Third-line treatment of biliary tract cancers in a real-life setting Authors: A. Boilève, A. Mihaele, M. Khoury, A. Hollebecque, C. Smolenschi, L. Antoun, N. Masri, M. Valéry, T. Pudlarz, A. Fuerea, F. Facchinetti, A. Tarabay, D. Malka, V. Boige, M.-L. Tanguy, M. Ducreux Published: 2026 ABSTRACT BackgroundBiliary tract cancers represent rare, aggressive malignancies of the bile duct epithelium. Only 20% cases are eligible for curative surgery because most cases are diagnosed at advanced stages, and prognosis is overall poor. The systemic treatment landscape has evolved, with gemcitabine cisplatin plus immunotherapy being now first-line standard. For second-line treatment, options include targeted therapies or chemotherapy for patients with or without specific actionable genetic alterations, respectively. However, no third-line treatment standard has been validated by randomized clinical trials.Patients and methodsThis retrospective analysis from a specialized French center aimed to assess patient outcomes receiving third-line chemotherapy, from 2012 to 2023.ResultsOverall, 256 patients receiving third-line treatment were included. Treatment distribution was as follows: FOLFIRI (n = 51, 20%), paclitaxel (n = 43, 17%), combinations of 5-FU cisplatin/FOLFOX (n = 32, 13%), molecularly targeted therapy (n = 65, 25%) and others (n = 63, 25%). In patients without actionable alterations, the median overall survival (OS) for patients receiving third-line treatment was 10.2 months. OS was 13 months with paclitaxel and 7 months with FOLFIRI (P = 0.03). The median progression-free survival for the third-line treatments was 3.3 months for paclitaxel and 2.6 months with FOLFIRI.ConclusionPaclitaxel seemed better than FOLFIRI as third-line treatment in patients without actionable alteration. This highlights the need for further research, especially in identifying prognostic factors and refining treatment strategies. Introduction Biliary tract cancers (BTC) comprise a group of rare and highly invasive malignancies including cholangiocarcinoma, which arises from the epithelium of the bile ducts and are further categorized into intrahepatic (iCCA), perihilar (pCCA), and distal (dCCA) subtypes including gallbladder carcinoma (GBC).1 In addition to the anatomical heterogeneity, there is also a genomic diversity, with genetic alterations that differ according to the anatomical location. Regarding actionable alterations, IDH1/2 mutations and FGFR2 fusions are mostly found in iCCA,2,3 and HER2 alterations are more frequently detected in gallbladder cancer and dCCA.3, 4, 5 Microsatellite instability can also be identified in some rare cases.6 Additional genetic alterations include BRAF mutations, TP53 mutation, CDKN2A/B loss, mutations in c-MET, NRAS, and PI3K, along with mutations in the ARID1A and BAP1 genes and rare NTRK fusions.3, 4, 5 KRAS mutations are predominantly observed in dCCA.3, 4, 5 Most patients with BTC are diagnosed at an advanced stage, with the majority presenting with unresectable or metastatic tumors.1,4,7,8 Systemic treatment in the first-line setting is based on a combination chemotherapy [cisplatin gemcitabine (GemCis)] and immunotherapy, from the TOPAZ-1 study9 and the KEYNOTE 966 study.10 Despite these advancements in survival with the addition of immunotherapy, many patients will still experience disease progression during first-line treatment. The introduction of targeted therapies for specific genetic alterations has revolutionized the management of advanced-stage disease.3, 4, 5 For fit patients without actionable alteration in second line, FOLFOX is a possible regimen, with a moderate efficacy [overall survival (OS) of 6.2 months compared with 5.3 months with best supportive care].11,12 In case of actionable alterations, several studies demonstrated the benefit of a targeted treatment. In case of FGFR2 fusion Pemigatinib13 or Futibatinib14 should be considered, Ivosidenib in case of IDH1 mutation,15 antiHER2 in case of HER2 amplification,16 a combination of