Single-agent cytotoxic therapies are now considered historical and are primarily discussed here to contextualize the development of modern combination regimens. Early studies comparing single-agent chemotherapies, such as fluoropyrimidines and gemcitabine, to best supportive care established their role as foundational treatments for combination regimens. For example, the oral fluoropyrimidine agent S1 achieved a 45% objective response rate (ORR) and a mOS of 8.1 months in a phase II study of 20 patients (13). Similarly, 5-fluorouracil (5-FU) with leucovorin yielded a 5% partial response (PR) and a 55% disease-control rate (DCR) in another 20-patient study (14). Gemcitabine monotherapy demonstrated response rates of 7–36% across multiple studies, with one trial reporting a 35.7% PR, 78.6% DCR, and a median time to progression (mTTP) of 6.4 months (95% CI: 5.8-7.1) and mOS of 17.1 months (95% CI: 15.8-18.5) (15–17). Geographic differences in efficacy have been noted, such as lower response rates in Japan (ORR: 8.8%) compared to other regions (15, 18). Other single agents, including irinotecan (ORR: 8.7%; median progression-free survival (mPFS): 2.7 months; mOS: 7.0 months; n=23) (19) and capecitabine (DCR: 19%; mOS: 8.1 months; n=26) (20), have also been evaluated, though with limited success, highlighting the need for more effective therapeutic strategies.

Non-randomized clinical studies demonstrated ORRs ranging from 22.7% to 36.7% (21–23). The ABC-01 trial was the first randomized study to evaluate cisplatin and gemcitabine in advanced/metastatic BTC, leading to the phase III ABC-02 trial, which compared gemcitabine monotherapy to the combination therapy (11). ABC-02 showed a significant survival benefit for the doublet over single agent gemcitabine, with mPFS of 8.0 months versus 5.0 months and mOS of 11.7 months versus 8.1 months. Subset analysis revealed that patients with GBC had improved outcomes with the doublet, showing higher PR rates (37.7% vs. 21.4%), DCR (85.2% vs. 76.8%), and lower rates of disease progression (14.8% vs. 23.2%). The BT22 trial, a phase II study, supported these findings, reporting a tendency toward longer survival for patients receiving the doublet compared to gemcitabine alone (mOS: 9.1 months vs. 6.7 months; p=0.675) (24). Gemcitabine and cisplatin became the standard first-line treatment for metastatic or unresectable BTC. It is currently the cytotoxic platform of choice to which other agents are added.

Gemcitabine combined with oxaliplatin (GEMOX) has shown varied efficacy across studies (25–28). In 33 advanced GBC patients, GEMOX demonstrated an ORR of 36%, stable disease in 88%, mTTP of 5.3 months, and mOS of 6.8 months (25). A smaller study (n=10) reported a 40% PR rate, DCR of 70%, and mOS of 11.1 months (26). However, a 3-weekly regimen in 48 BTC patients showed lower ORR (21.2%) and DCR (56.6%) with mOS of 7.5 months (27). Similarly, an international phase II trial reported an ORR of only 4.3% and mPFS of 2.5 months (28).

Gemcitabine combined with capecitabine yielded comparable ORRs to GEMOX, with PR rates of 33%, DCR of 64–75%, mPFS of 4.4 months, and mOS of 7.7–16 months (29, 30). In combination with pemetrexed, gemcitabine showed a CR rate of 6.3% and ORR of 12.5% (31). Historically, a phase III trial comparing modified GEMOX to gemcitabine/cisplatin reported similar ORR, PFS and OS between the two doublets. However, subsequent pooled analyses and real-world data have generally favored cisplatin/gemcitabine, particularly once immunotherapy (durvalumab or pembrolizumab) was added to this backbone. As a result, GEMOX is now typically reserved for cisplatin-ineligible or frail patients rather than considered an equivalent first-line standard (32).

Non-gemcitabine doublets include capecitabine/cisplatin, which showed a PR rate of 53.3% in one study but lower ORRs of 14.2–32% in others (33–35). Cisplatin with 5-FU demonstrated PR rates of 46.7% and 36% in Japanese and French studies, respectively, but short survival durations (mOS: 4.9 months) in the Japanese cohort (36, 37). A combination of S1 and oxaliplatin achieved a PR rate of 40% in a small cohort (n=10) (38). Finally, oxaliplatin with capecitabine reported an ORR of 30%, DCR of 63%, mTTP of 4.7 months, and mOS of 8.0 months in a German prospective multicenter phase II study (39).

