groups were evaluated using the log-rank test. Univariate and multivariate Cox proportional hazard analyses were performed for OS and PFS. Variables with P < 0.10 in univariate analysis were included in the multivariable model. HRs with corresponding 95% confidence intervals (CIs) were calculated using Cox proportional hazard regression. ORR and DCR were compared between groups using χ2 or Fisher’s exact tests. Data analyses were performed using SPSS version 23.0 (IBM Corp, Armonk, NY). Statistical significance was defined as a two-sided P value <0.05.
Results
Patient characteristics
In total, 175 patients with advanced biliary tract cancer who received first-line GCD were included, including 154 with CCA and 21 with cHCC-CCA (Table 1). In the TNM-aligned staging, 30.5% of patients with CCA and 23.8% of those with cHCC-CCA had unresectable, locally advanced disease, whereas 69.5% and 76.2%, respectively, presented with M1 metastatic disease. Before PSM, patients with cHCC-CCA were more likely to have elevated AFP levels (47.6% versus 12.3%, P < 0.001) and less likely to undergo biliary drainage (14.3% versus 42.9%, P = 0.016). Other demographic, clinical, and laboratory characteristics, including age, sex, ECOG PS, stage, hepatitis status, and liver function test results, were generally balanced between the groups.
Table 1.
Baseline characteristics
| Characteristics | Before PSM |
After PSM |
||||
|---|---|---|---|---|---|---|
| CCA (n = 154) | cHCC-CCA (n = 21) | P value | CCA (n = 42) | cHCC-CCA (n = 21) | P value | |
| Sex | 0.24 | 0.59 | ||||
| Male | 94 (61.0%) | 10 (47.6%) | 23 (54.8%) | 10 (47.6%) | ||
| Female | 60 (39.0%) | 11 (52.3%) | 19 (45.2%) | 11 (52.3%) | ||
| Age, median (range) | 64 (35-88) | 61 (44-78) | 0.30 | 60 (43-83) | 61 (44-78) | 0.62 |
| ECOG PS | 0.80 | 0.88 | ||||
| 0 | 30 (19.5%) | 5 (23.8%) | 10 (23.8%) | 5 (23.8%) | ||
| 1 | 108 (70.1%) | 15 (71.4%) | 31 (73.8%) | 15 (71.4%) | ||
| ≥2 | 16 (10.3%) | 1 (4.8%) | 1 (2.4%) | 1 (4.8%) | ||
| Primary tumor location | 0.38 | 0.52 | ||||
| Intrahepatic | 97 (63.0%) | 17 (81.0%) | 36 (85.7%) | 17 (81.0%) | ||
| Extrahepatic | 35 (22.7%) | 3 (14.3%) | 3 (7.1%) | 3 (14.3%) | ||
| Gall bladder | 22 (14.3%) | 1 (4.8%) | 3 (7.1%) | 1 (4.8%) | ||
| Stage | 0.62 | 0.99 | ||||
| Locally advanced | 47 (30.5%) | 5 (23.8%) | 10 (23.8%) | 5 (23.8%) | ||
| Metastatic (M1) | 107 (69.5%) | 16 (76.2%) | 32 (76.2%) | 16 (76.2%) | ||
| Recurrent pattern | 0.56 | 0.88 | ||||
| Recurrent disease | 37 (24.0%) | 7 (33.3%) | 13 (31.0%) | 7 (33.3%) | ||
| De novo | 117 (76.0%) | 14 (66.7%) | 29 (69.0%) | 14 (66.7%) | ||
| Surgery | 0.14 | 0.72 | ||||
| Yes | 42 (27.3%) | 9 (42.9%) | 16 (38.1%) | 9 (42.9%) | ||
| No | 112 (72.7%) | 12 (57.1%) | 26 (61.9%) | 12 (57.1%) | ||
| Biliary drainage | 0.016 | 0.68 | ||||
| Yes | 66 (42.9%) | 3 (14.3%) | 4 (9.5%) | 3 (14.3%) | ||
| No | 88 (57.1%) | 18 (85.7%) | 38 (90.5%) | 18 (85.7%) | ||
| Hepatitis B carrier | 0.59 | 0.99 | ||||
| Yes | 30 (19.5%) | 6 (28.6%) | 12 (28.6%) | 6 (28.6%) | ||
| No | 124 (80.5%) | 15 (71.4%) | 30 (71.4%) | 15 (71.4%) | ||
| Hepatitis C carrier | 0.87 | 0.74 | ||||
| Yes | 11 (7.1%) | 2 (9.5%) | 3 (7.1%) | 2 (9.5%) | ||
| No | 143 (92.9%) | 19 (90.5%) | 39 (92.9%) | 19 (90.5%) | ||
| Albumin | 0.65 | 0.51 | ||||
| Normal value | 110 (71.4%) | 16 (76.2%) | 35 (83.3%) | 16 (76.2%) | ||
| Decreased value | 44 (28.6%) | 5 (2.8%) | 7 (16.7%) | 5 (2.8%) | ||
