Research on Targeted Therapy for Malignant Tumors of the Biliary Tract

Introduction

Biliary tract cancer (BTC) encompasses a group of aggressive adenocarcinomas, including cholangiocarcinoma (intrahepatic cholangiocarcinoma and extrahepatic cholangiocarcinoma. Extrahepatic cholangiocarcinoma can be further classified into perihilar cholangiocarcinoma and distal cholangiocarcinoma) and gallbladder cancer.¹ The incidence of cholangiocarcinoma in high-income countries is relatively low (0.35 to 2 cases per 100,000 annually); however, in endemic regions such as Thailand and China, the incidence is up to 40-fold higher.^2,3 For gallbladder cancer, an estimated 219,420 new cases and 165,087 deaths were reported globally in 2018, with significant variations by sex and geographic region.⁴ The incidence of BTC is rising, primarily driven by an increase in intrahepatic cholangiocarcinoma, while rates of extrahepatic cholangiocarcinoma and gallbladder cancer remain stable.^2,5,6 The prognosis of BTC is poor, with a 5-year survival rate of approximately 5–15%.^7,8 This is attributed to the nonspecific early symptoms of BTC, leading to low diagnostic rates, with most patients diagnosed at an unresectable or metastatic stage.⁹ Only 30–40% of patients undergo surgical resection.¹⁰ However, even among those diagnosed at an early stage, the recurrence rate remains high, at approximately 50%.¹¹

Since most patients are diagnosed at an advanced stage, improving outcomes is critically important.¹² In this context, palliative care becomes the primary approach, necessitating not only advances in early detection technologies and novel biomarker development but also the creation of new therapeutic strategies.^10,13,15 For a prolonged period, palliative chemotherapy has been the mainstay of treatment for patients with advanced disease.^16,17 However, the advent of targeted therapies has introduced a new paradigm for BTC treatment.

This review is written with the aim of providing an overview of the current development of targeted therapy for biliary tract cancer.

Targeted Therapies for Biliary Tract Cancer

Overview of Molecular Targets

Data from the prospective MOSCATO-01 trial strongly support the “precision medicine” strategy for advanced biliary tract cancer (BTC). The results demonstrated a significant improvement in median overall survival (OS) for patients receiving targeted therapy compared to those without targeted therapy (17 months vs 5 months; p = 0.008).¹⁸ The success of this precision medicine strategy relies on identifying actionable therapeutic targets, thereby enabling the selection of appropriate patients. Recent genomic analyses reveal marked heterogeneity in cholangiocarcinoma, closely associated with tumor location, where biliary tract cancers at different sites exhibit diverse genomic alterations.^19,23 Consequently, identifying targetable alterations has become increasingly critical in BTC management, highlighting the potential of precision medicine as a promising therapeutic approach.^24,25 Currently, isocitrate dehydrogenase (IDH) mutations and fibroblast growth factor receptor 2 (FGFR2) fusions are the most extensively studied targetable alterations, predominantly associated with intrahepatic cholangiocarcinoma (iCCA).^26,31 Approximately 10–20% of iCCAs harbor IDH1 mutations, while 10–20% carry FGFR2 gene fusions.^21,32,36 In contrast, HER2 aberrations are most prevalent in extrahepatic cholangiocarcinoma (eCCA) and gallbladder cancer (GBC).^24,37,38 Prior systematic reviews and meta-analyses indicate a higher HER2 overexpression rate in extrahepatic BTC compared to iCCA (19.9% [95% CI 12.8–27.1%] vs 4.8% [95% CI 0–14.5%]; p = 0.0049).²⁵ Additionally, numerous other therapeutic targets and approaches are under active investigation. Examples include B-Raf proto-oncogene serine/threonine kinase (BRAF),³⁹ vascular endothelial growth factor (VEGF),³² neurotrophic tropomyosin kinase receptor (NTRK),⁴⁰ RET,⁴¹ RNF43,⁴² KRAS,⁴³ MET,⁴⁴ which may offer novel therapeutic avenues. Furthermore, significant progress has been made in novel therapies such as Immune Checkpoint Inhibitors,⁴⁵ Antibody–drug conjugates (ADCs),⁴⁶ Adoptive Cellular Therapies,⁴⁷ tumor microenvironment modulation,⁴⁸ MSI-high/dMMR,⁴⁹ and cancer vaccines.⁵⁰ This review will summarize and discuss these targets and therapeutic strategies.

Isocitrate Dehydrogenase as a Target (IDH)

Among the three IDH isoforms, IDH3 is located in the mitochondrial matrix,^51,53 IDH1 in the cytoplasm and peroxisomes, and IDH2 in the mitochondria,⁵⁴ These enzymes catalyze the decarboxylation of isocitrate to generate α-ketoglutarate (α-KG),⁵⁵ with varying expression levels across tissues and diverse physiological functions (Figure 1).^56,58 IDH1 and IDH2 participate in cellular metabolism and are the isoforms most frequently associated with cancer.⁵¹ Mutant IDH (mIDH) utilizes NADPH to reduce α-KG to the R-enantiomer of 2-hydroxyglutarate [(R)-2HG] (Figure 1).⁵⁹ mIDH exhibits reduced affinity for isocitrate and enhanced binding to NADPH and α-KG, resulting in a 1000-fold decrease in the catalytic rate of the forward oxidative decarboxylation reaction (which produces α-KG).⁵⁹ These somatic gain-of-function mutations occur early in tumorigenesis. The mutant enzyme aberrantly induces alterations in cellular metabolism and accumulation of the oncometabolite (R)-2HG,⁶⁰ disrupting epigenetics and metabolic pathways, leading to differentiation defects, immune suppression, and ultimately tumorigenesis.^61,69 The vast majority of these mutations occur at residue R132 in IDH1 and residues R140 or R172 in IDH2.^36,51,70,74 mIDH1 typically coexists with wild-type IDH1 (WT IDH1) in a heterozygous configuration, without significant amplification of the mutant allele.⁷⁵ Heterodimers formed between WT IDH1 and mIDH1 are far more active in producing (R)-2HG than mIDH1 homodimers. In this complex, the WT subunit converts isocitrate to α-KG and generates NADPH, while the mutant subunit utilizes NADPH to reduce α-KG to (R)-2HG.^76,78 Functional studies indicate that the oncogenic activity of mIDH1 requires WT IDH1.^79,80 In contrast to IDH1 mutants, mIDH2 functions as a homodimer and efficiently produces (R)-2HG, despite the typical retention of the WT IDH2 allele.⁸⁰ Because (R)-2HG is neither an intermediate nor a substrate in central carbon metabolism, it accumulates to exceedingly high levels in mIDH cancer cells.⁸¹ Common co-occurring genomic alterations in mIDH iCCAs include inactivating mutations/deletions in ARID1A (22.7%), PBRM1 (20%), and BAP1 (13%).⁸² Regarding mutational relationships, TP53 loss-of-function and activating FGFR2 fusions tend to be mutually exclusive with IDH1 mutations.¹⁹ IDH2 mutations are mutually exclusive with IDH1 mutations and most frequently co-occur with inactivating mutations/deletions in BAP1 (29.7%), PBRM1 (17.1%), and ARID1A (16.2%).⁸³ The correlation between IDH mutations and clinical outcomes in iCCA is inconsistent across studies, potentially due to sampling bias, population heterogeneity, varying clinical stages, and limited sample sizes.^74,84,88 Histologically, iCCA can be classified into small duct and large duct subtypes, with multiple studies associating IDH mutations with small duct features.^89,92 mIDH iCCA cells exhibit distinct morphological features, mitochondrial enrichment, and increased fibrosis,⁹³ Some mIDH hepatic tumors are histologically indistinguishable from hepatocellular carcinoma but display iCCA transcriptional signatures, with IDH mutations linked to enhanced mitochondrial gene expression and copy number gains.⁹⁴ Studies utilizing immune profiling, transcriptomics, and multi-region analyses indicate that IDH mutations in iCCA correlate with altered immune cell infiltration and activity. For example, some reports show IDH-mutant tumors are enriched in immune subtypes with lower CD8+ T cell infiltration.^95,98

Figure 1 Biologic function of wild-type IDH isomers and pathogenic function of mutant IDH isomers in malignant cells. (A) Biologic function of IDH isomers. (B) Pathogenic function of mutant IDH isomers in malignant cells.81,99

Ivosidenib (AG-120) is an oral, targeted inhibitor of mIDH1.¹⁰⁰ The ClarIDHy trial remains the only randomized Phase III study of targeted therapy for BTC. In this trial, 187 patients with histologically confirmed advanced IDH1-mutant cholangiocarcinoma, who had progressed after one or two prior lines of systemic therapy, were randomized (2:1) to receive oral ivosidenib 500 mg (n=126) or matched placebo (n=61) once daily in 28-day cycles. Crossover to ivosidenib was permitted upon radiographic progression assessed by investigators, and 43 patients (70%) in the placebo group subsequently received ivosidenib.

