Molecularly guided therapy in gastrointestinal malignancies: From established targets to resistance and emerging precision strategies

Abstract

Gastrointestinal (GI) malignancies comprise a heterogeneous group of cancers in which molecular alterations increasingly guide therapeutic selection. Biomarker-driven therapies are established in selected colorectal, gastric and gastroesophageal junction, biliary, pancreatic, and GI stromal tumors. However, primary and acquired resistance, tumor heterogeneity, evolving molecular profiles, and unequal access to molecular diagnostics limit durable benefit. This narrative review evaluates the current landscape of molecularly targeted therapies across major GI malignancies, focusing on validated molecular targets, clinical evidence, biomarker selection, resistance mechanisms, and emerging therapeutic strategies. A structured literature search was conducted in PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar for publications from January 2000 to 30 July 2026. Evidence was narratively synthesized according to tumor type and molecular target and graded as A (randomized phase III evidence or established standard of care), B (prospective phase II or substantial clinical evidence), C (early-phase, retrospective, or limited clinical evidence), or D (preclinical or experimental evidence). Molecularly targeted therapy is now an established component of treatment for selected molecular subgroups. Key advances include genotype-directed therapy in colorectal cancer and cholangiocarcinoma, human epidermal growth factor receptor 2- and claudin 18.2-directed approaches in gastric cancer, KIT proto-oncogene receptor tyrosine kinase/platelet-derived growth factor receptor alpha-directed treatment in GI stromal tumor, and molecularly selected strategies in pancreatic cancer. Resistance, intratumoral heterogeneity, molecular evolution, and inconsistent biomarker assessment remain major barriers. Emerging approaches include next-generation inhibitors, antibody–drug conjugates, bispecific antibodies, cellular therapies, circulating tumor DNA-guided treatment, targeted–immunotherapy combinations, and tissue-agnostic strategies. Further progress will require accessible biomarker testing, longitudinal molecular assessment, rational treatment sequencing, and prospective evaluation of emerging precision therapies.

Key Words: Gastrointestinal neoplasms; Precision medicine; Tumor biomarkers; Neoplasm drug resistance; Antineoplastic agents

Core Tip: Molecularly targeted therapy has transformed the management of selected gastrointestinal malignancies by linking specific genomic or protein alterations to individualized treatment. This narrative review summarizes established and emerging molecular targets across colorectal, gastric and gastroesophageal junction, biliary, pancreatic, and gastrointestinal stromal tumors, while contextualizing therapies according to clinical evidence, biomarker requirements, and resistance mechanisms. Emerging approaches—including circulating tumor DNA, antibody–drug conjugates, bispecific antibodies, KRAS-directed therapies, and targeted–immunotherapy combinations—may help address treatment resistance, although their clinical roles vary in maturity and require continued prospective evaluation.

INTRODUCTION

Gastrointestinal (GI) malignancies, including esophageal, gastric, colorectal, pancreatic, hepatobiliary cancers, and GI stromal tumors (GISTs), remain major causes of cancer morbidity and mortality. Advanced disease continues to have poor outcomes despite improvements in multimodal treatment, reflecting late presentation, molecular heterogeneity, treatment resistance, and recurrence. Molecular characterization has identified actionable alterations and enabled biomarker-directed treatment strategies across several GI malignancies[1].

Targeted therapy is now established in selected molecular subgroups. Human epidermal growth factor receptor 2 (HER2)-directed therapy is used in gastric and gastroesophageal junction (GC/GEJ) cancers, while epidermal growth factor receptor (EGFR)- and BRAF-directed strategies are established in selected colorectal cancers. Fibroblast growth factor receptor 2 (FGFR2)- and IDH1-directed therapies provide genotype-directed treatment options for cholangiocarcinoma (CCA), and KIT- and platelet-derived growth factor receptor alpha (PDGFRA)-directed therapy forms the basis of treatment for GIST. Vascular endothelial growth factor (VEGF)-directed agents have demonstrated benefit across several GI cancers, while poly (ADP-ribose) polymerase (PARP) inhibition provides a biomarker-selected option in pancreatic cancer with homologous recombination repair (HRR) defects. Biomarker-defined immunotherapy, particularly for microsatellite instability-high (MSI-H)/deficient mismatch repair (dMMR) and tumor mutational burden-high (TMB-H) tumors, has further expanded precision treatment. These approaches should be distinguished from conventional targeted therapy because the biomarkers primarily identify tumors likely to benefit from immune checkpoint blockade rather than directly identifying a druggable oncogenic driver. Emerging strategies include KRAS-directed therapy, antibody-drug conjugates (ADCs), bispecific antibodies, and targets such as claudin 18.2 (CLDN18.2) and FGFR2b. Next-generation sequencing (NGS) and liquid biopsy have further expanded molecular profiling and longitudinal disease assessment[2,3].

However, the presence of an actionable molecular alteration does not uniformly translate into a clinically meaningful or durable therapeutic benefit. The maturity of evidence varies substantially between tumor types and molecular subgroups, ranging from randomized trials supporting established standards of care to early-phase studies, basket trials, retrospective analyses, and preclinical investigations. Accordingly, regulatory approval, guideline endorsement, prospective clinical activity, and experimental promise should be interpreted as distinct levels of evidence. The clinical relevance of a biomarker also depends on the assay platform, tissue adequacy, positivity thresholds, concordance between tissue and circulating tumor DNA (ctDNA), and the availability of validated companion diagnostic approaches. These considerations are particularly important for biomarkers such as HER2, CLDN18.2, FGFR2 fusions, and other alterations for which testing methodology and interpretation may influence treatment selection.

The benefit of targeted therapy remains constrained by primary and acquired resistance, intratumoral heterogeneity, evolving molecular alterations, limited predictive biomarkers, toxicity, and cost. Resistance may result from secondary alterations in the therapeutic target, bypass signaling, downstream pathway reactivation, histological or phenotypic changes, or inadequate drug delivery. Although some resistance mechanisms are supported by clinical or translational evidence, others remain proposed or preclinical and should not be considered equivalent. In addition, limitations of molecular testing - including false-negative results in low-shedding or localized tumors and reduced sensitivity of plasma-based assays for certain structural alterations - may affect the identification of potentially actionable disease. Variability in biomarker testing and access to molecular diagnostics further affects treatment selection and treatment availability[4,5].

This narrative review evaluates molecularly targeted therapies across selected major GI malignancies, with emphasis on colorectal cancer, CCA and biliary tract malignancies, GC/GEJ cancers, GIST, and pancreatic cancer. These tumor groups were selected to represent a spectrum of precision-oncology development, from established biomarker-treatment associations to emerging strategies with early clinical or preclinical evidence. Esophageal and hepatocellular malignancies were included in the broader search strategy but are not discussed as separate comprehensive disease-specific sections; their omission reflects the defined scope of this review rather than their lack of clinical or biological importance. The review focuses on established targets, clinical evidence and evidence maturity, biomarker selection and testing considerations, mechanisms of resistance, and emerging strategies for treatment sequencing and precision oncology. The principal targets, biomarkers, therapeutic approaches, and corresponding evidence categories are summarized in Table 1.

Table 1 Literature search strategy used for the narrative review.

Component	Description
Review type	Narrative review
Databases	PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar
Search period	January 2000 to July 30, 2026
Language	English
Search approach	Structured keyword-based searches using Boolean operators (AND/OR), with MeSH terms used where applicable
Core search terms	“Gastrointestinal cancer”, “gastrointestinal malignancy”, “targeted therapy”, “molecular targeted therapy”, and “precision oncology”
Cancer-specific terms	Esophageal cancer, gastric cancer, gastroesophageal junction cancer, colorectal cancer, pancreatic cancer, hepatocellular carcinoma, cholangiocarcinoma, biliary tract cancer, and GIST
Molecular target terms	EGFR, HER2, VEGF, FGFR, KIT, PDGFRA, BRAF, KRAS, IDH1, CLDN18.2, FGFR2b, PARP, NTRK, RET, MET, ALK, ROS1, PI3K, AKT, and mTOR
Additional therapeutic terms	Antibody-drug conjugates, bispecific antibodies, DNA damage repair, genomic profiling, next-generation sequencing, and liquid biopsy
Additional searches	Cancer-specific and drug-specific searches addressing pivotal trials, regulatory evidence, resistance mechanisms, and emerging therapies
Evidence prioritized	Randomized clinical trials, phase II/III studies, prospective studies, meta-analyses, international guidelines, landmark trials, and high-quality reviews
Study selection	Studies were selected based on relevance, methodological quality, clinical significance, molecular specificity, and contribution to established or emerging targeted treatment
Screening process	Independent screening of titles and abstracts, followed by full-text assessment by two reviewers
Disagreement resolution	Discussion and consensus, with consultation of a third reviewer when required
Evidence synthesis	Narrative synthesis organized by gastrointestinal malignancy and molecular target, addressing mechanism of action, clinical role, evidence maturity, biomarkers, resistance mechanisms, and emerging strategies
Evidence grading	Grade A: Randomized phase III evidence and/or established standard-of-care treatment. Grade B: Prospective phase II evidence or substantial clinical evidence. Grade C: Early-phase, retrospective, or otherwise limited clinical evidence. Grade D: Preclinical or experimental evidence

MeSH: Medical Subject Headings; GIST: Gastrointestinal stromal tumor; EGFR: Epidermal growth factor receptor; HER2: Human epidermal growth factor receptor 2; VEGF: Vascular endothelial growth factor; FGFR: Fibroblast growth factor receptor; PDGFRA: Platelet-derived growth factor receptor alpha; PARP: Poly (ADP-ribose) polymerase; NTRK: Neurotrophic tyrosine receptor kinase; RET: Rearranged during transfection; MET: Mesenchymal-epithelial transition factor; ALK: Anaplastic lymphoma kinase; ROS1: ROS proto-oncogene 1; PI3K: Phosphatidylinositol 3-kinase; AKT: Protein kinase B; mTOR: Mechanistic target of rapamycin; KIT: KIT proto-oncogene receptor tyrosine kinase; CLDN18.2: Claudin 18.2.

LITERATURE REVIEW

Literature search strategy

A structured literature search was conducted to identify clinical and translational evidence on molecularly targeted therapies in GI malignancies. PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar were searched for publications from January 2000 through 30 July 2026. The search strategy combined disease-specific terms, molecular alterations, therapeutic targets, and targeted-treatment terms using Boolean operators (AND/OR).

The core search strategy included combinations of “gastrointestinal cancer”, “gastrointestinal malignancy”, colorectal, gastric, gastroesophageal junction, pancreatic, cholangiocarcinoma, “biliary tract”, and gastrointestinal stromal tumor (GIST) with terms including “targeted therapy”, “molecular targeted therapy”, “precision oncology”, EGFR, HER2, VEGF, FGFR, KIT, PDGFRA, BRAF, KRAS, IDH1, CLDN18.2, FGFR2b, PARP, NTRK, and RET. Additional target-, biomarker-, drug-, and disease-specific searches were performed to identify pivotal clinical trials, regulatory evidence, companion diagnostic information, resistance mechanisms, and emerging therapeutic approaches.

Search terms were adapted to the indexing system of each database. Reference lists of relevant clinical trials, systematic reviews, meta-analyses, guidelines, and key review articles were also screened manually to identify additional relevant publications. The final database search was completed on 30 July 2026.

Literature selection

Publications were considered eligible when they provided clinically or biologically relevant information regarding molecularly targeted treatment of GI malignancies, including evidence on therapeutic efficacy, safety, biomarker selection, molecular mechanisms, resistance, treatment sequencing, or emerging therapeutic strategies.

Priority was given to randomized phase II/III clinical trials, prospective clinical studies, pivotal regulatory studies, international guidelines, and high-quality systematic reviews or meta-analyses. Original clinical studies were preferentially used when evaluating the efficacy or safety of individual therapies. Early-phase studies, retrospective studies, case series, and preclinical investigations were included when addressing emerging targets, resistance mechanisms, or therapeutic strategies for which mature clinical evidence was unavailable.

Publications were excluded when they were unrelated to molecularly targeted treatment, did not provide relevant molecular or therapeutic information, or did not contribute meaningfully to the objectives of the review. Only English-language publications were considered.

Screening and data synthesis

Titles and abstracts identified through the database searches were screened for relevance, followed by assessment of potentially eligible full-text publications. Two reviewers independently assessed study relevance and eligibility. Disagreements were resolved by discussion and consensus; when consensus could not be reached, a third reviewer was consulted.

Data were synthesized narratively rather than through quantitative meta-analysis because of substantial heterogeneity in tumor types, molecular alterations, treatment regimens, lines of therapy, comparator groups, biomarker definitions, and clinical endpoints. For each major malignancy and molecular target, the review considered the underlying molecular alteration, biological rationale, representative targeted therapies, biomarker-selection approach, clinical efficacy and safety where available, regulatory or clinical role, mechanisms of primary and acquired resistance, and emerging treatment strategies.