Dabrafenib and Trametinib in case of BRAFV600E mutation,17 Larotrectinib or Entrectinib in case of NTRK fusion,18,19 immunotherapy in case of microsatellite instability-High (MSI-H)/Deficient Mismatch Repair (dMMR) tumors,6,20 as recommended by ESMO guidelines.7,8 Approximately 20%-40% of the patients could potentially receive a third-line therapy after second-line treatment.21,22 This group has a potential opportunity for tumor control, maintenance of an acceptable quality of life, and possibly prolonged survival, which highlights the critical need to identify patients who can benefit from or tolerate third-line therapies.23 Although a standard palliative regimen has not been established due to the lack of randomized studies, trials are challenging because of the disease’s heterogeneity, its rarity, and the varying performance status of patients at progression after second-line therapy.21,22 The actual benefit of third-line and subsequent chemotherapy remains unproven and must be carefully weighed against potential treatment-related toxicities, biliary complications, and overall deterioration in health.7 This study aimed to identify the systemic treatments commonly used for patients undergoing third-line palliative care and evaluate outcomes in a tertiary care center. Patients and methods Patients This single-center retrospective analysis was conducted at Gustave Roussy, a referral center for oncology in France. Patients were identified through a review of institutional registries at the study site between 2012 and 2023. Data were collected by hospital chart review and managed using REDCap electronic data capture tools hosted at Gustave Roussy. This retrospective study complies with the French MR004 methodology regarding general data protection regulation for noninterventional retrospective health research and was approved by our institutional review board, in compliance with Helsinki declaration (IRB 2024-392). Inclusion criteria were as follows: (i) age ≥18 years; (ii) diagnostic of BTC (iCCA, pCCA, dCCA) or GBC; (iii) receipt of at least one cycle of systemic treatment in the third line of palliative therapy; and (iv) medical oncological care (consultation and/or treatment administration) received at this oncology center. Patients with ampullary carcinoma or mixed hepatocellular-cholangiocarcinoma were excluded from the database. Objectives The primary objectives of this analysis were to characterize the third line of systemic treatment administered and to analyze the OS and the PFS from the date of the first dose of third line. Secondary objectives included the description of the clinical and molecular characteristics of patients who received third-line treatment, to identify prognostic factors and to describe subsequent lines of therapy administered to patients who progressed beyond third-line treatment. Of note, due to the long period of inclusion, targeted therapies were administered in various lines and not in second line as recommended in the ESMO guidelines.8 We analyzed the potential impact on survival of various parameters: age at diagnosis (<65 years or ≥65 years), anatomical location of the tumor (intrahepatic versus other), metastatic status at diagnosis, mode of diagnosis (jaundice, angiocholitis, or cholestasis), treatment, age at the start of treatment (<65 years or ≥65 years), the number of metastatic sites at beginning of L3 (<2 versus ≥2) and hepatic metastases at beginning of L3. Additionally, the analysis was adjusted for the year when third-line treatment was administered, categorized as either 2016 or earlier, or later than 2016. Statistical analysis OS was defined as the time from the first dose of third-line therapy to death from any cause. Patients alive at the last follow-up were censored at the date of last contact. PFS was defined as the time from the first dose of third-line therapy to the time of clinical and/or radiological progression or death. Patients who were alive without progression at the last date of follow-up were