Abstract
Background
Biliary tract cancer (BTC) comprises a heterogeneous group of malignancies, often diagnosed at advanced stages. Phase III trials have shown that adding immune checkpoint inhibitors (ICIs) to gemcitabine and cisplatin improves survival, supporting their use as first-line therapy; however, real-world evidence remains limited.
Materials and methods
We conducted a multicenter observational study including patients with advanced BTC treated with ICI-based regimens plus chemotherapy. Clinical data were retrospectively collected. Overall survival (OS) was estimated using the Kaplan–Meier method, and prognostic factors were evaluated with multivariate Cox regression. A historical control cohort without ICI exposure was analyzed using propensity score matching.
Results
A total of 152 patients received ICI-based therapy (median age 66.5 years, 57.9% men, and 68.4% metastatic). The most common primary sites were intrahepatic cholangiocarcinoma (52.6%) and gallbladder carcinoma (23.7%). Durvalumab plus gemcitabine/cisplatin was used in 88.1% of cases. After a median follow-up of 8.3 months, the median OS was 15.5 months (95% confidence interval 12.3-21.5), and the response rate was 38.8%. Maintenance therapy was given to 37.5%, mainly as ICI monotherapy. Independent adverse prognostic factors included Eastern Cooperative Oncology Group performance status of >0 [hazard ratio (HR) 3.91], neutrophil-to-lymphocyte ratio of ≥3 (HR 1.98), and vascular invasion (HR 1.83). Ten (6.6%) grade ≥2 immune-related adverse events were reported. A propensity score-matched analysis showed improved survival with ICI versus the historical cohort without ICI (HR 0.61, P = 0.025).
Conclusion
In this real-world cohort, ICI plus chemotherapy achieved outcomes comparable with pivotal trials, supporting its effectiveness in broader populations and highlighting the need for improved access.
Key words: advanced disease, biliary tract cancer, cholangiocarcinoma, immunotherapy, real-world evidence
Graphical abstract
Highlights
-
•
Chemoimmunotherapy achieved a median survival of 15.5 months in this BTC cohort.
-
•
ECOG PS of ≥1, NLR of ≥3, and vascular invasion predicted worse OS.
-
•
Response and survival matched phase III trial outcomes.
-
•
Matched analysis showed better survival compared with the preimmunotherapy cohort.
-
•
A subset treated with immunotherapy showed a durable survival benefit.
Introduction
Biliary tract cancer (BTC) comprises a heterogeneous group of malignancies arising from the gallbladder and bile ducts. Collectively, they are characterized by an aggressive clinical course, with most diagnoses occurring at advanced stages.1,2 Managing BTC is complicated by impaired performance status (PS), biliary obstruction, infectious complications, and treatment-related toxicities.1,2 Since 2010, cisplatin plus gemcitabine has become the backbone systemic treatment for BTC, demonstrating improved survival compared with gemcitabine.3
After more than a decade, recent clinical trials have highlighted the role of immune checkpoint inhibitors (ICIs). The addition of durvalumab to gemcitabine plus cisplatin has improved progression-free survival and overall survival (OS) compared with chemotherapy alone.4,5 Similarly, pembrolizumab has improved OS, leading to regulatory approvals and widespread adoption of these regimens.6 However, the clinical benefit is heterogeneous across subgroups, with risks of adverse events and financial toxicity.
Real-world evidence increasingly supports decision making outside of clinical trials. Since 2022, data on ICI in BTC have emerged, particularly from Europe, Asia, and China.7, 8, 9, 10, 11 However, the epidemiology of BTC varies across regions, with particularities regarding comorbidities, presentation, and ethnicity in Latin America.12 In this study, we describe characteristics and outcomes of patients receiving ICI in a multicenter Brazilian cohort, compared with a historical cohort treated before ICI incorporation.
Materials and methods
Study design and eligibility criteria
We conducted an observational multicenter study within a Brazilian oncology network (Oncologia D’Or). The study focused on patients with BTC who received at least one dose of ICI as part of routine clinical care. The included population was selected from medical records from December 2018 to July 2025.
The centers receive patients from other health care services (primary or secondary units or private physicians) or through self-referral. Patients with BTC are managed by multidisciplinary teams. Whenever there is an indication for systemic treatment—whether due to advanced-stage disease or contraindication to surgery or locoregional therapy—the patient is evaluated by clinical oncologists. Participating centers are part of the Brazilian complementary health system, which serves ∼25% of the population. Public and complementary systems have separate sets of covered medications and are regulated by different sectors within the Ministry of Health. Immunotherapy for BTC is incorporated only into the latter.13
Inclusion criteria included patients aged at least 18 years with a histological diagnosis of BTC, including intrahepatic cholangiocarcinoma (iCCA), distal cholangiocarcinoma (dCCA), perihilar cholangiocarcinoma (pCCA), gallbladder carcinoma, and ampullary carcinoma with immunohistochemical staining consistent with pancreatobiliary subtype, and with diseases not amenable to curative therapies. Patients with insufficient data were excluded. A historical cohort of patients with BTC consecutively treated at the same centers before the regulatory approval of ICI was also analyzed.