Established therapies were distinguished from approaches supported primarily by early-phase, retrospective, or preclinical evidence. Where appropriate, evidence from tissue-agnostic studies was identified separately from tumor-specific evidence. Biomarker-defined immunotherapies, particularly therapies selected by MSI-H/dMMR or TMB-H, were considered separately from conventional molecularly targeted therapies because these biomarkers primarily identify tumors likely to benefit from immune checkpoint blockade rather than directly identifying a druggable oncogenic driver.

Evidence grading

To facilitate consistent interpretation across tumor types and molecular subgroups, therapeutic approaches were categorized according to the highest level of supporting evidence:

Grade A: Randomized phase III evidence and/or established standard-of-care treatment supported by robust prospective clinical evidence.

Grade B: Prospective phase II or other substantial clinical evidence demonstrating therapeutic activity or clinical utility.

Grade C: Early-phase, retrospective, case-based, or otherwise limited clinical evidence.

Grade D: Preclinical, experimental, or hypothesis-generating evidence without established clinical efficacy.

The evidence grade was assigned to the therapeutic approach represented in each table entry and was intended to describe evidence maturity rather than regulatory approval status. Regulatory approval, guideline endorsement, prospective clinical activity, and experimental promise were therefore considered distinct dimensions and were not assumed to be interchangeable.

Scope of the review

The final disease-specific synthesis focused on colorectal cancer, GC/GEJ cancers, CCA and other biliary tract malignancies, pancreatic ductal adenocarcinoma (PDAC), and GIST. Esophageal and hepatocellular malignancies were captured within the broader search strategy but were not developed as separate comprehensive disease-specific sections because they were outside the defined scope of the final synthesis. Their exclusion from the detailed analysis does not imply absence of clinically relevant molecularly targeted therapies in these malignancies.

This approach was designed to provide a clinically oriented synthesis spanning established biomarker-therapy associations, mechanisms of treatment resistance, biomarker-testing considerations, and emerging precision-treatment strategies, while clearly distinguishing mature clinical evidence from investigational approaches.

Biomarker testing and interpretation

Biomarker assessment was considered an integral component of molecularly guided treatment selection. For each clinically relevant biomarker, the review considered the specimen type, assay platform, definition of positivity, and clinical context in which testing was performed. Depending on the biomarker, clinically used approaches included immunohistochemistry (IHC), fluorescence or chromogenic in situ hybridization, polymerase chain reaction-based assays, and NGS. The interpretation of biomarkers was considered in the context of the assay and threshold used in the supporting clinical studies rather than assuming interchangeability between platforms.

Particular attention was given to biomarkers for which assay methodology can influence treatment eligibility, including HER2, CLDN18.2, FGFR2 fusions/rearrangements, and other genomic alterations. For protein-expression biomarkers, IHC scoring and predefined positivity thresholds were considered, whereas genomic biomarkers were evaluated according to the molecular alteration and the assay's ability to detect relevant substitutions, amplifications, fusions, or rearrangements. Where available, the review also considered the distinction between validated companion or complementary diagnostic approaches and broader molecular profiling assays.

Tissue-based testing remains the principal approach for many GI malignancies; however, ctDNA and other liquid-biopsy approaches may provide complementary information for molecular profiling and longitudinal assessment. Tissue and plasma results were not considered inherently interchangeable. Potential discordance may arise from tumor heterogeneity, low tumor shedding, limited tissue availability, treatment-related clonal evolution, and differences in assay sensitivity. In particular, plasma-based testing may have reduced sensitivity for alterations in low-shedding or localized tumors and for certain structural variants, including some gene fusions. A negative plasma result was therefore not considered sufficient to exclude an actionable alteration when clinical suspicion remained high and tissue testing was feasible.

Repeat or orthogonal testing may be appropriate when the initial result is technically inadequate, discordant with the clinical or pathological context, or obtained at a time when molecular evolution is likely to have occurred. Biomarker results were therefore interpreted as context-dependent treatment-selection tools rather than static characteristics of the tumor.

CONCEPTUAL FRAMEWORK OF TARGETED THERAPY IN GI MALIGNANCIES

Targeted therapy in GI malignancies is based on identifying molecular alterations that contribute to tumor initiation, proliferation, survival, angiogenesis, metastatic progression, and therapeutic resistance, followed by the selection of treatments directed against these vulnerabilities. This approach differs from conventional cytotoxic chemotherapy, which primarily targets rapidly proliferating cells without requiring a specific molecular alteration. Targeted agents act on defined oncogenic drivers, receptor tyrosine kinases, intracellular signaling networks, DNA damage-response pathways, or tumor-associated antigens, thereby providing a framework for more selective and biomarker-informed treatment.

The principal therapeutic platforms include monoclonal antibodies directed against extracellular receptors or tumor-associated antigens, small-molecule inhibitors that penetrate cells and modulate intracellular kinases or signaling pathways, ADCs, and emerging bispecific or multispecific agents. Across GI malignancies, clinically relevant molecular targets include the EGFR/HER family, VEGF/VEGFR axis, FGFR, KIT, PDGFRA, BRAF, KRAS, IDH1, and DNA damage-repair pathways, as well as lineage-associated surface antigens such as CLDN18.2. The therapeutic relevance of these targets varies considerably among tumor types. Molecular prevalence alone is insufficient to establish actionability; clinical utility also depends on oncogenic dependence, the predictive performance of the biomarker, availability of an effective agent, treatment-line context, safety, and the strength and maturity of supporting clinical evidence.

The increasing incorporation of NGS, IHC, in situ hybridization, and other molecular diagnostic platforms has progressively shifted treatment selection from an exclusively anatomical or histopathological model toward biomarker-defined therapeutic stratification. However, the interpretation of molecular results requires attention to assay characteristics, tissue adequacy, biomarker thresholds, tumor heterogeneity, and, where applicable, concordance between tissue and ctDNA. These considerations are particularly relevant for biomarkers such as HER2, CLDN18.2, FGFR2 fusions, and other alterations for which differences in testing methodology may influence patient selection and treatment outcomes.

Precision oncology has also expanded beyond direct inhibition of an oncogenic protein or signaling pathway. Biomarkers such as MSI-H, dMMR, and TMB-H identify tumors that may benefit from immune checkpoint blockade and should therefore be considered biomarker-defined immunotherapy indications rather than conventional molecularly targeted therapies. Similarly, tissue-agnostic targets such as neurotrophic tyrosine receptor kinase (NTRK) fusions and selected rearranged during transfection (RET) alterations illustrate how molecular characteristics can support treatment approaches across different anatomical tumor sites, although the magnitude and maturity of evidence may vary between tumor types.

Thus, the contemporary framework of targeted therapy integrates molecular characterization, validated biomarker assessment, treatment selection, evaluation of clinical benefit and toxicity, and longitudinal monitoring of molecular evolution. Because intratumoral heterogeneity, pathway redundancy, and acquired resistance may limit the durability of response, effective precision oncology increasingly requires sequential or combination treatment strategies. The clinical interpretation of each molecularly guided approach should therefore distinguish established standard-of-care treatment from emerging clinical activity and experimental strategies, while considering the quality of evidence, regulatory status, and applicability to the relevant patient population (Figure 1).

Figure 1 Molecular targets and therapeutic strategies in gastrointestinal malignancies. A: Key molecular targets and their associated oncogenic or biological pathways, including EGFR/RAS, HER2, FGFR2, BRAF, KIT/PDGFRA, KRAS, IDH1, CLDN18.2, and BRCA1/2; B: Representative targeted therapies corresponding to the identified molecular targets; C: Tumor-specific applications of molecularly targeted therapies across colorectal cancer, cholangiocarcinoma/biliary tract cancer, gastric/gastroesophageal junction cancer, gastrointestinal stromal tumors (GIST), and pancreatic ductal adenocarcinoma; D: Proposed clinical pathway for molecularly guided treatment, encompassing biomarker testing, selection of targeted therapy, response monitoring, evaluation at disease progression or intolerance, and subsequent treatment adaptation based on molecular reassessment. EGFR: Epidermal growth factor receptor; HRR: Homologous recombination repair; IHC: Immunohistochemistry; PCR: Polymerase chain reaction; NGS: Next-generation sequencing; ctDNA: Circulating tumor DNA; FGFR: Fibroblast growth factor receptor; BRAF: B-Raf proto-oncogene; PDGFRA: Platelet-derived growth factor receptor alpha; PI3K: Phosphatidylinositol 3-kinase; IDH1: Isocitrate dehydrogenase 1; ADC: Antibody-drug conjugate; AKT: Protein kinase B; mTOR: Mechanistic target of rapamycin; MAPK: Mitogen-activated protein kinase; HER2: Human epidermal growth factor receptor 2; VEGFR: Vascular endothelial growth factor receptor; NTRK: Neurotrophic tyrosine receptor kinase; RET: Rearranged during transfection; KIT: KIT proto-oncogene receptor tyrosine kinase; TKI: Tyrosine kinase inhibitor; CLDN18.2: Claudin 18.2.

Targeted therapy in colorectal cancer

EGFR-targeted therapy: EGFR-directed therapy represents an established component of precision treatment for molecularly selected metastatic colorectal cancer (mCRC). Cetuximab and panitumumab are the principal anti-EGFR monoclonal antibodies used in clinical practice. Both inhibit ligand-dependent EGFR signaling by preventing receptor activation and downstream pathway stimulation. Cetuximab is a chimeric IgG1 antibody and may additionally mediate antibody-dependent cellular cytotoxicity, whereas panitumumab is a fully human IgG2 antibody with comparatively limited immune-effector activity. The clinical efficacy of both agents is predominantly confined to tumors without activating RAS alterations[6].

Molecular selection is therefore central to the use of anti-EGFR therapy. Activating mutations in KRAS or NRAS maintain downstream MAPK signaling despite EGFR blockade and represent major mechanisms of primary resistance. Accordingly, assessment of KRAS and NRAS mutations involving exons 2, 3, and 4 remains an essential component of treatment selection. BRAF status provides an additional layer of molecular stratification. A meta-analysis by Pietrantonio et al[7] demonstrated that anti-EGFR therapy did not significantly improve progression-free survival, overall survival, or response rate in RAS wild-type, BRAF-mutant CRC, supporting the assessment of BRAF status before initiating EGFR-directed treatment.

Primary tumor sidedness has emerged as an important clinical modifier of anti-EGFR efficacy. Kasi et al[8] and Khelwatty et al[9] reported greater benefit from EGFR-directed therapy in left-sided RAS wild-type tumors than in right-sided tumors. This distinction reflects the biological heterogeneity between proximal and distal colorectal cancers and has increasingly influenced therapeutic selection. Consequently, anti-EGFR therapy has its strongest established role in patients with left-sided, RAS wild-type mCRC, particularly when an EGFR-dependent molecular phenotype is present. Nevertheless, tumor sidedness should be interpreted alongside disease burden, treatment line, clinical status, coexisting molecular alterations, and the intended treatment regimen rather than as an isolated predictive marker.

Despite initial responses, resistance to EGFR inhibition frequently develops through both pre-existing and acquired mechanisms. Primary resistance may be associated with undetected or coexisting RAS alterations, BRAF activation, HER2 or mesenchymal-epithelial transition factor (MET) amplification, phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT) pathway activation, and oncogenic kinase fusions. Acquired resistance may involve secondary KRAS/NRAS alterations, EGFR extracellular-domain mutations, amplification of alternative receptor tyrosine kinases, downstream pathway reactivation, and clonal selection under therapeutic pressure. Intratumoral heterogeneity and clonal evolution further contribute to treatment failure. Leite et al[10] describe the concept of negative hyperselection, whereby molecular selection extends beyond conventional RAS and BRAF testing to exclude additional alterations associated with EGFR resistance. Although this approach may improve the identification of patients most likely to benefit, the optimal panel, assay methodology, and clinical integration of negative hyperselection remain under evaluation.

Liquid biopsy has provided an important means of monitoring dynamic molecular evolution during EGFR-directed treatment. Serial ctDNA analysis can detect emerging RAS, EGFR, and other resistance-associated alterations during treatment, potentially identifying molecular progression before radiological progression. However, ctDNA may produce false-negative results in patients with low-shedding or localized disease, and a negative plasma result should not automatically be interpreted as the absence of an actionable alteration. Conversely, resistant RAS-mutant clones may decline after discontinuation of EGFR inhibition, providing a biological rationale for molecularly selected anti-EGFR rechallenge. The clinical utility of this approach depends on the timing of ctDNA assessment, assay sensitivity, definition of molecular clearance, and prospective validation[11].