censored at the date of last contact. OS and PFS were estimated using the Kaplan–Meier method. Median survival times were obtained from the Kaplan–Meier estimates and 95% confidence intervals (CIs) were calculated using the Brookmeyer–Crowley method. A multivariable cox regression model with backward selection was used to identify prognostic factors. No imputation of missing data was carried out and a complete-case analysis was carried out. A log-rank test was used to compare OS between patients receiving FOLFIRI and those receiving other treatments. Dunnett’s correction was applied to control for multiple comparisons. Statistical analyses were conducted using SAS 9.4 software (version 7.15). A P value < 0.05 was considered statistically significant. Results Patient characteristics From 594 eligible patients, a total of 256 patients received third-line systemic therapy and were included in the study (excluded patients: 189 received only one-line, 130 only two-lines, 19 with missing data) (Figure 1). All of them presented with metastatic disease [metachronous (n = 61), 28% or synchronous (n = 156)]. The cohort was predominantly female (61%), with a median age at diagnosis of 60 years. The majority of patients had tumors anatomically localized to the intrahepatic region (n = 174, 69%), followed by GBC (n = 34, 13%), dCCA (n = 26, 10%) and pCCA (n = 20 (8%). Among patients with metastatic disease, the most common sites of metastasis included liver metastases (multifocal disease n = 120, 55%), lymph node involvement (n = 65, 30%), peritoneal metastases (n = 40, 18%) and lung metastases (n = 30, 14%). The baseline clinicopathological features of the cohort are summarized in Table 1 and Supplementary Table S1, available at https://doi.org/10.1016/j.esmogo.2026.100345, according to treatment group.Table 1Patients characteristicsVariableNumber (%)Age, years median (range)60 (27; 84)Sex Female157 (61.3%) Male99 (38.6%)ECOG PS at diagnosis 050 (23.4%) 1158 (73.8%) 25 (2.3%) 31 (0.5%) Missing42Primary tumor site iCCA174 (68.5%) pCCA20 (7.9%) dCCA26 (10.2%) GBC34 (13.3%) Missing2Metastatic setting at diagnosis Yes156 (71.9%) No61 (28.1%) Missing39Primary surgery of the primary tumor Yes103 (40.2%) No153 (59.8%)Presence of hepatic metastases at L3 Yes153 (60%) No103 (40%)Number of metastatic sites at L3 199 (39%) >2157 (61%)ECOG at L3 0 or 1165 (82%) >236 (18%) Missing55Locoregional therapy received Radiofrequency ablation8 (3.1%) Radiotherapy83 (32.4%) Radioembolization16 (6.3%)Outcomes of the patients Death199 (85%) Lost to follow-up22 (9%) On-going treatment at the time of data cutoff28 (10.9%) Synchronous liver metastases/multifocal disease120 (46.9%) Synchronous bilobar liver disease97 (37.9%) Synchronous extrahepatic metastases111 (43.4%) Lymph node metastases65 (25.4%) Peritoneal metastases40 (15.6%) Lung metastases30 (11.7%) Bone metastases10 (3.9%) Other metastases10 (3.9%)dCCA, distal subtypes; ECOG PS, performans status; iCCA, categorized into intrahepatic; GBC, gallbladder carcinoma; pCCA, perihilar categorized into intrahepatic. Table 1Patients characteristicsVariableNumber (%)Age, years median (range)60 (27; 84)Sex Female157 (61.3%) Male99 (38.6%)ECOG PS at diagnosis 050 (23.4%) 1158 (73.8%) 25 (2.3%) 31 (0.5%) Missing42Primary tumor site iCCA174 (68.5%) pCCA20 (7.9%) dCCA26 (10.2%) GBC34 (13.3%) Missing2Metastatic setting at diagnosis Yes156 (71.9%) No61 (28.1%) Missing39Primary surgery of the primary tumor Yes103 (40.2%) No153 (59.8%)Presence of hepatic metastases at L3 Yes153 (60%) No103 (40%)Number of metastatic sites at L3 199 (39%) >2157 (61%)ECOG at L3 0 or 1165 (82%) >236 (18%) Missing55Locoregional therapy received Radiofrequency ablation8 (3.1%) Radiotherapy83 (32.4%) Radioembolization16 (6.3%)Outcomes of the patients Death199 (85%) Lost to follow-up22 (9%) On-going treatment at the time of data cutoff28 (10.9%) Synchronous liver metastases/multifocal disease120 (46.9%) Synchronous bilobar liver disease97 (37.9%) Synchronous extrahepatic metastases111 (43.4%) Lymph node metastases65 (25.4%) Peritoneal metastases40 (15.6%) Lung metastases30 (11.7%) Bone metastases10 (3.9%) Other metastases10 (3.9%)dCCA, distal subtypes; ECOG PS, performans status; iCCA, categorized into intrahepatic; GBC, gallbladder carcinoma; pCCA, perihilar categorized into intrahepatic. Patients characteristics dCCA, distal subtypes; ECOG PS, performans status; iCCA, categorized into intrahepatic; GBC, gallbladder carcinoma; pCCA, perihilar categorized into intrahepatic. Among the 256 patients who received at least a third line of therapy, 206 (80%) had successful molecular profiling carried out. Analysis of genomic alterations in this specific cohort revealed that KRAS mutations were the most frequent, occurring in 32% of patients (n = 66). The second most common alteration was IDH1 mutation, present in 24% of patients (n = 49), followed by FGFR2 fusions 20% (n = 42) and FGFR2 mutations at 14% (n = 28). HER2 amplifications were observed in 9% (n = 18). A comprehensive overview of the molecular alterations can be found in Table 2. An actionable alteration was identified in 112 patients (44%). Of the 112 patients with an actionable alteration, 80 (71%) received a targeted treatment during the disease course [33 (29%) in second line, 65 (58%) in third line].Table 2Molecular alterations in patients receiving third-line systemic treatment with successful molecular profilingGenomic alterationNumber (N = 206)KRAS66 (32%)NRAS10 (4.9%)BRAF13 (6.3%)FGFR1 fusion1 (0.5%)FGFR2 fusion42 (20.4%)FGFR3 fusion6 (2.9%)FGFR4 fusion1 (0.5%)FGFR1 mutation4 (1.9%)FGFR2 mutation28 (13.6%)FGFR3 mutation6 (2.9%)FGFR4 mutation2 (0.9%)HER2 amplification18 (8.7%)HER3 amplification3 (1.5%)HER2 mutation8 (3.9%)HER3 mutation12 (5.8%)IDH149 (23.8%)IDH212 (5.8%)BRCA115 (7.3%)BRCA223 (11.2%) Table 2Molecular alterations in patients receiving third-line systemic treatment with successful molecular profilingGenomic alterationNumber (N = 206)KRAS66 (32%)NRAS10 (4.9%)BRAF13 (6.3%)FGFR1 fusion1 (0.5%)FGFR2 fusion42 (20.4%)FGFR3 fusion6 (2.9%)FGFR4 fusion1 (0.5%)FGFR1 mutation4 (1.9%)FGFR2 mutation28 (13.6%)FGFR3 mutation6 (2.9%)FGFR4 mutation2 (0.9%)HER2 amplification18 (8.7%)HER3 amplification3 (1.5%)HER2 mutation8 (3.9%)HER3 mutation12 (5.8%)IDH149 (23.8%)IDH212 (5.8%)BRCA115 (7.3%)BRCA223 (11.2%) Molecular alterations in patients receiving third-line systemic treatment with successful molecular profiling Systemic therapy regimens The median follow-up time was 42 months. In the first-line palliative systemic therapy, the majority of patients received a combination of gemcitabine and a platinum salt, with 49% of patients treated with gemcitabine plus cisplatin (n = 126), and 35% with gemcitabine plus oxaliplatin (n = 89). FOLFIRINOX was the third most frequently administered chemotherapy regimen, used in 7% of patients (n = 17), followed by gemcitabine monotherapy in 4% of cases (n = 9) and combination of fluoropyrimidine and a platinum salt (2% of patients, n = 5) and 10 other (Table 3).Table 3Systemic therapy in third lineNumber of patients 256 (100%)FOLFIRI51 (19.9%)TAXANE43 (16.8%)LV5FU2 Cisplatin/ FOLFOX32 (12.5%)5FU/ XELODA10 (3.9%)GemOx8 (3.1%)GemCis4 (1.6%)Gemcitabine2 (0.8%)Targeted therapya66 (25.8%)Other systemic treatment combinationsb40 (15.6%)aSee the Supplementary Table S3, available at https://doi.org/10.1016/j.esmogo.2026.100345.bDifferent combinations of chemotherapy/immunotherapy/targeted treatments and agents explored in clinical trials (Supplementary Table S4, available at https://doi.org/10.1016/j.esmogo.2026.100345). Table 3Systemic therapy in third lineNumber of patients 256 (100%)FOLFIRI51 (19.9%)TAXANE43 (16.8%)LV5FU2 Cisplatin/ FOLFOX32 (12.5%)5FU/ XELODA10 (3.9%)GemOx8 (3.1%)GemCis4 (1.6%)Gemcitabine2 (0.8%)Targeted therapya66 (25.8%)Other systemic treatment combinationsb40 (15.6%)aSee the Supplementary Table S3, available at https://doi.org/10.1016/j.esmogo.2026.100345.bDifferent combinations of chemotherapy/immunotherapy/targeted treatments and agents