The study protocol complied with international ethical standards. The study was approved by the local ethics committee (Hospital e Maternidade São Luiz—HMSL, ethical report: 7.782.318; amendment 1; 7 May 2024), and informed consent was waived. Study data were recorded in a web-based electronic case report (REDCap; Vanderbilt University, Nashville, TN) hosted by the Instituto D’Or. Data were last updated on 1 August 2025. The study was reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines and to the European Society for Medical Oncology Guidance for Reporting Oncology real-world evidence.14
Data collection
Clinical data from both the ICI and historical control cohorts were retrospectively collected from medical records by two investigators (L.G.D.F. and C.S.F.) using standardized case report forms. The remaining coauthors were consulted whenever necessary for data clarification. Variables included demographics, comorbidities, laboratory values, and site of origin. Disease characteristics were evaluated through imaging and serum carbohydrate antigen [(CA) 19-9] and carcinoembryonic antigen values. PS was assessed using the Eastern Cooperative Oncology Group (ECOG) classification.15 Missing data were infrequent and were handled through complete-case analysis without imputation.
Treatment administration
Patients received treatment from the assistant physicians and were administered according to established protocols. Durvalumab plus gemcitabine and cisplatin were administered at the following doses: durvalumab 1500 mg on day 1 of each cycle, plus gemcitabine 1000 mg/m2 and cisplatin 25 mg/m2 on days 1 and 8 of each cycle (every 3 weeks). Chemotherapy was administered for eight cycles, followed by maintenance durvalumab every 4 weeks until progression or limiting toxicity.4 Pembrolizumab plus gemcitabine and cisplatin were administered as follows: pembrolizumab 200 mg on day 1, plus gemcitabine 1000 mg/m2 and cisplatin 25 mg/m2 on days 1 and 8 of each cycle (every 3 weeks).6 The decision on maintenance after cycle 8, as well as the treatment schedule and dosing, could be adapted according to the physician’s discretion and depended on multiple factors, including progression during the chemoimmunotherapy phase, clinical deterioration, and treatment-free interval strategies. Other backbone chemotherapies could be used, such as carboplatin plus gemcitabine, oxaliplatin plus gemcitabine, or gemcitabine alone.16,17 Alternative regimens were included because the study aimed to reflect real-world practice, particularly in patients considered unfit for cisplatin or requiring adapted regimens due to renal dysfunction, frailty, or toxicity concerns. Clinical monitoring and assessments were carried out according to the local practice. Treatment was continued until disease progression, loss of clinical benefit, or limiting toxicity. According to the local protocol, locally advanced BTC is managed with upfront systemic treatment, and locoregional therapies, including intra-arterial, radiation, and surgical therapies, are considered exceptional only after a response assessment and multidisciplinary evaluation.
Radiological response to treatment was assessed per RECIST criteria v1.1.18 Images were carried out as part of periodic restaging and analyzed locally. RECIST v1.1 assessments were based on imaging reviews carried out locally at participating institutions. Treatment-related adverse events were registered according to the National Cancer Institute Common Terminology Criteria for Adverse Events v5.0.19 Only adverse events that were considered related to the ICI were described.
Statistical analysis
The primary endpoint was OS. Secondary endpoints included objective response rate, treatment patterns, safety, prognostic factor analyses, and exploratory comparisons with a historical cohort.
We used descriptive statistics to summarize demographic characteristics, including frequencies, medians, and interquartile ranges (IQRs). OS curves were calculated using the Kaplan–Meier method and compared using the log-rank test. Survival was defined as the time between the first ICI treatment and death. Cox regression analyses were carried out to identify prognostic variables and to estimate hazard ratios (HRs) with 95% confidence intervals (CIs). The multivariable model was based on variables identified as statistically significant in the univariate analysis. All relevant baseline characteristics were included as follows: primary site, disease, previous treatment, disease status, ECOG PS, vascular invasion, ascites, albumin (< or ≥3.5 g/dl), neutrophil-to-lymphocyte ratio (NLR) (< or ≥3), bilirubin (< or ≥1.5), and CA 19-9 (< or ≥37 U/ml). ECOG PS, vascular invasion, NLR, and serum albumin were statistically significant in the univariate analysis (P < 0.05) and were included in the multivariable model. Response rate was analyzed separately as a time-dependent variable, adjusted for the independent variables.
A propensity score was estimated using logistic regression, including ECOG PS and chemotherapy backbone. These variables were selected because they represented clinically relevant factors with minimal missingness across cohorts. ECOG PS was independently associated with prognosis in the multivariable analysis, while chemotherapy backbone represented a potential source of imbalance between groups, given the established superiority of platinum doublets over single agents.3 Given the moderate sample size and number of events, a propensity score model was preferred to reduce overfitting and instability. Patients were matched 1:1 using nearest-neighbor matching with replacement and a caliper of 0.2 standard deviations of the logit of the propensity score. Inverse probability weighting was evaluated, but matching was chosen due to concerns about extreme weights and limited cohort overlap. It was also regarded as a more clinically interpretable method. Robust variance estimation was applied to account for the nonindependence introduced by replacement matching.
Given visual evidence of violation of the proportional hazards assumption, the restricted mean survival time (RMST) was used as a complementary measure. RMST was estimated up to a prespecified truncation time of 40 months, corresponding to the area under the Kaplan–Meier survival curves within this time horizon. All statistical analyses were carried out using STATA 17.0 (Stata, TX). Forest plots were generated using Google Colab (https://colab.research.google.com).
Results
Patients and disease presentation
After screening 179 patients for the ICI cohort, 27 were excluded due to insufficient data, and 152 were included. The baseline characteristics are summarized in Table 1. The cohort had a median age of 66.5 years, with a 57.9%:42.1% male-to-female distribution. Three patients had autoimmune disorders—one with Crohn disease, one with psoriasis plus ankylosing spondylitis, and one with psoriasis. The primary sites were intrahepatic (52.8%) and gallbladder (23.7%), and most patients had metastatic disease (68.4%). Within the cohort, a subset of patients had a somatic genetic profile available and presented molecular alterations of therapeutic importance: nine with isocitrate dehydrogenase 1 mutations, two with deficient mismatch repair proteins, two with fibroblast growth factor receptor 2 fusions, and three with B-Raf proto-oncogene (BRAF) mutations.