Results demonstrated that ivosidenib significantly improved progression-free survival (PFS) compared to placebo (median PFS 2.7 months vs 1.4 months; hazard ratio [HR], 0.37; 95% CI, 0.25–0.54; one-sided p < 0.0001). In the final overall survival (OS) analysis, the median OS was 10.3 months (95% CI, 7.8–12.4) in the ivosidenib group versus 7.5 months (95% CI, 4.8–11.1) in the placebo group (HR, 0.79; 95% CI, 0.56–1.12; one-sided p = 0.09). These results, combined with supportive quality-of-life data and a manageable safety profile, establish the clinical benefit of ivosidenib in patients with advanced IDH1-mutant cholangiocarcinoma.²⁷ The selective mutant IDH1 (mIDH1) inhibitor AG-120/ivosidenib has been approved by the US FDA and European Medicines Agency (EMA) for adult patients with previously treated, locally advanced or metastatic cholangiocarcinoma harboring IDH1 mutations.^26,28

Olutasidenib is an IDH1 inhibitor previously studied in acute myeloid leukemia (AML). It is currently being evaluated in a phase I/II clinical trial involving 44 patients with relapsed or refractory IDH1-mutant solid tumors, including 26 with iCCA. This preclinical study demonstrated the safety and tolerability of olutasidenib in relapsed/refractory IDH1-mutant solid tumors.¹⁰¹

Additionally, mechanisms of ivosidenib resistance have been investigated in preclinical animal models of advanced iCCA.⁹⁵ Ivosidenib treatment induced rapid recruitment and enhanced effector function of CD8+ T cells, including increased interferon-gamma (IFN-γ) production (Figure 2). Conversely, CD8+ T cell exhaustion contributed to ivosidenib resistance. Functional studies revealed that at baseline, transcriptional response to IFN-γ is impaired in mIDH1 iCCA cells due to (R)-2-hydroxyglutarate [(R)-2HG]-mediated suppression of ten-eleven translocation 2 (TET2) demethylase, which is required for epigenetic activation of IFN-γ target genes. Genetic ablation of IFN-γ receptor or TET2 in tumor cells also conferred ivosidenib resistance in vivo. However, loss of IFN-γ receptor or TET2 specifically compromised immune responses in tumor cells without affecting CD8+ T cell recruitment or IFN-γ secretion capacity. Thus, mIDH orchestrates immune evasion in iCCA models through distinct yet interconnected mechanisms. Although mIDH1 inhibition initially restored antitumor immunity in murine models, efficacy was unsustainable due to immune checkpoint activation and other immunosuppressive processes. In combination studies of AG-120 plus anti-CTLA-4 antibody, the two agents acted synergistically in mIDH1 allograft models, inducing complete responses and cure in a subset of animals (Figure 2). Preliminary evidence suggests that ivosidenib also promotes immune checkpoint activation in human mIDH1 cholangiocarcinoma. A clinical trial evaluating the selective mIDH1 inhibitor FT-2102 (olutasidenib) combined with nivolumab (anti-PD-L1) in IDH1-mutant solid tumors (NCT03684811) is currently ongoing. Future research on resistance mechanisms should focus on comprehensively characterizing the immunologic effects of mIDH1 inhibition in patients and comparing the efficacy of combination regimens involving different checkpoint inhibitors (eg, CTLA4-blocking antibodies). Additionally, exploring IDH inhibitor combinations with chemotherapeutic agents or other strategies to enhance antitumor efficacy represents a key future direction. Clinical trials targeting this pathway are summarized in Table 1.

Figure 2 Mechanism of mutant IDH1 inhibition and immune modulation in intrahepatic cholangiocarcinoma. (A) (R)-2HG inactivates TET2 in mIDH1 iCCA cells. On AG-120 treatment, TET2 is reactivated and drives therapeutic efficacy by epigenetically activating IFN-γinduced target genes. (B) Murine tumors with mIDH1 mutations exhibit a scarcity of CD8⁺ T cells. Treatment with the mIDH1 inhibitor AG120 (ivosidenib) rapidly stimulates CD8⁺ T cell recruitment and effector function, including IFN-γ production, while inducing proliferative arrest and death of tumor cells. However, treatment efficacy is short-lived due to subsequent activation of immune checkpoints and stimulation of regulatory T cells (Tregs). In contrast, combination treatment with AG120 and an anti-CTLA4 antibody confers complete and durable therapeutic responses.81

Table 1 Clinical Trials by Targeted Agent102

The Role of Fibroblast Growth Factor Receptor (FGFR)

The FGFR family comprises four transmembrane receptors with intracellular tyrosine kinase domains (RTKs), designated FGFR1 to FGFR4, which share similar structural organization.¹⁰³ The fifth member (FGFR5/FGFRL1) acts as a co-receptor for FGFR1, retaining ligand-binding capacity but lacking an intracellular tyrosine kinase domain; thus, it is not considered oncogenic.¹⁰⁴ FGFR activity is regulated by 18 secreted fibroblast growth factors (FGFs). Most FGFs interact with heparan sulfate proteoglycans (HSPGs) on the cell surface and extracellular matrix, functioning locally as autocrine/paracrine growth factors.¹⁰⁴ In the ligand-free inactive state, FGFRs exist as monomers or transient dimers on the cell surface, with kinase domains maintained in a default autoinhibited conformation via intramolecular interactions.¹⁰⁵ FGF binding to HSPGs induces receptor dimerization, triggering trans-autophosphorylation.^106,107 This process overcomes energetic barriers from autoinhibitory interactions, enabling FGFR catalytic activation. Subsequent intracellular pathway activation (eg, Ras/MAPK, JAK/STAT, PI3K/AKT) drives transcription of genes regulating proliferation, migration, differentiation, and survival (Figure 3).^108,113 In cholangiocarcinoma, FGFR alterations predominantly affect FGFR2 (6.1%), with rare FGFR1 involvement (0.9%) and no other family members.¹¹⁴ Rearrangements/fusions (3.5%) exceed amplifications (2.6%), while point mutations are infrequent (0.9%).¹¹⁵ Alternative splicing is a key feature of FGFR2 expression. Splicing at the third immunoglobulin-like domain generates IIIb (epithelial-specific) and IIIc (mesenchymal-specific) isoforms (Figure 4). FGFR2 fusions in intrahepatic cholangiocarcinoma (iCCA) exclusively select the IIIb isoform, consistent with iCCA’s epithelial origin and isoform-restricted expression. Thus, isoform selectivity represents a potential therapeutic target.

Figure 3 Schematic overview of the FGFR signaling pathway. Ligand binding induces FGFR dimerization and transphosphorylation, activating downstream RAS/MAPK, PI3K/AKT, and JAK/STAT signaling cascades to promote cell proliferation, migration, differentiation, and survival.99

Figure 4 Alternative splicing and isoform expression pattern of FGFR2. Exon 8 is included in the epithelial-specific IIIb isoform, whereas exon 9 is included in the mesenchymal-specific IIIc isoform. Ig, immunoglobulin-like domain; TM, transmembrane domain Red arrow indicates exon 8; pink arrow indicates exon 9.114

Cancer-associated FGFR2 structural alterations disrupt regulatory principles by: (1) enhancing ligand affinity to overcome limited FGF availability, (2) inducing ligand-independent dimerization, or (3) perturbing kinase domain autoinhibition.¹¹⁶ C-terminal truncations of FGFR2 cause loss of regulatory motifs critical for FGFR control. Such truncations are established drivers in gastric carcinogenesis.^112,117 In these gastric cancers, alternatively spliced FGFR2 IIIb yields three variants: C1, C2, and C3. C1 retains an intact C-terminus and represents canonical FGFR2 IIIb. C2 truncates 34 C-terminal amino acids, whereas C3 truncates 53 residues, eliminating the entire terminal exon. The domains retained in C3 mirror those preserved in iCCA-associated FGFR2 fusion proteins. Both C2 and C3 variants exhibit transforming activity in gastric cancer. C3 demonstrates greater transforming potency than C2, indicating that the terminal exon exerts regulatory/inhibitory effects on FGFR2 IIIb activity (Figure 5).¹¹⁸ The 53-residue truncation in C3 abolishes the YLDL motif (Y770–L773) and a proline-rich motif (aa 774–806) harboring the Grb2-binding site (PPPVPPK). (Figure 5) Y770 directly binds phospholipase Cγ (PLCγ), activating downstream PKC and Ca²⁺ signaling. L773 mediates receptor internalization, regulating FGFR2 membrane recycling and signal duration. L773 deletion traps FGFR2 at the membrane, prolonging RAS/MAPK and PI3K/AKT signaling. In C1, the Y770F mutation enhances binding to FGFR substrate 2 (FRS2), activating RAS/MAPK signaling and increasing transforming capacity.^112,119 Grb2 is a critical adaptor protein that regulates RTK signaling dynamics. In the inactive state, Grb2 binds the FGFR C-terminal proline-rich motif via its SH3 domain, forming a pre-dimer complex that sterically hinders dimerization and autoactivation. Loss of the Grb2-binding site enables spontaneous dimerization, causing constitutive phosphorylation and hyperactivation of pro-proliferative pathways (eg, RAS-MAPK, PI3K-AKT), thereby driving malignant transformation.^120,121

Figure 5 C-terminal structural variants of FGFR2 IIIb. C1 represents the full-length canonical FGFR2 IIIb isoform; C2 and C3 represent C-terminally truncated variants with progressive loss of regulatory motifs, including the Grb2-binding site and internalization signals.114

Beyond truncation variants, FGFR2 fusion proteins offer novel therapeutic avenues for iCCA. Genomic alterations activating FGFR2 occur in nearly 20% of iCCAs, with numerous FGFR2 translocations identified as oncogenic drivers.^{21,32,33,122,123} FGFR2 fusions arise from chromosomal translocations that break FGFR2 and fuse it to partner genes (eg, PPHLN1, BICC1, AHCYL1, TACC3). These rearrangements retain the N-terminal domain (including the intact kinase domain) of FGFR2, while partner genes contribute C-terminal domains driving constitutive dimerization.^109,124 C-terminal domains of partners (eg, coiled-coil in PPHLN1, SAM domain in BICC1) provide dimerization interfaces, enabling ligand-independent dimerization and sustained FGFR2 kinase activation.¹²⁵ FGFR2 fusions typically retain the IIIb isoform, which is highly expressed in cholangiocytes and may confer tissue-specific oncogenic signaling.¹¹⁴ Breakpoints commonly occur near exon 17 or 18, deleting C-terminal regulatory regions (eg, Ser780 and Grb2-binding site encoded by exon 18). This deletion derepresses kinase activity, enhancing signaling output.^126,127 In wild-type (WT) FGFR2, Ser780 is phosphorylated by ERK1/2. FGFR2 activation triggers ERK1/2-mediated Ser780 phosphorylation, which induces conformational changes to reduce catalytic activity, establishing a negative feedback loop that prevents signal hyperactivation.¹²⁶ Loss of Ser780 in FGFR2 fusions disrupts this feedback, resulting in sustained RAS-ERK pathway activation that promotes tumor proliferation and survival.¹²⁷ Grb2 regulatory mechanisms are described in the truncation variant section. FGFR2 fusion proteins are summarized in (Table 2).