explored in clinical trials (Supplementary Table S4, available at https://doi.org/10.1016/j.esmogo.2026.100345). Systemic therapy in third line See the Supplementary Table S3, available at https://doi.org/10.1016/j.esmogo.2026.100345. Different combinations of chemotherapy/immunotherapy/targeted treatments and agents explored in clinical trials (Supplementary Table S4, available at https://doi.org/10.1016/j.esmogo.2026.100345). In second line, the most frequently administered regimens were FOLFIRI (29% of patients, n = 73), FOLFOX (20% of patients, n = 51), GEMOX (n =25, 10%), gemcitabine alone (n = 12, 5% of patients) and fluoropyrimidine monotherapy in 4% (n = 10; Supplementary Table S2, available at https://doi.org/10.1016/j.esmogo.2026.100345). Targeted treatments were administered as single agents to 13% of the patients (n = 33), predominantly within clinical trials. The remaining 21% of patients received various other chemotherapy regimens or immunotherapy. The third line of systemic treatment was diverse, with targeted therapies the most frequently employed modality (25% of cases, n = 65), the agents predominantly utilized being IDH1 and FGFR inhibitors (Supplementary Table S3, available at https://doi.org/10.1016/j.esmogo.2026.100345). Regarding chemotherapy regimens specifically used in the third line, FOLFIRI was administered to 20% of patients (n = 51), taxane-based regimens in 17% of cases (n = 43; Table 4). Combinations of fluoropyrimidine with a platinum salt were used in 13% of patients (n = 32; Table 4). Gemcitabine-based treatments were also utilized, albeit less frequently (n = 12; Table 4). Other systemic treatments were used in 10 patients (n = 4%; Supplementary Table S4, available at https://doi.org/10.1016/j.esmogo.2026.100345).Table 4Univariate analysisLog-rank testP valueHR95% CITumor localization (ref: iCCA)0.21.2150.9021.636Metastatic at diagnosis (ref: no metastases)0.031.4641.0292.082Discovery mode (icterus/angiocholitis or cholestasis) (ref: no)0.91.0270.7231.461Year of third-line treatment (ref ≤2016)0.141.2920.9181.819Age at diagnosis (ref <65 years)0.061.3370.9921.804Age at beginning of L3 (ref <65 years)0.071.3080.9781.749Number of metastatic sites at beginning of L3 (ref: ≤1)0.11.2800.9561.712Hepatic metastases at beginning of L3 (ref: no metastases)0.0071.5011.1172.016Treatment received in L30.07 FOLFIRI (ref)1 5FU CisPt/FOLFOX0.7020.4101.201 Taxane0.5280.3330.836 Targeted therapy alone0.6340.4200.956 Other0.6430.4230.977Bold indicates P-value < 0.05.CI, confidence interval; HR, hazard ratio; iCCA, categorized into intrahepatic. Table 4Univariate analysisLog-rank testP valueHR95% CITumor localization (ref: iCCA)0.21.2150.9021.636Metastatic at diagnosis (ref: no metastases)0.031.4641.0292.082Discovery mode (icterus/angiocholitis or cholestasis) (ref: no)0.91.0270.7231.461Year of third-line treatment (ref ≤2016)0.141.2920.9181.819Age at diagnosis (ref <65 years)0.061.3370.9921.804Age at beginning of L3 (ref <65 years)0.071.3080.9781.749Number of metastatic sites at beginning of L3 (ref: ≤1)0.11.2800.9561.712Hepatic metastases at beginning of L3 (ref: no metastases)0.0071.5011.1172.016Treatment received in L30.07 FOLFIRI (ref)1 5FU CisPt/FOLFOX0.7020.4101.201 Taxane0.5280.3330.836 Targeted therapy alone0.6340.4200.956 Other0.6430.4230.977Bold indicates P-value < 0.05.CI, confidence interval; HR, hazard ratio; iCCA, categorized into intrahepatic. Univariate analysis Bold indicates P-value < 0.05. CI, confidence interval; HR, hazard ratio; iCCA, categorized into intrahepatic. Outcomes and treatment sequences The median survival from the start of the first cycle of systemic treatment in the third line was 10.2 months (95% CI 8.3-11.6; Figure 2A).Figure 1Flow chart.Figure 2Study flow chart. (A) Survival for the whole cohort of patients that received a third line of treatment. The survival could not be estimated for 21 patients due to missing data on the date of first administration. (B) Median survival under different third line treatment