Cholangiocarcinoma (CCA) is a malignant tumor arising from the bile ducts and is anatomically divided into intrahepatic CCA (iCCA) and extrahepatic CCA (eCCA), which is further subclassified into peri-hilar and distal CCA.1,2 CCA is a rare cancer, accounting for 15% of all primary liver cancers and 3% of gastrointestinal malignancies, but its incidence and mortality are increasing worldwide.3,4 The most common risk factors for CCA development include conditions associated with biliary tract inflammation, such as primary sclerosing cholangitis, biliary cystic diseases, and calculi in the biliary tree, metabolic and endocrine disorders, viral hepatitis B and C, hepatobiliary fluke, as well as lifestyle factors including alcohol consumption, cigarette smoking, and exposure to environmental pollutants.3 Specific genetic polymorphisms are also associated with an increased risk of developing CCA; however, CCA is generally sporadic and arises without well-defined risk factors. It usually develops insidiously, and early disease stages are not associated with specific symptoms. Surgery followed by adjuvant capecitabine is the only curative treatment but unfortunately, most patients experience local or distance disease recurrence after surgery.5 Moreover, CCA is frequently diagnosed at a locally advanced or metastatic stage, at which point systemic therapy represents the backbone of treatment.6,7 Historically, the ABC-02 trial defined the combination of cisplatin and gemcitabine as first-line standard of care, with a median overall survival (OS) of 11.7 months vs. 8.1 months with single-agent gemcitabine (hazard ratio [HR], 0.64; 95% confidence interval [CI], 0.52-0.80; P<0.001).8 The efficacy and tolerability of the combination was also confirmed by a Japanese study and a subsequent metanalysis.9,10 Moreover, gemcitabine plus S1 demonstrated non-inferiority to cisplatin plus gemcitabine.11

Intensification of chemotherapy is currently under evaluation and improved survival has been demonstrated with cisplatin and gemcitabine plus S1 vs. cisplatin and gemcitabine (median OS, 13.5 vs. 12.6 months; HR, 0.79; 95% CI, 0.628-0.996; P=0.045).12 Conversely, modified FOLFIRINOX was not found to be superior to standard of care chemotherapy.13 The phase III SWOG 1815 study evaluating the addition of nab-paclitaxel to cisplatin plus gemcitabine did not demonstrate a statistically significant improvement in OS (14.0 vs. 12.7 months; HR, 0.93; 95% CI, 0.74-1.19; P=0.47).14

Recently, the ABC-07 study evaluating stereotactic body radiotherapy and cisplatin plus gemcitabine in locally advanced inoperable CCA, did not show an advantage in progression-free survival (PFS) over standard chemotherapy, even though demonstrating a longer median OS and a better primary tumor control.15

In the second-line setting, the ABC-06 trial set FOLFOX as the recommended regimen, even though demonstrating only a modest OS benefit (median OS, 6.2 vs. 5.3 months; HR, 0.69; 95% CI, 0.50-0.97; P=0.031).16 FOLFIRI showed comparable efficacy and tolerability to FOLFOX in a randomized phase II study.17 In a phase IIb Korean study, 5-fluorouracil (5-FU) plus nano-liposomal irinotecan showed improved median PFS vs. 5-FU (7.1 vs. 1.4 months; HR, 0.56; 95% CI, 0.39-0.81; P=0.0019).18 However, in Western patients the combination of 5-FU and irinotecan did not show a survival benefit over 5-FU alone and was associated with increased rate of adverse events.19

Considering the urgent need of new treatment options, the direction of research has been twofold: on the one hand, studies on the tumor microenvironment (TME) have provided the basis for the use of immunotherapy; on the other hand, advances in genomic profiling have led to the identification of druggable alterations and paved the way for targeted agents. In this review, we aim to provide an overview of the latest therapies and future directions for the treatment of patients with advanced CCA.