Table 2 FGFR2 Fusion proteins109,114,128

Early treatments targeting FGFR aberrations relied on non-selective multi-target tyrosine kinase inhibitors (TKIs), including dovitinib, ponatinib, lucitanib, lenvatinib, pazopanib, and regorafenib. Although these agents exert antitumor effects by inhibiting FGFR signaling, their non-selective mechanism leads to concurrent suppression of other kinase families such as c-kit, FLT3, vascular endothelial growth factor receptor (VEGFR), platelet-derived growth factor receptor (PDGFR), c-SRC, and RET.^103,110,129 These molecules demonstrate modest activity in FGFR2 fusion-positive iCCA.^130,131 Ponatinib (AP24534), a multi-target TKI initially developed for BCR-ABL fusion kinase inhibition in chronic myeloid leukemia (CML), also exhibits broad multi-kinase inhibitory properties, including targeting FGFR family members.^132,133 In a prior study, ponatinib demonstrated clinical benefit in two FGFR2 fusion-positive iCCA patients.¹³⁰ One patient resistant to standard chemotherapy achieved tumor reduction after 6 weeks of ponatinib without significant toxicity. Another patient progressing on gemcitabine-cisplatin combination therapy received pazopanib for 4 months followed by ponatinib for 2 months, resulting in reduced tumor volume.¹³¹ In a phase II trial (NCT02265341), ponatinib showed efficacy in FGFR fusion-positive cholangiocarcinoma, with an objective response rate (ORR) of 15.6% (5 partial responses), disease control rate (DCR) of 53.1%, and median progression-free survival (mPFS) of 3.7 months.

Subsequently, selective FGFR tyrosine kinase inhibitors (TKIs) have been developed, with multiple compounds demonstrating acceptable safety profiles in Phase I trials and preliminary efficacy in phase II studies involving patients with refractory iCCA.^29,134,141 Among these, pemigatinib, infigratinib, and futibatinib have received FDA approval.^137,142,143

Pemigatinib (INCB054828) is an oral small-molecule TKI that selectively targets FGFR1/2/3, inhibiting tumor cell proliferation, survival, and angiogenesis by blocking FGFR signaling. The phase II FIGHT-202 trial (NCT02924376) evaluated pemigatinib in a multicenter, open-label study.¹³⁷ This study enrolled 107 patients with locally advanced/metastatic cholangiocarcinoma harboring FGFR2 fusions/rearrangements who had progressed after ≥1 prior systemic therapy. Results showed an objective response rate (ORR) of 35.5% (38/107), including 2.8% complete responses (CR) and 32.7% partial responses (PR), with a disease control rate (DCR) of 82%. Median progression-free survival (PFS) was 6.9 months, and median overall survival (OS) was 21.1 months. The most common adverse events were hyperphosphatemia (63%), diarrhea (51%), and fatigue (47%), mostly grade 1–2 and manageable/reversible. Among patients with FGFR2 fusions, ORR was higher (40.6%), and median duration of response (DOR) reached 7.5 months, significantly superior to conventional chemotherapy (median OS ~6–7 months). The significant efficacy of FIGHT-202 (ORR 35.5%, OS 21.1 months) formed the basis for pemigatinib’s accelerated FDA approval. The ongoing phase III FIGHT-302 trial (NCT03656536) is a randomized, multicenter, open-label study comparing pemigatinib versus gemcitabine-cisplatin (GemCis) chemotherapy as first-line therapy for advanced iCCA patients with FGFR2 rearrangements.¹⁴⁴

Infigratinib (BGJ398) is an oral, selective FGFR1–3 tyrosine kinase inhibitor that, similar to pemigatinib, competitively binds the ATP-binding pocket of FGFR, inhibiting FGFR signaling and thereby blocking tumor cell proliferation and angiogenesis. A phase I trial involving 3 cholangiocarcinoma patients with FGFR2 alterations (2 fusions, 1 mutation) showed stable disease and reduced tumor burden in all patients.¹⁴⁵ Infigratinib received accelerated FDA approval on May 28, 2021, for previously treated, unresectable locally advanced or metastatic cholangiocarcinoma. Approval was based on the phase II trial NCT02150967—an open-label, single-arm, multicenter study evaluating infigratinib in advanced cholangiocarcinoma with FGFR2 fusions/rearrangements.¹⁴³ Among 108 enrolled patients (median 2 prior therapies), objective response rate (ORR) was 23.1% (1 CR, 24 PR), disease control rate (DCR) 84.3%, median duration of response (DOR) 5.0 months (32% with DOR ≥6 months), median overall survival (OS) 12.2 months, and median progression-free survival (PFS) 7.3 months. These data supported accelerated FDA approval for infigratinib as the second targeted therapy after pemigatinib in FGFR2 fusion/rearrangement-positive advanced cholangiocarcinoma. Toxicity profiles mirrored pemigatinib: electrolyte imbalances, stomatitis, dry eye, retinopathy, fatigue, and alopecia. Hyperphosphatemia occurred in 76%, with grade 3–4 hyponatremia and hypophosphatemia each in 13%. The ongoing phase III trial NCT03773302 compares infigratinib versus gemcitabine-cisplatin as first-line therapy for advanced cholangiocarcinoma with FGFR2 abnormalities.¹⁴⁶

 Futibatinib (TAS-120) is the only irreversible FGFR TKI approved globally for advanced FGFR2 fusion/rearrangement-positive iCCA. It also shows activity in broader FGF/FGFR aberrations (ORR 17.6% in non-fusion cohorts).¹⁴⁷ A phase I trial provided preliminary evidence of tolerability and clinical efficacy in iCCA patients.¹⁴⁸ In the phase II FOENIX-CCA2 trial (NCT02052778; n=103, all with confirmed FGFR2 aberrations), ORR was 42% (2% CR, 40% PR), and DCR 84.3%, indicating potent disease stabilization. Median DOR was 9.7 months (72% with DOR ≥6 months), median PFS 9.0 months (significantly longer than chemotherapy [4–6 months]), and median OS 21.7 months—a breakthrough in survival extension.¹⁴² Adverse events included hyperphosphatemia (85%), nail toxicity (33%), musculoskeletal pain (30%), and fatigue (25%), mostly grade 1–2. Grade 3 hyperphosphatemia (30%) and retinopathy (8%) were manageable via dose adjustment/supportive care. Discontinuation due to toxicity was only 6%, confirming favorable tolerability. Based on the above data, the drug was approved by the FDA on September 30, 2022. Compared to reversible FGFR inhibitors (pemigatinib ORR 35%, infigratinib ORR 23%), futibatinib showed superior ORR (42%) and longer DOR (9.7 months), potentially due to reduced resistance via irreversible binding. The global phase III trial NCT04093362 is evaluating futibatinib plus gemcitabine-cisplatin as first-line therapy for advanced/metastatic FGFR2-rearranged iCCA.¹⁴⁹ These three oral agents—pemigatinib, infigratinib, and futibatinib—are FDA-approved as subsequent treatment options for advanced FGFR2-altered cholangiocarcinoma after disease progression. Hyperphosphatemia is a frequently observed adverse event, likely associated with FGF23 blockade, which regulates renal tubular phosphate reabsorption.^150,152 Hyperphosphatemia is rarely symptomatic and typically manageable through dietary phosphate restriction and phosphate-binding agents. Hypophosphatemia occurs in 7–15% of patients, potentially due to overcorrection of hyperphosphatemia or persistent use of phosphate binders after FGFR inhibitor discontinuation.^29,137 Although generally well-tolerated (treatment discontinuation due to toxicity in only ~6% of patients), chronic adverse events—including fatigue, musculoskeletal pain, and nail toxicity—may impair quality of life and require active management.