Current research is increasingly focused on resistance-directed and pathway-combination strategies rather than isolated EGFR inhibition. Combinations involving EGFR blockade with MAPK-pathway inhibitors, HER2-directed agents, immune-based therapies, and other targeted treatments are being evaluated in molecularly defined resistant disease. These strategies aim to suppress bypass signaling, prevent downstream pathway reactivation, or exploit vulnerabilities created by acquired resistance. Tardito et al[5] also highlight ADCs and patient-derived organoids as emerging platforms for investigating resistance and evaluating individualized therapeutic combinations. However, these approaches remain predominantly investigational and should be distinguished from established anti-EGFR therapy[12].

VEGF-targeted therapy: The VEGF/VEGFR signaling axis represents a major therapeutic target in mCRC. VEGF-A promotes endothelial proliferation, migration, vascular permeability, and neovascularization, primarily through VEGFR-2, thereby supporting tumor growth and metastatic dissemination. Unlike EGFR-directed treatment, the clinical use of anti-VEGF therapy is not restricted by RAS or BRAF mutation status and can therefore be applied across a broader molecular spectrum of CRC[13].

Bevacizumab, a monoclonal antibody targeting VEGF-A, remains a central component of anti-angiogenic therapy and is commonly combined with cytotoxic chemotherapy in different treatment settings. By neutralizing VEGF-A, bevacizumab suppresses angiogenic signaling and may transiently normalize abnormal tumor vasculature, potentially improving the delivery and efficacy of concomitant systemic therapy. Clinical studies have demonstrated improved outcomes with bevacizumab-containing chemotherapy in selected treatment settings, supporting its continuing role in mCRC[13,14]. However, the magnitude of benefit varies according to treatment line, chemotherapy backbone, patient selection, and disease characteristics.

Additional agents have expanded therapeutic options for patients progressing after initial treatment. Aflibercept, a recombinant fusion protein that acts as a decoy receptor for VEGF-A, VEGF-B, and placental growth factor, provides broader ligand sequestration, whereas ramucirumab is a monoclonal antibody directed against VEGFR-2. Both agents have demonstrated clinical benefit when combined with chemotherapy in subsequent-line treatment strategies, supporting continued inhibition of the VEGF pathway after disease progression. Their use, however, should be interpreted in the context of prior treatment exposure, performance status, toxicity profile, and the availability of alternative systemic options.

Later-line treatment has further expanded through oral tyrosine kinase inhibitors. Regorafenib inhibits VEGFR1-3 and multiple additional kinases involved in angiogenesis, tumor-cell signaling, and the tumor microenvironment. Fruquintinib is a more selective inhibitor of VEGFR-1, VEGFR-2, and VEGFR-3. Both agents have established roles in previously treated or refractory mCRC and illustrate the continued therapeutic relevance of angiogenic signaling despite exposure to chemotherapy and earlier anti-angiogenic therapy[15]. Their clinical benefit is generally modest but clinically relevant in appropriately selected patients, and treatment decisions require careful consideration of toxicity, including hypertension, hand-foot skin reaction, fatigue, diarrhea, and hepatic or vascular adverse effects.

A major limitation of VEGF-directed treatment is the absence of a validated predictive biomarker that can reliably identify patients most likely to benefit. Circulating VEGF concentrations, VEGFR expression, angiogenic cytokines, and tissue-based vascular markers have not demonstrated sufficient consistency for routine patient selection. Jacobsen et al[16] emphasize that vascular heterogeneity and endothelial plasticity contribute substantially to interpatient variability in treatment response. Resistance may develop through activation of alternative angiogenic pathways involving fibroblast growth factors, platelet-derived growth factors, angiopoietins, hepatocyte growth factor, and inflammatory mediators. Hypoxia-driven adaptation, extracellular matrix remodeling, pericyte-mediated vascular support, and vascular reprogramming may further reduce the effectiveness of VEGF inhibition[17]. These mechanisms suggest that angiogenesis is regulated through a complex and adaptable network rather than a single dominant pathway.

The close relationship between angiogenesis and antitumor immunity has generated interest in combining VEGF-directed therapy with immune checkpoint blockade. Excessive VEGF signaling can contribute to an immunosuppressive tumor microenvironment by impairing dendritic-cell maturation, promoting regulatory immune populations, and limiting effective immune-cell trafficking. Conversely, vascular normalization induced by VEGF inhibition may improve immune-cell access to the tumor and potentially enhance the activity of immune checkpoint inhibitors. These observations provide the biological rationale for combination strategies involving anti-angiogenic agents and immune checkpoint blockade, as discussed by Jiao et al[18]. However, the clinical efficacy of such combinations is dependent on molecular and clinical context, and their role in CRC remains under evaluation rather than representing a universal treatment strategy.

Combination of VEGF-directed therapy with other molecularly targeted agents is also being explored. Li et al[14] describe case-based evidence supporting cetuximab plus fruquintinib in selected patients with refractory RAS/BRAF wild-type mCRC. Such an approach is biologically plausible because simultaneous inhibition of EGFR-mediated tumor signaling and VEGF-mediated angiogenesis may address complementary mechanisms of tumor progression. Nevertheless, case-based or retrospective evidence cannot establish comparative efficacy, and the available data remain insufficient to define the optimal patient population, treatment sequence, or toxicity profile. Prospective clinical studies are therefore required before these combinations can be considered broadly established treatment strategies[15,17-19].

BRAF, HER2, and other actionable alterations: The increasing availability of comprehensive molecular profiling has expanded precision treatment in mCRC beyond the EGFR and VEGF pathways. Clinically relevant alterations include BRAF V600E mutations, HER2 amplification or overexpression, KRAS G12C mutations, NTRK fusions, RET rearrangements, and other rare kinase alterations. Although many of these abnormalities occur at relatively low frequencies, their therapeutic relevance makes broad molecular profiling important for identifying patients who may benefit from molecularly matched treatment[20-22].

(1) BRAF V600E. BRAF V600E occurs in approximately 8%-12% of mCRC and is associated with an aggressive clinical phenotype and unfavorable prognosis. BRAF functions downstream of RAS within the MAPK pathway, and constitutive BRAF activation promotes persistent signaling. Importantly, inhibition of BRAF alone can induce feedback activation of EGFR and restoration of MAPK signaling, explaining the limited efficacy of single-agent BRAF inhibition in CRC[23,24].

The BEACON CRC study established combined BRAF and EGFR inhibition with encorafenib and cetuximab, with or without binimetinib, as an effective targeted strategy for previously treated BRAF V600E-mutant mCRC[25]. This represents an important example of rational combination therapy in which simultaneous inhibition of the oncogenic driver and a compensatory upstream pathway is used to overcome adaptive resistance[26].

The relevance of BRAF-directed therapy has also been explored in molecularly complex diseases. Ambrosini et al[22] reported clinical activity of BRAF and EGFR inhibition in patients with BRAF V600E-mutant, dMMR/MSI-H mCRC following progression on immune checkpoint inhibitors, with outcomes broadly comparable to those observed in pMMR/MSS disease, although response rates were numerically lower. Evidence for non-V600E BRAF alterations remains considerably more limited. Sasaki et al[26] described an initial response followed by progression in a patient with BRAF K601E-mutant mCRC treated with encorafenib, binimetinib, and cetuximab. Such observations support continued molecular investigation but do not establish a standard treatment strategy for non-V600E BRAF variants[22-27].

(2) HER2. HER2 amplification or overexpression occurs in approximately 2%-4% of mCRC and is particularly relevant in RAS wild-type tumors. HER2 signaling can activate both MAPK and PI3K/AKT pathways and represents an important mechanism of resistance to EGFR inhibition. Thus, HER2 assessment may provide both a predictive biomarker for HER2-directed treatment and information regarding potential resistance to anti-EGFR therapy.

The HERACLES study established the clinical activity of dual HER2 blockade with trastuzumab and lapatinib in heavily pretreated HER2-positive mCRC. Subsequent studies have evaluated alternative HER2-directed combinations, including trastuzumab with pertuzumab or tucatinib, while ADCs such as trastuzumab deruxtecan (T-DXd) have further expanded therapeutic options for HER2-positive disease. However, treatment selection remains influenced by differences in HER2 testing methodologies, expression thresholds, molecular heterogeneity, and the potential for acquired HER2 loss[28-37].

And (3) KRAS G12C. The emergence of KRAS G12C-directed therapy has altered the traditional view of KRAS mutations as purely negative predictive biomarkers for EGFR therapy. KRAS G12C represents a relatively small molecular subgroup of CRC, but the mutant protein contains a therapeutically exploitable binding site that permits direct pharmacological inhibition. Clinical studies have demonstrated activity when KRAS G12C inhibitors are combined with EGFR blockade, reflecting the importance of simultaneously suppressing mutant KRAS signaling and compensatory EGFR-mediated pathway reactivation[26].

NTRK, RET, and other rare alterations: Rare molecular alterations, including NTRK gene fusions and RET rearrangements, provide additional opportunities for molecularly selected or tissue-agnostic treatment. Their low prevalence in CRC makes comprehensive genomic profiling particularly important, as conventional limited biomarker testing may fail to identify eligible patients. Other alterations involving PIK3CA, PTEN, MET, and FGFR remain primarily investigational either as therapeutic targets or as biomarkers of resistance to established targeted therapies[27].

Overall, the therapeutic landscape of mCRC increasingly reflects a transition from pathway-specific treatment toward multi-dimensional molecular stratification, in which RAS, BRAF, HER2, and rare actionable alterations are integrated with tumor sidedness, mismatch-repair status, and longitudinal molecular monitoring. The principal molecular targets, therapeutic agents, evidence grades, and emerging strategies are summarized in Table 2.

Table 2 Molecularly targeted therapies in colorectal cancer.

Molecular target/pathway	Biomarker/alteration	Therapeutic agent(s)	Current clinical role	Evidence grade	Key resistance mechanisms/emerging strategies	Representative sources
EGFR	RAS wild-type mCRC; BRAF status should also be considered for treatment selection and sequencing	Cetuximab; panitumumab	Established therapy in molecularly selected mCRC, used with chemotherapy or as monotherapy in appropriate later-line settings	A	Primary or acquired resistance may involve emergent KRAS/NRAS or BRAF alterations, HER2 or MET amplification, PI3K/PTEN pathway alterations, EGFR extracellular-domain mutations, gene fusions, and intratumoral heterogeneity. ctDNA-guided rechallenge is an emerging strategy in selected patients	Kasi et al[8]; Khelwatty et al[9]; Leite et al[10]; Li et al[14]
VEGF-A	No validated predictive biomarker	Bevacizumab	Established anti-angiogenic therapy, generally combined with chemotherapy in first-line and subsequent-line treatment strategies	A	Adaptive angiogenesis, vascular heterogeneity, hypoxia-driven adaptation, and activation of alternative pro-angiogenic pathways. Biomarker discovery and combinations with immunotherapy remain under investigation	Colombo and Porretto[13]; Li et al[14]; Jacobsen et al[16]
VEGF-A/VEGF-B/PlGF	No validated predictive biomarker	Aflibercept	Established second-line treatment in combination with chemotherapy after progression on an oxaliplatin-containing regimen in appropriate patients	A	Alternative angiogenic pathway activation, vascular remodeling, and endothelial adaptation	Li et al[14]; Fakih[17]
VEGFR-2	No validated predictive biomarker	Ramucirumab	Established second-line treatment in combination with chemotherapy in selected patients previously treated with bevacizumab-, oxaliplatin-, and fluoropyrimidine-based therapy	A	Compensatory angiogenic signaling, vascular remodeling, and microenvironmental adaptation	Li et al[14]
VEGFR1–3 and multiple angiogenic/tumor-signaling kinases	No validated predictive biomarker	Regorafenib	Established later-line treatment for previously treated or refractory mCRC	A	Multikinase pathway escape, adaptive angiogenesis, and tumor–microenvironment interactions. Toxicity and performance status are important treatment considerations	Li et al[14]; Jacobsen et al[16]
VEGFR1–3	No validated predictive biomarker	Fruquintinib	Established later-line treatment for previously treated or refractory mCRC	A	Adaptive angiogenesis and vascular pathway redundancy; combination strategies are under investigation	Li et al[14]; Jiao et al[18]
BRAF/MAPK	BRAF V600E mutation	Encorafenib + cetuximab, with or without binimetinib depending on the treatment context	Established targeted therapy for previously treated BRAF V600E-mutant mCRC	A	MAPK pathway reactivation, feedback signaling, and bypass activation. Combined BRAF/EGFR inhibition, with or without MEK inhibition, is intended to suppress pathway escape	Kopetz et al[24]; Ambrosini et al[22]; Li et al[14]
HER2	HER2 amplification and/or overexpression; assessment requires an appropriate validated assay	Trastuzumab-based dual HER2 blockade; tucatinib-based combinations; trastuzumab deruxtecan and other HER2-directed ADCs	Clinically actionable target in selected HER2-positive mCRC, particularly in RAS wild-type disease and appropriate later-line settings	B	HER2 heterogeneity, HER2 loss, bypass signaling, downstream pathway reactivation, and acquired resistance. Next-generation ADCs and combination HER2-directed strategies are under investigation	Sartore-Bianchi et al[34]; Pizzamiglio et al[33]; Benli et al[38]
KRAS G12C	KRAS G12C mutation	KRAS G12C inhibitor combined with EGFR inhibition	Molecularly selected targeted therapy for KRAS G12C-mutant mCRC, generally in previously treated disease	A	EGFR-mediated feedback, MAPK pathway reactivation, secondary KRAS alterations, and downstream pathway escape. Combination strategies aim to improve pathway suppression and delay resistance	Li et al[14]; Manzi et al[39]
NTRK	NTRK gene fusion	Larotrectinib; entrectinib	Tissue-agnostic targeted therapy for fusion-positive tumors, including rare CRC cases	B	Acquired kinase-domain mutations and bypass signaling may lead to resistance; next-generation TRK inhibitors are being developed for resistant disease	Li et al[40]
RET	RET fusion	Selective RET inhibitors	Rare molecularly selected indication; CRC-specific evidence remains limited	B	Acquired resistance, limited tumor-specific clinical evidence, and uncertainty regarding optimal sequencing; prospective evaluation continues	Li et al[40]
BRAF/MAPK	Non-V600E BRAF alterations	Molecularly matched BRAF-, MEK-, and/or EGFR-directed combinations	Limited clinical evidence in selected patients; not equivalent to the established BRAF V600E treatment pathway	C	Biological heterogeneity and variable sensitivity according to the specific BRAF alteration. Molecularly matched combinations require further clinical validation	Sasaki et al[26]; Ambrosini et al[22]
MET/PI3K/PTEN and related bypass pathways	MET amplification, PIK3CA alterations, PTEN loss, or other putative resistance-associated alterations	Molecularly matched combinations	Investigational strategies aimed at overcoming resistance to EGFR inhibition	C	These alterations may contribute to intrinsic or acquired resistance, although their predictive and therapeutic value is context-dependent. Multi-target inhibition and biomarker-guided approaches remain under investigation	Khelwatty et al[9]; Leite et al[10]