Immunotherapy has recently changed the treatment landscape of advanced CCA, as immune checkpoint inhibitors (ICIs) have yielded the first improvement in OS in the first-line setting after over a decade of chemotherapy-based systemic treatment.8

The preclinical rationale supporting the efficacy of immunotherapy in CCA lies in the characteristics of the TME, which consists of a dense desmoplastic stroma containing both nonimmune and immune cell types, such as cancer-associated fibroblasts and tumor-associated macrophages. Four TME-based molecular subtypes have been identified using 14 gene signatures, corresponding to the primary cell populations and the elements involved in tumor-stroma interactions (Fig. 1). The genomic signatures suggest that the inflamed I2 immunogenic subtype responds to single-agent ICIs, as it overexpresses immune checkpoints, while the other three subtypes require treatment combinations that can induce sensitivity to immunotherapy through their activity on the TME.20

Pembrolizumab, an anti-programmed death-1 (PD-1) antibody, was the first ICI evaluated in advanced CCA. In the phase Ib KEYNOTE-028 trial including 24 patients with biliary tract cancer (BTC) (20 CCA), the objective response rate (ORR) was 13%, with median PFS and OS of 1.8 (95% CI, 1.4-3.1) and 5.7 months (95% CI, 3.1-9.8), respectively.21 Moreover, pembrolizumab was the first ICI approved for patients with pretreated microsatellite instability-high (MSI-H) advanced CCA, based on the results of the KEYNOTE-158 phase II basket trial. The study enrolled 351 patients with MSI-H tumors, including 22 with CCA/BTC. After a median follow-up of 37.5 months, single-agent pembrolizumab yielded an ORR of 40.9% and a median duration of response (DOR) of 30.6 months (range, 6.2-over 40.5). Median PFS was 4.2 months (95% CI, 2.1-24.9) and median OS was 19.4 months (95% CI, 6.5-not reached [NR]).22,23

Similarly, nivolumab was evaluated in previously treated patients with advanced CCA, reporting an ORR of 22%, a median PFS of 3.7 months (95% CI, 2.3-5.7) and median OS of 14.2 months (95% CI, 6.0-NR).24

Despite these positive results, single-agent ICIs and combinatorial ICIs strategies have shown only modest efficacy in unselected patients. Therefore, research has focused on combinations of systemic chemotherapies and ICIs.25,26

Preliminary encouraging data on the efficacy of chemoimmunotherapy in advanced CCA came from an open-label, single-center phase II trial. The study initially evaluated the addition of durvalumab, an anti-programmed death ligand 1 (PD-L1) antibody, and tremelimumab, an anti-cytotoxic T-lymphocyte antigen 4 (CTLA-4) antibody, starting from the second cycle of first-line chemotherapy. Following protocol amendment, patients were enrolled to receive cisplatin, gemcitabine and durvalumab with or without tremelimumab from the first cycle. Among 128 enrolled patients, ORR was 72% in the chemotherapy plus durvalumab group, and 70% in the chemotherapy plus durvalumab and tremelimumab group. Median PFS was 11.8 months (95% CI, 6.9-16.6) with chemotherapy plus durvalumab, and 12.3 months (95% CI, 9.3-15.2) with the addition of tremelimumab; median OS was 20.2 months (95% CI, 12.8-27.6) and 18.7 months (95% CI, 14.1-23.2), respectively. High tumor mutational burden (TMB) and baseline PD-L1 expression did not appear to influence outcomes.27