| Bilirubin | 0.98 | 0.15 | ||||
| Normal value | 117 (76.0%) | 16 (76.2%) | 38 (90.5%) | 16 (76.2%) | ||
| Increased value | 37 (24.0%) | 5 (23.8%) | 4 (9.5%) | 5 (23.8%) | ||
| AST | 0.48 | 0.15 | ||||
| Normal value | 86 (55.8%) | 10 (47.6%) | 28 (66.7%) | 10 (47.6%) | ||
| Increased value | 68 (44.2%) | 11 (52.4%) | 14 (33.3%) | 11 (52.4%) | ||
| ALT | 0.87 | 0.99 | ||||
| Normal value | 85 (55.2%) | 12 (57.1%) | 24 (57.1%) | 12 (57.1%) | ||
| Increased value | 69 (44.8%) | 9 (42.9%) | 18 (42.9%) | 9 (42.9%) | ||
| ALP | 0.64 | 0.29 | ||||
| Normal value | 65 (42.2%) | 10 (47.6%) | 24 (57.1%) | 10 (47.6%) | ||
| Increased value | 89 (57.8%) | 11 (52.4%) | 18 (42.9%) | 11 (52.4%) | ||
| CA 19-9 | 0.17 | 0.72 | ||||
| Normal value | 50 (32.5%) | 10 (47.6%) | 21 (50.0%) | 10 (47.6%) | ||
| Increased value | 104 (67.5%) | 11 (52.4%) | 21 (50.0%) | 11 (52.4%) | ||
| CEA | 0.38 | 0.42 | ||||
| Normal value | 87 (56.5%) | 14 (66.7%) | 22 (52.4%) | 14 (66.7%) | ||
| Increased value | 67 (43.5%) | 7 (33.3%) | 20 (46.7%) | 7 (33.3%) | ||
| AFP | <0.001 | 0.42 | ||||
| Normal value | 135 (87.7%) | 11 (52.4%) | 27 (64.3%) | 11 (52.4%) | ||
| Increased value | 19 (12.3%) | 10 (476%) | 15 (35.7%) | 10 (476%) | ||
| NLR | 0.24 | 0.17 | ||||
| <3 | 37 (37.0%) | 5 (23.8%) | 19 (45.2%) | 5 (23.8%) | ||
| ≥3 | 97 (63.0%) | 16 (76.2%) | 23 (54.8%) | 16 (76.2%) | ||
AFP, α-fetoprotein; ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; CA 19-9, carbohydrate antigen 19-9; CCA, cholangiocarcinoma; CEA, carcinoembryonic antigen; cHCC-CCA, combined hepatocellular and cholangiocarcinoma; ECOG PS, Eastern Cooperative Oncology Group performance status; NLR, neutrophil-lymphocyte ratio; PSM, propensity score matching.
After 2:1 PSM, 42 patients with CCA were compared with 21 patients with cHCC-CCA. The baseline covariates, including ECOG PS, primary tumor site, disease stage, liver function, and tumor markers, were well balanced between the groups. Notably, differences in biliary drainage and AFP elevation observed before matching were no longer significant in the matched cohort.
In the overall cohort, the CCA group consisted of 97 patients with iCCA (63.0%), 35 with eCCA (22.7%), and 22 with gallbladder cancer (14.3%). Among the 21 patients with cHCC-CCAs, 17 (81.0%) presented with intrahepatic tumors. After 2:1 PSM, the matched CCA cohort included 36 iCCA (85.7%), 3 eCCA (7.1%), and 3 gallbladder cancers (7.1%), resulting in a tumor location distribution closely aligned with the cHCC-CCA cohort.
In total, 117 patients with CCA (76.0%) and 14 with cHCC-CCA (66.7%) presented with de novo advanced disease, whereas 37 patients with CCA (24.0%) and seven with cHCC-CCA (33.3%) had recurrent disease following prior resection. After PSM, the distribution remained balanced, with recurrent disease observed in 31.0% of matched CCA cases and 33.3% of cHCC-CCA cases.
Survival outcomes
At a median follow-up of 16.2 months (range 2.6-35.4), OS did not differ significantly between the two groups (Figure 2A), with a median OS of 13.3 (range 9.5-17.1) for CCA and 13.8 (range 2.1-25.6) for cHCC-CCA (log-rank P = 0.60). Histological subtype (CCA versus cHCC-CCA) did not significantly impact OS.
Figure 2.
Kaplan–Meier curve for overall survival in patients with pure cholangiocarcinoma and combined hepatocellular-cholangiocarcinoma. Before (A) and after (B) propensity score matching.