Evidence classification: Evidence levels were assigned according to the maturity and clinical validation of the available data. Established: Supported by prospective randomized trials and/or incorporated into clinical guidelines or routine standard-of-care practice; Clinically supported: Supported by prospective clinical studies, non-randomized comparative studies, or consistent clinical evidence but with more limited validation; Emerging/investigational: Supported by early-phase clinical trials, small prospective or retrospective studies, or preliminary clinical data. Preclinical: Supported primarily by in vitro, animal, organoid, or other laboratory-based studies without sufficient clinical validation. Evidence grades describe the maturity of the evidence and do not necessarily indicate the magnitude of treatment benefit. ADC: Antibody–drug conjugate; BRAF: B-Raf proto-oncogene; EGFR: Epidermal growth factor receptor; FGFR: Fibroblast growth factor receptor; MAPK: Mitogen-activated protein kinase; mCRC: Metastatic colorectal cancer; PDGFR: Platelet-derived growth factor receptor; PI3K: Phosphatidylinositol-3-kinase; PlGF: Placental growth factor; PTEN: Phosphatase and tensin homolog; RET: Rearranged during transfection; T-DXd: Trastuzumab deruxtecan; TRK: Tropomyosin receptor kinase; VEGF: Vascular endothelial growth factor; VEGFR: Vascular endothelial growth factor receptor; ctDNA: Circulating tumor DNA.

Targeted therapy in CCA

FGFR2-targeted therapy: FGFR2 fusions and rearrangements are among the most actionable molecular alterations in CCA, occurring in approximately 10%-16% of intrahepatic CCAs (iCCA). These alterations define a distinct molecular subgroup characterized by constitutive FGFR signaling through downstream pathways including MAPK and PI3K/AKT. Their identification has established FGFR inhibition as an important example of biomarker-directed treatment in CCA[38-43].

Pemigatinib, a reversible FGFR1-3 inhibitor, demonstrated clinically meaningful tumor responses and disease control in previously treated patients with FGFR2 fusion-positive CCA and established FGFR inhibition as a targeted treatment option for this molecular subgroup. Futibatinib, an irreversible FGFR inhibitor, subsequently expanded the therapeutic options available for FGFR2-altered disease. In the FOENIX-CCA2 study, Goyal et al[42] reported an objective response rate of 42% in patients with FGFR2-rearranged iCCA, with a median duration of response of 9.7 months. These findings established FGFR inhibition as an important targeted treatment strategy for appropriately selected patients with previously treated FGFR2-altered iCCA[44-46].

Acquired resistance remains a major limitation of FGFR-directed therapy. Secondary FGFR2 kinase-domain mutations represent an important mechanism of resistance, while activation of bypass signaling pathways and intratumoral clonal heterogeneity may further contribute to therapeutic escape[41,47]. Multiple resistance mutations may emerge across different tumor sites, potentially limiting the effectiveness of sequential treatment with inhibitors of the same therapeutic class.

ctDNA provides an opportunity for longitudinal assessment of these molecular changes during treatment. Serial plasma-based profiling may identify emerging FGFR2 resistance alterations and potentially assist in subsequent treatment selection or clinical-trial enrollment. However, the clinical utility of ctDNA-guided FGFR inhibitor sequencing remains investigational[44-46,48,49].

Current research is focused on developing next-generation FGFR inhibitors capable of overcoming resistance mutations, improving detection of FGFR2 fusions and rearrangements, and evaluating rational combination strategies. Comprehensive NGS, including RNA-based approaches, may improve detection of fusion events that can be missed by limited molecular assays. Earlier-line and perioperative incorporation of FGFR inhibition is also being investigated, although the optimal timing, sequencing, and combination of FGFR-directed therapy remain to be established.

IDH1-targeted therapy: IDH1 mutations occur predominantly in iCCA and represent an important actionable molecular alteration. Mutant IDH1 produces the oncometabolite 2-hydroxyglutarate, which disrupts epigenetic regulation and cellular differentiation. This distinctive biological dependency provides the rationale for selective IDH1 inhibition and represents an important example of precision medicine in CCA[50,51].

Ivosidenib is an oral selective inhibitor of mutant IDH1 and represents an established targeted therapy for previously treated IDH1-mutated CCA. The phase III ClarIDHy trial demonstrated significantly improved progression-free survival compared with placebo, with an overall survival benefit becoming evident after adjustment for crossover. Objective tumor responses were relatively uncommon, with the principal clinical benefit arising from prolonged disease stabilization[52,53]. Buccella[54] described patients maintaining clinical benefit for more than one year, consistent with the durable disease-control pattern observed in ClarIDHy.

Real-world evidence has further supported the clinical use of ivosidenib following prior systemic therapy. Buccella[54] reported durable disease control and clinical benefit in individual patients, while Chmielewska and Kubiatowski[52] described ivosidenib as a treatment option following progression after standard therapy. These observations further emphasize the importance of early molecular testing, as identification of an IDH1 mutation can provide access to subsequent genotype-directed treatment.

Acquired resistance remains an important limitation of IDH1 inhibition. Proposed mechanisms include secondary IDH1 alterations, metabolic adaptation, epigenetic remodeling, clonal evolution, and activation of alternative signaling pathways. Bray and Sahai[51] and Salati et al[50] emphasize that understanding these mechanisms will be important for developing strategies capable of extending the duration of disease control.

Combination approaches involving IDH1 inhibitors with immune checkpoint inhibitors, epigenetic agents, PARP inhibitors, or chemotherapy are being investigated, although these strategies remain experimental. Liquid biopsy may also provide a means of monitoring molecular evolution when repeat tissue sampling is difficult. Earlier incorporation of ivosidenib and combination treatment therefore require prospective validation before their use can be extended beyond the current post-progression setting[53,55,56].

HER2, BRAF, KRAS, NTRK, RET, and other emerging molecular targets: Comprehensive genomic profiling has identified an expanding spectrum of potentially actionable alterations in CCA, including HER2 abnormalities, BRAF V600E mutations, KRAS G12C mutations, NTRK and RET fusions, MSI-H, and TMB-H. Mie et al[57] reported that nearly half of biliary tract cancers may harbor potentially targetable alterations, supporting comprehensive molecular profiling in patients with advanced disease.

HER2 alterations are reported more frequently in extrahepatic CCA and gallbladder cancer than in iCCA. HER2-directed treatment has demonstrated clinical activity with trastuzumab-based combinations, including trastuzumab plus pertuzumab and trastuzumab plus tucatinib. More recent evidence has expanded to ADCs such as T-DXd. A 2025 review reported clinical benefit with several HER2-directed regimens[58], while Yarlagadda et al[59] described a durable response to trastuzumab plus pertuzumab in an ERBB2-amplified CCA identified by ctDNA. T-DXd has also demonstrated activity in individual cases, although the CCA-specific evidence base remains less mature than that established for other HER2-driven malignancies.

BRAF V600E represents another actionable molecular alteration in CCA. Mie et al[57] and Yang et al[60] identify BRAF V600E among the molecular abnormalities for which targeted treatment has demonstrated clinically meaningful benefit in advanced CCA. In contrast, evidence for KRAS G12C, NTRK fusions, and RET fusions remains limited by their low prevalence and the relative paucity of CCA-specific clinical studies. Nevertheless, identification of these alterations can provide access to mutation- or fusion-directed therapies and therefore remains clinically relevant to comprehensive molecular profiling[61,62].

Other molecular features, including MSI-H and TMB-H, provide additional biomarker-defined treatment opportunities. These biomarkers are important to distinguish from conventional molecularly targeted therapy because they primarily identify patients who may benefit from immune checkpoint blockade. Similarly, BRCA and other DNA damage-repair alterations are being investigated as potential therapeutic targets, although their clinical role in CCA remains investigational[63-66].

Overall, the expanding molecular landscape of CCA illustrates the transition from histology-based treatment toward increasingly biomarker-defined therapeutic strategies. FGFR2 and IDH1 alterations currently represent the most established molecular targets in iCCA, while BRAF V600E provides another clinically validated target and HER2-directed therapy is rapidly evolving. Rare alterations such as KRAS G12C, NTRK, and RET remain important primarily because comprehensive profiling can identify small patient subsets eligible for matched or tissue-agnostic therapies. The major actionable molecular alterations, targeted therapies, evidence grades, resistance mechanisms, and emerging strategies are summarized in Table 3.

Table 3 Molecularly targeted therapies in cholangiocarcinoma: Clinical role, evidence, resistance, and emerging strategies.

Molecular target/alteration	Approximate frequency	Representative therapy	Current clinical role	Evidence grade	Major resistance mechanisms/emerging strategies	Representative sources
FGFR2 fusion/rearrangement	10%-16%, predominantly intrahepatic CCA	Pemigatinib; futibatinib	Established targeted therapy for previously treated, unresectable, or metastatic FGFR2 fusion/rearranged CCA	A	Acquired FGFR2 kinase-domain mutations, bypass signaling, polyclonal resistance, and intratumoral heterogeneity. Next-generation FGFR inhibitors, ctDNA-guided monitoring, combination approaches, and earlier-line treatment are under investigation	Li et al[65]; Mie et al[57]; Goyal et al[42]; Ellis et al[41]; Xin and Miao[47]
IDH1 mutation	10%–20%, predominantly intrahepatic CCA	Ivosidenib	Established targeted therapy for previously treated IDH1-mutated CCA	A	Clonal evolution, metabolic adaptation, secondary pathway alterations, epigenetic remodeling, and bypass signaling. Combinations with immunotherapy, epigenetic agents, or PARP-directed therapy, as well as earlier-line treatment and ctDNA-guided monitoring, remain investigational	Abou-Alfa et al[66]; Bray and Sahai[51]; Salati et al[50]; Li et al[65]
HER2 amplification/overexpression	5%-15%; more frequent in extrahepatic CCA and gallbladder cancer	Trastuzumab-based combinations; pertuzumab; tucatinib-based combinations; trastuzumab deruxtecan	Molecularly selected treatment option for HER2-positive biliary tract cancers; clinical evidence is emerging and varies according to the agent and study population	B	HER2 heterogeneity, HER2 loss, bypass signaling, and downstream pathway reactivation. Next-generation ADCs, bispecific antibodies, and combination HER2-directed therapies are under investigation	Inoue et al[62]; Theocharopoulos et al[58]
BRAF V600E mutation	3%-5%	Dabrafenib + trametinib	Established molecularly selected targeted therapy for BRAF V600E-mutated biliary tract cancer	A	MAPK pathway reactivation, adaptive signaling, and bypass pathway activation. Additional combination strategies are under investigation	Mie et al[57]; Ellis et al[41]

Evidence grading: Grade A, randomized phase III evidence and/or established standard-of-care treatment; Grade B, prospective phase II evidence or substantial clinical evidence; Grade C, early-phase, retrospective, or otherwise limited clinical evidence; Grade D, preclinical or experimental evidence. Evidence grades reflect the maturity and clinical validation of the available evidence and do not directly indicate the magnitude of treatment benefit. ADC: Antibody–drug conjugate; CCA: Cholangiocarcinoma; ctDNA: Circulating tumor DNA; DDR: DNA damage response; dMMR: Mismatch repair deficient; FGFR2: Fibroblast growth factor receptor 2; IDH1: Isocitrate dehydrogenase 1; MSI-H: Microsatellite instability-high; NTRK: Neurotrophic tyrosine receptor kinase; PARP: Poly (ADP-ribose) polymerase; RET: Rearranged during transfection; TMB-H: Tumor mutational burden-high; TRK: Tropomyosin receptor kinase.