These promising results, coupled with a manageable safety profile, represented the basis for the TOPAZ-1 trial, which was the first phase III study that demonstrated a benefit in survival by adding durvalumab to standard of care chemotherapy with cisplatin and gemcitabine.28 Intravenously durvalumab/placebo was combined with cisplatin and gemcitabine and administered on a 21-day cycle for up to eight cycles, on day 1 of each cycle. Subsequently, 1,500 mg of durvalumab or placebo monotherapy was administered once every 4 weeks until progression or unacceptable toxicity. Data from a preplanned interim analysis demonstrated a benefit in PFS, with a median PFS of 7.2 months (95% CI, 6.7-7.4) with durvalumab vs. 5.7 months (95% CI, 5.6-6.7) with placebo (HR, 0.75; 95% CI, 0.64-0.89; P=0.001).28 At the updated analysis conducted after a median follow-up of 23.4 months in the durvalumab arm and 22.4 months in the placebo arm, median OS was 12.9 months (95% CI, 11.6-14.1) with the experimental combination vs. 11.3 months (95% CI, 10.1-12.5) with placebo (HR, 0.74; 95% CI, 0.63-0.87), with a 12-month OS rate of 54.3% (95% CI, 48.8-59.4) and 47.2% (95% CI, 41.7-52.3), a 24-month OS rate of 23.6% (95% CI, 18.7-28.9) and 11.5% (95% CI, 7.6-16.2), and a 36-month OS rate of 14.6% (95% CI, 11.0-18.6) and 6.9% (95% CI, 4.5-10.0), respectively.29 26% of participants in the experimental arm and 19% in the control arm were long-term survivors, defined as patients who survived at least 18 months after randomization. Long-term survivors who received durvalumab had more frequently recurrent disease at start of first-line systemic therapy; additionally, they more often had a carbohydrate antigen 19-9 level of less than 500 U/mL, a carcinoembryonic antigen level of less than 5 ng/mL, and a neutrophil-to-lymphocyte ratio (NLR) of less than 3. Besides, long-term survivors had a higher ORR than patients in the intention-to-treat population: ORR was 44% in the experimental arm and 34% in the control arm for long-term survivors, whereas it was 27% and 19% in the intention-to-treat population, respectively.29 Unfortunately, there are no predictive biomarkers. PD-L1 expression did not influence survival outcomes: the HR for OS with durvalumab was 0.79 (95% CI, 0.61-1.00) among patients with a PD-L1 tumor area positivity (TAP) score of 1% or greater (≥1% tumor area occupied by tumor and/or immune cells with PD-L1 staining at any intensity), while it was 0.86 (95% CI, 0.60-1.23) in patients with a TAP score less than 1%. As far as microsatellite status is concerned, 46.9% of patients in the durvalumab arm and 48.8% in the control arm had microsatellite stable (MSS) tumors, and approximately 50% had an unknown status. Therefore, MSI-H could not be assessed as a predictive factor.

The most common adverse events (AEs) were anemia (48.2%), nausea (40.2%), constipation (32.2%), and neutropenia (31.7%). Grade 3-4 toxicity rates were comparable between the two treatment groups (61% and 63%), suggesting that toxicity was largely due to chemotherapy.29 Moreover, there was no difference in time to deterioration or adjusted mean changes from baseline for global health status or quality of life (QoL), functioning, and symptoms between the two study arms.30

Two observational Italian studies confirmed the efficacy and safety of the combination of cisplatin, gemcitabine and durvalumab in a real-life setting. A prospective analysis conducted on 149 patients treated with chemoimmunotherapy showed a median OS of 12.9 months (95% CI, 10.9-12.9) and a median PFS of 8.9 months (95% CI, 7.4-11.7), with an ORR of 34.5%.31 Another large retrospective study evaluating the benefit of adding durvalumab to standard chemotherapy showed a statistically significant increase in both OS (HR, 0.63; 95% CI, 0.50-0.80; P=0.0002) and PFS (HR, 0.57; 95% CI, 0.47-0.70; P<0.0001).32 Additionally, a German real-life retrospective observational study showed that patients beyond the inclusion criteria of the TOPAZ-1 trial, either due to comorbidities, poor performance status or having received previous chemo- or radiotherapy, had a similar benefit in terms of OS and PFS with chemoimmunotherapy, with a comparable toxicity profile.33