CCA, cholangiocarcinoma; cHCC-CCA, combined hepatocellular and cholangiocarcinoma; CI, confidence interval.
PFS was also comparable between groups (Figure 3A), with median PFS of 5.1 (range 4.7-5.5) for CCA and 5.3 (range 3.1-7.5) for cHCC-CCA, with no significant difference (log-rank P = 0.40).
Figure 3.
Kaplan–Meier curve for progression-free survival in patients with pure cholangiocarcinoma and combined hepatocellular-cholangiocarcinoma. Before (A) and after (B) propensity score matching.
CCA, cholangiocarcinoma; cHCC-CCA, combined hepatocellular and cholangiocarcinoma; CI, confidence interval.
Survival outcomes remained consistent after PSM, with no significant differences in median OS (log-rank P = 0.78; Figure 2B) or PFS (log-rank P = 0.39; Figure 3B) between the CCA and cHCC-CCA groups.
Univariate and multivariate analysis of OS
In univariate analysis, older age, poor ECOG PS, metastatic stage, decreased albumin, elevated AST, elevated ALP, and NLR ≥3 were associated with worse OS (Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2026.108246). The histological subtype (pure CCA versus cHCC-CCA: HR 0.83, 95% CI 0.42-1.66, P = 0.60) did not independently predict OS. In the multivariable model, several factors remained independently prognostic: older age (adjusted HR 1.03, 95% CI 1.01-1.05), ECOG PS ≥2 (adjusted HR 2.86, 95% CI 1.09-7.56), metastatic stage (adjusted HR 1.62, 95% CI 1.22-2.17), decreased albumin (adjusted HR 2.30, 95% CI 1.39-3.79), elevated AST (adjusted HR 2.03, 95% CI 1.27-3.25), and NLR ≥3 (adjusted HR 1.69, 95% CI 1.01-2.82).
Univariate and multivariate analysis of PFS
In univariate analysis, ECOG PS ≥2, metastatic stage, reduced albumin, elevated AST, elevated ALT, elevated ALP, and NLR ≥3 were associated with shorter PFS (Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2026.108246). Histological type (cHCC-CCA versus CCA) was not associated with PFS (HR 0.79, P = 0.40). In the multivariable model, several factors remained independently predictive of PFS, including ECOG PS ≥2 (adjusted HR 2.31, 95% CI 1.17-4.56), metastatic disease (adjusted HR 1.24, 95% CI 1.01-1.53), and decreased albumin (adjusted HR 1.81, 95% CI 1.22-2.70).
Tumor response
Before PSM, the best overall responses in the CCA group were CR in 1.9%, PR in 21.4%, SD in 46.8%, and progressive disease (PD) in 29.9% (Table 2). Among patients with cHCC-CCA, no CR was observed; PR occurred more frequently (38.1% versus 21.4%), SD was less common (28.6% versus 46.8%), and PD rates were comparable (33.3% versus 29.9%; P = 0.26).
Table 2.
Summary of treatment response
| Parameter, n (%) | Before PSM |
After PSM |
||||
|---|---|---|---|---|---|---|
| CCA (n = 154) | cHCC-CCA (n = 21) | P value | CCA (n = 42) | cHCC-CCA (n = 21) | P value | |
| Best tumor response | 0.26 | 0.42 | ||||
| Complete response | 3 (1.9%) | 0 | 0 | 0 | ||
| Partial response | 33 (21.4%) | 8 (38.1%) | 10 (23.8%) | 8 (38.1%) | ||
| Stable disease | 72 (46.8%) | 6 (28.6%) | 18 (42.9%) | 6 (28.6%) | ||
| Progressive disease | 46 (29.9%) | 7 (33.3%) | 14 (33.3%) | 7 (33.3%) | ||
| Objective response rate | 36 (23.4%) | 8 (38.1%) | 0.15 | 10 (23.8%) | 8 (38.1%) | 0.24 |
| Disease control rate | 108 (70.1%) | 14 (66.7%) | 0.75 | 28 (66.7%) | 14 (66.7%) | 0.99 |
| Ongoing treatment | 38 (24.7%) | 7 (33.3%) | 9 (21.4%) | 7 (33.3%) | ||
| Treatment discontinuation | 116 (75.3%) | 14 (66.7%) | 0.75 | 33 (78.6%) | 14 (66.7%) | 0.72 |
| Due to progression | 94 (81.0%) | 11 (78.6%) | 27 (64.3%) | 11 (78.6%) | ||
| Due to toxicity | 12 (10.3%) | 1 (7.1%) | 3 (7.1%) | 1 (7.1%) | ||
| Other reasons | 10 (8.6%) | 2 (14.3%) | 3 (7.1%) | 2 (14.3%) | ||
CCA, cholangiocarcinoma; cHCC-CCA, combined hepatocellular and cholangiocarcinoma; PSM, propensity score matching.