regimens. (C) Median progression-free survival under different third line treatment regimens. Flow chart. Study flow chart. (A) Survival for the whole cohort of patients that received a third line of treatment. The survival could not be estimated for 21 patients due to missing data on the date of first administration. (B) Median survival under different third line treatment regimens. (C) Median progression-free survival under different third line treatment regimens. Patients receiving taxane monotherapy had a higher median OS (mOS) of 13 months (95% CI 7.3-19.8) compared with those treated with FOLFIRI, who had a mOS of 7 months (95% CI 4.3-8.9, P = 0.03) and hazard ratio (HR) was 0.58 (95% CI 0.4-0.9), after adjustment for multiple comparisons using Dunnett’s correction. Patients receiving third-line molecularly targeted therapy had a median survival of 11 months (95% CI 7.4-15.6; Figure 2B). The category labeled ‘Other Treatments’ encompasses a diverse range of therapeutic agents, detailed in Supplementary Table S5 (available at https://doi.org/10.1016/j.esmogo.2026.100345), with a median OS of 11.02 months [6.7-14.4]. The median progression-free survival (mPFS) for the third-line treatments was as follows: taxane 3.3 months (95% CI 2.4-5.5), FOLFIRI 2.6 months (95% CI 1.89-3.4), 5-FU cisplatin/FOLFOX: 3.9 months (95% CI 2-5.7) and targeted treatment: 5.3 months (95% CI 3.5-7.2; Figure 2C). Overall response rate (ORR) was 45% with taxane, 35% with FOLFIRI, 42% with 5-FU cisplatin/FOLFOX and 71% with targeted treatments. Analysis of the sequence of treatment revealed distinct patterns (Supplementary Figure S1 and Table S5, available at https://doi.org/10.1016/j.esmogo.2026.100345). Patients who received taxane monotherapy in third line typically had a treatment sequence of gemcitabine cisplatin or gemcitabine oxaliplatin in the first line, followed by FOLFIRI in the second line. In this group, FOLFOX or other fluoropyrimidine platinum salts combinations were less frequently used in the second line. Conversely, patients who received FOLFIRI in the third line generally had gemcitabine oxaliplatin, or gemcitabine cisplatin as their first-line treatments, succeeded by mostly FOLFOX in the second line. Patients treated with LV5FU2 cisplatin or FOLFOX as third-line therapy usually had gemcitabine cisplatin in the first line and gemcitabine oxaliplatin or FOLFIRI in the second line. Treatments beyond third-line A subset of patients continued to receive treatments beyond three lines (n = 136, 53%). In the fourth line, fluoropyrimidine-based chemotherapy was predominantly administered, notably FOLFIRI (18%), or taxanes (13%). Molecular targeted therapies constituted the majority of treatments in this line (25%). In the fifth line of therapy, taxanes were most frequently employed (18%), followed by fluoropyrimidine-based combinations such as FOLFIRI (12%), and then LV5FU2 with platinum derivatives (9%). Detailed information regarding the specific treatments chosen for other lines can be found in Supplementary Table S6, available at https://doi.org/10.1016/j.esmogo.2026.100345. Prognostic factors In the univariate analysis for OS, metastatic setting at diagnosis (P = 0.03) and the presence of hepatic metastases at the start of L3 (P = 0.007) were the only significant factors (Table 4). In the multivariable analysis, only hepatic metastases remained significantly associated with OS, with a HR of 1.5, 95% CI 1.1-2.1, indicating better survival in patients without hepatic metastases (P value = 0.02). Type of treatment, age at diagnosis, anatomical localization, metastatic setting at diagnosis, mode of diagnosis (jaundice, angiocholitis, or cholestasis versus other), age at the start of treatment, number of metastatic sites at the beginning of L3 and year when third line was administered were not significantly associated with survival. Discussion This single-center retrospective analysis of third-line systemic therapy in patients with BTC presents an evaluation of a highly