Derazantinib (ARQ 087), another pan-FGFR inhibitor with low IC50 against FGFRs, showed statistically significant serum phosphate elevation in phase I trials but a lower incidence of hyperphosphatemia. Reduced FGF19 levels confirmed adequate target engagement. Derazantinib’s multi-target inhibition (eg, concurrent CSF1R targeting) may modulate phosphate metabolism pathways, mitigating severe hyperphosphatemia risk.^134,153,154 In the phase I/II ARQ 087–101 trial (NCT01752920), ORR 20.7%, DCR 82.8%, and mPFS 5.7 months.¹³⁴ A phase II study (NCT03230318) demonstrated significantly lower risks of hyperphosphatemia and retinopathy versus other FGFR inhibitors, with superior tolerability and a DCR of 79% in non-fusion variant cohorts.¹⁵⁵

Lirafugratinib (RLY-4008) is currently under evaluation in a global phase I/II multicenter trial (NCT04526106) for patients with cholangiocarcinoma harboring FGFR2 fusions or rearrangements.¹⁵⁶ Bemarituzuma is primarily targeting FGFR2b-overexpressing gastric cancer, with studies in cholangiocarcinoma remaining at an early exploratory stage.¹⁵⁷ FP1039 (FivePrime Therapeutics) is a soluble fusion protein comprising the extracellular domain of FGFR1 linked to the Fc region of human IgG1, which inhibits ligand binding to FGFR1 and is under clinical development.¹⁵⁸

Resistance to FGFR inhibitors encompasses primary and acquired resistance. Definitive conclusions on primary resistance remain elusive due to limited prospective data, small patient cohorts for comparative profiling, and insufficient understanding of how specific genetic alterations impact clinical outcomes. Acquired resistance is primarily driven by mutations in the FGFR2 kinase domain and/or the mitogen-activated protein kinase (MAPK) pathway.^122,159,172 Common FGFR2 mutations affect gatekeeper residue V565 (regulating access to the ATP-binding pocket back cleft) or auto-inhibitory brake residues (N550, K642, E566) that stabilize the inactive conformation.^105,173 In a resistance study, FGFR2 kinase domain mutations were detected in 65% of patients with prior clinical benefit (post-progression), compared to only 10% in primary resistance cases (P < 0.0001), confirming these mutations as key markers of acquired resistance. Twenty-six distinct FGFR2 kinase domain mutations spanning 15 residues were identified, with N550 (auto-inhibitory brake) and V565 (gatekeeper) mutations accounting for 50% and 41% of cases, respectively. Sixty-three percent of patients harbored ≥2 FGFR2 mutations, with co-occurring alterations (eg, N550K + V565F) observed in individual tumors. Higher variant allele frequency (VAF) for V565F (median 3.2%) versus N550K (0.8%) suggested greater fitness advantage for V565F mutants. Longitudinal cfDNA monitoring validated the diversity and polyclonality of resistance mutations.¹⁷⁴ N550 mutations disrupt auto-inhibition, releasing kinase activity. N550K increases the half-maximal inhibitory concentration (IC50) 2–5-fold against pemigatinib but enhances ATP affinity (55-fold lower dissociation constant, Kd), amplifying downstream signaling. V565 mutations (eg, V565F) cause steric hindrance blocking drug access to the ATP-binding pocket, reducing drug binding by 90% in NanoBRET assays. Mutants with IC50 below trough plasma concentration (Ctrough) may retain sensitivity, whereas those with IC50 exceeding peak plasma concentration (Cmax) confer resistance—establishing thresholds based on unbound drug levels.^139,175,177 Integrating in vitro IC50 and in vivo pharmacokinetics (Ctrough/Cmax) quantifies mutant resistance, revealing that subtherapeutic drug concentrations permit coexistence of low-fitness mutants (eg, N550K, V565F), driving polyclonal resistance. The next-generation reversible multi-kinase inhibitor tinengotinib (TT-00420) lacks the dimethoxyphenyl moiety—present in infigratinib, pemigatinib, and futibatinib—reducing dependence on the ATP back pocket and bypassing gatekeeper steric hindrance (eg, V565F). Three hydrogen bonds (with FGFR2 E566/A568 residues) enhance binding to the active conformation, exhibiting “fast-on/slow-off” kinetics that prolong target engagement. It is the first and only inhibitor evaluated in a registrational phase III trial (NCT05948475) for patients with FGFR2-altered cholangiocarcinoma after prior FGFR inhibitor therapy.

FGFR-targeted therapy represents a major breakthrough in advanced cholangiocarcinoma, transitioning from single-target inhibition to a new era of resistance decoding and multimodal intervention. Future directions may include: combination strategies to overcome tumor heterogeneity; machine learning models to predict mutational fitness; screening for potent drug combinations; developing high-selectivity inhibitors with optimized pharmacokinetics and safety profiles to combat polyclonal resistance.

Therapeutic responses in intrahepatic cholangiocarcinoma (iCCA) patients with non-fusion FGF/FGFR alterations (eg, mutations, amplifications) warrant further investigation. Clinical trials targeting this pathway are summarized in (Table 1).

HER Receptor Family

The human epidermal growth factor receptor (HER/ERBB) family comprises four members: HER1 (EGFR/ERBB1), HER2 (ERBB2), HER3 (ERBB3), and HER4 (ERBB4), all belonging to the receptor tyrosine kinase (RTK) superfamily. Each receptor contains an extracellular ligand-binding domain, a transmembrane domain, and an intracellular tyrosine kinase domain (TKD).¹⁷⁸ HER2 is the only member without known ligands, requiring homodimerization or heterodimerization (eg, HER2-HER3) with other HER members to activate downstream signaling pathways. Physiological HER activation initiates with ligand binding to the extracellular domain, inducing homodimeric or heterodimeric dimerization. Dimerization triggers conformational changes that activate the intrinsic TKD, leading to autophosphorylation of intracellular tyrosine residues. Phosphorylated tyrosine residues serve as hubs for recruiting SH2 domain-containing effectors (eg, Grb2, PI3K), initiating multistep cascades including MAPK/ERK, PI3K/AKT/mTOR, and JAK/STAT pathways (Figure 6). Fine-tuned regulation of these signaling networks is essential for maintaining the dynamic equilibrium of cell proliferation, differentiation, and apoptosis, whereas dysregulation (eg, sustained receptor activation or phosphatase dysfunction) drives uncontrolled neoplastic growth.^179,180 While all four HER members can dimerize, HER2 exhibits a preference for heterodimerization with HER3, forming the most potent oncogenic dimer.¹⁸¹ Despite lacking kinase activity, HER3 harbors multiple PI3K-binding sites in its intracellular domain, synergizing with HER2 to activate the PI3K/AKT pathway and promote tumor cell survival and drug resistance. Therefore, specifically targeting and blocking the HER2 signaling pathway is an effective strategy for treating cancers that express HER2.HER2 alterations are rare in intrahepatic cholangiocarcinoma (iCCA, 2.8%) but more prevalent in gallbladder cancer (GBC, 17%) and extrahepatic cholangiocarcinoma (eCCA, 20%). Mutation rates in eCCA reach 5.2%.^20,182,186 HER2 positivity correlates with shorter progression-free survival (PFS) and overall survival (OS), particularly in GBC.^20,44,187 HER2-targeted therapies include four classes: ①Monoclonal antibodies②Bispecific antibodies ③Antibody-drug conjugates (ADCs) ④Tyrosine kinase inhibitors (TKIs):

Figure 6 Overview of the HER2 signaling network. HER2 forms homodimers or heterodimers with EGFR, HER3, or HER4 to activate downstream MAPK and PI3K/AKT signaling, driving cell proliferation, migration, and survival.188

Monoclonal Antibodies

Trastuzumab is a recombinant humanized IgG1 monoclonal antibody (mAb) that binds with high affinity and specificity to extracellular domain IV (ECD IV) of HER2.¹⁸⁹ It effectively blocks ligand-independent heterodimerization of HER2/HER3 and EGFR/HER2, suppressing downstream signaling and limiting cancer cell proliferation.^190,191 In addition, Trastuzumab induces antibody-dependent cellular cytotoxicity (ADCC) by engaging FcγRIII on natural killer cells via its Fc domain, directly lysing HER2-positive tumor cells.¹⁹² In HER2-overexpressing BC patients, ADCC is induced in 83% of cases and significantly correlates with treatment response.^193,194 In cholangiocarcinoma cell lines, studies confirm robust ADCC activity, with concurrent activation of antibody-dependent cellular phagocytosis (ADCP) and complement-dependent cytotoxicity (CDC).¹⁹⁵ According to existing research, Trastuzumab monotherapy exhibits limited activity in HER2-positive cholangiocarcinoma, whereas combination regimens demonstrate favorable outcomes in HER2-altered BTC patients. MyPathway (NCT 02091141) is a phase II basket study evaluating the efficacy of trastuzumab plus pertuzumab in previously treated HER2-positive biliary tract cancer.¹⁹⁶ The results showed an overall objective response rate (ORR) of 23% (9/39), with all responses being partial responses (PR). The disease control rate (DCR) was 51% (20/39, including patients with stable disease ≥4 months). The median duration of response (DOR) was 10.8 months (95% CI 0.7–25.4), the median progression-free survival (PFS) was 4.0 months, and the median overall survival (OS) was 10.9 months. The 1-year OS rate was 50% (95% CI 33–64%), significantly superior to traditional second-line chemotherapy. Notably, iCCA showed a lower response rate, with no ORR achieved in 7 iCCA patients, while patients with GBC and ampullary cancer had the best ORRs (31% and 40%, respectively). This study first demonstrated the activity of HER2 dual-target therapy in biliary tract cancer, providing a new option for second-line and above treatment. Additionally, the phase II basket study TAPUR (NCT02693535) enrolled 29 previously treated advanced biliary tract cancer patients (including 15 gallbladder cancers, 11 cholangiocarcinomas, and 3 ampullary cancers) to explore the potential of trastuzumab plus pertuzumab dual-target therapy. The results showed a DCR of 40% (95% CI 27–100%), an ORR of 32%, a median PFS of 11 weeks (approximately 2.5 months), and a median OS of 30 weeks (approximately 7 months).¹⁹⁷

Beyond dual-target therapy, a single-arm phase II study (NCT 04722133) evaluated trastuzumab combined with FOLFOX chemotherapy in 34 HER2-positive biliary tract cancer (BTC) patients who progressed after first-line chemotherapy.¹⁹⁸ The study showed an overall ORR of 29.4% and a DCR of 79.4% (27/34 patients with stable disease or response). The median PFS was 5.1 months, and the median OS was 10.7 months. GBC patients had the highest ORR (38.9%), while 6 iCCA patients did not achieve ORR. This study indicates that trastuzumab combined with FOLFOX demonstrates promising efficacy and manageable safety in second-line treatment of HER2-positive BTC.