The KEYNOTE-966 phase III study provided further confirmation of the efficacy of chemoimmunotherapy for the treatment of patients with advanced BTC.34 Two hundred mg of intravenous pembrolizumab or placebo was administered once every 3 weeks, whereas intravenous cisplatin and gemcitabine were administered on days 1 and 8 of 21-day cycles. Pembrolizumab/placebo was limited to 35 cycles, whereas cisplatin was limited to 8 cycles. Gemcitabine had no limit in the number of cycles. Pembrolizumab plus cisplatin and gemcitabine yielded a median OS of 12.7 months (95% CI, 11.5-13.6) vs. 10.9 months (95% CI, 9.9-11.6) with chemotherapy (HR, 0.83; 95% CI, 0.72-0.95; P=0.0034), with a 12- and 24-month OS rate of 52% (95% CI, 47-56) and 25% (95% CI, 21-29), respectively. Subgroup analysis showed an OS benefit for both patients with PD-L1 combined positive score less than 1 (HR, 0.84; 95% CI, 0.62-1.14) and equal or greater than 1 (HR, 0.85; 95% CI, 0.72-1.00). As in the TOPAZ-1 trial, most patients had MSS tumors, therefore outcomes by microsatellite status could not be assessed. However, although a benefit in median PFS was observed, it did not reach statistical significance: median PFS was 6.5 months (95% CI, 5.7-6.9) in the experimental arm and 5.6 months (95% CI, 5.1- 6.6) in the control arm (HR, 0.86; 95% CI, 0.75-1.00; P=0.023). The ORR was 29% in both cohorts, even though the median DOR was 9.7 months (95% CI, 6.9-12.2) in the pembrolizumab arm vs. 6.9 months (95% CI, 5.7-8.2) in the placebo arm. After a median follow-up of 36.6 months, the OS (HR, 0.86; 95% CI, 0.75-0.98) and PFS (HR, 0.85; 95% CI, 0.75-0.97) benefit was maintained.35 Treatment was well tolerated, and grade 3-4 toxicity rates were similar (70% and 69%); the most common grade 3-4 treatment-related AEs (TRAEs) in the pembrolizumab arm were decreased neutrophil count (61%) and anemia (53%). There was no difference in terms of QoL between the study arms.34

Tables 1 and 2 summarize the results of these two clinical trials compared to the historical ABC-02 study. Based on this data, the United States (US) Food and Drug Administration (FDA), the European Medicines Agency (EMA), and other regulatory agencies have approved both combinations for the first-line treatment of patients with advanced BTC.

Currently, there are no criteria that guide the choice between these two regimens. Therefore, treatment selection depends exclusively on drug availability and approvals by national regulatory agencies.

ICIs have been and are currently being studied in combination with systemic therapies other than chemotherapy, such as anti-vascular endothelial growth factor (VEGF) agents, tyrosine kinase inhibitors (TKIs), and targeted agents, with the aim of increasing treatment responses and widening the population that can benefit from immunotherapy.

Anti-VEGF agents such as bevacizumab can increase responses to PD-L1 inhibitors by inducing immune-permissive TME changes, such as the maturation of dendritic cells and the reduction in the activity of immunosuppressive cells, like regulatory T lymphocytes and myeloid-derived suppressor cells.36 The IMbrave151 trial was a multicenter, randomized, double-blind, placebo-controlled, non-comparative study that evaluated the addition of bevacizumab to first-line chemotherapy with cisplatin and gemcitabine plus the anti-PD-L1 atezolizumab. Median PFS was 8.35 months in the bevacizumab arm vs. 7.90 months in the placebo arm (HR, 0.67; 95% CI, 0.46-0.95), with a 6-month PFS rate of 78% vs. 63%. Median OS was 14.9 vs. 14.6 months (HR, 0.97; 95% CI, 0.64-1.47). ORR was 26.6% with the quadruplet regimen and 26.5% with the triplet regimen, with a median DOR of 10.28 months (95% CI, 6.7-16.7) and 6.18 months (95% CI, 4.3-6.7), respectively. At the exploratory biomarker analysis, patients with high expression of VEGF A and hepatocytes high gene signature had better PFS and OS with bevacizumab rather than with placebo. Moreover, patients with mutations of genes related to the phosphoinositide 3-kinase/protein kinase B pathway had worse outcomes with atezolizumab plus bevacizumab compared to those with a wild-type profile.37,38