The cHCC-CCA group had a higher ORR than the CCA group (38.1% versus 23.4%), although the difference was statistically nonsignificant (P = 0.15). The DCR was similar between the groups (66.7% versus 70.1%, P = 0.75).
After PSM, the results were consistent with those of the unadjusted analysis. In the matched cohort, PR was observed in 38.1% and 23.8% of patients with cHCC-CCA and CCA, respectively, whereas SD and PD occurred at similar frequencies in both patient groups (P = 0.42). The ORR was higher in the cHCC-CCA group (38.1% versus 23.8%, P = 0.24), whereas the DCR was similar in both groups (66.7% in both groups, P = 0.99).
At the time of analysis, 24.7% of patients with CCA and 33.3% with cHCC-CCA were receiving GCD before matching, with 21.4% and 33.3%, respectively, continuing therapy after matching. Among patients who discontinued treatment, disease progression was the most common cause, followed by toxicity and other causes, with no significant differences observed between the groups. The postprogression therapy rates were comparable.
Safety profile
The incidence of grade ≥3 treatment-related AEs is summarized in Supplementary Table S3, available at https://doi.org/10.1016/j.esmoop.2026.108246. Before PSM, any grade ≥3 toxicity was observed in 70.1% and 57.1% of patients with CCA and cHCC-CCA, respectively (P = 0.23). Hematologic toxicities were the most common in both groups, with similar rates of anemia (26.6% versus 28.6%), thrombocytopenia (19.5% versus 23.8%), leukopenia (14.9% versus 28.6%), and neutropenia (18.2% versus 23.8%). Neutropenic fever was rare (2.6% versus 4.8%). Nonhematologic toxicities, including electrolyte disturbances, liver enzyme elevation, fatigue, anorexia, and gastrointestinal events, were infrequent and balanced between the groups. Grade ≥3 infection-related events occurred in 38.3% and 33.3% of patients with CCA and cHCC-CCA, respectively (P = 0.66). Immune-related AEs (irAEs) were recorded in 4.2% of patients with CCA, but not in those with cHCC-CCA.
The safety profiles remained consistent after matching. Any grade ≥3 toxicity occurred in 42.9% and 57.1% of patients with CCA and cHCC-CCA, respectively (P = 0.30). Hematologic events were the most frequent, with no significant between-group differences. Nonhematologic toxicities and irAEs were rare, with no clear imbalance observed.
The median number of durvalumab cycles administered was similar across the groups, with a median of seven cycles in both the CCA and cHCC-CCA cohorts (Supplementary Table S4, available at https://doi.org/10.1016/j.esmoop.2026.108246). The median treatment duration was 5.1 months (range 1.6-32.1) for CCA and 5.3 months (range 1.4-21.4) for cHCC-CCA, comparable to prior real-world reports. Dose reductions occurred in 29.2% of patients with CCA and 23.8% of those with cHCC-CCA, while dose delays were observed in 11.7% and 14.3% of patients, respectively. Discontinuation due to AEs was infrequent (7.8% in CCA and 4.8% in cHCC-CCA), and no treatment-related deaths were observed in either group.
Postprogression treatments
Postprogression treatments were administered to 42.2% of patients with CCA and 42.9% of those with cHCC-CCA before matching and to 47.6% and 42.9% of patients, respectively, after matching (Supplementary Table S5, available at https://doi.org/10.1016/j.esmoop.2026.108246). Chemotherapy was the most common subsequent therapy in both groups (29.2% in CCA versus 38.1% in cHCC-CCA), followed by targeted agents (6.5% versus 14.3%), immunotherapy (4.5% versus 4.8%), and, less commonly, antibody‒drug conjugates. Overall, the distribution of subsequent therapies was broadly comparable between groups.
Discussion
To our knowledge, this multicenter retrospective study represents one of the largest real-world cohort studies to directly compare the efficacy and safety of GCD between patients with cHCC-CCA and those with pure CCA in an era of immunochemotherapy. Our findings address the substantial gap in clinical knowledge created by the exclusion of patients with cHCC-CCA from pivotal trials.
First-line treatment with GCD yielded comparable OS and PFS between CCA and cHCC-CCA in both the overall cohort and the PSM analysis. Our findings demonstrate that GCD therapy in patients with cHCC-CCA yields outcomes comparable to those observed in patients with pure CCA, reinforcing a unified therapeutic approach. The safety profiles were comparable to those of the TOPAZ-1 study,5 with no new or unexpected toxicities identified. The most common AEs were hematologic toxicities, including anemia, thrombocytopenia, leukopenia, and neutropenia.