Targeted therapy in GC/GEJ adenocarcinoma

HER2-targeted therapy: HER2 amplification or overexpression occurs in approximately 15%-20% of GC/GEJ adenocarcinomas and represents the best-established molecular target in advanced disease. HER2 activation promotes downstream MAPK and PI3K/AKT signaling, supporting tumor proliferation and survival. Accurate assessment of HER2 status is therefore central to treatment selection, although intratumoral heterogeneity and temporal changes in HER2 expression may complicate classification and treatment sequencing[67-71].

The ToGA trial[72] established trastuzumab combined with platinum- and fluoropyrimidine-based chemotherapy as a first-line standard for HER2-positive advanced GC/GEJ adenocarcinoma. Bang et al[72] demonstrated improved overall survival with trastuzumab-containing therapy compared with chemotherapy alone, with greater benefit observed in tumors with higher HER2 expression. This trial provided the foundation for biomarker-directed treatment in advanced gastric cancer.

The first-line treatment paradigm subsequently evolved with the KEYNOTE-811 trial, in which pembrolizumab was added to trastuzumab and chemotherapy for HER2-positive advanced disease. This strategy expanded HER2-directed treatment from antibody-chemotherapy combinations to chemoimmunotherapy and further established the importance of integrating molecular and immune biomarkers in treatment selection[73].

T-DXd, an ADC linking trastuzumab to a topoisomerase I inhibitor, has become an important treatment following trastuzumab-based therapy. In DESTINY-Gastric01, Shitara et al[74] demonstrated higher objective response rates and longer overall survival with T-DXd than with physician's choice chemotherapy. Its potent cytotoxic payload and bystander effect may provide activity in tumors with heterogeneous HER2 expression, although acquired resistance and HER2 heterogeneity remain important limitations.

Several additional HER2-directed strategies are being investigated, including disitamab vedotin, zanidatamab, tucatinib-based combinations, pyrotinib, and HER2-directed CAR-T therapy. These approaches remain at different stages of clinical development. Shao et al[75] identify HER2 heterogeneity and the complex tumor microenvironment as important determinants of therapeutic resistance. The 2025 HER2 guideline also emphasizes standardized tissue-based assessment and highlights the potential role of liquid biopsy in addressing spatial and temporal heterogeneity.

Resistance to HER2-directed therapy may arise through HER2 heterogeneity or loss, activation of alternative signaling pathways, and clonal evolution. MET, EGFR, FGFR, and PI3K/AKT signaling have been implicated in therapeutic resistance[70]. Serial tissue assessment and ctDNA analysis may provide opportunities to identify changes in HER2 status and emerging resistance alterations, although the clinical utility of dynamic biomarker-guided treatment selection remains under investigation.

Anti-angiogenic therapy (VEGF/VEGFR-targeted therapy): The VEGF/VEGFR pathway is an established therapeutic target in advanced GC/GEJ adenocarcinoma, although the magnitude of clinical benefit has generally been more modest than that observed in colorectal cancer. Among anti-angiogenic agents, ramucirumab, a monoclonal antibody targeting VEGFR-2, has the strongest clinical evidence[76].

The REGARD trial demonstrated improved overall survival with ramucirumab monotherapy compared with best supportive care in patients whose disease had progressed following first-line chemotherapy[77]. The subsequent RAINBOW trial showed that ramucirumab combined with paclitaxel improved objective response rate, progression-free survival, and overall survival compared with paclitaxel alone[78]. These studies established ramucirumab, either as monotherapy or in combination with paclitaxel, as an important second-line treatment option for advanced GC/GEJ adenocarcinoma.

Other VEGFR-directed agents have demonstrated more variable efficacy. Apatinib, a selective VEGFR-2 tyrosine kinase inhibitor, provides modest benefit in previously treated disease and is approved for advanced gastric cancer in China. Regorafenib has also demonstrated activity in pretreated patients, although neither agent has established a role as standard first-line therapy. Differences in toxicity, patient selection, and the absence of reliable predictive biomarkers may contribute to the variable efficacy observed with these agents[79-81].

A major limitation of anti-angiogenic therapy is the absence of a validated predictive biomarker. Circulating VEGF concentrations, VEGFR expression, angiogenic cytokines, and other candidate biomarkers have been evaluated, but none has demonstrated sufficient reproducibility for routine patient selection. Consequently, anti-angiogenic treatment remains primarily clinically rather than molecularly selected, in contrast to HER2-directed therapy[82,83].

Resistance may develop through activation of alternative angiogenic pathways and adaptive remodeling of the tumor microenvironment. Hypoxia, vascular remodeling, and compensatory signaling can reduce the durability of VEGFR inhibition. These limitations have stimulated investigation of combination strategies, particularly combinations of VEGF/VEGFR inhibition with immune checkpoint blockade. Butters et al[79] identified VEGF-immunotherapy combinations as an important area of ongoing investigation, while the 2026 review by Raimbourg et al[84] noted that newer bispecific approaches simultaneously targeting VEGF and PD-1/PD-L1 remain investigational and require mature survival data.

Combination strategies targeting additional signaling pathways are also being explored. Preclinical work by Jung et al[82] demonstrated greater tumor growth inhibition with simultaneous VEGFR and EGFR blockade than with either treatment alone; however, these findings remain preclinical and do not establish a clinical treatment strategy[85].

CLDN18.2, FGFR2b, and other emerging molecular targets: Molecular profiling has identified CLDN18.2, FGFR2b, MET, and selected genomic alterations as additional therapeutic targets in GC/GEJ adenocarcinoma. Among these, CLDN18.2 and FGFR2b have progressed furthest clinically, whereas MET and other molecular alterations remain primarily investigational[86,87].

CLDN18.2 is a gastric lineage-associated tight-junction protein that remains accessible on the surface of tumor cells following malignant transformation. Its expression in a substantial proportion of gastric cancers has established CLDN18.2 as a clinically relevant biomarker. Zolbetuximab, a monoclonal antibody directed against CLDN18.2, significantly improved progression-free and overall survival when added to platinum-fluoropyrimidine chemotherapy in patients with CLDN18.2-positive, HER2-negative advanced GC/GEJ adenocarcinoma in the phase III SPOTLIGHT and GLOW trials[88-90]. These studies established CLDN18.2 as a clinically actionable target beyond HER2.

CLDN18.2-directed treatment is now expanding beyond zolbetuximab. ADCs, bispecific antibodies, and CAR-T therapies are being evaluated, with early evidence suggesting activity across selected levels of CLDN18.2 expression. Liu et al[91] reported a 29% confirmed objective response rate and a median progression-free survival of 5.5 months with the CLDN18.2-directed ADC IBI343 in CLDN18.2-high GC/GEJ adenocarcinoma in a phase I study. Important challenges include variation in biomarker thresholds, heterogeneous antigen expression, GI toxicity, and uncertainty regarding optimal sequencing of CLDN18.2-directed therapies[92].

FGFR2b represents another clinically relevant target. FGFR2 amplification and FGFR2b overexpression identify a subset of GC/GEJ adenocarcinomas characterized by activation of downstream MAPK, PI3K/AKT, and STAT signaling. Kawakami[93] noted that FGFR2b IHC may identify FGFR2-dependent tumors more effectively than broader receptor tyrosine kinase testing. The phase II FIGHT study provided the clinical basis for development of bemarituzumab, particularly in tumors with higher FGFR2b expression, while subsequent phase III evaluation has sought to establish its definitive clinical role[94-100].

MET amplification and alterations involving KRAS, PIK3CA, and DNA damage-repair pathways remain investigational targets. Their relatively low frequency, biological heterogeneity, and lack of consistently validated predictive biomarkers have limited the development of established targeted therapies. Kawakami[93] also describes ADCs and bispecific antibodies as emerging approaches for MET and other receptor tyrosine kinase-driven tumors.

Overall, the evolution of targeted therapy in GC/GEJ adenocarcinoma illustrates a shift from a predominantly HER2-centered treatment paradigm toward a broader biomarker landscape incorporating HER2, CLDN18.2, FGFR2b, and other emerging molecular targets. Importantly, biomarker selection increasingly relies not only on genomic alterations but also on protein expression and quantitative immunohistochemical assessment. The established and emerging molecular targets, therapeutic agents, evidence grades, resistance mechanisms, and investigational strategies are summarized in Table 4.

Table 4 Molecularly targeted therapies in gastric and gastroesophageal junction adenocarcinoma.

Molecular target/pathway	Biomarker/alteration	Representative therapy	Current clinical role	Evidence grade	Major resistance mechanisms/emerging strategies	Representative sources
HER2	HER2 overexpression and/or amplification, assessed using validated IHC and ISH-based methods	Trastuzumab + platinum/fluoropyrimidine chemotherapy ± pembrolizumab; T-DXd	Established first-line HER2-directed therapy based on ToGA and KEYNOTE-811. T-DXd is an established treatment option after trastuzumab-based therapy based on DESTINY-Gastric01	A	HER2 heterogeneity or loss, MET/EGFR/FGFR bypass signaling, PI3K pathway activation, and clonal evolution. HER2-low disease, next-generation ADCs, bispecific antibodies, CAR-T approaches, ctDNA-guided monitoring, and combination immunotherapy remain under investigation	Bang et al[72]; Shitara et al[74]; Nakamura et al[95]; Janjigian et al[73]
VEGFR-2	Not biomarker selected	Ramucirumab ± paclitaxel	Established second-line therapy based on REGARD and RAINBOW	A	Adaptive angiogenic resistance, alternative pro-angiogenic pathway activation, and vascular remodeling. Combinations with chemotherapy and immune checkpoint inhibitors are under investigation	Fuchs et al[77]; Wilke et al[78]; Nakamura et al[95]
VEGFR-2	Not biomarker selected	Apatinib	Later-line anti-angiogenic therapy, with regulatory approval in China	B	Alternative angiogenic pathway activation and adaptive vascular signaling	Wang et al[96]; Yuan et al[97]
VEGFR1–3	Not biomarker selected	Fruquintinib	Emerging anti-angiogenic therapy under clinical evaluation in gastric and GEJ cancer	C	Adaptive angiogenesis and pathway redundancy; combination approaches remain investigational	Wang et al[96]
CLDN18.2	CLDN18.2-positive, generally HER2-negative disease	Zolbetuximab + chemotherapy	Established first-line targeted therapy for appropriately selected CLDN18.2-positive, HER2-negative advanced gastric or GEJ adenocarcinoma based on SPOTLIGHT and GLOW	A	Heterogeneous antigen expression, variable assay thresholds, antigen loss, and treatment-related gastrointestinal toxicity. CLDN18.2 ADCs, bispecific antibodies, CAR-T therapy, and improved biomarker classification are under investigation	Shitara et al[74]; Nakamura et al[95]; Derks and van Laarhoven[88]
FGFR2b	FGFR2b overexpression	Bemarituzumab + chemotherapy	Clinical evidence from the phase II FIGHT study; further validation and treatment positioning remain necessary	B	FGFR2b expression heterogeneity, pathway bypass, and uncertainty regarding optimal biomarker thresholds. Earlier-line treatment and combination strategies are under investigation	Wainberg et al[100]; Yuan et al[97]
MET	MET amplification or other activating MET alterations	MET inhibitors	Early clinical or investigational development; no established routine role	C	Biomarker heterogeneity, pathway redundancy, and limited predictive consistency. Biomarker refinement and combination targeted therapy are under investigation	Wang et al[96]; Hsu et al[98]
HER2	HER2-positive refractory disease	Disitamab vedotin	Clinical activity reported in selected HER2-positive disease, with ongoing development	B	HER2 heterogeneity, antigen loss, payload resistance, and altered intracellular trafficking. ADC-based strategies are being developed to improve activity in heterogeneous or treatment-resistant disease	Shao et al[75]; Zhang et al[99]
HER2	HER2-positive disease	Zanidatamab; tucatinib-based therapy; pyrotinib	Emerging or region-specific clinical evidence; not uniformly established across gastric/GEJ cancer treatment settings	C	HER2 heterogeneity, downstream pathway reactivation, and acquired resistance. Dual or multi-agent HER2 blockade and treatment-sequencing strategies remain under investigation	Hu et al[68]; Shao et al[75]; Zhang et al[99]
KRAS/PI3K and related signaling pathways	KRAS or PI3K pathway alterations	Selective pathway inhibitors and molecularly matched combinations	Early clinical or investigational development	C	Pathway redundancy, feedback activation, and compensatory signaling. Combination pathway inhibition requires biomarker refinement and prospective validation	Yuan et al[97]; Kawakami[93]
VEGF/VEGFR and immune pathways	Not biomarker selected	VEGF/VEGFR inhibitor + immune checkpoint inhibitor	Investigational strategies with preliminary activity	C	Uncertain biomarker selection, adaptive angiogenesis, immune resistance, and overlapping toxicity. Optimal sequencing and combination partners remain under investigation	Yuan et al[97]; Kawakami[93]
VEGF + PD-1/PD-L1	Not biomarker selected	VEGF/PD-1 or PD-L1 bispecific antibodies	Early clinical investigation	C	Requires prospective validation of efficacy, safety, biomarker selection, and treatment sequencing	Yuan et al[97]; Kawakami[93]
MET/EGFR/FGFR/PI3K	Molecularly selected or pathway-driven disease	Combination targeted strategies	Investigational	C	Bypass signaling, clonal evolution, pathway redundancy, and intratumoral heterogeneity	Nakamura et al[95]