In the SGNTUC-019 (NCT 04579380) open-label, single-arm phase II basket study, the efficacy and safety of trastuzumab combined with tucatinib (a HER2-specific TKI) were evaluated in patients with HER2-positive metastatic biliary tract cancer (mBTC).¹⁹⁹ In the BTC cohort (30 previously treated patients), the confirmed ORR (cORR) was 46.7% (14/30), including 1 CR and 13 PRs—one of the highest response rates reported for HER2-targeted BTC therapy. DCR reached 76.7%. Median PFS was 5.5 months (90% CI: 3.9–8.1), median OS was 15.5 months (90% CI: 6.5–16.7), 12-month OS rate was 53.6% (90% CI: 36.8–67.8%), and median DOR was 6.0 months. Grade ≥3 AEs occurred in 60% (18/30) of patients, with cholangitis, decreased appetite, and nausea each accounting for 10%, though most were unrelated to treatment. Only 1 patient discontinued due to interstitial lung disease and liver disorders, with no treatment-related deaths. This study first validated the significant efficacy and manageable safety of tucatinib + trastuzumab in HER2-positive mBTC, offering a new precision therapy direction for this refractory tumor. Two ongoing trials are exploring trastuzumab’s potential in advanced BTC. The HERBOT trial (NCT05749900), an open-label phase I/II study, assesses trastuzumab + gemcitabine + cisplatin (GemCis) in HER2-positive BTC, focusing on dose-escalation tolerability and expansion-cohort ORR. The TRAP-BTC trial (NCT06178445), a randomized phase II study, compares trastuzumab combined with nivolumab vs pembrolizumab in advanced HER2-positive BTC, with PFS as the primary endpoint. These studies will provide critical evidence for HER2-targeted therapy combined with chemotherapy or immunotherapy in BTC.

Bispecific Antibodies

Zanidatamab is a novel humanized biparatopic anti-HER2 IgG1-like antibody that targets two non-overlapping HER2 epitopes simultaneously: ECD4 (trastuzumab binding site) and ECD2 (pertuzumab binding site), forming biparatopic binding. This binding induces HER2 receptor crosslinking and internalization, significantly reducing cell-surface HER2 protein expression, blocking downstream signaling pathways, and inhibiting tumor growth. Compared with trastuzumab and pertuzumab, zanidatamab exhibits 1.3–1.6-fold higher maximal binding capacity to HER2-positive cancer cells, leading to enhanced Fc effector functions, including ADCC, ADCP, and CDC.²⁰⁰ Results from the global multicenter, single-arm phase IIb trial HERIZON-BTC-01 (NCT04466891) showed an ORR of 41.3% (33/80), including 1.3% complete response (CR) and 40% partial response (PR). Median PFS was 5.5 months, and median DOR was 14.9 months. The treatment demonstrated manageable safety, with grade 3 treatment-related adverse events (TRAE) occurring in only 18% of patients, no grade 4 TRAEs observed, and no treatment-related deaths reported.²⁰¹ Based on these data, Zanidatamab was granted accelerated approval by the FDA in November 2024 for second-line treatment of HER2-positive biliary tract cancer, becoming the first bispecific antibody drug indicated for this indication.

Antibody-Drug Conjugates (ADCs)

Targeted HER2 antibody-drug conjugates (ADCs) exert efficacy through three core steps: HER2 receptor binding, internalization-induced cytotoxic drug release, and tumor cell killing.

Fam-Trastuzumab Deruxtecan (T-DXd, Enhertu^®)

Fam-Trastuzumab Deruxtecan (T-DXd, DS-8201a) comprises trastuzumab targeting the HER2 extracellular domain (ECD-IV), linked via a cleavable tetrapeptide linker (Gly-Gly-Phe-Gly) to the topoisomerase I inhibitor DXd (Exatecan derivative) at a 1:8 ratio. With a drug-to-antibody ratio (DAR) of 8—one of the highest among current ADCs.²⁰² DXd is a lipophilic small molecule that induces tumor cell apoptosis by inhibiting DNA repair, exhibiting a bystander effect to penetrate cell membranes and kill adjacent HER2-low/negative tumor cells.²⁰³ Despite its high payload, T-DXd features uniformly distributed DAR and exceptional circulatory stability: after 3 weeks of in vitro culture, <4% DXd is released, demonstrating target-specific inhibition against various HER2-positive cell lines.²⁰⁴ Efficacy and safety were validated in two pivotal trials: HERB (JMA-IIA00423) and DESTINY-PanTumor02 (NCT04482309). The HERB phase II trial (JMA-IIA 00423) investigated T-DXd monotherapy in 32 gemcitabine-refractory/intolerant BTC patients.²⁰⁵ In HER2-positive patients (n=24), ORR was 35.4%, DCR 81.8%, mPFS 4.4 months, and mOS 7.1 months. In HER2-low patients (n=8), ORR was 12.5%, mPFS 4.2 months, and mOS 8.9 months. ctDNA analysis showed HER2-amplified ctDNA-positive patients had an ORR of 50% and mOS of 10.8 months, highlighting liquid biopsy’s prognostic value.²⁰⁶ This first confirmed T-DXd’s efficacy in HER2-positive cholangiocarcinoma among Asian populations.

The DESTINY-PanTumor02 trial (NCT04482309), a basket study validating HER2-targeted therapy across tumor types, showed an overall ORR of 37.1% (including 22% in cholangiocarcinoma patients). HER2-high (IHC3+) subgroups had an ORR of 56.3% and mOS of 12.4 months, significantly outperforming HER2-low (IHC1+/2+) groups. In the cholangiocarcinoma subgroup, mPFS was 4.6 months and mOS 7.0 months.²⁰⁷ These results supported T-DXd as the first ADC approved for HER2-positive cholangiocarcinoma.

Ado-Trastuzumab Emtansine (Kadcyla^®)

Ado-trastuzumab emtansine (T-DM1) consists of Emtansine (DM1), a microtubule inhibitor, conjugated to trastuzumab via an uncleavable thioether linker (MCC) with a DAR of 3.5 (3.5 DM1 molecules per antibody on average).²⁰² T-DM1 specifically binds HER2-overexpressing tumor cells via trastuzumab, releasing DM1 upon internalization. DM1 inhibits tubulin polymerization, blocks cell division, and induces apoptosis, while preserving trastuzumab-mediated ADCC.^208,209 Unlike cleavable ADCs, the released drug does not exert bystander effects after effective ADC internalization and lysosomal degradation.²¹⁰ Although FDA-approved, its primary indications focus on HER2-positive breast cancer, with no FDA approval for cholangiocarcinoma. The phase II basket KAMELEON trial (NCT02999672) evaluating T-DM1 in HER2-positive BTC was prematurely terminated due to recruitment difficulties, yielding suboptimal results. An ongoing phase II single-arm trial (NCT02675829) aims to assess T-DM1 monotherapy in HER2-amplified/mutated advanced solid tumors, including cholangiocarcinoma.

Disitamab Vedotin (Aidexi^®)

Disitamab vedotin (RC-48, DV) comprises Hertuzumab—a humanized anti-HER2 IgG1 mAb with 3.7-fold higher HER2 affinity than trastuzumab—linked via a cleavable Val-Cit linker to monomethyl auristatin E (MMAE), a microtubule inhibitor that blocks cell division by inhibiting microtubule polymerization. DV has an average DAR of 4, carrying four MMAE molecules per antibody. MMAE’s lipophilicity confers bystander effect, enabling it to penetrate cell membranes and kill adjacent HER2-negative cells, thus overcoming tumor heterogeneity.²¹¹ Approved in China for third-line treatment of HER2-positive advanced gastric cancer (ORR 24.8%, mPFS 4.1 months), DV is not FDA-approved, with gallbladder cancer indications under clinical investigation. A phase I study (NCT 04280341) evaluating DV combined with Toripalimab (PD-1 inhibitor) showed that among 24 enrolled “other solid tumor” patients, 2 cholangiocarcinoma patients achieved stable disease (SD), but no partial (PR) or complete response (CR) was observed.²¹² Based on phase I results, DV plus PD-1/PD-L1 inhibitor regimens have entered multiple phase II trials (eg, NCT05540483, NCT05417230) to further validate efficacy in cholangiocarcinoma.

In addition, some new unapproved ADC drugs are under research. MRG002, composed of hyper-fucosylated anti-HER2 mAb (MAB802) conjugated to MMAE via a cleavable linker, with a DAR of 3.8. Preclinical studies show efficacy against trastuzumab-resistant models, outperforming T-DM1.²¹³ Results from the phase I basket trial (CTR20181778) demonstrated an ORR of 43% and DCR of 81% in 25 HER2-positive solid tumor patients, with manageable safety (≥3 grade TRAEs in 28%).²¹⁴ A phase II trial (NCT04837508) in pretreated HER2-positive cholangiocarcinoma has been completed, but results remain unpublished.