Treatment-exposure patterns in our cohort further support the feasibility of administering GCD in routine practice. The median number of durvalumab cycles (7 cycles) and median treatment duration (5.1-5.3 months) were slightly lower than those observed in TOPAZ-1, in which the median number of durvalumab cycles was 10 (range 1-29) and the median treatment duration was 7.3 months (range 0.1-24.5).5,6 This difference is expected because real-world cohorts often include patients with poorer performance status, biliary obstruction, prior drainage, or comorbidities who would not have met eligibility criteria for clinical trial enrollment. Despite these differences, the rates of dose reduction, dose delay, and AE-related discontinuation in our study remained modest and comparable between CCA and cHCC-CCA and no treatment-related deaths occurred.
Our multivariable analyses for OS and PFS showed a consistent pattern: poor performance status, metastatic disease, and hypoalbuminemia were key predictors of worse outcomes, with elevated AST and NLR ≥3 further associated with inferior OS. These findings reflect the combined effect of tumor burden and host vulnerability on treatment responses. Importantly, cHCC-CCA histology was not independently associated with OS or PFS, indicating that, after accounting for clinical factors, first-line GCD provides comparable efficacy to cHCC-CCA and CCA.
The interpretation of OS in real-world studies of biliary tract cancers is often complicated by the heterogeneity of postprogression treatment strategies, particularly for cHCC-CCA, in which both CCA-like and HCC-like systemic therapies may be considered.26,27 However, in our cohort, subsequent therapy patterns were largely homogeneous, with the vast majority of patients continuing CCA-directed systemic regimens after progression to first-line GCD. HCC-like agents,36 such as sorafenib or lenvatinib, were rarely administered to patients with cHCC-CCA in our cohort. This relative uniformity in subsequent treatments reduces the risk of differential postprogression care influencing OS estimates and enhances the internal validity of our survival comparisons.
Existing evidence for systemic therapy in cHCC-CCA is limited to small retrospective studies and case series, with highly variable outcomes (ORR 0%-50%, median PFS 2.8-9.0 months, and median OS 5.3-17.8 months; Supplementary Table S6, available at https://doi.org/10.1016/j.esmoop.2026.108246). Platinum-based chemotherapy has typically elicited better outcomes than nonplatinum regimens, as illustrated by Salimon et al.22 and Kobayashi et al.23 In contrast, sorafenib and other tyrosine kinase inhibitors have shown limited efficacy, with ORRs of 0%-14% and median OS ranging from 4.8 to 11.9 months, consistently inferior to chemotherapy in comparative series.23, 24, 25, 26 Recent studies suggest that immune checkpoint inhibitor-based therapy can yield encouraging outcomes; for example, Jang et al.28 reported an ORR of 20% and OS of 8.3 months with programmed cell death protein 1 or programmed death-ligand 1 (PD-L1) inhibitors, while Gigante et al.29 and Satake et al.30 described ORRs of 33%-50% and OS up to 13 months with atezolizumab–bevacizumab. A real-world study from China further reported ORR 41.2%, PFS 8.9 months, and OS 17.8 months with combined immunotherapy and interventional treatment.31 These findings, together with our observation that patients with cHCC-CCA achieved outcomes comparable with those of patients with CCA when treated with GCD, underscore the rationale of managing cHCC-CCA within the CCA paradigm rather than as a distinct orphan disease.
Although not statistically significant, the numerically higher ORR in cHCC-CCA (38.1%) than in CCA (23.8%) suggests important biological considerations. The comparable outcomes between the two groups may partly reflect the cholangiocytic component shared by both tumor types, which drives the sensitivity to gemcitabine–platinum chemotherapy and PD-L1 blockade.5,6 Interestingly, gemcitabine–cisplatin has also demonstrated modest clinical benefits in patients with advanced HCC, with a phase II study reporting PR in 20% of patients and SD in 43%.37 This suggests that the hepatocellular component of cHCC-CCA may also contribute to immunochemotherapy responsiveness, consistent with the efficacy of immune checkpoint inhibitors in HCC—including durvalumab monotherapy, which achieved noninferior survival versus sorafenib in the HIMALAYA trial,38 as well as combination regimens such as durvalumab–tremelimumab and atezolizumab–bevacizumab.37,38 Collectively, these observations underscore the heterogeneous biology of cHCC-CCA, which likely reflects varying contributions of hepatocellular and cholangiocytic components, and suggest that future molecular studies are needed to clarify whether distinct subgroups may respond differently to immunochemotherapy.39,40
The management of cHCC-CCA has long been challenged by its rarity, histological heterogeneity, and exclusion from pivotal clinical trials such as TOPAZ-1.5 More recently, KEYNOTE-966 included a small proportion of patients with cHCC-CCA (∼1%-2%),7 making it one of the few phase III studies to include this patient population. However, subgroup analyses have not been reported, leaving the efficacy of immune checkpoint inhibitor plus GC in this population undefined. Against this backdrop, our study provides complementary real-world evidence that directly compares GCD outcomes in cHCC-CCA and CCA.