Building on the success of doublet therapies, triplet combinations were explored for advanced BTC. A regimen of 5-FU, gemcitabine, and cisplatin demonstrated an ORR of 40% with one CR in 15 patients, but with significant grade 4 neutropenia (71.4%) (40). A triplet of gemcitabine, 5-FU, and oxaliplatin showed a PR rate of 23%, DCR of 69%, mTTP of 5.7 months, and mOS of 9.9 months in 35 patients with advanced GBC (41). Another triplet of gemcitabine, leucovorin, and 5-FU reported a PR rate of 21.4%, mPFS of 5.2 months, and mOS of 7.2 months in 14 patients (42).

Fluoropyrimidines are the only monotherapies evaluated for unresectable and advanced GBC in the second-line setting. S-1 monotherapy (80 mg/m² for 28 days followed by 14 days of rest) was first studied in 2009, showing a response rate of 18.8%, mPFS of 5.5 months, mOS of 8.0 months, and a 1-year survival rate of 38.2% in 16 second-line patients (52). A subsequent multicenter phase II study in 22 gemcitabine-refractory BTC patients reported an ORR of 22.7%, DCR of 50%, mOS of 13.5 months, and mTTP of 5.4 months (53). However, a more recent study in 14 GBC patients showed lower efficacy, with an ORR of 7.1%, mPFS of 1.4 months, and mOS of 4.7 months (54).

Given the modest efficacy of single-agent fluoropyrimidine, fluoropyrimidine-based combination regimens were explored for advanced BTC.

A phase IIb trial conducted in South Korea evaluated liposomal irinotecan plus fluorouracil/leucovorin versus fluorouracil/leucovorin alone in 174 patients with metastatic BTC who had progressed on gemcitabine and cisplatin. The experimental arm demonstrated a significantly longer PFS in the initial analysis (7.1 vs. 1.4 months; HR 0.56, p=0.0019) (57). However, subsequent analyses incorporating blinded independent central radiologic review reported a more modest PFS benefit, and no consistent overall survival advantage was observed. These findings raise uncertainty regarding the magnitude of benefit of liposomal irinotecan–based regimens in the second-line setting. Notably, the phase II NALIRICC-AIO-HEP-0116 trial conducted in Germany failed to confirm these results (58). In that study of 100 patients, liposomal irinotecan plus fluorouracil/leucovorin was associated with an mOS of 6.9 months and an mPFS of 2.76 months, compared with 8.21 months and 2.3 months, respectively, for fluorouracil/leucovorin alone, while also demonstrating higher rates of grade ≥3 toxicity (70.8% vs. 50%). Collectively, the available data suggest that the role of liposomal irinotecan plus fluorouracil/leucovorin in BTC remains uncertain, particularly in the absence of direct comparative data against FOLFOX, which remains the most commonly used second-line chemotherapy backbone.

A phase II trial evaluated FOLFIRINOX in 40 patients who progressed on gemcitabine/cisplatin, using standard and modified dosages (59). mPFS and OS were 6.2 and 10.7 months, respectively, with common grade 3–4 adverse events including neutropenia, diarrhea, nausea, vomiting, and mucositis. Other regimens, such as irinotecan monotherapy, XELIRI (irinotecan and capecitabine), gemcitabine monotherapy, and 5-FU with doxorubicin and mitomycin-C, have been reported (60–65). However, these studies are small and retrospective, lacking sufficient evidence to impact clinical practice. In real-world practice, treatment sequencing following first-line chemo-immunotherapy is often individualized and influenced by performance status, comorbidities, and access to molecular testing. Early comprehensive genomic profiling is increasingly favored to identify actionable alterations before clinical deterioration, although testing may also be performed at progression when tissue is limited at diagnosis. For patients with preserved performance status, fluoropyrimidine-based chemotherapy (most commonly FOLFOX) remains the most widely used second-line option in the absence of a targetable alteration. In older patients, those with impaired renal function, or those who are cisplatin-ineligible, treatment decisions frequently diverge from trial-defined pathways and prioritize tolerability and symptom control. Across all lines of therapy, enrollment in clinical trials remains strongly encouraged whenever available. To facilitate clinical interpretation of the evolving treatment landscape, Figure 2 provides a schematic overview of contemporary systemic treatment approaches in advanced biliary tract cancer, integrating chemo-immunotherapy, subsequent cytotoxic options, and biomarker-informed targeted therapies.

While biomarker testing in BTC holds promise for personalized targeted therapy, its real-life application faces significant challenges, particularly with tissue acquisition. Advanced disease often precludes surgical resection, and diagnostic needle biopsies are limited by insufficient tumor cells for molecular characterization, as well as inadequate sample quality (75). Failure rates for tissue-biopsy profiling in metastatic BTCs are reported to reach 26.8% (76). Given the increasing therapeutic relevance of actionable genomic alterations in BTC, obtaining adequate tissue for molecular profiling is essential whenever feasible. Core needle biopsy is generally preferred over fine-needle aspiration alone, as limited cytologic samples may be insufficient for comprehensive next-generation sequencing and biomarker assessment.