Lenvatinib is an oral multitargeted TKI, which has shown synergistic effect with anti-PD-1 agents in other cancer types including hepatocellular carcinoma, mainly due to its anti-VEGF receptor activity.39 LEAP-005 was an open-label, multi-cohort, phase II trial, evaluating lenvatinib plus pembrolizumab in patients with previously treated solid tumors, including 31 patients with BTC. ORR was 10%, and disease control rate (DCR) was 68%, with a median DOR of 5.3 months (range, 2.1-6.2). Median PFS was 6.1 months (95% CI, 2.1-6.4) and median OS was 8.6 months (95% CI, 5.6-NR). 48% of patients had grade 3-4 TRAEs, the most common were hypertension (42%), dysphonia (39%), and diarrhea (32%).40 Similar results have been observed in the first-line setting with the combination of lenvatinib and other anti-PD-1 agents including tislelizumab, camrelizumab, and sintilimab.41

All these results are hypothesis-generating and contribute to shed light on the complex interactions between TME and cancer cells and on how responses to ICIs can be increased through TME immunomodulation, but would need further confirmation in larger, global, phase III studies.

The results achieved with immunotherapy represent an essential step forward, considering the paucity of effective therapeutic options for patients with advanced CCA. Indeed, even though the median OS improvement yielded by the recently approved chemoimmunotherapy regimens might seem modest, it is worth noting that the 12-month, 24-month, and 36-month OS rates are unprecedented, suggesting that there is a subset of patients that derives a prolonged benefit from ICIs.

However, the lack of predictive biomarkers that could help the identification of responders to immunotherapy still poses a crucial challenge. While the aforementioned TME-based classification of CCAs holds therapeutic implications, it cannot be easily implemented in routine clinical practice. Typically, TMB, PD-L1 expression and MSI-H status are used to guide access to immunotherapy. However, in CCA only the latter has demonstrated predictive value in patients treated with single-agent ICIs, while TMB and PD-L1 expression are not predictive of response to treatment with ICIs. As approximately only 1.2% of CCAs are MSI-H, for the vast majority of patients the benefit of immunotherapy cannot be predicted.42

For this reason, current research aims to identify new biomarkers that can help tailor treatment and more accurately select patients. Moreover, research has also been focusing on the development of new immunotherapeutic strategies, including bispecific antibodies, bifunctional fusion proteins, tumor vaccines, adoptive cell therapy, and cytokine therapy (Fig. 2).

Bispecific antibodies are agents with two binding sites directed at two different antigens or epitopes. GEMINI-hepatobiliary (NCT05775159) is an ongoing phase II, open-label, uncontrolled trial, evaluating the addition of either rilvegostomig (an anti-PD-1 and anti-T cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibitory motif domains bispecific antibody) or volrustomig (an anti-PD-1 and anti-CTLA-4 antibody) to cisplatin and gemcitabine as first-line therapy.