GQ1001

A next-generation HER2-targeted ADC using site-specific conjugation to link trastuzumab with DM1 (DAR undefined). Compared to marketed T-DM1, GQ1001 exhibits superior plasma stability, effectively reducing off-target toxicity risk.²¹⁵ The phase I trial (NCT04450732) showed good tolerability in various HER2-positive solid tumors (including cholangiocarcinoma), with ≥3 grade TRAEs in 28.1% of patients; some received treatment for >12 months. Cholangiocarcinoma subgroup data have not been reported separately, though preclinical studies indicate activity in cholangiocarcinoma models.²¹⁶

Trastuzumab Rezetecan (SHR-A1811)

Trastuzumab conjugated to topoisomerase I inhibitor SHR9264 via a cleavable linker (DAR 5.7), featuring a potent bystander effect.²¹⁷ The phase I trial (NCT04446260) in 98 HER2-expressing/mutated solid tumor patients (16 cholangiocarcinoma) showed an overall ORR of 45.9%, with the cholangiocarcinoma subgroup achieving an ORR of 56.3% (9/16) and a 6-month PFS rate of 52.1%. Preliminary data suggest significant efficacy in HER2-positive cholangiocarcinoma, though sample size is small. High response rates and bystander effects may overcome HER2 heterogeneity, but large-scale trials are needed to validate safety and survival benefits.²¹⁸ These ADCs provide new treatment strategies for cholangiocarcinoma through innovative designs (eg, high DAR, potent bystander effect), but their clinical translation requires more data support.

TKI

Pyrotinib (Irene^®)

Pyrotinib (SHR1258) is an irreversible EGFR/HER2 dual-target tyrosine kinase inhibitor (TKI) that covalently binds to the intracellular tyrosine kinase domain of HER2, blocking downstream signaling pathways (eg, PI3K/AKT and MAPK/ERK) to inhibit tumor cell proliferation and survival. Approved in China for second-line treatment of HER2-positive advanced breast cancer, preclinical in vitro studies show significant proliferative inhibition of HER2-high cholangiocarcinoma cell lines (eg, KMBC, TFK-1). In HER2-amplified cholangiocarcinoma xenograft mouse models, monotherapy or combination with chemotherapy (eg, gemcitabine) significantly reduces tumor volume.²¹⁹ The phase I basket trial NCT02500199 evaluated Pyrotinib monotherapy in HER2-mutated/amplified solid tumors (including cholangiocarcinoma). Results showed an ORR of 19% and DCR of 68% in non-breast cancer solid tumors (including 1 cholangiocarcinoma patient).²²⁰ A phase II trial (NCT04571710) of Pyrotinib monotherapy in pretreated HER2-positive cholangiocarcinoma has been completed, pending results. Two case reports demonstrate promising efficacy: a HER2-overexpressing cholangiocarcinoma patient achieved PR with 15-month PFS under second-line Pyrotinib monotherapy, with good tolerance;²²¹ a 64-year-old patient with HER2-positive advanced intrahepatic cholangiocarcinoma achieved PR and 17-month PFS after failing first-line gemcitabine + cisplatin, treated with pyrotinib plus pembrolizumab and lenvatinib combination.²²²

Neratinib (Nerlynx^®)

Neratinib is an irreversible pan-HER tyrosine kinase inhibitor (TKI) targeting HER1 (EGFR), HER2, and HER4, inhibiting receptor phosphorylation to block downstream pathways (eg, PI3K/AKT, MAPK/ERK) and tumor proliferation. FDA-approved for adjuvant treatment of HER2-positive breast cancer.²²³ the phase II basket SUMMIT trial (NCT01953926) evaluated Neratinib monotherapy in HER2-mutated solid tumors (including cholangiocarcinoma). The cholangiocarcinoma cohort included 25 pretreated patients (10 gallbladder, 11 cholangiocarcinoma, 4 ampullary cancer); excluding 6 early discontinuations, ORR was 16% (4 PRs), mPFS 2.8 months (3.7 months for gallbladder cancer, 1.4 months for cholangiocarcinoma), and mOS 5.4 months.²²⁴ The phase III SAFIR-ABC10 trial (NCT05615818) is ongoing, comparing maintenance therapy with targeted agents (including Neratinib) vs continued chemotherapy in patients with disease control after first-line chemotherapy.

Lapatinib (Tykerb^®)

Lapatinib is a reversible dual tyrosine kinase inhibitor (TKI) targeting EGFR (HER1) and HER2 (ERBB2), blocking intracellular kinase domain phosphorylation to inhibit downstream pathways (eg, PI3K/Akt, MAPK/ERK) and tumor cell viability. Approved for second-line treatment of HER2-positive breast cancer (in combination with chemotherapy or other targeted agents), it is not approved for cholangiocarcinoma. Preclinical studies show significant proliferation inhibition in HER2-overexpressing cholangiocarcinoma cell lines; Lapatinib monotherapy is effective against early-stage but not advanced tumors in xenograft models, while combination with gemcitabine shrinks tumor volume and prolongs survival.^225,226 However, two monotherapy trials (NCT00107536, NCT00101036) showed 0% ORR and poor efficacy.^227,228 Combination therapy case reports show better outcomes: a HER2-mutated gallbladder cancer patient achieved PR with Lapatinib + sirolimus (mTOR inhibitor),²²⁹ and a HER2-mutated cholangiocarcinoma patient had 7.1-month SD with Lapatinib + trastuzumab.²³⁰

HER2-targeted therapy provides breakthrough options for cholangiocarcinoma, with ADCs and bispecific antibodies initially outperforming traditional chemotherapy. Future directions include precision subtyping, combination strategies, and new drug development to expand the beneficiary population, while addressing resistance and toxicity. With advancing phase III trials and testing standardization, HER2-targeted therapy is poised to become a first/second-line standard for cholangiocarcinoma. Clinical trials for this target are summarized in (Table 1).

v-Raf Murine Sarcoma Viral Oncogene Homolog B1 (BRAF)

BRAF mutations can be detected in a small number of patients with biliary tract malignancies, with an incidence of approximately 5%. These mutations mainly occur in iCCA, and about 80% of them are of the V600E type.^21,231,232 BRAF is a key kinase in the RAS-RAF-MEK-ERK signaling pathway (MAPK pathway), responsible for transducing extracellular growth signals to the nucleus and regulating cell proliferation, differentiation, survival, and metabolism (Figure 7).²³³ Under normal conditions, this pathway is transiently activated to promote physiological growth, while BRAF mutations (especially V600E) lead to constitutive activation, driving tumorigenesis.²³⁴ The substitution of valine (Val) with glutamic acid (Glu) at position 600 of the BRAF protein (V600E) results in more than a 500-fold increase in kinase activity. The mutant does not require activation by upstream RAS proteins and continuously phosphorylates downstream MEK and ERK, thereby causing uncontrolled cell proliferation.

Figure 7 BRAF V600E mutation and targeted inhibition in the MAPK pathway. The V600E mutation causes constitutive BRAF activation; dual inhibition with dabrafenib (BRAF inhibitor) and trametinib (MEK inhibitor) blocks signaling and suppresses tumor cell proliferation and survival.235

The efficacy of BRAF inhibitor monotherapy in CCA is limited, which may be due to drug resistance caused by reactivation of the MAPK pathway through EGFR signal feedback.²³⁶ In the ACSE basket trial (NCT02304809), vemurafenib (a selective inhibitor of BRAF V600E kinase) showed an ORR of only 18.2% in the cholangiocarcinoma cohort.²³⁷ In contrast, dual inhibition of BRAF and MEK has achieved better efficacy. The phase II ROAR basket trial (NCT02034110) evaluated dabrafenib (a BRAF inhibitor) combined with trametinib (a MEK inhibitor) in BRAF V600E-mutated advanced solid tumors, including cholangiocarcinoma. Results in the cholangiocarcinoma cohort (N=43) showed an ORR of 58.1% (25/43), including 3 CRs. Median PFS was 9.0 months (95% CI: 5.5–9.4), median OS was 13.5 months (95% CI: 10.4–17.6), and median DoR was 8.9 months, with some patients maintaining responses for over 1 year. Common adverse events included fever (56%), fatigue (44%), rash (37%), and diarrhea (33%); grade ≥3 adverse events included hypertension (14%) and neutropenia (9%). These were effectively controlled through dose adjustment and supportive care, with no treatment-related deaths reported.²³⁸ Based on the significant benefits in ORR and DoR, the FDA approved this combination therapy in June 2022 for advanced solid tumors with BRAF V600E mutation (including cholangiocarcinoma) when no alternative treatments are available, making it the first combined targeted regimen for this mutation.

Another phase I trial (NCT04190628) is ongoing, aiming to evaluate the safety, tolerability, pharmacokinetics, and preliminary efficacy of ABM-1310 monotherapy and its combination with MEK inhibitor (Cobimetinib) in advanced solid tumors with BRAF V600E mutation.²³⁹ ABM-1310 is a novel BRAF inhibitor that targets BRAF V600E mutant protein, blocks the MAPK signaling pathway, and inhibits tumor proliferation. Its molecular structure has been optimized to effectively penetrate the blood-brain barrier, potentially exerting efficacy on brain metastases. In addition, studies evaluating the synergistic effects of BRAF/MEK inhibitors combined with PD-1 inhibitors in immunotherapy combinations are also underway. Clinical trials targeting this target are summarized in (Table 1).