The strengths of the present study include its multicenter design, expert pathological confirmation of cHCC-CCAs, and the use of PSM to mitigate baseline imbalances. Nonetheless, the limitations also need to be addressed. A key limitation is the marked imbalance in group size (154 CCA versus 21 cHCC-CCA), which may reduce the stability of treatment-effect estimates and introduce the possibility of residual confounding despite PSM; therefore, comparisons should be interpreted cautiously. Another limitation relates to heterogeneity within the CCA cohort, as intrahepatic, extrahepatic, and gallbladder tumors are biologically distinct diseases. Because cHCC-CCA more closely resembles iCCA in its evolution and molecular features, clarifying the tumor location is essential when making comparative inferences. To address this, we explicitly incorporated tumor location into the matching algorithm and reported the exact pre- and post-PSM proportions. Although this approach is consistent with the TOPAZ-1 stratification strategy, the limited number of eCCA cases in our study may influence interpretability. Another limitation is the inclusion of both de novo and recurrent cases, which may differ in biological behavior, prior treatment exposure, and prognosis. Although recurrence status was balanced between the groups after matching and is unlikely to account for the comparable outcomes observed, heterogeneity between these disease trajectories may still influence the generalizability of our findings. Furthermore, safety profiles based on retrospective chart reviews may have been underestimated. Finally, the absence of biomarker data (e.g. PD-L1 expression, genomic alterations, or immune microenvironment features) limited our ability to explore the biological correlations of the observed outcomes.
Our findings suggest that patients with cHCC-CCA can be managed under the same first-line GCD paradigm as those with CCA, providing an evidence-based approach for this rare entity. Given the difficulty of conducting large prospective trials in such uncommon tumors, real-world data remain crucial. Future collaborative efforts, such as multicenter registries incorporating clinical outcomes and biomarker information, may help refine treatment strategies and identify subgroups that respond differently to immunochemotherapy.
Conclusion
This multicenter, real-world study demonstrated that patients with cHCC-CCA achieved survival, tumor response, and safety outcomes comparable to those of patients with pure CCA treated with GCD. These findings support the GCD as a feasible first-line treatment option for cHCC-CCA and highlight the urgent need for dedicated prospective trials to validate its efficacy and define precision-based strategies for this rare and heterogeneous malignancy.
Acknowledgement
The authors gratefully acknowledge the support from the Cancer Centerof Chang Gung Memorial Hospital, Taoyuan, Taiwan.
Ethical approval
This study was approved by the Institutional Review Board of the Chang Gung Memorial Hospital (approval no. 202501211B0), and all procedures were conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the analysis.
Funding
None declared.
Disclosure
The authors have declared no conflicts of interest.
Supplementary data
References
- 1.Valle J.W., Kelley R.K., Nervi B., Oh D.Y., Zhu A.X. Biliary tract cancer. Lancet. 2021;397:428–444. doi: 10.1016/S0140-6736(21)00153-7. [DOI] [PubMed] [Google Scholar]
- 2.Brindley P.J., Bachini M., Ilyas S.I., et al. Cholangiocarcinoma. Nat Rev Dis Primers. 2021;7:65. doi: 10.1038/s41572-021-00300-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lee J.Y., Kim J.W. Recent 5-year trends in biliary tract cancer survival rates: an analytical big data survey. Med Int (Lond) 2025;5:15. doi: 10.3892/mi.2025.214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Valle J., Wasan H., Palmer D.H., et al. Cisplatin plus gemcitabine versus gemcitabine for biliary tract cancer. N Engl J Med. 2010;362:1273–1281. doi: 10.1056/NEJMoa0908721. [DOI] [PubMed] [Google Scholar]
- 5.Oh D.Y., Ruth He A., Qin S., et al. Durvalumab plus gemcitabine and cisplatin in advanced biliary tract cancer. NEJM Evid. 2022;1 doi: 10.1056/EVIDoa2200015. [DOI] [PubMed] [Google Scholar]
- 6.Oh D.Y., He A.R., Qin S., et al. Durvalumab plus chemotherapy in advanced biliary tract cancer: 3-year overall survival update from the phase III TOPAZ-1 study. J Hepatol. 2025;83:1092–1101. doi: 10.1016/j.jhep.2025.05.003. [DOI] [PubMed] [Google Scholar]