Table 1.
Baseline characteristics
| Variable | ICI cohort (n = 152) | Historical cohort (n = 94) |
|---|---|---|
| Age (years) | 66.5 (55.1-73.1) | 63.8 (53.5-74.3) |
| Sex | ||
| Male | 88 (57.9) | 54 (57.5) |
| Female | 64 (42.1) | 40 (42.5) |
| Primary site | ||
| Intrahepatic | 80 (52.6) | 56 (59.6) |
| Perihilar | 14 (9.2) | 13 (13.8) |
| Distal | 16 (10.5) | 9 (9.6) |
| Gallbladder | 36 (23.7) | 10 (10.6) |
| Ampullary | 6 (4.0) | 6 (6.4) |
| Comorbidities | ||
| Arterial hypertension | 60 (39.5) | 41 (43.6) |
| Diabetes | 38 (25.2) | 23 (24.4) |
| Overweight or obesity | 28 (18.4) | 23 (24.4) |
| Hepatitis B | 3 (2) | 1 (1.1) |
| Hepatitis C | 3 (2) | 1 (1.1) |
| Cirrhosis | 8 (5.3) | 6 (6.4) |
| Alcohol abuse | 29 (19.1) | 8 (8.5) |
| Liver steatosis | 21 (13.8) | 7 (7.4) |
| Tobacco | 29 (19.1) | 14 (14.9) |
| Cholelithiasis | 21 (13.8) | 18 (19.1) |
| Cholecystectomy | 31 (29.4) | 16 (17.0) |
| Autoimmune disorder | 3 (2.0) | 3 (3.2) |
| Previous surgery | 50 (32.8) | 25 (26.6) |
| Previous ablation | 4 (2.6) | 1 (1.1) |
| Previous biliary drainage or stent | 26 (17.1) | 26 (27.6) |
| Disease status | ||
| Locally advanced | 48 (31.6) | 37 (39.4) |
| Metastatic | 104 (68.4) | 57 (60.4) |
| ECOG PS | ||
| 0 | 73 (48.0) | 52 (55.3) |
| 1 | 59 (38.8) | 32 (34.0) |
| 2-3 | 20 (13.1) | 10 (10.7) |
| Vascular invasion | 32 (21.0) | 18 (19.1) |
| Ascites | 11 (7.2) | 5 (5.3) |
| Total bilirubin (mg/dl) | 0.6 (0.4-1.1) | 0.7 (0.4-0.8) |
| Albumin (g/dl) | 3.9 (3.1-4.3) | 3.9 (3.5-4.2) |
| CA 19.9 (U/ml) | 55 (15-660) | 47 (22-111) |
| CEA (μg/l) | 3 (1.8-11) | 3.2 (3.0-6.0) |
| NLR | 3 (1.7-5.4) | 3.1 (2.3-3.8) |
Values are n (%) or median (IQR).
CA, carbohydrate antigen; CEA, carcinoembryonic antigen; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; IQR, interquartile range; NLR, neutrophil-to-lymphocyte ratio; PS, performance status.
Treatment characteristics and outcomes
After a median follow-up of 8.3 months (IQR 3.7-14.9), 84 patients (55.2%) died, whereas 63 (41.4%) were alive (18 under treatment, and 5 were lost to follow-up). The median duration of ICI treatment was 4.4 months (IQR 2.4-8.2). Durvalumab was the most used ICI (n = 145; 95.4%), followed by pembrolizumab (n = 6; 3.9%) and nivolumab (n = 1; 0.7%). The most common backbone chemotherapy was cisplatin plus gemcitabine (90.1%), with a median of six cycles. Maintenance was applied in 57 (37.5%) patients after chemotherapy de-escalation, while 36 patients received a treatment-free interval after treatment de-escalation. Maintenance strategies included ICI monotherapy (n = 46) or gemcitabine plus ICI (n = 11). Table 2 provides details on the treatment characteristics and duration. The overall response rate was 38.8%, including 55 patients with partial responses and 4 with complete responses (Supplementary Table S1, available at https://doi.org/10.1016/j.esmorw.2026.100749). A significant association was found between the best response and type of maintenance (P = 0.001). Maintenance was less frequent among patients with stable disease (50.0%) than among those with partial response (72.1%). For patients with a partial response, maintenance with immunotherapy alone was more common than with immunotherapy plus gemcitabine (56.5% versus 19.6%), while 23.9% received a treatment-free interval. Similarly, for patients with stable disease, maintenance with immunotherapy was more common than with gemcitabine plus immunotherapy (43.7% versus 6.3%), while 50% received a treatment-free interval. Three patients (2.0%) with locally advanced disease underwent surgical resection, three (2.0%) received external radiation, and two (1.3%) received intra-arterial therapies.
Table 2.