Bintrafusp alfa (M7824) is a first-in-class bifunctional fusion protein composed of an anti-PD-L1 antibody and the extracellular domain of transforming growth factor-β receptor II. Single-agent treatment showed an ORR of 20%, a median PFS of 2.5 months (95% CI, 1.3-5.6) and a median OS of 12.7 months (95% CI, 6.7-15.7) in pretreated patients.43 Therefore, a phase II/III study evaluating bintrafusp alfa in combination with cisplatin and gemcitabine in first line (NCT04066491) was opened but later discontinued. Of note, a review of the data conducted by the Independent Data Monitoring Committee stated that it seemed unlikely to meet the primary endpoint (OS).44

Further studies have tested multiple peptide vaccines directed against mucin 1 (MUC-1) and Wilms tumor protein 1, with unsatisfactory results.45,46 Messenger ribonucleic acid vaccines hold more promise, as they have demonstrated efficacy in other tumor types in combination with ICIs.47 Potential targets in CCA have already been identified in cluster of differentiation 247, Fc gamma receptor Ia, and transformation/transcription domain associated protein, but prospective studies are needed to assess their efficacy.48

Adoptive cell therapy aims to increase T lymphocytes reactivity to tumor cells using patient-derived T lymphocytes that are expanded in vitro. Subsequently, cells can be administered back to the patient without undergoing any modifications (tumor-infiltrating lymphocytes [TILs]), or they can be engineered to express receptors that recognize tumor antigens (chimeric antigen receptor T [CAR-T] cells). Evidence regarding the efficacy of TILs for the treatment of CCA is limited to the early-stage setting, where a combined approach with a dendritic cell vaccine has shown effectiveness in 36 CCA patients who underwent surgery.49 In the advanced setting, a phase II trial is currently ongoing (NCT03801083). On the other hand, CAR-T cells have been used in patients with BTC overexpressing the epidermal growth factor receptor: one patient out of 17 achieved a complete response (CR), whereas 10 patients had stable disease (SD), with a median PFS of 4.0 months. An ongoing trial is evaluating MUC-1 targeting CAR-T cells in patients with iCCA (NCT03633773).50

Lastly, there is preliminary evidence on the efficacy of cytokine therapy: even though the application of cytokines has been limited due to their short half-lives and their toxicity profile, they can be used to promote the growth and activity of immune cells, either as single-agents or in combination with ICIs. It seems that granulocyte macrophage colony stimulating factor has synergistic effect with pembrolizumab. However, further studies are warranted.51

Based on a precision medicine approach, the development of next generation sequencing technologies and molecularly driven agents has revolutionized the treatment landscape of many solid cancers, including CCA, and particularly iCCA. Around 40% of CCAs harbor a genetic alteration that is potentially druggable.52 Therefore, a molecular analysis should be carried in all patients diagnosed with CCA as early as possible, especially in advanced disease. The most common clinically relevant molecular alterations include isocitrate dehydrogenase 1 (IDH1) mutations (15-20%), fibroblast growth factor receptor 2 (FGFR2) fusions (10%), human epidermal growth factor receptor 2 (HER2) amplifications and mutations (5-10%), and proto-oncogene B-Raf (BRAF) mutations (3%). Other less frequent alterations involve rearranged during transfection (RET), neurotrophic tyrosine receptor kinase (NTRK), the Kirsten rat sarcoma virus (KRAS) G12C mutation, and mouse double minute 2 homolog (MDM2) amplifications.53 Table 3 reports the efficacy of the most common targeted agents evaluated in patients with pretreated, advanced CCA.

NTRK fusions are associated with the activation of an oncogenic pathway that promotes cell transformation, growth, and proliferation. Even though NTRK fusions are rare in CCA (<1%), two oral NTRK inhibitors (entrectinib and larotrectinib) are currently available. In a phase I/II basket study, entrectinib showed an ORR of 57%, with 7% of CRs.100 Larotrectinib demonstrated an ORR of 79%, with 16% of CRs. Considering these results, entrectinib and larotrectinib have obtained the US FDA, EMA, and other regulatory agencies tumor-agnostic approval.101

Similarly, RET fusions act as oncogenic drivers in various solid tumors. The phase I/II study evaluating pralsetinib in non-thyroid and non-non-small cell lung cancer patients showed good efficacy (ORR, 57%) as well as manageable toxicity.102 Likewise, selpercatinib has shown promising results in the phase I/II LIBRETTO-001 basket trial, yielding an ORR of 43.9%.103 However, both studies enrolled very few patients with BTC (three and two, respectively), and only selpercatinib is a viable option, as it has received the US FDA approval in the tumoragnostic setting.