- 7.Kelley R.K., Ueno M., Yoo C., et al. Pembrolizumab in combination with gemcitabine and cisplatin compared with gemcitabine and cisplatin alone for patients with advanced biliary tract cancer (KEYNOTE-966): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2023;401:1853–1865. doi: 10.1016/S0140-6736(23)00727-4. [DOI] [PubMed] [Google Scholar]
- 8.Komuta M., Yeh M.M. A review on the update of combined hepatocellular cholangiocarcinoma. Semin Liver Dis. 2020;40:124–130. doi: 10.1055/s-0039-3402515. [DOI] [PubMed] [Google Scholar]
- 9.Wege H., Campani C., de Kleine R., et al. Rare primary liver cancers: an EASL position paper. J Hepatol. 2024;81:704–725. doi: 10.1016/j.jhep.2024.06.004. [DOI] [PubMed] [Google Scholar]
- 10.Terashima T., Harada K., Yamashita T. Diagnosis, clinical characteristics, and treatment of combined hepatocellular-cholangiocarcinoma. Jpn J Clin Oncol. 2025;55:327–333. doi: 10.1093/jjco/hyaf029. [DOI] [PubMed] [Google Scholar]
- 11.Satake T., Morizane C., Rikitake R., Higashi T., Okusaka T., Kawai A. The epidemiology of rare types of hepatobiliary and pancreatic cancer from national cancer registry. J Gastroenterol. 2022;57:890–901. doi: 10.1007/s00535-022-01920-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tang Y., Wang L., Teng F., Zhang T., Zhao Y., Chen Z. The clinical characteristics and prognostic factors of combined hepatocellular carcinoma and cholangiocarcinoma, hepatocellular carcinoma and intrahepatic cholangiocarcinoma after surgical resection: a propensity score matching analysis. Int J Med Sci. 2021;18:187–198. doi: 10.7150/ijms.50883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yin X., Zhang B.H., Qiu S.J., et al. Combined hepatocellular carcinoma and cholangiocarcinoma: clinical features, treatment modalities, and prognosis. Ann Surg Oncol. 2012;19:2869–2876. doi: 10.1245/s10434-012-2328-0. [DOI] [PubMed] [Google Scholar]
- 14.Iijima H., Kudo M., Kubo S., et al. Report of the 23rd nationwide follow-up survey of primary liver cancer in Japan (2014-2015) Hepatol Res. 2023;53:895–959. doi: 10.1111/hepr.13953. [DOI] [PubMed] [Google Scholar]
- 15.Ye L., Schneider J.S., Ben Khaled N., et al. Combined hepatocellular-cholangiocarcinoma: biology, diagnosis, and management. Liver Cancer. 2024;13:6–28. doi: 10.1159/000530700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Koh K.C., Lee H., Choi M.S., et al. Clinicopathologic features and prognosis of combined hepatocellular cholangiocarcinoma. Am J Surg. 2005;189:120–125. doi: 10.1016/j.amjsurg.2004.03.018. [DOI] [PubMed] [Google Scholar]
- 17.Goodman Z.D., Ishak K.G., Langloss J.M., Sesterhenn I.A., Rabin L. Combined hepatocellular-cholangiocarcinoma. A histologic and immunohistochemical study. Cancer. 1985;55:124–135. doi: 10.1002/1097-0142(19850101)55:1<124::aid-cncr2820550120>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
- 18.Chi C.T., Chau G.Y., Lee R.C., et al. Radiological features and outcomes of combined hepatocellular-cholangiocarcinoma in patients undergoing surgical resection. J Formos Med Assoc. 2020;119:125–133. doi: 10.1016/j.jfma.2019.02.012. [DOI] [PubMed] [Google Scholar]
- 19.Jarnagin W.R., Weber S., Tickoo S.K., et al. Combined hepatocellular and cholangiocarcinoma: demographic, clinical, and prognostic factors. Cancer. 2002;94:2040–2046. doi: 10.1002/cncr.10392. [DOI] [PubMed] [Google Scholar]
- 20.Wakizaka K., Yokoo H., Kamiyama T., et al. Clinical and pathological features of combined hepatocellular-cholangiocarcinoma compared with other liver cancers. J Gastroenterol Hepatol. 2019;34:1074–1080. doi: 10.1111/jgh.14547. [DOI] [PubMed] [Google Scholar]
- 21.Nolasco F., Fonseca G.M., de Mello E.S., et al. Prognostic impact of the cholangiolar component in combined hepatocellular-cholangiocarcinoma: insights from a Western single-center study. J Surg Oncol. 2025;131:427–434. doi: 10.1002/jso.27955. [DOI] [PubMed] [Google Scholar]
- 22.Salimon M., Prieux-Klotz C., Tougeron D., et al. Gemcitabine plus platinum-based chemotherapy for first-line treatment of hepatocholangiocarcinoma: an AGEO French multicentre retrospective study. Br J Cancer. 2018;118:325–330. doi: 10.1038/bjc.2017.413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kobayashi S., Terashima T., Shiba S., et al. Multicenter retrospective analysis of systemic chemotherapy for unresectable combined hepatocellular and cholangiocarcinoma. Cancer Sci. 2018;109:2549–2557. doi: 10.1111/cas.13656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Trikalinos N.A., Zhou A., Doyle M.B.M., et al. Systemic therapy for combined hepatocellular-cholangiocarcinoma: a single-institution experience. J Natl Compr Canc Netw. 2018;16:1193–1199. doi: 10.6004/jnccn.2018.7053. [DOI] [PubMed] [Google Scholar]