Treatment characteristics and duration
| Variable | ICI cohort (n = 152) | Historical cohort (n = 94) |
|---|---|---|
| Backbone chemotherapy | ||
| Gemcitabine plus cisplatin | 137 (90.1) | 70 (74.5) |
| Gemcitabine plus carboplatin | 5 (3.3) | 1 (1.1) |
| Gemcitabine | 3 (2.0) | 6 (6.4) |
| Gemcitabine plus oxaliplatin | 1 (0.7) | 9 (9.6) |
| Other | 6 (4.0) | 8 (8.5) |
| No. of chemotherapy cycles | 6 (4-8) | 6 (4-7) |
| No. of immunotherapy cycles | 7 (4-11) | — |
| Maintenance after chemotherapy discontinuationa | ||
| None | 36 (38.7) | — |
| Gemcitabine plus immunotherapy | 11 (11.8) | — |
| Immunotherapy alone | 46 (49.4) | — |
| Reason for treatment discontinuationb | ||
| Progression | 133 (87.5) | 78 (82.9) |
| Toxicity | 6 (3.9) | 9 (9.6) |
| Maximum benefit | 8 (5.3) | 7 (7.4) |
| Line after immunotherapy-based treatment | ||
| 5-Fluorouracil plus oxaliplatin | 36 (23.7) | — |
| 5-Fluorouracil plus irinotecan | 5 (3.3) | — |
| 5-Fluorouracil monotherapy | 1 (0.6) | — |
| Locoregional | 5 (3.3) | — |
| Other | 12 (13.1) | — |
Values are n (%) or median (IQR).
ICI, immune checkpoint inhibitor; IQR, interquartile range.
For patients who were considered for treatment de-escalation without progression or limiting toxicity (n = 93).
Excluding five patients who were lost to follow-up in the ICI cohort.
The safety profile was favorable, with 10 (6.6%) cases of ICI-related adverse events of grade ≥2 (Supplementary Table S2, available at https://doi.org/10.1016/j.esmorw.2026.100749). Treatment discontinuation due to toxicity occurred in six patients (3.9%).
The median OS (n = 152) was 15.5 months (95% CI 12.3-21.5) (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmorw.2026.100749). Fifty-nine patients received second-line therapies, with a median OS from the initiation of first-line therapy of 18.0 months (95% CI 14.1-30.9).
Subgroup analysis, prognostic factors, and treatment response
In the univariate analysis, ECOG PS of >0 (HR 2.40, 95% CI 1.75–3.31, P < 0.001), vascular invasion (HR 2.48, 95% CI 1.51-4.08, P < 0.001), ascites (HR 3.18, 95% CI 1.56-6.48, P = 0.001), NLR ≥ 3 (HR 3.25, 95% CI 1.46-7.25, P = 0.004), and albumin > 3.5 g/dL (HR 0.26, 95% CI 0.11-0.60, P = 0.002) correlated with prognosis. The multivariate analysis confirmed the negative prognostic roles of ECOG PS of >0 (HR 3.91, 95% CI 1.17-13.10, P = 0.026), NLR ≥ 3 (HR 1.98, 95% CI 1.06-5.35, P = 0.017), and vascular invasion (HR 1.83, 95% CI 1.14-4.41, P = 0.006) (Table 3 and Figure 1).
Table 3.
Univariate and multivariate analyses in subgroups
| Subgroup | Median OS (95% CI) | Univariate P | Multivariate HR (95% CI); P |
|---|---|---|---|
| ECOG PS | |||
| 0 | 22.9 (15.6-NR) | <0.001 | 3.91 (1.17-13.10); 0.026 |
| ≥1 | 9.8 (8.0-13.6) | ||
| Vascular invasion | |||
| Yes | 12.2 (5.2-15.5) | 0.001 | 1.83 (1.14-4.41); 0.006 |
| No | 18.0 (13.6-30.9) | ||
| NLR | |||
| ≥3 | 14.0 (5.7-18) | 0.004 | 1.98 (1.06-5.35); 0.017 |
| <3 | 30.9 (12.6-NR) | ||
| Serum albumin (g/dl) | |||
| <3.5 | 12.3 (3.3-14.2) | 0.002 | 1.22 (0.27-5.41); 0.792 |
| ≥3.5 | 20.5 (12.6-NR) | ||
CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; NLR, neutrophil-to-lymphocyte ratio; NR, not reached; OS, overall survival; PS, performance status.
Figure 1.
Kaplan–Meier curves showing the overall survival in the subgroups. (Left) Performance status (PS), 0 versus ≥1; (center) vascular invasion, no versus yes; and (right) neutrophil-to-lymphocyte ratio (NLR), ≥3 versus 3.
There was no significant difference in OS by tumor site. The median OS was 22.9 months (95% CI 8.0-not available [NA]) for gallbladder carcinoma; 15.6 months (95% CI 12.2-21.6) for iCCA; 13.7 months (95% CI 9.4-NA) for dCCA; 11.6 months (95% CI 2.2-NA) for pCCA; and 5.9 months (95% CI 1.1-NA) for ampullary tumors (P = 0.683) (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmorw.2026.100749). The patients with deficient mismatch repair proteins (n = 2) had survival times of 26.3 and 16.1 months. Similarly, there was no significant difference according to disease extent between metastatic (median OS 17 months, 95% CI 12.6-28.8) versus locally advanced (median OS 12.6, 95% CI 10.1-18.0, HR 0.73, 95% CI 0.47-1.12, P = 0.151).
Radiological response is a postbaseline variable and is reported as an exploratory descriptive analysis. Patients with a partial response had a median OS of 27.5 months (95% CI 20.9-NA), while those with stable disease had a median OS of 14.3 months (95% CI 11.6-18.0) and those with progressive disease had a median OS of 5.7 months (95% CI 2.2-8.6). Low response rates were observed in patients with vascular invasion (28.1%), NLR ≥ 3 (19.3%), and low albumin (25.0%) (Supplementary Figure S3, available at https://doi.org/10.1016/j.esmorw.2026.100749).
