Omics-driven networks herald a new era of multidimensional therapy for cholangiocarcinoma

For too long, cholangiocarcinoma (CCA), a formidable and aggressive cancer of the bile ducts, has presented a grim prognosis, leaving patients and clinicians with limited therapeutic options. Characterized by its insidious nature and often late-stage diagnosis, CCA has stubbornly resisted significant advancements in treatment. Surgical resection benefits only a small subset of patients with CCA. The landmark ABC-02 trial in 2010 established cisplatin plus gemcitabine (GEMCIS) as the non-surgical standard of care. Since 2020, however, the advent of targeted and immunotherapeutic strategies has rapidly reshaped the treatment paradigm. Within 5 years, 3 FGFR2-directed tyrosine-kinase inhibitors—pemigatinib, infigratinib, and futibatinib—have been approved for CCA harboring FGFR2 fusions or rearrangements. Additional approvals during the same period include the IDH1 inhibitor ivosidenib, the BRAF V600E-directed combination of dabrafenib plus trametinib, the HER2-targeted antibody–drug conjugate zanidatamab, and checkpoint-inhibitor regimens combining GEMCIS with either nivolumab or pembrolizumab. Collectively, these advances provide new therapeutic avenues for patients who derive limited benefit from the traditional "surgery + chemotherapy" approach.1 Likewise, the majority of ongoing clinical trials in CCA are being designed around the core modalities of targeted therapy, cytotoxic chemotherapy, and immunotherapy. The highly heterogeneous nature of CCA is reflected in its multi-level and multidimensional aberrant regulation. Multi-omics approaches are precisely suited to unraveling these molecular changes from various angles, connecting previously considered "independent" biological processes into an integrated network. The advent of artificial intelligence further promises to capture the most critical driving events within this network. These profound transformations are prompting a shift in CCA's treatment paradigm from "one-size-fits-all" to precision individualized therapy. The therapeutic scope is deepening from macro-level phenotypes to the micro-molecular level, and the treatment range is extending from the tumor cells themselves to targeting the interactions between tumor cells and their microenvironment. Molecular diagnostics for CCA are now explicitly recommended in both the NCCN and ESMO guidelines, and drugs targeting genomic drivers have entered routine clinical practice. Nevertheless, fewer than half of patients harbor actionable genomic events. FGFR2 fusions or rearrangements are seen in only 9%–15% of intrahepatic cholangiocarcinoma (ICC), IDH1 mutations in 10%–20%, and they are rare in extrahepatic cholangiocarcinoma (ECC); HER2 overexpression or amplification is reported in merely 5%–20% of CCA, and the prevalence of most other mutations is below 5%. Hence, multi-omics efforts that look beyond the genome are essential for uncovering alternative drivers and for developing therapies that either enhance current regimens or benefit a broader patient population. Wu et al2 used a multi-omics framework to show that IDH-mutant tumors display an immunologically "cold" microenvironment; pharmacological IDH1 inhibition rapidly recruits CD8+ T cells, converting the milieu to "hot" and thereby potentiating antitumor immunity. Combining IDH1 inhibitors with checkpoint blockade could therefore amplify therapeutic benefit. In another study, Liao et al3 performed whole-genome methylation profiling of ICC and stratified patients into 4 prognostic subtypes (S1–S4), and found that chemotherapy, immunotherapy, and IDH/FGFR2 inhibitors appeared most effective in the good-prognosis S1 and S4 groups, whereas the poor-prognosis S2 and S3 groups lacked canonical genomic targets; here, novel strategies against aberrant GBP4 methylation may hold greater promise. Notwithstanding these advances, omics-driven discovery outside the genome remains nascent. Future work should prioritize therapies tailored to specific non-genomic aberrations. Another promising strategy for increasing objective response rates and decreasing recurrence rates involves combination therapeutic strategies based on multi-omics approaches.4 Consequently, the integration of multi-omic mechanism-based combination drug therapies with local treatments has become a prominent area of research. However, the mechanisms of synergistic drug interactions and the controllability of adverse reactions still require further elucidation. Beyond the limited breadth of available therapeutic modalities, drug resistance constitutes a second major challenge in the management of CCA. Once resistance emerges, salvage options are scant, and patients who exhibit primary refractoriness to first-line therapy typically garner even less benefit from subsequent lines. Accordingly, pre-treatment multi-omics screening to identify and comprehensively characterize resistant disease—followed, within the limits of tolerability, by multidimensional and intensive regimens—may induce deeper responses before high-level resistance becomes entrenched, a strategy of particular importance for patients undergoing downstaging or awaiting liver transplantation. When resistance does arise, tumors should again be profiled from both molecular and microenvironmental perspectives. As in triple-negative breast cancer, where therapeutic avenues are similarly constrained, molecular subtyping informed by integrated omics may offer a foothold for breakthrough strategies. At the genomic level, early evidence links resistance to FGFR2 inhibitors with emergent secondary mutations; distinct agents select for different mutation spectra, highlighting the necessity of real-time molecular reassessment