- 25.Kim E.J., Yoo C., Kang H.J., et al. Clinical outcomes of systemic therapy in patients with unresectable or metastatic combined hepatocellular-cholangiocarcinoma. Liver Int. 2021;41:1398–1408. doi: 10.1111/liv.14813. [DOI] [PubMed] [Google Scholar]
- 26.Gigante E., Hobeika C., Le Bail B., et al. Systemic treatments with tyrosine kinase inhibitor and platinum-based chemotherapy in patients with unresectable or metastatic hepatocholangiocarcinoma. Liver Cancer. 2022;11:460–473. doi: 10.1159/000525488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pomej K., Balcar L., Shmanko K., et al. Clinical characteristics and outcome of patients with combined hepatocellular-cholangiocarcinoma–a European multicenter cohort. ESMO Open. 2023;8 doi: 10.1016/j.esmoop.2023.100783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Jang Y.J., Kim E.J., Kim H.D., et al. Clinical outcomes of immune checkpoint inhibitors in unresectable or metastatic combined hepatocellular-cholangiocarcinoma. J Cancer Res Clin Oncol. 2023;149:7547–7555. doi: 10.1007/s00432-023-04704-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gigante E., Bouattour M., Bedoya J.U., et al. Atezolizumab and bevacizumab for non-resectable or metastatic combined hepatocellular-cholangiocarcinoma: a multicentric retrospective study. United European Gastroenterol J. 2024;12:429–439. doi: 10.1002/ueg2.12503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Satake T., Shibuki T., Watanabe K., et al. Case report: atezolizumab plus bevacizumab for combined hepatocellular-cholangiocarcinoma. Front Oncol. 2023;13 doi: 10.3389/fonc.2023.1234113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Lin Y.S., Wu L.Y., Lin L.H., et al. Interventional treatment combined with immunotargeted therapy in unresectable combined hepatocellular-cholangiocarcinoma: a real-world retrospective cohort study. Front Immunol. 2025;16 doi: 10.3389/fimmu.2025.1591127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.WHO Classification of Tumours Editorial Board . 5th ed. International Agency for Research on Cancer; Lyon, France: 2019. International Agency for Research on Cancer WHO Classification of Tumours of the Digestive System. [Google Scholar]
- 33.Eisenhauer E.A., Therasse P., Bogaerts J., et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1) Eur J Cancer. 2009;45:228–247. doi: 10.1016/j.ejca.2008.10.026. [DOI] [PubMed] [Google Scholar]
- 34.US Department of Health and Human Services, National Institutes of Health, National Cancer Institute . US Department of Health and Human Services; Rockville, MD: November 27, 2025. Common Terminology Criteria for Adverse Events (CTCAE). Version 5. [Google Scholar]
- 35.Amin M.B., Edge S.B., Greene F.L., et al. AJCC Cancer Staging Manual. 8th ed. Springer; New York, NY: 2017. [Google Scholar]
- 36.Finn R.S., Qin S., Ikeda M., et al. Atezolizumab plus bevacizumab in unresectable hepatocellular carcinoma. N Engl J Med. 2020;382:1894–1905. doi: 10.1056/NEJMoa1915745. [DOI] [PubMed] [Google Scholar]
- 37.Parikh P.M., Fuloria J., Babu G., et al. A phase II study of gemcitabine and cisplatin in patients with advanced hepatocellular carcinoma. Trop Gastroenterol. 2005;26:115–118. [PubMed] [Google Scholar]
- 38.Abou-Alfa G.K., Lau G., Kudo M., et al. Tremelimumab plus durvalumab in unresectable hepatocellular carcinoma. NEJM Evid. 2022;1 doi: 10.1056/EVIDoa2100070. [DOI] [PubMed] [Google Scholar]
- 39.Murugesan K., Sharaf R., Montesion M., et al. Genomic profiling of combined hepatocellular cholangiocarcinoma reveals genomics similar to either hepatocellular carcinoma or cholangiocarcinoma. JCO Precis Oncol. 2021;5 doi: 10.1200/PO.20.00397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Yang C.M., Lim J., Noh M.G., et al. Comprehensive molecular profiling of combined hepatocellular carcinoma and cholangiocarcinoma reveals distinct Notch signaling subgroups with prognostic significance. Virchows Arch. 2025 doi: 10.1007/s00428-025-04172-9. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.