Neurotrophic Tyrosine Receptor Kinase (NTRK)

The NTRK gene encodes tropomyosin receptor kinase (TRK), a membrane-bound receptor whose activated state functionally couples with the mitogen-activated protein kinase (MAPK) signaling pathway.²⁴⁰ The NTRK gene encodes tropomyosin receptor kinase (TRK), a membrane-bound receptor whose activated state functionally couples with the mitogen-activated protein kinase (MAPK) signaling pathway.²⁴¹ Larotrectinib and entrectinib are two FDA-approved TRK inhibitors.²⁴² In three clinical trials (NAVIGATE, SCOUT, LOXO-TRK-14001) including 339 NTRK fusion solid tumor patients, larotrectinib showed an ORR of 75%, median PFS of 28.3 months, and median OS of 44.4 months. In the cholangiocarcinoma subgroup (predominantly colorectal cancer, with 4 cholangiocarcinoma patients), ORR was 44%.²⁴³ The ALKA-372-001, STARTRK-1, and STARTRK-2 studies showed entrectinib achieved an ORR of 69.2% in Asian patients with NTRK fusion solid tumors, with median PFS of 14.9 months and median OS of 13.5 months. In cholangiocarcinoma, ORR was 53%, and median PFS was 9 months.²⁴⁴ Current research on NTRK-targeted therapy for cholangiocarcinoma focuses on developing next-generation inhibitors and evaluating combination therapy efficacy, with related trials ongoing. Clinical trials for this target are summarized in (Table 1).

V-Ki-ras2 Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS)

KRAS gene mutation leads to loss of its encoded GTPase activity, causing KRAS protein to persistently bind GTP and remain in an activated state, thereby promoting cell proliferation, survival, and metastasis via downstream MAPK, PI3K-AKT, and other signaling pathways.²⁴⁵ In BTC, KRAS gene mutation occurs in approximately 20% of cases, but only KRAS G12C mutation is currently the only targetable mutation subtype among the gene’s various variants, accounting for ~5% of all KRAS mutations (ie, 1% of BTC).²⁴⁶ The inhibitor adagrasib shows promising antitumor efficacy in clinical trials. It covalently binds to the GTP-binding pocket of KRAS G12C mutant protein, locking mutant KRAS in an inactive state and blocking downstream MAPK and PI3K-AKT signaling to inhibit tumor cell proliferation and metastasis.²⁴⁷ The phase I/II trial KRYSTAL-1 (NCT03785249) showed an overall ORR of 35.1%, with an ORR of 41.7% in biliary tract cancer, median PFS of 8.6 months, and median OS of 15.1 months.²⁴⁸ Based on KRYSTAL-1 data, the 2024 NCCN guidelines recommended adagrasib for second-line treatment of cholangiocarcinoma. Another inhibitor, sotorasib (AMG-510), demonstrates activity against KRAS G12C mutations in gastrointestinal tumors, but data in cholangiocarcinoma remain limited and require further clinical validation. Targeted drugs for other KRAS mutation types (eg, G12D), such as AST2169 liposomes and PROTAC technology, are under development.²⁴⁹ And other combination treatment methods are also in clinical trials. Clinical trials for this target are summarized in (Table 1).

Vascular Endothelial Growth Factor (VEGF)

VEGF promotes endothelial cell proliferation, migration, and survival by binding to VEGFR2 (KDR), activating downstream pathways (eg, PI3K-AKT, RAS-MAPK) to form immature tumor vascular networks that nourish tumors and support metastasis. VEGF is overexpressed in 42–76% of BTC cases.²⁵⁰ Compared with other BTC subtypes, intrahepatic CCAs are associated with high VEGF-A expression and increased microvessel density.²⁵¹ Bevacizumab, a recombinant humanized anti-VEGF mAb, neutralizes VEGF-A to block receptor binding and inhibit tumor angiogenesis. While monotherapy efficacy is limited, combination therapy demonstrates significant antitumor activity. The COMBATBIL imCORE trial (ESMO 2024) evaluated bevacizumab combined with atezolizumab (PD-L1 inhibitor) and mFOLFOX6 chemotherapy as second-line treatment for advanced BTC progressing after first-line therapy. Among 35 enrolled patients, ORR was 31.4%, median OS was 13.8 months—significantly exceeding traditional chemotherapy expectations—with good tolerability and no new severe AEs. Tovecimig, a bispecific antibody, simultaneously blocks DLL4 and VEGF-A signaling to inhibit tumor angiogenesis and promote vascular normalization, enhancing chemotherapeutic penetration and immune cell infiltration. Its efficacy in cholangiocarcinoma is under clinical validation. Current VEGF-targeted research for cholangiocarcinoma focuses on dual-target agents and combination therapies, aiming to improve outcomes through multi-mechanistic synergy. Clinical trials for this target are summarized in (Table 1).

Rearranged During Transfection (RET)

The rearranged during transfection (RET) gene encodes a receptor tyrosine kinase, whose constitutive activation of the kinase domain drives tumorigenesis. RET gene fusion, an important oncogenic driver event, has been identified in multiple malignancies, including ~1% of biliary tract cancer (BTC) cases.^252,253 The global multicenter phase I/II LIBRETTO-001 trial showed that oral RET inhibitor selpercatinib exhibits activity in RET fusion-positive solid tumors (including BTC), with an overall ORR of 43.9%, median duration of response (DoR) of 24.5 months, median PFS (mPFS) of 13.2 months, and median OS (mOS) of 18 months.²⁵⁴ FDA granted accelerated approval to selpercatinib in 2022 for RET fusion-positive locally advanced or metastatic solid tumors, excluding cholangiocarcinoma. The 2024 NCCN guidelines list selpercatinib as a second-line treatment option for RET fusion-positive cholangiocarcinoma. The multicenter phase I/II ARROW trial evaluated RET inhibitor praseltinib in RET-mutated solid tumors, including cholangiocarcinoma patients, showing an ORR of 57%, median PFS of 7 months, and median OS of 14 months.²⁵⁵ Although RET represents a rare driver in cholangiocarcinoma, its well-defined oncogenic mechanism and existing clinical efficacy data position it as an important direction for precision medicine. Future efforts should expand clinical trial scale and optimize combination strategies to ultimately provide better treatment options for RET-altered cholangiocarcinoma patients.

Ring Finger Protein 43 (RNF43)

RNF43 is a transmembrane E3 ubiquitin ligase that inhibits Wnt/β-catenin pathway activity by ubiquitinating the Wnt receptor Frizzled (FZD), promoting its lysosomal degradation.²⁵⁶ Loss-of-function mutations in RNF43 lead to FZD receptor accumulation, causing excessive Wnt pathway activation that drives tumor cell proliferation and invasion. RNF43 mutations are rare in BTC (<5%). A study has evaluated the safety and activity of Wnt inhibitors in previously treated advanced solid tumor patients with RNF43 mutations (NCT 03447470). However, research on RNF43-targeted therapy for cholangiocarcinoma remains in the early stage, and future efforts should accelerate drug development through multicenter clinical trials.

c-Met Inhibitors

The MET gene encodes the hepatocyte growth factor receptor (c-Met), a high-affinity receptor for hepatocyte growth factor (HGF). Upon binding, HGF/c-Met axis activation initiates intracellular signaling that regulates cell growth, migration, differentiation, and tissue repair. In cancer cells, aberrant HGF/c-Met axis activation promotes tumor progression by stimulating PI3K/AKT/mTOR, MAPK, and STAT pathways.^257,258 c-MET overexpression in BTC is associated with poor prognosis, detected in 12–25% of intrahepatic CCAs and 0–16% of extrahepatic CCAs.^259,261 A case report showed symptom relief in a patient with refractory metastatic cholangiocarcinoma carrying CAPZA2-MET fusion and amplification treated with capmatinib (a highly selective MET inhibitor).²⁶² Another case study reported favorable efficacy in tumor markers, size, and clinical assessment with good safety in an intrahepatic cholangiocarcinoma patient with RBPMS-MET fusion treated with MET inhibitor crizotinib.²⁶³ Although clinical research on MET inhibitors in cholangiocarcinoma remains in the early stage, significant potential has been demonstrated. Future multicenter clinical trials are needed to validate efficacy and promote clinical translation of MET-targeted therapy.