Historical control group: characteristics and outcomes
Data from 94 patients consecutively treated over the 3 years preceding local approval of ICI (2018-2021) were collected. The baseline characteristics are summarized in Table 1. There were no differences in baseline characteristics between the historical and the ICI cohorts after adjusting for multiple comparisons using an independent-samples (unpaired) t-test for continuous variables and a two-sided Fisher’s exact test for categorical variables. The median age of the historical cohort was 63.8 years (IQR 53.5-74.3; t-test P = 0.361), and ECOG PS was 0-1 in 89.3% (P = 0.653). The main sites were intrahepatic (n = 56, 59.6%), followed by perihilar (n = 13, 13.8%, P = 0.093). The presentation was locally advanced (n = 37, 39.4%) or metastatic (n = 57, 60.6%, P = 0.234). The most used chemotherapy regimen was gemcitabine plus cisplatin (n = 70, 74.5). Second-line therapies were received by 64 patients (68.1%), mainly oxaliplatin plus 5-fluorouracil (FOLFOX) (n = 36, 38.3%) and irinotecan plus 5-fluorouracil (FOLFIRI) (n = 11, 11.7%).
After a median follow-up of 11.3 months (IQR 7.3-21.3), the median OS in the historical group was 14.5 months (95% CI 9.8-17.6). A Cox regression showed no difference in survival between the historical and ICI cohorts (HR 0.79, 95% CI 0.57-1.09, P = 0.13). The Kaplan–Meier curves show an initial crossing of the curves and a subsequent separation, with a plateau at 35-40 months in favor of the ICI group (Figure 2). Due to visual violation of the proportional hazards assumption, RMSTs were compared between groups: the ICI cohort had an estimated RMST of 22.1 months (95% CI 18.1-25.9) and the historical cohort 17.7 months (95% CI 14.7-20.7, P = 0.08).
Figure 2.
Kaplan–Meier curves. (Left) The survival of the immune checkpoint inhibitor (ICI) cohort (n = 152) and the historical cohort (n = 94); (right) the curves after propensity score matching comparing the ICI cohort (n = 77) versus the historical cohort (n = 78).
Propensity score matching was carried out 1 : 1 with replacement using nearest-neighbor matching based on the backbone chemotherapy and ECOG PS, resulting in the matched ICI (n = 77) and historical cohorts (n = 78). Main baseline characteristics are shown in Table 4. After propensity score matching, baseline covariates were well balanced between groups, with all standardized mean differences <0.10. ECOG PS and backbone chemotherapy distribution showed a substantial reduction in baseline imbalance. Subsequent lines of therapy were used in 69.2% of patients in the historical cohort and in 51.9% in the ICI cohort, after matching.
Table 4.
Baseline characteristics after propensity score matching
| Variable | ICI cohort (n = 77) | Historical cohort (n = 78) |
|---|---|---|
| Age (years) | 62.6 (52.7-71.6) | 63.0 (53.1-71.7) |
| Sex | ||
| Male | 45 (58.4) | 42 (53.8) |
| Female | 32 (41.6) | 36 (46.2) |
| Primary site | ||
| Intrahepatic | 46 (59.7) | 48 (61.5) |
| Perihilar | 6 (7.8) | 10 (12.8) |
| Distal | 7 (9.1) | 8 (10.2) |
| Gallbladder | 14 (18.2) | 8 (10.2) |
| Ampullary | 4 (5.2) | 4 (5.1) |
| Comorbidities | ||
| Arterial hypertension | 31 (40.3) | 31 (39.7) |
| Diabetes | 18 (23.4) | 19 (24.3) |
| Overweight or obesity | 21 (27.3) | 20 (25.6) |
| Disease status | ||
| Locally advanced | 28 (36.4) | 30 (38.5) |
| Metastatic | 49 (63.6) | 48 (61.5) |
| ECOG PS | ||
| 0 | 68 (88.3) | 68 (87.2) |
| 1 | 8 (10.4) | 9 (11.5) |
| 2-3 | 1 (1.3) | 1 (1.3) |
| Vascular invasion | 12 (15.6) | 16 (20.5) |
| CA 19.9 (U/ml) | 57 (31-119) | 60 (29-104) |
| NLR | 3.6 (1.9-5.5) | 3.4 (2.9-3.9) |
Values are n (%) or median (IQR).
CA, carbohydrate antigen; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; IQR, interquartile range; NLR, neutrophil-to-lymphocyte ratio; PS, performance status.
Survival analyses adjusted for matching weights derived from the propensity score model showed a significant benefit of ICI, with median OS of 20.5 months (95% CI 14.3-27.1) in the ICI cohort versus 13.7 (95% CI 8.0-17.9) months in the historical control cohort (HR 0.61, 95% CI 0.39-0.93, P = 0.025).
Discussion
This study provides real-world evidence on the use of chemoimmunotherapy in patients with advanced BTC, corroborating the outcomes in the pivotal trials.4,6 Among 152 patients, we observed a median OS of 15.5 months, which aligns with the 12.8- to 15.8-month range reported in clinical and real-world studies worldwide.9, 10, 11,20
Notably, our cohort comprised a more heterogeneous population than those typically enrolled in clinical trials. For instance, 12.6% of patients had an ECOG PS of ≥2, and comorbidities such as autoimmune disorders were included. Besides, we observed a notable prevalence of diabetes and overweight, while a low prevalence of cirrhosis. A previous Brazilian study suggested an association of metabolic disorders and cholangiocarcinoma.21 On the contrary, an analysis of 109 studies from different regions revealed that 10.2% of patients with cholangiocarcinoma had cirrhosis.22 This suggests distinct regional risk factors.