selected cohort from a referral hospital in France. The study highlights several noteworthy aspects regarding patient characteristics, third-line treatment selection, and survival outcomes in this rare pathology. Our findings provide valuable insights into the management of advanced BTC in a tertiary care setting. Considering the methodological limitations of this study, the results lead us to believe that targeted therapies are the best third-line treatment for patients with actionable alterations (OS 11 months), provided they have not previously received such treatment in second line. For patients without actionable alterations, our study suggests that it is better to prescribe taxanes as monotherapy (OS 13 months) rather than FOLFIRI or FOLFOX/5FU cisplatin, if not received in second line. The mOS for the patients receiving a third line in this study is 10.2 months (95% CI 8.3-11.6), higher than the one reported in other analyses, probably because the study was done in a referral center21,22 with access to therapeutic trials and targeted therapies. In this study, PFS showed a low correlation with OS, as recently shown for first and second line in BTC, as well as a low correlation with ORR and DCR.24 PFS is currently the best candidate surrogate marker for OS, but our results also indicate an inadequacy of PFS/ORR as surrogate endpoints of OS. In terms of molecular characteristics, the vast majority of patients had a molecular profile carried out at some stage of their disease. Compared with the literature, the frequency of genetic alterations in this patient cohort is higher.2 This is likely because these patients specifically sought molecular profiling at this hospital, resulting in a more selected group. Nonetheless, the overall proportion of molecular alterations remains relatively low, only a small number of patients had genetic anomalies that could be targeted by specific therapies,25 emphasizing the need to evaluate cytotoxic treatments in the third line, as there is no established standard in this setting. The inclusion period for patients in this study spanned several therapeutic advancements, including the introduction of immunotherapy in the first-line setting alongside chemotherapy, and the significant impact of molecular profiling and targeted therapies. GemCis was the predominant first-line regimen, administered to 50% of patients, reflecting its status as the standard of care since the ABC-02 trial in 2010.26 Recent approvals of immunotherapy agents (durvalumab and pembrolizumab) in combination with GemCis occurred in 2022-20239,10; consequently, only a small subset of our cohort received these newer combination therapies. In second-line treatment, most patients in this cohort received FOLFIRI, followed by FOLFOX or 5-FU-based platinum chemotherapy, this might explain a lower efficacy of FOLFIRI in third line in our cohort. This practice shows a slight deviation from the currently recommended standard established by the ABC-06 trial, which demonstrated FOLFOX’s efficacy as a possible second-line therapy since 202111 probably because some patients were treated before FOLFOX became a standard or because patients were unfit for FOLFOX.12 Of note, ABC-06 trial included mostly very fit and young patients which does not represent the majority of patients with advanced BTC, and the benefit in OS was limited, making other treatment options possible. FOLFIRI’s manageable toxicity profile and potential benefits in mOS and mPFS may explain its continued use, especially in patients with sequelae neuropathy after first-line cisplatin. Specifically, the NIFTY trial, a Phase IIb study conducted in Korea, assessed the efficacy and safety of liposomal irinotecan (Nal-IRI) in combination with 5-FU and leucovorin versus 5-FU and leucovorin alone in the second line. This study reported a mOS of 8.6 months for the Nal-IRI combination, compared with 5.5 months for the control group, and a mPFS of 3.9 months versus 1.4 months in the control group.27 In 2023, the NALIRICC