The KRAS G12C mutation, although rare in patients with CCA, is worth investigating: in the KRISTAL-1 phase II basket trial, among 12 patients with pretreated BTC receiving adagrasib, ORR was 41.7%, median PFS was 8.6 months (95% CI, 2.7-11.3) and median OS was 15.1 months (95% CI, 8.6-NR).104 No regulatory entity has approved adagrasib in this setting yet, however enrollment of this specific group of patients in clinical trials should be encouraged, considering these promising results.

Lastly, MDM2 amplifications are uncommon and have only recently become of interest. MDM2 is a negative regulator of p53, which acts by inhibiting its transcriptional activity and promoting its degradation. In patients with wild-type tumor protein 53 (TP53), blocking the interaction between these two proteins restores the proapoptotic function of p53. This is the mechanism of action of brigimadlin, which has been evaluated both as single-agent (NCT03449381) and in combination with the anti-PD-1 antibody ezabenlimab (NCT03964233) in two different phase Ia/Ib trials. Across the two studies, among 16 patients with BTC, seven (three in the monotherapy study and four in the combination one) achieved a PR, while seven (five treated with single-agent brigimadlin and two with the combination treatment) had a SD. The safety profile was manageable, with nausea being the most common TRAE.105 Based on these results, a phase IIa/IIb trial of single-agent brigimadlin in advanced BTC, pancreatic ductal adenocarcinoma and other solid tumors is ongoing (Brightline-2, NCT05512377).

While targeted therapy has widened the therapeutic landscape for patients with druggable alterations, there are still unanswered questions.

In the MOSCATO-01 trial, a longer OS was achieved in the subgroup of patients receiving a targeted therapy compared to those receiving unselected treatment.106 However, despite the identification of a potentially druggable alteration, there is no certainty of therapeutic success, and this could be related to either co-mutations or intrinsic resistance.53 Studies on genomic profiling are warranted to better understand the interrelations between different molecular alterations and their impact on treatment response, and when more than one targeted drug is available for a specific mutation, treatment selection should ideally take into account the spectrum of activity of each drug.

At the time of progression to targeted therapy, a re-biopsy to identify acquired on-target mutations or novel off-target resistance mechanisms could guide the subsequent treatment choices. However, its implementation in clinical practice is challenging also because of financial issues. While liquid biopsy could represent a potential alternative, prospective studies are needed to confirm whether it has comparable sensitivity to tissue biopsy in detecting all mutations of interest and whether its results can actually change therapeutic choices.

Moreover, the correct collocation of targeted therapy in the treatment algorithm is far to be determined yet. Even though these drugs are available only for pretreated patients, studies evaluating them in first line are ongoing, although some of them had to be discontinued due to slow accrual. One interesting question is whether targeted therapy could be used in first line after a 4-cycle course of chemotherapy with cisplatin and gemcitabine, if either PR or SD are achieved: the SAFIR-ABC10 trial (NCT05615818) will try to test this hypothesis, but recruitment has not begun yet.

New systemic treatment options including immunotherapy and targeted agents have demonstrated clinical efficacy and manageable safety profiles, surpassing chemotherapy-based regimens, and constitute the foundation of a novel treatment algorithm for the management of advanced CCA. Despite the improvement in survival, many challenges remain. On the immunotherapy front, attention is directed towards the identification of predictive biomarkers, as well as new agents or combinations that can overcome resistance either by inducing immune-permissive TME changes or through novel mechanisms of action. Instead, on the targeted therapy front, research is focusing not only on the identification of new druggable alterations, but also on understanding primary and secondary resistance mechanisms, in order to improve patients’ selection and define the most appropriate therapeutic sequence.