Latin America is underrepresented in BTC trials, although the local incidence is high.23 In the TOPAZ-1 trial, only Argentina (n = 22) and Chile (n = 9) were represented.4 In the KEYNOTE-966, the inclusion of patients from Latin America was not detailed, although an unbalanced representation of Asia was observed (45%).6 Both trials were statistically positive, leading to global regulatory approval, including Latin American countries. While these regimens are associated with modest benefits, regional evidence is needed to confirm their applicability.
The median OS and response rate were equivalent across primary sites and disease extent, although we acknowledge a higher proportion of locally advanced tumors (versus metastatic) than in the pivotal clinical trial.4,6 Multivariate analysis confirmed that ECOG PS of >0, NLR of ≥3, and vascular invasion were negative prognostic factors. Vascular invasion is a surrogate of local aggressiveness and frequently induces complications such as portal hypertension and strictures of adjacent structures, such as the bile ducts.24 Studies suggest that NLR may predict outcomes in liver tumors.25,26 NLR reflects the balance between neutrophils, which promote inflammation and cancer growth, and lymphocytes, which drive antitumor immunity. It is suggested that an inverse relationship exists between NLR and CD8+-cell infiltration.27, 28, 29 In BTC, however, the link between systemic and local immunity remains unclear.
In patients with initial disease control, the pivotal trials were designed to deliver a maximum of eight cycles of chemotherapy; however, the benefit of maintenance beyond eight cycles remains uncertain. The TOPAZ-1 trial was intended for an ICI-only maintenance,4 while the KEYNOTE-966 included gemcitabine with ICI.6 We observed that maintenance strategies are used in over half of patients who completed chemoimmunotherapy, and the response to initial treatment may influence the decision to use maintenance therapy. This trend varies across reports favoring less-intensive maintenance for responders and prospective studies are warranted to confirm the optimal strategy.
We found a favorable safety profile, showing discontinuations due to toxicity in 6%. The incidence of immune-related adverse events was low, although it is potentially underestimated due to limited follow-up or underreporting.
In addition to descriptive outcomes, we compared the effectiveness of chemoimmunotherapy versus a historical cohort treated before the local approval of ICI. After propensity score matching, a statistically significant difference in OS was observed. Visual inspection of Kaplan–Meier curves revealed a late separation, with a survival plateau emerging beyond 35 months, accounting for 25%-30% of patients in the ICI cohort. Although the between-group difference in RMST up to 40 months did not reach conventional statistical significance, the numerically longer RMST observed in the ICI cohort suggests a potential late survival advantage consistent with the durable benefit pattern reported with immunotherapy across solid tumors.
This study has limitations inherent to its retrospective design and the context of a segmented health care system, including potential selection and reporting biases, as well as variability in imaging assessments and follow-up intervals. Twenty-seven patients were initially excluded for insufficient data. We found no clear evidence of systematic exclusion of patients with more aggressive disease; however, selection bias cannot be ruled out. Molecular testing was not uniformly carried out, reflecting access barriers, and programmed death-ligand 1 expression was not consistently available.
One important limitation of this study is the relatively short follow-up in comparison with the observed OS. This reflects the recent incorporation of ICI into clinical practice and the consequent limited duration of available observation. Therefore, the estimated long-term survival outcomes should be interpreted with caution, as additional follow-up may modify the magnitude of the observed benefit. Furthermore, the current follow-up duration limits a more comprehensive assessment of late recurrences, long-term disease control, and delayed toxicities. Despite these limitations, the study provides relevant real-world evidence regarding the early effectiveness and safety in a population that is frequently underrepresented in clinical trials.
An additional limitation inherent to historical comparisons relates to the temporal separation between the historical and ICI cohorts. Because patients in the historical cohort were treated predominantly during earlier years of the study period, differences in outcomes may have been influenced not only by ICI exposure but also by evolving supportive care practices, availability of molecular profiling, and increased access to subsequent lines of therapy during more recent years. Furthermore, follow-up duration differed between cohorts, and a higher proportion of censored observations in the ICI cohort may have introduced informative censoring and affected long-term estimates. Although propensity score matching was carried out to reduce baseline imbalances, residual confounding related to calendar-time effects cannot be fully excluded and should be considered when interpreting the comparative results.
Despite these limitations, this study is among the first analyses of immunotherapy in BTC conducted in Latin America, offering insights from a health care system with unique structural and access challenges. Only a small proportion of patients in our region have access to ICI due to the financial burden and barriers to early diagnosis. The findings support the need to discuss measures to reduce financial toxicity and to explore biomarkers to identify patients who are more likely to benefit.
Conclusion
Our real-world cohort supports the efficacy and safety of combining chemoimmunotherapy in BTC, with survival outcomes comparable to those in pivotal trials. Prognostic factors, including ECOG PS, NLR, and vascular invasion, were associated with worse outcomes, underscoring the need for individualized approaches. Low-resource adaptations of ICI-based regimens remain a critical next step in optimizing care across regions.