after resistance develops.5 Encouragingly, lenvatinib has demonstrated potential efficacy in patients with FGFR2-altered disease who experience either adverse events or secondary resistance to prior FGFR2-targeted therapy. Moreover, whole-genome sequencing has revealed that mesenchymal–epithelial transition factor (MET) rearrangements do occur in CCA and that secondary mutations within the MET kinase domain can drive resistance to MET inhibition. MET inhibitors themselves are generally divided, on the basis of binding mode and structural features, into type I (subclassified as Ia and Ib) and type II compounds. The type Ib agent vebreltinib has shown pronounced activity in MET-fusion–positive tumors, whereas resistance substitutions—such as D1228V—can be overcome with type II inhibitors like cabozantinib, broadening the palette of options for follow-on combination strategies.6 From a tumor-microenvironment (TME) perspective, Lu et al7 evaluated the immune landscape that emerges after resistance to a quadruple regimen comprising gemcitabine, oxaliplatin, lenvatinib, and PD-1 blockade, and demonstrated that CTLA-4 inhibition can reprogram this resistance-associated immunosuppressive niche—providing preliminary proof-of-concept that microenvironment-targeted interventions can overturn drug resistance. Notably, the uptake of molecular profiling and the prevailing spectra of actionable mutations differ markedly between ECC and ICC, underscoring the need to tailor therapeutic strategies to these 2 anatomical subtypes.8 Beyond enabling the precision deployment of already-approved therapies, multi-omic profiling can also catalyze the development of leading-edge interventions—tumor vaccines, chimeric antigen receptor T cells (CAR-T), and NK cell therapies, and antibody–drug conjugates (ADCs)—for CCA. In contrast to traditional chemotherapy, which exerts a broad, indiscriminate cytotoxic effect, or classical targeted/immunotherapeutic approaches directed at a defined subgroup, these frontier modalities demand a far more granular characterization of each patient and tumor at the molecular level. At the bench, the discovery of novel targets, identification of synergistic sensitizing agents, and design of resistance-bypassing strategies rely on a panoramic view of molecular changes across the continuum of tumor evolution. Clinically, the engineering of cellular therapies, the selection of microenvironment-specific adjuvants, the mechanistic evaluation of acquired resistance, and the rational choice of later-line regimens likewise hinge on continuous, system-wide molecular assessment. CCA vaccines can be designed against protein neoantigens or their corresponding messenger RNAs.9 Comprehensive neoantigen discovery is foundational, and clarifying the relationship between neoantigen mRNA and its translated protein is also critical. Multi-omic approaches uniquely bridge these 2 layers, furnishing the foundational data needed to refine neoantigen-prediction algorithms. ADCs and CAR-T/NK cell therapies have already been shown to improve outcomes in EGFR-positive, late-stage CCA.10 Yet, grades 3–4 toxicities arising from off-target effects, along with the spatiotemporal heterogeneity of target-antigen expression within tumor tissue, continue to undermine both safety and efficacy. Consequently, a comprehensive and dynamic appraisal of the TME and adjacent normal tissues—leveraging spatial transcriptomics, spatial proteomics, and related platforms—is essential so that, even before treatment commences, therapeutic benefit can be forecast and adverse events anticipated. Multi-omics approaches possess inherent and unique advantages in revealing immunosuppressive factors within the TME and elucidating complex mechanisms of drug resistance. By integrating diverse data types such as genomics, transcriptomics, proteomics, and metabolomics, multi-omics enables a comprehensive and in-depth exploration of key immunosuppressive elements, immune evasion pathways, and resistance-associated biomarkers within the TME. This, in turn, facilitates the optimization of cancer therapies and enhances treatment efficacy. Finally, patients who develop severe toxicities should likewise undergo deep multi-omic interrogation, including functional and molecular profiling of the affected organs, to elucidate the mechanisms driving adverse reactions and inform subsequent management. Translating omics research into clinical applications necessitates new demands on the research itself. Firstly, multi-omics subtyping studies should be organically integrated with the development of corresponding therapies, including novel treatments such as ADC, CAR-T/NK, tumor vaccine, and oncolytic virus. Multi-omics plays a crucial role throughout the entire process of developing new therapies, from the initial screening of therapeutic targets and in vivo drug efficacy evaluation to monitoring drug-host co-evolution, tracking changes in the TME, developing efficacy biomarkers, patient stratification, and optimizing therapies across multiple levels. In clinical applications, this translates into a dynamic closed loop: in-course phenotyping informs the selection of the optimal therapy, followed by monitoring and re-phenotyping to continuously adapt and refine treatment strategies. Secondly, existing molecular classifications should be integrated with cutting-edge methodologies, such as artificial intelligence, to establish a comprehensive mapping from clinical subtypes to specific therapeutic strategies. Finally, reducing technological costs is crucial to ultimately ensure that these advancements genuinely benefit patients (Figure 1).FIGURE 1: The role of multi-omics in overcoming therapeutic dilemmas in CCA. Abbreviation: CCA, cholangiocarcinoma.