Liquid biopsy assays have been proposed as an alternative and complement to tissue-based sampling for BTC and are now supported by a growing body of prospective data. Recent studies and reviews demonstrate acceptable concordance between ctDNA and tissue NGS for identifying actionable alterations in advanced BTC and support ctDNA as a feasible platform for genomic profiling when tissue is inadequate or unobtainable. At the same time, important limitations remain, including variable ctDNA shedding, lower sensitivity for certain genomic events (particularly low-VAF mutations and some gene fusions), and the need for standardized assay platforms and reporting frameworks. Emerging data also suggest that ctDNA dynamics may have prognostic value—higher baseline cfDNA/ctDNA levels and persistent or rising ctDNA during therapy have been associated with worse PFS/OS or earlier recurrence after resection in BTC (77–80).

FGFR2 fusions, observed in 10–20% of iCCA cases, activate downstream signaling pathways that drive tumorigenesis, promoting cell proliferation, survival, and angiogenesis (81–83). These fusions are more common in younger women and are a defining molecular feature of iCCA (2). Multiple FGFR inhibitors have been developed, transforming the treatment landscape. Pemigatinib, the first FDA-approved FGFR-targeted therapy, demonstrated an ORR of 35.5%, mPFS of 6.9 months, and mOS of 21.1 months in the FIGHT-202 trial for previously treated FGFR2 fusion-positive iCCA and is currently being studied in the first-line setting (FIGHT-302) (84, 85). Futibatinib, a covalent FGFR1–4 inhibitor, showed superior efficacy in the FOENIX-CCA2 trial, achieving an ORR of 42%, mPFS of 9.0 months, and mOS of 21.7 months. Unlike its predecessors, futibatinib targets resistance mutations commonly associated with ATP-competitive inhibitors, reducing drug-resistant clones (86–88). Other inhibitors, such as derazantinib and erdafitinib, have shown promise in early studies (89, 90). Despite these advancements, challenges remain, including understanding the clinical implications of more than 150 identified FGFR2 fusion partners and co-occurring mutations. Initial genomic analyses suggest limited impact of these factors on treatment efficacy, but larger studies are needed to confirm these findings (91). FGFR inhibitors are now integral in managing FGFR2 fusion-positive iCCA, with ongoing trials evaluating their potential in the first line setting.

cMET amplification and overexpression, observed in 2–8% of BTCs, are linked to tumor invasion, angiogenesis, and poor prognosis. Elevated cMET expression is reported in 15% of eCCA and 12% of iCCA (125, 126). Cabozantinib, a multikinase inhibitor targeting MET, showed limited activity and significant toxicity in a phase II trial of 19 CCA patients (127).

Aberrations in this pathway occur in up to 25% of iCCA, 40% of eCCA, and 4–16% of GBC (128). While inhibitors targeting PI3K, AKT, and mTOR have been studied in first- and second-line settings, they have shown minimal efficacy (129–131). Combination strategies with chemotherapy or other targeted therapies are being investigated.

DNA damage repair gene alterations, including BRCA1/2 mutations, occur in up to 63% of BTCs (132–134). Although retrospective data on PARPi in BRCA-mutant BTCs are inconclusive, no clinical trials have evaluated their efficacy in first- or second-line settings (135). Recently, however, a phase II study evaluated the efficacy of niraparib, a PARP inhibitor, in patients with BRCA-mutated unresectable or recurrent biliary tract, pancreatic, and other gastrointestinal cancers. The study found that niraparib demonstrated promising antitumor activity in this patient population (136).

NTRK1/2/3 gene fusions, activating PI3K and MAPK pathways, are found in 2–3% of CCA cases (137–139). Larotrectinib, a pan-NTRK inhibitor, achieved an ORR of 80% and a 1-year PFS of 55% in a phase II trial, including 2 CCA patients (140, 141). Entrectinib demonstrated an ORR of 57%, mPFS of 11.2 months, and mOS of 21 months in the phase II STARTRK-2 study, leading to FDA approval for NTRK fusion-positive solid tumors (142).

RET fusions drive oncogenesis through downstream pathway activation. Prevalence in BTCs is unclear (143). In the phase I/II ARROW trial, pralsetinib achieved an ORR of 57%, mPFS of 7.4 months, and mOS of 13.6 months in RET fusion-positive solid tumors, including 3 CCA patients (144, 145). Selpercatinib, evaluated in the LIBRETTO-001 trial, showed an ORR of 43.9%, mPFS of 13.2 months, and mOS of 18 months, with 2 CCA patients included (146, 147). Further studies are needed to clarify the role of RET inhibitors in BTCs.