Evidence grading: Grade A, randomized phase III evidence and/or established standard-of-care treatment; Grade B, prospective phase II evidence or substantial clinical evidence; Grade C, early-phase, retrospective, or otherwise limited clinical evidence; Grade D, preclinical or experimental evidence. Evidence grades reflect the maturity and clinical validation of the available evidence and do not directly indicate the magnitude of treatment benefit. ADC: Antibody–drug conjugate; CAR-T: Chimeric antigen receptor T-cell therapy; CLDN18.2: Claudin 18.2; ctDNA: Circulating tumor DNA; EGFR: Epidermal growth factor receptor; FGFR: Fibroblast growth factor receptor; FGFR2b: Fibroblast growth factor receptor 2b; GC/GEJ: Gastric/gastroesophageal junction; HER2: Human epidermal growth factor receptor 2; MET: Mesenchymal–epithelial transition factor; PD-1: Programmed cell death protein 1; PD-L1: Programmed death-ligand 1; PI3K: Phosphatidylinositol-3-kinase; T-DXd: Trastuzumab deruxtecan; VEGF: Vascular endothelial growth factor; VEGFR: Vascular endothelial growth factor receptor.

Targeted therapy in GIST

KIT- and PDGFRA-targeted therapy: GIST represents one of the clearest examples of genotype-directed cancer therapy. Activating alterations in KIT or PDGFRA account for approximately 85%-90% of GISTs and drive downstream signaling through the MAPK, PI3K/AKT/mechanistic target of rapamycin (mTOR), and JAK/STAT pathways. KIT alterations are present in approximately 75%-80% of tumors, most commonly involving exon 11, whereas PDGFRA alterations occur in approximately 5%-10%, with the PDGFRA D842V variant representing the most clinically important subtype because of its characteristic resistance to imatinib. Accordingly, molecular genotyping is central to therapeutic selection in GIST[101,102].

Imatinib, a tyrosine kinase inhibitor targeting KIT and PDGFRA, remains the standard first-line systemic treatment for most patients with unresectable or metastatic GIST and has established roles in selected neoadjuvant and adjuvant settings. Clinical studies have demonstrated high response rates in KIT-mutant disease, although therapeutic sensitivity varies according to the underlying genotype. Tumors harboring KIT exon 11 mutations generally demonstrate the greatest sensitivity to standard-dose imatinib, whereas selected patients with KIT exon 9 mutations may derive additional benefit from dose escalation to 800 mg/day. In contrast, PDGFRA D842V-mutant GIST is intrinsically resistant to imatinib because of structural changes that impair effective drug binding[103,104].

The development of avapritinib illustrates the value of genotype-specific treatment. Avapritinib provides highly effective targeted therapy for PDGFRA exon 18-mutant GIST, particularly tumors harboring the D842V mutation. Wang et al[105], Mazzocca et al[106], and Schaefer et al[107] reported substantial response rates and durable disease control in this molecular subgroup, establishing avapritinib as the preferred targeted treatment for PDGFRA D842V-mutant disease. This experience demonstrates how molecular characterization can move beyond predicting response to a drug and directly determine the most appropriate therapeutic agent.

Despite the high initial efficacy of KIT-directed therapy, acquired resistance remains a major therapeutic challenge. Primary resistance is generally associated with intrinsically less-sensitive molecular subtypes, whereas secondary resistance commonly develops through additional KIT or PDGFRA kinase-domain mutations. Different secondary mutations may emerge in separate metastatic deposits, resulting in substantial intratumoral and intermetastatic heterogeneity. Duensing and Duensing[102] further demonstrated that KIT inhibition may induce cellular quiescence rather than complete tumor-cell eradication, potentially allowing residual tumor cells to persist and contribute to subsequent disease progression[102,103].

Molecular reassessment at progression can therefore provide clinically relevant information regarding the mechanisms underlying treatment failure. NGS can identify primary and secondary KIT or PDGFRA alterations and may assist in selecting subsequent tyrosine kinase inhibitors according to the resistance genotype. Similarly, ctDNA offers a minimally invasive approach for longitudinal assessment of clonal evolution and emerging resistance, although its routine role in treatment selection remains insufficiently validated[104,105,107].

The therapeutic paradigm in GIST has consequently evolved from a uniform TKI-based approach toward molecularly stratified treatment and genotype-directed sequencing. Genotyping at diagnosis remains essential for identifying patients likely to benefit from imatinib or avapritinib, while molecular reassessment at progression may help characterize acquired resistance and inform subsequent treatment strategies. Emerging approaches incorporating serial ctDNA analysis and NGS-guided treatment selection may further enable dynamic adaptation of therapy as the molecular landscape evolves.

The principal genotype-directed therapeutic strategies, their clinical evidence, and emerging approaches are summarized in Table 5.

Table 5 KIT proto-oncogene receptor tyrosine kinase- and platelet-derived growth factor receptor alpha-targeted therapies in gastrointestinal stromal tumors.

Molecular alteration	Approximate frequency	Preferred targeted therapy	Clinical role	Evidence grade	Resistance/emerging strategy	Representative sources
KIT exon 11 mutation	60%–70%	Imatinib 400 mg/day	Established first-line therapy; generally associated with the greatest sensitivity to imatinib	A	Secondary KIT mutations, clonal evolution, and acquired resistance; molecular reassessment may guide subsequent therapy	Li and Raut[104]; Wang et al[105]
KIT exon 9 mutation	10%–15%	High-dose imatinib 800 mg/day	Dose escalation may improve disease control, particularly in advanced disease	A	Secondary KIT mutations and clonal evolution; NGS-guided treatment selection and sequencing	Li and Raut[104]; Jakhetiya et al[103]
PDGFRA D842V mutation	5%–6%	Avapritinib	Preferred targeted therapy; this genotype is characteristically resistant to imatinib	A	Secondary resistance and clonal evolution; molecular monitoring and resistance-directed strategies remain under investigation	Mazzocca et al[106]; Wang et al[105]
Other PDGFRA mutations	Rare	Imatinib or alternative therapy according to mutation subtype	Variable sensitivity based on the specific PDGFRA alteration	B	Mutation-specific sensitivity and resistance require detailed molecular characterization	Li and Raut[104]
Wild-type GIST, including SDH-deficient, NF1-associated, and BRAF-associated tumors	10%–15%	Individualized treatment and clinical-trial enrollment	Limited response to conventional KIT-directed TKIs; management depends on the underlying molecular subtype	C	Alternative molecular drivers and genotype-specific treatment strategies are under investigation	Schaefer et al[107]; Wang et al[105]
Secondary KIT/PDGFRA resistance mutations	Variable	Sequential TKI therapy guided by molecular profiling	Molecular characterization can support selection and sequencing of subsequent therapies	B	Intratumoral heterogeneity, polyclonal resistance, and clonal evolution	Li and Raut[104]; Wang et al[105]

Evidence classification: Grade A, established therapy supported by prospective randomized evidence and/or incorporated into clinical guidelines or routine standard-of-care practice; Grade B, clinically supported therapy based on prospective clinical studies: Non-randomized comparative studies: Or consistent clinical evidence with more limited validation; Grade C, emerging or investigational strategy supported by early-phase trials: Small prospective or retrospective studies: Or preliminary clinical data; Grade D, preclinical or experimental strategy supported primarily by laboratory or animal studies without sufficient clinical validation. Evidence grades describe the maturity and clinical validation of the evidence and do not necessarily indicate the magnitude of treatment benefit. BRAF: B-Raf proto-oncogene; GIST: Gastrointestinal stromal tumor; KIT: KIT proto-oncogene receptor tyrosine kinase; NF1: Neurofibromatosis type 1; NGS: Next-generation sequencing; PDGFRA: Platelet-derived growth factor receptor alpha; SDH: Succinate dehydrogenase; TKI: Tyrosine kinase inhibitor.

Targeted therapy in pancreatic cancer

KRAS-targeted therapy: KRAS mutations occur in approximately 90%-95% of PDAC, making KRAS the principal oncogenic driver and a major focus of precision therapeutic development. The most frequent variants are G12D, G12V, and G12R, which activate downstream RAF-MEK-ERK, PI3K-AKT-mTOR, and related signaling pathways. Despite its central biological role, direct KRAS inhibition remained challenging for decades because of the high affinity of KRAS for GTP and the lack of suitable drug-binding pockets. Li et al[108], Leroux and Konstantinidou[109], and Xing et al[110] identify KRAS-directed treatment as a major area of contemporary precision oncology research in PDAC.

The development of KRAS G12C inhibitors provided the first clinical proof that direct pharmacological inhibition of mutant KRAS is feasible. Sotorasib and adagrasib irreversibly inhibit the G12C-mutant protein and have demonstrated objective responses and disease stabilization in molecularly selected patients with PDAC. Although KRAS G12C occurs in only approximately 1%-2% of PDAC, these findings established proof of concept and stimulated development of inhibitors against the substantially more prevalent KRAS variants[111].

KRAS G12D, present in approximately 40% of PDACs, has consequently become a major focus of next-generation inhibitor development. Mutation-specific small molecules, RNA-based and antisense approaches, and proteolysis-targeting chimeras are being investigated. Xing et al[110] also describe ADCs, aptamer-drug conjugates (ApDCs), peptide-drug conjugates (PDCs), and nanoparticle-based delivery systems as emerging approaches intended to improve tumor-specific KRAS-directed drug delivery. These strategies remain predominantly investigational.

Durable control with direct KRAS inhibition may be limited by adaptive and acquired resistance. Potential mechanisms include reactivation of downstream signaling, receptor tyrosine kinase activation, PI3K/AKT signaling, YAP activation, metabolic adaptation, and interactions with the tumor stroma. These mechanisms have prompted investigation of rational combinations involving MEK, SHP2, EGFR, immune checkpoint, and stromal targets. However, these approaches remain investigational because robust evidence demonstrating durable clinical benefit is currently limited[108].

A clinically important consequence of the predominance of KRAS mutations is the identification of a smaller KRAS-wild-type PDAC subgroup, estimated at approximately 8%-10% of tumors. Li et al[108] describe potentially actionable alterations in this subgroup, including BRAF mutations, RET and NTRK fusions, anaplastic lymphoma kinase (ALK) rearrangements, ROS proto-oncogene 1 (ROS1) fusions, and EGFR alterations. Comprehensive genomic profiling is therefore particularly relevant in KRAS-wild-type disease to identify candidates for matched therapy and biomarker-directed clinical trials.

DNA damage repair pathway targeting (PARP inhibitors): Defects in HRR represent one of the most clinically established actionable molecular subgroups in PDAC. Germline BRCA1/2 mutations occur in approximately 5%-10% of patients, while alterations involving PALB2, ATM, CHEK2, and other DNA damage repair genes define additional, less well-validated subgroups. Loss of BRCA-mediated homologous recombination creates a dependence on alternative DNA repair mechanisms and provides the biological rationale for PARP inhibition through synthetic lethality[112,113].

Olaparib remains the best-established PARP inhibitor in PDAC. In the phase III POLO trial, Golan et al[114] demonstrated significantly prolonged progression-free survival with maintenance olaparib compared with placebo in patients with metastatic PDAC harboring germline BRCA mutations whose disease had not progressed following platinum-based chemotherapy. The trial established maintenance olaparib as a biomarker-directed treatment option in this molecular subgroup, although a definitive overall survival advantage was not demonstrated.