Acknowledgments
Data availability
The datasets generated and analyzed during the current study are derived from institutional clinical records and are not publicly available due to privacy and regulatory restrictions, but may be made available from the corresponding author upon reasonable request and with permission of the participating institutions.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Disclosure
LGDF—honoraria: AstraZeneca, Roche, SERVIER, Bayer, and Merck Sharp & Dohme (MSD). RAVG—honoraria: Novartis, Roche, AstraZeneca, MSD Oncology, Bristol Myers Squibb (BMS) Brazil, Astellas Pharma, and Pfizer; consulting or advisory role: Novartis, Roche, MSD Oncology, and BMS Brazil; and travel, accommodations, and expenses: AstraZeneca, MSD Oncology, and Roche. MIFMB—honoraria: MSD, BMS Brazil, Eurofarma, Takeda, Amgen, Bayer, AstraZeneca, and Daiichi Sankyo/Astra Zeneca; consulting or advisory role: MSD, Lilly, Roche, Merck, SERVIER, BeiGene, Johnson & Johnson/MedTech, AstraZeneca, and BMS Brazil; research funding: BMS Brazil (Inst), MSD (Inst), BeiGene (Inst), AstraZeneca (Inst), and Daiichi Sankyo/Astra Zeneca (Inst); and travel, accommodations, and expenses: MSD, SERVIER, BMS Brazil, and Bayer. MBS—consulting or advisory role: Janssen-Cilag, Bayer, Pfizer, and Astellas Pharma; speakers’ bureau: Bayer, Janssen-Cilag, Pfizer, BMS/Pfizer, and AstraZeneca; and travel, accommodations, and expenses: Janssen Oncology, SERVIER, and Bayer. AdMP—honoraria: SERVIER, Bayer, Roche, Amgen, Bristol Myers Squibb Foundation, and MSD; consulting or advisory role: Amgen; and travel, accommodations, and expenses: Merck. DRDRF—speakers’ bureau: Ipsen, Bayer, Merck Serono, United Medical, SERVIER, AstraZeneca, and Roche and travel, accommodations, and expenses: Roche. MCG—honoraria: AstraZeneca; consulting or advisory role: Pfizer; and research funding: Merck. GLSP—consulting or advisory role: SERVIER, Pfizer, Daiichi Sankyo/Lilly, Bayer, and AstraZeneca; speakers’ bureau: SERVIER/Pfizer, Daiichi Sankyo/Lilly, AstraZeneca, and Bayer; and travel, accommodations, and expenses: Daiichi Sankyo/Lilly, SERVIER, and AstraZeneca. PMH—leadership: Oncologia D’Or; stock and other ownership interests: Oncostar; honoraria: Bayer Health (Inst) and United Medical (Inst); consulting or advisory role: United Health Group (Inst), Bayer (Inst), Lilly (Inst), Exelixis (Inst), EMS, AstraZeneca (Inst), and EMS; research funding: Bayer (Inst), MSD Oncology (Inst), Novartis (Inst), Exelixis (Inst), Roche/Genentech (Inst), AstraZeneca/MedImmune (Inst), Lilly (I), BMS Brazil (Inst), Daiichi Sankyo/Astra Zeneca (Inst), and AbbVie (Inst). The other authors have declared no conflicts of interest.
Supplementary data
References
- 1.Forner A., Vidili G., Rengo M., Bujanda L., Ponz-Sarvisé M., Lamarca A. Clinical presentation, diagnosis and staging of cholangiocarcinoma. Liver Int. 2019;39(suppl 1):98–107. doi: 10.1111/liv.14086. [DOI] [PubMed] [Google Scholar]
- 2.Celotti A., Solaini L., Montori G., Coccolini F., Tognali D., Baiocchi G. Preoperative biliary drainage in hilar cholangiocarcinoma: systematic review and meta-analysis. Eur J Surg Oncol. 2017;43(9):1628–1635. doi: 10.1016/j.ejso.2017.04.001. [DOI] [PubMed] [Google Scholar]
- 3.Valle J., Wasan H., Palmer D.H., et al. Cisplatin plus gemcitabine versus gemcitabine for biliary tract cancer. N Engl J Med. 2010;362(14):1273–1281. doi: 10.1056/NEJMoa0908721. [DOI] [PubMed] [Google Scholar]
- 4.Oh D.-Y., Ruth He A., Qin S., et al. Durvalumab plus gemcitabine and cisplatin in advanced biliary tract cancer. NEJM Evid. 2022;1(8) doi: 10.1056/EVIDoa2200015. [DOI] [PubMed] [Google Scholar]
- 5.Oh D.-Y., He A.R., Bouattour M., et al. Durvalumab or placebo plus gemcitabine and cisplatin in participants with advanced biliary tract cancer (TOPAZ-1): updated overall survival from a randomised phase 3 study. Lancet Gastroenterol Hepatol. 2024;9(8):694–704. doi: 10.1016/S2468-1253(24)00095-5. [DOI] [PubMed] [Google Scholar]
- 6.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(10391):1853–1865. doi: 10.1016/S0140-6736(23)00727-4. [DOI] [PubMed] [Google Scholar]
- 7.Deng M., Li S., Wang Q., et al. Real-world outcomes of patients with advanced intrahepatic cholangiocarcinoma treated with programmed cell death protein-1-targeted immunotherapy. Ann Med. 2022;54(1):803–811. doi: 10.1080/07853890.2022.2048416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rimini M., Foti S., Camera S., et al. Survival outcomes of durvalumab in combination with cisplatin and gemcitabine in advanced biliary tract cancer: real-world results from a single Italian institution. Oncology. 2025;103(8):677–698. doi: 10.1159/000541891. [DOI] [PubMed] [Google Scholar]
- 9.Gerhardt F., Müller C., Venerito M., et al. Treatment with gemcitabine/cisplatin and durvalumab for advanced biliary tract cancer—real-world data from a multicenter German patient population. J Cancer Res Clin Oncol. 2025;151(6):191. doi: 10.1007/s00432-025-06239-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rimini M., Fornaro L., Rizzato M.D., et al. Durvalumab plus gemcitabine and cisplatin in advanced biliary tract cancer: a large real-life worldwide population. Eur J Cancer. 2024;208 doi: 10.1016/j.ejca.2024.114199. [DOI] [PubMed] [Google Scholar]