Immunotherapy

Immune Checkpoint Inhibitors

Immune Checkpoint Inhibitors (ICIs) are a class of antibody drugs targeting immune checkpoint proteins, including cytotoxic T lymphocyte antigen-4 (CTLA-4) and programmed death-1/programmed death ligand-1 (PD-1/PD-L1), which can suppress antitumor immune responses in solid tumors.²⁶⁴ PD-L1 expression in intrahepatic CCAs is as high as 30–40%.²⁶⁵ Microsatellite Instability-High (MSI-H) refers to length abnormalities in DNA microsatellite sequences due to unrepaired replication errors, defined when ≥2 microsatellite loci are unstable. Mismatch Repair Deficiency (dMMR) occurs due to mutations or epigenetic inactivation of MMR genes (MLH1, MSH2, MSH6, PMS2), leading to loss of repair protein expression and inability to correct DNA replication errors, thereby inducing a high genomic mutation phenotype. dMMR is highly consistent with MSI-H, often collectively referred to as MSI-H/dMMR, which increases tumor-associated antigen expression.²⁶⁶ MSI-H/dMMR is observed in 1–10% of advanced BTC cases.^32,267,268 The KEYNOTE-158 study evaluated the efficacy and safety of the PD-1 inhibitor pembrolizumab in various advanced solid tumors.²⁶⁹ In the cholangiocarcinoma cohort, ORR was 5.8%, median PFS 2.0 months, and median OS 7.4 months; the MSI-H/dMMR subgroup showed an ORR of 40.9%, median PFS of 4.2 months, and median OS of 24.3 months. This supports pembrolizumab as effective therapy for MSI-H/dMMR tumor patients. Cholangiocarcinoma patients with Tumor Mutational Burden-high (TMB-H) also demonstrate good response to pembrolizumab, which has been FDA-approved for second-line treatment of metastatic MSI-H/dMMR or TMB-H solid tumors (including cholangiocarcinoma).²⁷⁰

CTLA-4 inhibitor monotherapy has limited efficacy, with current cholangiocarcinoma immunotherapy research focusing on combination strategies. The CA209-538 prospective multicenter phase 2 non-randomized trial enrolled advanced rare cancer patients, including those with biliary tract cancer, using a dual immunotherapy strategy of PD-1 inhibitor nivolumab + CTLA-4 inhibitor ipilimumab. Subgroup analysis showed a disease control rate (DCR) of 44% (17 patients), ORR of 23% (9 patients), median OS of 5.7 months, and median PFS of 2.9 months.²⁷¹ Intrahepatic cholangiocarcinoma and gallbladder cancer patients had higher response rates, outperforming PD-1 inhibitor monotherapy, though sample size was small (n=39), requiring further study of long-term efficacy and biomarker predictive value.

Additionally, the TOPAZ-1 and KEYNOTE-966 studies confirmed that PD-L1/PD-1 inhibitor combined with chemotherapy significantly prolongs overall survival. TOPAZ-1, a phase III study evaluating gemcitabine + cisplatin plus PD-1 inhibitor durvalumab or placebo, showed median OS: 12.9 months vs 11.3 months in the placebo group (HR=0.74, 26% reduction in death risk), and 3-year OS rates: 14.6% vs 6.9% (112% improvement), demonstrating robust and sustained OS benefit with good safety.²⁷² KEYNOTE-966 (NCT04003636), a global multicenter, randomized double-blind, placebo-controlled phase III trial, evaluated gemcitabine + cisplatin plus PD-1 inhibitor pembrolizumab vs chemotherapy alone for first-line treatment of advanced/non-resectable BTC. Following TOPAZ-1, this was the second phase III study confirming survival improvement with immune-combined chemotherapy, showing median OS: 12.7 months vs 10.9 months (HR=0.83, 17% reduction in death risk), 3-year OS rates: 13% vs 11% (no significant difference), and median DOR in the experimental group of 8.3 months, significantly longer than 6.9 months in the control group (12-month durable response rate 38% vs 27%).²⁷³ Based on KEYNOTE-966, the pembrolizumab + chemotherapy regimen was approved by the FDA in 2023 for first-line treatment of advanced BTC. Both studies establish the role of immune-combined chemotherapy in first-line advanced biliary tract cancer, with TOPAZ-1 data more strongly supporting it as the preferred regimen.

Adoptive Cell Therapy

Adoptive Cell Therapy (ACT) is a treatment approach that involves extracting, expanding, or genetically engineering a patient’s own or donor-derived immune cells, then reinfusing them to enhance antitumor immune responses. Its core mechanism utilizes optimized immune cells to precisely recognize and kill tumor cells while minimizing damage to normal tissues.²⁷⁴ In BTC, research in this field remains in the early stage, primarily including three therapies: tumor-infiltrating lymphocytes (TILs), chimeric antigen receptor T (CAR-T) cells, and cytokine-induced killer (CIK) cells.

TILs are T cells isolated from tumor tissue, expanded in vitro, and reinfused. Studies suggest TILs may have positive effects in patients with locally advanced or distant metastatic BTC. An iCCA patient with lymph node metastasis and portal vein invasion received CD3-activated T cells and tumor lysate- or peptide-pulsed dendritic cells immunotherapy after surgery, remaining relapse-free for 3 years and 6 months.²⁷⁵ Another clinical trial using TILs in patients with ERBB2 interacting protein mutations also showed significant inhibition of BTC.²⁷⁶ A case-control adjuvant study explored the efficacy of dendritic cell vaccine + TIL in long-term survival and recurrence prevention for post-surgical iCCA patients, showing median PFS (mPFS) and median OS (mOS) of 18.3 and 31.9 months in 36 patients receiving adjuvant immunotherapy, versus 7.7 and 17.4 months in 26 surgery-alone patients.²⁷⁷

CAR-T cells are genetically engineered T cells expressing chimeric antigen receptors (CARs) for specific tumor antigen recognition. In a phase I trial of HER2-targeted CAR-T cell therapy, 1 out of 9 advanced BTC patients showed partial response, and 3 had stable disease (NCT01935843).²⁷⁸ Another phase I trial of EGFR-targeted CAR-T cell therapy in advanced biliary tract cancer showed 1 complete response and 10 patients with disease control among 17 enrolled (NCT01869166).²⁷⁹ Suimon et al constructed a fourth-generation CAR-T cell carrying an anti-MUC1 single-chain antibody fragment, demonstrating significant destructive and direct cytotoxic effects on cholangiocarcinoma cells (CCA) via specific MUC1 antigen recognition, confirming MUC1’s potential clinical value as a therapeutic target for cholangiocarcinoma.⁴⁷ An ongoing phase I/II trial (NCT03633773) aims to evaluate the safety and antitumor efficacy of MUC1-targeted CAR-T cell therapy in intrahepatic cholangiocarcinoma (ICC) patients, with results pending.

CIK cells are generated by activating peripheral blood lymphocytes ex vivo to enhance non-specific antitumor activity, then reinfused for antitumor effects. A study showed that among 85 BTC patients treated with CIK combined with dendritic cell vaccine, 2 had complete responses, 14 had partial responses, and 54 had stable disease.²⁸⁰ ACT provides new treatment options for advanced BTC patients, but most studies remain in early stages. High treatment costs and technical barriers limit clinical popularization, requiring technological innovation and clinical validation to overcome current limitations for broader application.

Cancer Vaccines

Cancer vaccines, which activate the patient’s own immune system to induce specific antitumor immune responses, represent an important direction in immunotherapy for biliary tract cancer (BTC). Current tumor vaccines are primarily divided into polypeptide vaccines, dendritic cell (DC) vaccines, and nucleic acid vaccines.

Polypeptide vaccines utilize short peptides of tumor-associated antigens (TAAs) or tumor-specific neoantigens to directly present to T cells, activating CD8⁺ and CD4⁺ T cell responses—for example, targeting antigens highly expressed in cholangiocarcinoma such as WT1 and MUC1. Wilms tumor protein 1 (WT1) mutations and mucin 1 (MUC1) overexpression occur in 80–90% of BTC cases.²⁸¹ In a phase I trial (16 advanced BTC patients), WT1 vaccine combined with gemcitabine led to disease stability ≥2 months in 8 patients.²⁸² However, an early study showed limited efficacy of MUC1 vaccine in advanced cholangiocarcinoma.²⁸³ Recent research has shifted to multi-target vaccines: a phase II trial of a tripeptide vaccine targeting VEGFR1, VEGFR2, and KIF20A induced peptide-specific T cell responses in 4 of 6 patients.²⁸⁴

DC vaccines involve loading tumor antigens onto DCs ex vivo, which are then reinfused to activate T cell antitumor immunity.²⁸⁵ DCs loaded with lysates from heat-shocked GBC cell lines can induce activation of patient CD4⁺ and CD8⁺ T cells.²⁸⁶ The combination of DC vaccine and TIL therapy has been described previously.

Nucleic acid vaccines deliver nucleic acid sequences encoding tumor antigens, inducing host cells to express antigens and activate specific antitumor immune responses. In a study screening cholangiocarcinoma tumor antigens and immune subtypes, three tumor antigens—CD247, FCGR1A, and TRRAP—were identified as potential targets for mRNA vaccine development.⁵⁰

Conclusions

The incidence of biliary tract cancer (BTC) is on the rise with poor prognosis, while the emergence of targeted therapy has brought new hope for its treatment. Under the precision medicine strategy, research and drug development for different molecular targets have made significant progress.

Targets such as IDH mutations and FGFR2 fusions have been widely studied, with related inhibitors approved or in clinical trials, demonstrating promising efficacy. HER2-targeted therapies are diverse, with ADCs and bispecific antibodies initially outperforming traditional chemotherapy in cholangiocarcinoma treatment. Targets like BRAF, NTRK, and KRAS also have corresponding treatment strategies, with some achieving favorable outcomes.

In immunotherapy, immune checkpoint inhibitors are effective in specific populations (eg, MSI-H/dMMR, TMB-H patients), while adoptive cell therapy and cancer vaccines, though in the early research stage, offer new treatment options for patients. Combining targeted therapy with immunotherapy, leveraging the interactions and crosstalk of various signaling pathways in tumors and the tumor microenvironment, may be an effective approach for BTC treatment.

However, targeted therapy for biliary cancer still faces numerous challenges. Tumor heterogeneity leads to significant differences in treatment responses among patients, and drug resistance limits therapeutic efficacy. In the future, it is necessary to further deepen the study of tumor molecular mechanisms, screen optimal populations through precision subtyping, develop highly selective drugs, optimize combination therapy strategies, and strengthen the exploration and overcoming of resistance mechanisms to improve the survival rate and quality of life of BTC patients, and promote the clinical application and development of targeted therapy for biliary cancer.