The potential applicability of PARP inhibition extends beyond BRCA1/2 alterations. PALB2 and other HRR alterations may confer sensitivity to PARP inhibition, but prospective evidence remains limited and the predictive significance of individual alterations is heterogeneous. The concepts of homologous recombination deficiency (HRD) and “BRCAness” have therefore generated interest in identifying patients who may benefit despite the absence of a BRCA mutation. Comprehensive molecular profiling may help characterize these subgroups, although functional HRD biomarkers require further clinical validation[115-117].

Resistance remains an important limitation to the durability of PARP inhibition. Reported mechanisms include BRCA reversion mutations restoring homologous recombination, altered replication-fork protection, restoration of DNA end resection, increased drug efflux, and activation of alternative DNA repair pathways. Tumor heterogeneity and clonal evolution may further contribute to treatment failure. Li et al[108] and Leroux and Konstantinidou[109] identify characterization of resistance biology as an important area for development of subsequent DNA damage repair-directed strategies.

Several combination strategies are being investigated to extend the therapeutic benefit of PARP inhibition. These include combinations with ATR or CHK1 inhibitors, platinum chemotherapy, immune checkpoint inhibitors, anti-angiogenic agents, and KRAS-directed therapies. Although biologically rational, these approaches remain investigational and require prospective clinical validation before incorporation into routine practice[118-120].

Molecular assessment is also expanding beyond germline BRCA testing to include tumor sequencing and characterization of broader HRR alterations. ctDNA and functional HRD assays are being evaluated for longitudinal treatment monitoring and detection of emerging resistance; however, their clinical utility for selecting patients for PARP inhibitor therapy remains to be established[121,122].

EGFR and other emerging molecular targets: Although most PDACs harbor KRAS alterations, approximately 8%-10% are KRAS wild-type and may contain alternative actionable molecular abnormalities. Li et al[108], Leroux and Konstantinidou[109], and Xing et al[110] describe potentially targetable alterations involving EGFR, NTRK, RET, ALK, ROS1, BRAF, HER2, and DNA damage repair pathways. Because each alteration is uncommon, comprehensive genomic profiling is preferable to isolated hotspot testing when evaluating advanced PDAC for precision treatment opportunities.

EGFR inhibition has demonstrated limited clinical value in PDAC. Erlotinib combined with gemcitabine produced a statistically significant but modest survival improvement and represents an early example of targeted therapy in pancreatic cancer. However, the high prevalence of activating KRAS mutations allows downstream signaling to remain active despite EGFR blockade, substantially limiting the therapeutic effect of receptor inhibition. Current research has therefore shifted toward combining EGFR inhibition with KRAS, SHP2, MEK, or PI3K pathway inhibitors rather than relying on EGFR blockade alone[123-126].

NTRK fusions occur in fewer than 1% of pancreatic cancers but represent an important actionable alteration because fusion-positive tumors may respond to selective TRK inhibitors. RET rearrangements and ALK or ROS1 fusions are similarly uncommon but clinically relevant when identified, particularly because they may provide access to matched molecular therapies. The evidence specific to PDAC remains limited because of the rarity of these alterations[127,128].

HER2 amplification and BRAF mutations represent additional uncommon alterations, particularly within KRAS-wild-type disease. HER2-directed antibodies and ADCs, as well as BRAF/MEK inhibition, have been evaluated in molecularly selected patients. However, the available evidence remains largely derived from early clinical studies and case-based reports, and these approaches should therefore be distinguished from established systemic treatments for PDAC[129-131].

ADCs and novel drug-delivery platforms: ADCs and other targeted drug-delivery platforms represent emerging approaches in PDAC. ADCs combine recognition of tumor-associated antigens with intracellular delivery of potent cytotoxic payloads, while ApDCs and PDCs are being explored as alternative strategies for targeted drug delivery. Xing et al[110] also describe nanoparticle-based systems designed to improve therapeutic delivery within the dense desmoplastic tumor microenvironment characteristic of PDAC. These approaches remain predominantly preclinical or in early clinical development, and their eventual clinical role remains to be established.

Overall, targeted therapy in PDAC remains characterized by a contrast between strong biological rationale and a relatively small number of clinically established biomarker-directed treatments. Direct KRAS inhibition has provided proof of concept, while maintenance olaparib represents the most established genotype-directed strategy beyond conventional chemotherapy. The major priorities for further development include effective inhibition of prevalent KRAS variants such as G12D, overcoming adaptive resistance, refining HRD biomarkers, and identifying rare actionable alterations through comprehensive molecular profiling[132].

The molecularly targeted treatment landscape of PDAC is summarized in Table 6.

Table 6 Molecularly targeted therapies in pancreatic ductal adenocarcinoma: Established targets, therapeutic agents, evidence, and emerging strategies.

Molecular target/pathway	Approximate frequency	Biomarker/molecular subgroup	Representative therapy	Evidence grade	Current clinical role	Major challenges/emerging strategies	Representative sources
KRAS	90%–95%	KRAS-mutant PDAC	Direct KRAS inhibitors and pathway-directed strategies	B	Principal molecular target under active clinical development	Adaptive pathway reactivation and stromal resistance; combinations with SHP2, SOS1, MEK or EGFR inhibition	Li et al[108]; Leroux and Konstantinidou[109]; Xing et al[110]
KRAS G12C	1%–2%	KRAS G12C	Sotorasib, adagrasib	B	Early clinical activity in molecularly selected PDAC	Adaptive resistance and rational combination strategies	Li et al[108]; Leroux and Konstantinidou[109]
KRAS G12D	Approximately 40%	KRAS G12D	Mutation-specific KRAS inhibitors	C	Early clinical development	Allele-specific inhibition, pathway reactivation, and acquired resistance	Li et al[108]; Xing et al[110]
KRAS G12V/G12R	Common	KRAS G12V/G12R	Allele-specific inhibitors	C	Preclinical and early clinical development	Development of effective mutation-specific inhibitors	Leroux and Konstantinidou[109]; Li et al[108]
BRCA1/2 and HRR	5%–10% germline BRCA1/2; other HRR alterations less common	Germline BRCA1/2 or selected HRR-deficient PDAC	Olaparib maintenance after platinum-based chemotherapy; other PARP/DDR-directed strategies	A	Olaparib is an established maintenance option for appropriately selected germline BRCA1/2-mutated disease; broader HRR-directed approaches remain investigational	Acquired resistance and identification of PARP-sensitive tumors beyond BRCA1/2	Golan et al[114]; Li et al[108]; Leroux and Konstantinidou[109]
PALB2	Rare	PALB2 mutation	PARP inhibitors and platinum-based strategies	C	Selected/investigational patients	Prospective validation of predictive value and definition of PARP-sensitive subgroups	Leroux and Konstantinidou[109]; Li et al[108]
EGFR	Common	EGFR-expressing PDAC	Erlotinib + gemcitabine	A	Randomized evidence demonstrates a modest benefit, with limited contemporary clinical role	Constitutive KRAS signaling limits efficacy; combinations with KRAS/SHP2/MEK/PI3K inhibition	Li et al[108]; Leroux and Konstantinidou[109]
NTRK	< 1%	NTRK fusion	TRK inhibitors	B	Tissue-agnostic targeted therapy for fusion-positive tumors	Comprehensive genomic profiling to identify rare fusion-positive patients	Li et al[108]
RET	< 1%	RET fusion	Selective RET inhibitors	B	Rare molecularly selected indication	Limited PDAC-specific evidence; prospective validation	Li et al[108]
HER2	Rare	HER2 amplification, particularly KRAS-wild-type PDAC	HER2-targeted therapy, trastuzumab-based regimens, ADCs	C	Early clinical/case-based evidence	Biomarker refinement and combination HER2-targeted strategies	Li et al[108]; Xing et al[110]
BRAF	Rare	BRAF-mutant PDAC	BRAF ± MEK inhibitors	C	Limited molecularly selected evidence	Combination MAPK inhibition and prospective validation	Leroux and Konstantinidou[109]; Li et al[108]

Evidence classification: Grade A, established therapy supported by prospective randomized evidence and/or incorporated into clinical guidelines or routine standard-of-care practice; Grade B, clinically supported therapy based on prospective clinical studies, non-randomized comparative studies, or consistent clinical evidence with more limited validation; Grade C, emerging or investigational strategy supported by early-phase trials, small prospective or retrospective studies, or preliminary clinical data; Grade D, preclinical or experimental strategy supported primarily by laboratory or animal studies without sufficient clinical validation. Evidence grades describe the maturity and clinical validation of the evidence and do not necessarily indicate the magnitude of treatment benefit. ADC: Antibody–drug conjugate; ALK: Anaplastic lymphoma kinase; ApDC: Aptamer–drug conjugate; ATR: Ataxia telangiectasia and Rad3-related protein; BRAF: B-Raf proto-oncogene; DDR: DNA damage response/repair; EGFR: Epidermal growth factor receptor; HRD: Homologous recombination deficiency; HRR: Homologous recombination repair; MEK: Mitogen-activated protein kinase kinase; NGS: Next-generation sequencing; PARP: Poly (ADP-ribose) polymerase; PDAC: Pancreatic ductal adenocarcinoma; PDC: Peptide–drug conjugate; PI3K: Phosphatidylinositol 3-kinase; PROTAC: Proteolysis-targeting chimera; ROS1: ROS proto-oncogene 1; SHP2: Src homology region 2 domain-containing phosphatase 2; SOS1: Son of sevenless 1; TKI: Tyrosine kinase inhibitor; TRK: Tropomyosin receptor kinase.

CLINICAL SYNTHESIS OF MOLECULARLY GUIDED THERAPIES

Across GI malignancies, the clinical value of molecularly guided therapy varies according to the strength of the biomarker-drug relationship, the availability of prospective clinical evidence, and the treatment setting. The most established examples include HER2-directed therapy in HER2-positive GC/GEJ adenocarcinoma, BRAF-directed therapy in BRAF V600E-mutated mCRC, FGFR inhibition in FGFR2-altered CCA, IDH1 inhibition in IDH1-mutated CCA, KIT/PDGFRA-directed therapy in molecularly defined GISTs, and olaparib maintenance in selected patients with germline BRCA1/2-mutated PDAC. These approaches demonstrate that biomarker selection can improve treatment efficacy and refine patient selection beyond histology alone.

In mCRC, the magnitude of benefit is particularly dependent on molecular context. EGFR-directed therapy is most relevant in RAS wild-type disease, whereas BRAF inhibition requires combination treatment to address pathway reactivation and has demonstrated clinically meaningful benefit in BRAF V600E-mutated tumors. HER2-directed therapies and KRAS G12C inhibitor-EGFR antibody combinations have expanded treatment options for molecularly selected, previously treated patients. However, the available evidence is heterogeneous across platforms, and response rates, progression-free survival, and overall survival outcomes are not directly comparable between trials conducted in different lines of therapy and patient populations.

In GC/GEJ adenocarcinoma, trastuzumab-based therapy established HER2 as a clinically actionable target, while T-DXd provided an additional treatment option after trastuzumab exposure. More recently, CLDN18.2-directed therapy has established a further biomarker-defined treatment strategy in appropriately selected patients. Nevertheless, HER2 heterogeneity, antigen expression thresholds, loss of target expression, and acquired resistance remain important limitations. Similarly, in CCA, FGFR2 and IDH1 alterations identify clinically actionable subgroups, although the evidence is often derived from molecularly selected or non-randomized populations and should not be extrapolated to unselected disease.

In GISTs, the relationship between genotype and treatment sensitivity is especially well defined. KIT exon 11-mutated tumors are generally highly sensitive to imatinib, KIT exon 9-mutated tumors may require dose escalation, and PDGFRA D842V-mutated tumors are preferentially treated with avapritinib because of their characteristic resistance to imatinib. These examples illustrate the importance of mutation-specific treatment selection and molecular reassessment at the time of progression.

In PDAC, the clinical impact of targeted therapy remains more restricted. Olaparib maintenance is an established option for selected patients with germline BRCA1/2-mutated disease who have not progressed during platinum-based treatment; however, the pivotal evidence demonstrated a progression-free survival benefit without a statistically significant overall survival improvement. KRAS-directed therapy, particularly for KRAS G12C-mutated tumors, has shown early clinical activity, while strategies targeting the more prevalent KRAS G12D, G12V, and G12R alterations remain under active development. The dense stromal environment, intratumoral heterogeneity, adaptive signaling, and limited frequency of actionable non-KRAS alterations continue to constrain the broader application of precision therapy in PDAC.

Overall, the available evidence supports a selective rather than universal model of precision oncology in GI cancer. Established benefit is concentrated in molecularly defined subgroups for which a target has been validated in prospective clinical studies. By contrast, many emerging strategies remain supported by early-phase, single-arm, retrospective, or preclinical evidence. Accordingly, molecular profiling should be integrated with tumor type, treatment line, performance status, prior therapy, resistance mechanisms, and the maturity of available clinical evidence rather than used as an isolated determinant of treatment selection.