- 11.Rimini M., Masi G., Lonardi S., et al. Durvalumab plus gemcitabine and cisplatin versus gemcitabine and cisplatin in biliary tract cancer: a real-world retrospective, multicenter study. Target Oncol. 2024;19(3):359–370. doi: 10.1007/s11523-024-01060-1. [DOI] [PubMed] [Google Scholar]
- 12.da Fonseca L.G., Izquierdo-Sanchez L., Hashizume P.H., et al. Cholangiocarcinoma in Latin America: a multicentre observational study alerts on ethnic disparities in tumour presentation and outcomes. Lancet Reg Health Am. 2024;40 doi: 10.1016/j.lana.2024.100952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Castelo-Branco L., Pellat A., Martins-Branco D., et al. ESMO Guidance for Reporting Oncology real-World evidence (GROW) ESMO Real World Data Digit Oncol. 2023;1 doi: 10.1016/j.esmorw.2023.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Oken M.M., Creech R.H., Tormey D.C., et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol. 1982;5(6):649–655. [PubMed] [Google Scholar]
- 15.Williams K.J., Picus J., Trinkhaus K., et al. Gemcitabine with carboplatin for advanced biliary tract cancers: a phase II single institution study. HPB (Oxford) 2010;12(6):418–426. doi: 10.1111/j.1477-2574.2010.00197.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lagenfelt H., Blomstrand H., Elander N.O. Real-world evidence on palliative gemcitabine and oxaliplatin (GemOx) combination chemotherapy in advanced biliary tract cancer. Cancers (Basel) 2021;13(14):3507. doi: 10.3390/cancers13143507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.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(2):228–247. doi: 10.1016/j.ejca.2008.10.026. [DOI] [PubMed] [Google Scholar]
- 18.Freites-Martinez A., Santana N., Arias-Santiago S., Viera A. Using the Common Terminology Criteria for Adverse Events (CTCAE – version 5.0) to evaluate the severity of adverse events of anticancer therapies. Actas Dermosifiliogr (Engl Ed) 2021;112(1):90–92. doi: 10.1016/j.ad.2019.05.009. [DOI] [PubMed] [Google Scholar]
- 19.Wang B.-C., Kuang B.-H., Lin G.-H., Fu C. Durvalumab and pembrolizumab in advanced biliary tract cancer: a reconstructed patient-level mimic head-to-head comparative analysis. Front Immunol. 2024;15 doi: 10.3389/fimmu.2024.1497415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Da Fonseca L.G., Hashizume P.H., de Oliveira I.S., et al. Association between metabolic disorders and cholangiocarcinoma: impact of a postulated risk factor with rising incidence. Cancers (Basel) 2022;14(14):3483. doi: 10.3390/cancers14143483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Tham E.K.J., Lim R.Y., Koh B., et al. Prevalence of chronic liver disease in cholangiocarcinoma: a meta-analysis. Clin Gastroenterol Hepatol. 2025;23(10):1710–1718. doi: 10.1016/j.cgh.2024.09.028. [DOI] [PubMed] [Google Scholar]
- 22.Chen Y., Liu H., Zhang J., et al. Prognostic value and predication model of microvascular invasion in patients with intrahepatic cholangiocarcinoma: a multicenter study from China. BMC Cancer. 2021;21(1):1299. doi: 10.1186/s12885-021-09035-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Da Fonseca L.G., Uratani L.F., Soares G.F., et al. Early variation of inflammatory indexes refines prognostic prediction in patients with hepatocellular carcinoma under systemic treatment. Mol Clin Oncol. 2023;18(4):29. doi: 10.3892/mco.2023.2625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Templeton A.J., McNamara M.G., Šeruga B., et al. Prognostic role of neutrophil-to-lymphocyte ratio in solid tumors: a systematic review and meta-analysis. J Natl Cancer Inst. 2014;106(6) doi: 10.1093/jnci/dju124. [DOI] [PubMed] [Google Scholar]
- 25.Nozawa H., Chiu C., Hanahan D. Infiltrating neutrophils mediate the initial angiogenic switch in a mouse model of multistage carcinogenesis. Proc Natl Acad Sci U S A. 2006;103(33):12493–12498. doi: 10.1073/pnas.0601807103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shi F., Shi M., Zeng Z., et al. PD-1 and PD-L1 upregulation promotes CD8(+) T-cell apoptosis and postoperative recurrence in hepatocellular carcinoma patients. Int J Cancer. 2011;128(4):887–896. doi: 10.1002/ijc.25397. [DOI] [PubMed] [Google Scholar]
- 27.Hedrick C.C., Malanchi I. Neutrophils in cancer: heterogeneous and multifaceted. Nat Rev Immunol. 2022;22(3):173–187. doi: 10.1038/s41577-021-00571-6. [DOI] [PubMed] [Google Scholar]
- 28.Szczerba B.M., Castro-Giner F., Vetter M., et al. Neutrophils escort circulating tumour cells to enable cell cycle progression. Nature. 2019;566(7745):553–557. doi: 10.1038/s41586-019-0915-y. [DOI] [PubMed] [Google Scholar]
- 29.Lamarca A., Adeva J., Ales Díaz I., Alvarez Gallego R., Muñoz Martín A.J., Macarulla Mercade T. Maintenance chemotherapy in biliary tract tumours in the era of immuno-chemotherapy. ESMO Gastrointest Oncol. 2025;7 doi: 10.1016/j.esmogo.2024.100116. [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.
Supplementary Materials
Data Availability Statement
The datasets generated and analyzed during the current study are derived from institutional clinical records and are not publicly available due to privacy and regulatory restrictions, but may be made available from the corresponding author upon reasonable request and with permission of the participating institutions.