study, another phase II trial, evaluated the efficacy of Nal-IRI combined with 5-FU and leucovorin.28 Although this study did not find a significant OS benefit, it observed PFS improvements similar to those reported in the NIFTY trial. ESMO guidelines recommend FOLFIRI only a treatment option in case oxaliplatin is contraindicated.7,8 In the third-line treatment, molecularly targeted monotherapy was administered more frequently than in the second-line setting—27% versus 13%, respectively, mostly due to inclusion in therapeutic trials and the fact that targeted therapies were not always available in second line at time of the study. The most frequently administered chemotherapy regimen was FOLFIRI, followed by taxane monotherapy. Additionally, combinations of 5-FU cisplatin or FOLFOX were also common. Patients receiving taxane monotherapy had a higher mOS compared with those treated with FOLFIRI or FOLFOX/5-FU cisplatin. Specifically, the taxane group demonstrated significantly prolonged OS relative to the FOLFIRI group (P = 0.03), with a 42% lower risk of death (HR= 0.58, 95% CI 0.36-0.92). We identified specific patterns in third-line treatment outcomes. In the analysis of treatment sequences, GemCis was the most commonly used regimen in the first line. Taxane was administered as the third-line treatment following FOLFIRI, whereas FOLFIRI was mostly employed when FOLFOX was used as the second-line therapy. The observed survival difference suggests that taxane-based chemotherapy may be more effective in the third line, especially for patients who have previously received FOLFIRI as second-line treatment because of a contraindication to FOLFOX. In this retrospective study, we aimed to identify prognostic factors for patients with cholangiocarcinoma who received third-line treatment.22,23 Several factors described in Table 4 were assessed as potential prognostic elements. The analysis was adjusted for the year of diagnosis, with 2016 serving as a cutoff. This year was chosen as a pivotal point in the management of advanced-stage cholangiocarcinoma due to significant advancements in molecular profiling and the identification of specific molecular targets with many patients enrolled in various clinical trials. A better OS was observed in patients without hepatic metastases. However, because of missing data in some prognostic factors, the multivariable analysis included only 80% of the study population. This may have led to biased results if missingness is associated with OS. Other limitations of this study include its retrospective and monocentric design, as well as the highly selected patient population. A high number of patients received four or five lines of treatment, which is selection bias, explained by the fact that patients come from a tertiary center with immortality bias (some patients were addressed for therapeutic trials after several lines of treatment). Moreover, due to the substantial amount of missing data regarding the safety profiles of the chemotherapy regimens administered in the third line, no definitive conclusions or assumptions could be made about their tolerability or adverse effects in this setting. Further studies with comprehensive safety data are required to assess the toxicity profiles of these treatments, as well as quality of life and patients’ preferences. Conclusion Although this study provides valuable insights regarding third-line chemotherapy regimen, it also emphasizes the need for further research to refine our understanding of optimal treatment strategies in advanced BTC care. It is highly recommended that molecular profiling should be conducted at the initial diagnosis of advanced, inoperable, or metastatic disease. Given the potential survival benefits observed, we could highlight the potential efficacy of taxane monotherapy as the third-line chemotherapy regimen, considering ECOG performance status, comorbidities and previous lines of treatment. This hypothesis should be further validated through randomized controlled trials.