Pembrolizumab was studied in the phase 1b KEYNOTE-028 trial with 24 PD-L1-positive BTC patients, showing a 17% PR and 17% SD rate (149). In KEYNOTE-158, 104 unselected pre-treated BTC patients had an ORR of 5.8%, with mOS and mPFS of 7.4 and 2.0 months, respectively. Retrospective analysis revealed that PD-L1 positivity (58.7%) was associated with a higher ORR (6.6% vs. 2.9%) but not superior survival outcomes (150). Nivolumab demonstrated promise in phase 1 and 2 trials, with a phase 2 study reporting an ORR of 22%, DCR of 59%, mPFS of 3.7 months, and mOS of 14.2 months in 46 advanced refractory BTC patients. PD-L1 positivity correlated with longer PFS but not OS (151, 152). Durvalumab, tested as monotherapy and combined with tremelimumab in a phase 1 trial of pre-treated Asian BTC patients, achieved mOS of 8.1 months and 10.1 months, respectively (153).

The combination of nivolumab and ipilimumab was evaluated in the phase 2 CA209–538 trial in 39 BTC patients (16 iCCA, 10 eCCA, 13 GBC). This combination achieved an ORR of 23%, DCR of 44%, mPFS of 2.9 months, and mOS of 5.7 months (154).

The potential synergy of chemotherapy and immunotherapy was explored extensively in BTC, with gemcitabine/cisplatin (GEMCIS) as the most common chemotherapy backbone. In a phase 2 study of 32 BTC patients, nivolumab combined with GEMCIS achieved an ORR of 61.9% in treatment-naïve patients and 33% in pre-treated patients, with no significant differences in PFS or OS (155). Toripalimab, an anti-PD1 inhibitor, was evaluated with S-1 and gemcitabine in a phase 2 trial of 48 treatment-naïve patients, achieving a mPFS of 7 months, mOS of 16 months, and an ORR of 27.1% with a DCR of 87.5% (156). Camrelizumab, combined with oxaliplatin-based regimens GEMOX and FOLFOX, demonstrated ORRs of 16.3–54%, mPFS of 5.3–6.1 months, and mOS of 11.8–12.4 months, with better responses in PD-L1-positive patients (ORR 80% vs. 53.8%) (157, 158).

Durvalumab-based and pembrolizumab-based chemo-immunotherapy regimens have demonstrated broadly comparable survival benefits, establishing PD-1/PD-L1 blockade plus gemcitabine/cisplatin as a class effect rather than a drug-specific advantage. In a phase 2 study of 121 chemotherapy-naïve BTC patients, GEMCIS plus durvalumab achieved an ORR of 73.4%, mPFS of 11 months, and mOS of 18.1 months. Comparable outcomes were seen with the addition of tremelimumab (159). The subsequent TOPAZ-1 phase 3 trial randomized 685 treatment-naïve patients with metastatic or recurrent BTC to GEMCIS plus durvalumab or placebo. Durvalumab improved ORR (26.7% vs. 18.7%), 2-year OS rates (24.9% vs. 10.4%), and reduced hazard ratios for OS (0.80; p=0.021) and PFS (0.75; p=0.001) (160). Based on these results, the FDA approved GEMCIS plus durvalumab for advanced BTC, establishing it as a new standard first-line therapy and a pivotal change in the treatment landscape since the ABC-02 trial (161).

The phase 3 Keynote-966 trial randomized 1,069 untreated BTC patients to receive gemcitabine/cisplatin with either pembrolizumab or placebo. Pembrolizumab significantly improved OS (12.7 vs. 10.9 months; p=0.0034) without increasing toxicity, establishing it as a new first-line option for BTC (162).

The phase 2 IMbrave151 trial evaluated the addition of bevacizumab to atezolizumab, gemcitabine, and cisplatin in 162 untreated BTC patients. PFS was slightly higher in the bevacizumab arm (8.4 vs. 7.9 months), but the difference was not clinically meaningful (163).

Racial differences may influence ICI efficacy in BTC. Cross-trial comparisons indicate higher response rates in Asian patients, as observed in studies like TOPAZ-1, where subgroup analysis showed greater benefit for Asian patients (150, 151, 153, 160). Geographic and molecular heterogeneity in BTC, including differences in tumor biology and risk factors, highlight the importance of considering these factors in trial design (164, 165). The neutral results of IMbrave151 suggest that the addition of anti-angiogenic therapy to chemo-immunotherapy may not confer incremental benefit in unselected BTC populations. Future studies may need to incorporate biomarker selection or alternative combination strategies to better define subsets most likely to benefit.