CROSS-TUMOR CHALLENGES AND FUTURE DIRECTIONS

Cross-tumor challenges

Biological resistance: Despite improving outcomes in selected GI cancer subgroups, targeted therapies frequently encounter primary or acquired resistance. Primary resistance may reflect inadequate target dependence, coexisting genomic alterations, intratumoral heterogeneity, or activation of compensatory pathways. Acquired resistance can arise through secondary target mutations, bypass signaling, phenotypic adaptation, and treatment-driven clonal evolution. Although several resistance mechanisms are clinically validated, others remain investigational and require prospective confirmation. In routine practice, resistance is further complicated by comorbidity, cumulative treatment toxicity, and the more heterogeneous clinical characteristics of patients compared with those enrolled in clinical trials[133].

Biomarker testing: Accurate biomarker identification is essential for effective targeted therapy, but testing remains limited by assay variability, differences in interpretation, tissue insufficiency, and intratumoral or temporal heterogeneity. These challenges are particularly relevant for biomarkers such as HER2, PD-L1, and CLDN18.2, where expression thresholds and scoring methods influence treatment eligibility. Comprehensive NGS may improve detection of actionable genomic alterations, although access remains uneven, particularly in resource-constrained settings. ctDNA can complement tissue testing and facilitate longitudinal assessment of molecular evolution, but its sensitivity may be limited in low-shedding tumors and for selected fusions or copy-number alterations[134]. Thus, biomarker testing is clinically useful when analytically validated and linked to an actionable treatment decision, whereas many emerging liquid-biopsy applications remain investigational. The principal biomarker-testing considerations relevant to the therapies discussed in this review are summarized in Table 7.

Table 7 Biomarker testing considerations for molecularly targeted therapy in gastrointestinal malignancies.

Biomarker	Principal GI malignancies	Preferred/commonly used assay	Key interpretation considerations	Tissue–liquid biopsy considerations	Role in treatment selection
HER2 amplification/overexpression	Gastric/GEJ, colorectal, biliary	IHC ± ISH/FISH; NGS may assess amplification	IHC scoring and predefined positivity thresholds should follow the assay/clinical trial context; heterogeneity may affect interpretation	Plasma NGS may identify amplification but does not necessarily replace tissue assessment	Selects patients for HER2-directed therapy
CLDN18.2 expression	Gastric/GEJ	IHC using validated antibody/scoring approach	Positivity depends on staining intensity, proportion of positive tumor cells, and the threshold used in the supporting clinical study; spatial heterogeneity may occur	Plasma testing is not an established substitute for tissue protein-expression assessment	Selects patients for CLDN18.2-directed therapy
FGFR2 fusion/rearrangement	Intrahepatic cholangiocarcinoma	NGS, RNA-based fusion testing, FISH or other validated molecular assays	Assay choice influences detection of fusion/rearrangement; DNA-only assays may have limitations for some structural alterations	Plasma ctDNA can identify some alterations but may have lower sensitivity for fusions/rearrangements, particularly in low-shedding tumors	Identifies candidates for FGFR-directed therapy
IDH1 mutation	Intrahepatic cholangiocarcinoma	PCR-based assay or NGS	Specific pathogenic IDH1 alterations should be distinguished from non-actionable variants	Plasma testing may be complementary but tissue remains important when plasma is negative or inconclusive	Selects patients for IDH1-directed therapy
BRAF V600E	Colorectal, biliary and other selected GI cancers	PCR, IHC where validated, or NGS	Specific V600E alteration should be confirmed using an appropriate validated assay	Plasma ctDNA can provide complementary mutation detection; negative results may require tissue confirmation	Selects patients for BRAF-directed therapy
KRAS/NRAS mutations	Colorectal	PCR-based hotspot testing or NGS	RAS status should include relevant KRAS and NRAS alterations when determining eligibility for EGFR-directed therapy	ctDNA may provide a non-invasive alternative or complementary assessment, but false negatives can occur with low tumor shedding	Determines eligibility for or resistance to selected targeted therapies
KRAS G12C	Colorectal; rare pancreatic and other GI tumors	PCR or NGS	Detection of the specific G12C alteration is required; broader KRAS positivity is not equivalent	Plasma NGS may detect G12C but negative plasma testing does not necessarily exclude a tissue alteration	Selects patients for KRAS G12C-directed strategies
NTRK fusion	Rare across GI malignancies	RNA-based NGS, DNA/RNA hybrid NGS, FISH or validated fusion assay	Fusion detection requires an assay capable of identifying relevant structural events; IHC may be useful for screening but generally requires molecular confirmation	Plasma detection is possible but less established than tissue-based fusion testing	Supports tissue-agnostic TRK-directed therapy
RET fusion	Rare GI malignancies	NGS or other validated fusion assays	Fusion detection requires appropriate structural-variant methodology	Plasma testing may be complementary but sensitivity depends on tumor shedding and assay characteristics	Supports RET-directed therapy in molecularly selected patients
KIT/PDGFRA mutations	GIST	NGS or targeted molecular testing	Mutation subtype and exon are clinically relevant because sensitivity to individual TKIs varies	Plasma ctDNA may help identify secondary resistance alterations but is not a universal substitute for tumor genotyping	Guides initial TKI selection and reassessment after resistance
BRCA1/2 and other HRR alterations	Pancreatic cancer and selected GI malignancies	Germline testing plus tumor NGS where appropriate	Germline and somatic alterations should be distinguished; pathogenicity and clinical relevance require appropriate interpretation	Plasma testing may complement tissue analysis but should not automatically replace germline assessment	Identifies selected patients for platinum/PARP or other DNA-repair-directed strategies
MSI-H/dMMR	Colorectal, gastric/GEJ and other GI malignancies	MMR IHC and/or MSI PCR/NGS	Assay concordance and appropriate interpretation of equivocal results are important	Plasma MSI assessment remains complementary and less established than validated tissue testing in many settings	Biomarker for precision immunotherapy rather than conventional targeted therapy
TMB-H	Multiple GI malignancies	Validated NGS-based genomic profiling	Thresholds are assay-specific and should not be assumed to be interchangeable across platforms	Plasma-derived TMB may differ from tissue TMB because of assay and biological factors	May identify selected patients for precision immunotherapy where clinically validated
ctDNA/molecular residual disease	Multiple GI malignancies	Plasma NGS or other validated ctDNA platforms	Primarily a monitoring/longitudinal biomarker rather than a therapeutic target; interpretation depends on assay sensitivity and timing	Particularly vulnerable to false negatives in low-shedding or localized disease; serial testing may improve longitudinal assessment	May support molecular monitoring and detection of emerging resistance; treatment-changing utility remains context dependent

Biomarker thresholds and assay requirements should be interpreted according to the validated assay, pivotal clinical study, and applicable regulatory/clinical diagnostic framework. Tissue and plasma assays are complementary rather than universally interchangeable. MSI-H/dMMR and TMB-H are included because they guide precision immunotherapy rather than conventional molecularly targeted therapy. CISH: Chromogenic in situ hybridization; ctDNA: Circulating tumor DNA; FISH: Fluorescence in situ hybridization; GEJ: Gastroesophageal junction; GI: Gastrointestinal; GIST: Gastrointestinal stromal tumor; HRR: Homologous recombination repair; IHC: Immunohistochemistry; ISH: In situ hybridization; MSI-H: Microsatellite instability-high; NGS: Next-generation sequencing; PCR: Polymerase chain reaction; TKI: Tyrosine kinase inhibitor; TMB-H: Tumor mutational burden-high; CLDN18.2: Claudin 18.2.

Toxicity, cost, and access: Targeted therapies are not uniformly less toxic than conventional treatment. Toxicity varies by drug class and may include dermatologic reactions, infusion-related events, hypertension, thromboembolic complications, endocrine or ocular effects, GI symptoms, cytopenias, and organ-specific adverse events. Combination regimens may increase treatment burden and compromise treatment continuity. High drug prices, the cost of companion diagnostics, limited molecular-testing infrastructure, and unequal access to specialist services further restrict implementation. Consequently, clinical utility depends not only on molecular efficacy but also on tolerability, cost-effectiveness, diagnostic availability, and equitable access[135].

Future directions

Rational targeted therapy plus immunotherapy: Combining targeted agents with immune checkpoint inhibitors is supported by the possibility that pathway inhibition may alter tumor vasculature, immune-cell infiltration, antigen presentation, or the tumor microenvironment. Clinical application is already evident in selected settings, including HER2-directed therapy combined with pembrolizumab in HER2-positive gastric cancer. However, the benefit of such combinations is not universal and may depend on the molecular subgroup, immune context, treatment sequence, and toxicity profile. At present, this approach has established utility in selected indications, while many other combinations remain investigational and require biomarker-driven prospective evaluation.

ADCs, bispecific antibodies, and next-generation agents: ADCs, bispecific antibodies, and next-generation kinase inhibitors may expand the therapeutic scope of precision oncology by improving target delivery, addressing antigen heterogeneity, or overcoming selected resistance mechanisms. T-DXd demonstrates the clinical potential of ADC technology in HER2-positive disease, while CLDN18.2-directed ADCs and cellular or bispecific approaches remain emerging strategies. Next-generation FGFR inhibitors, mutation-specific KRAS inhibitors, and improved IDH-directed therapies may provide options for tumors with acquired resistance or previously limited treatment alternatives. Tissue-agnostic therapies directed against alterations such as NTRK fusions further support molecularly defined treatment. Nevertheless, the clinical utility of many newer platforms remains dependent on confirmation in prospective studies, appropriate patient selection, and demonstration of meaningful patient benefit.

AI and real-world evidence: AI and machine-learning approaches may support precision oncology by integrating genomic, radiologic, histopathologic, and clinical data. Potential applications include biomarker classification, prediction of treatment response, identification of resistance-associated phenotypes, and longitudinal assessment of disease progression. Real-world evidence from registries, electronic health records, and molecular databases can complement randomized trials by evaluating treatment effectiveness, toxicity, treatment patterns, and outcomes in broader patient populations. However, AI-based tools and real-world datasets are currently most useful as decision-support and evidence-generation instruments. Their use for autonomous treatment selection or prediction of individual outcomes requires external validation, assessment of bias, interpretability, and prospective clinical evaluation.

Personalized sequencing and treatment adaptation: As tumors evolve under treatment pressure, the molecular profile obtained at diagnosis may not fully represent the biology of progressive disease. Repeat tissue profiling and serial ctDNA assessment may therefore help identify emerging resistance alterations and inform subsequent treatment selection. Molecularly guided sequencing, targeted rechallenge, switching between agents, and rational combination therapy may be useful when a specific resistance mechanism has been demonstrated. Maintenance strategies and treatment de-escalation may also reduce cumulative toxicity in patients with sustained disease control. However, these approaches currently have selective or investigational utility, and prospective studies are needed to establish whether biomarker-guided sequencing, adaptive treatment, or de-escalation improves survival, quality of life, or overall treatment burden.

LIMITATIONS

This review has several limitations. First, it was designed as a narrative review rather than a systematic review or meta-analysis; therefore, the literature search and selection process was structured but did not follow a formal systematic-review framework with protocol registration, formal risk-of-bias assessment, or quantitative pooling of treatment effects. Consequently, the findings should be interpreted as a clinically oriented synthesis rather than an estimate of comparative treatment efficacy.

Second, substantial heterogeneity across the available evidence limited direct comparison between therapies. Studies differed in tumor type and molecular subtype, treatment line, prior therapy, biomarker definitions, assay platforms, comparator arms, follow-up duration, and reported endpoints. In addition, several molecularly selected populations are uncommon, resulting in small prospective cohorts, single-arm studies, retrospective series, or basket trials. These factors make cross-trial comparisons of response rates, progression-free survival, overall survival, and toxicity potentially misleading.

Third, the maturity of evidence varies considerably between molecular targets. Some therapies are supported by randomized trials and established clinical practice, whereas others are supported primarily by phase II, early-phase, retrospective, or preclinical evidence. The author-defined evidence grades used in this review provide a framework for describing evidence maturity but are not a validated evidence-rating instrument and should not be interpreted as equivalent to formal GRADE assessments or regulatory classifications.

Fourth, biomarker assessment is itself a source of uncertainty. Differences in IHC scoring, positivity thresholds, sequencing platforms, tissue quality, and interpretation can affect identification of eligible patients. This is particularly relevant for HER2, CLDN18.2, FGFR2 fusions/rearrangements, and other heterogeneous biomarkers. Tissue and liquid-biopsy results may also be discordant, while ctDNA assays can have reduced sensitivity in low-shedding or localized tumors and for some structural alterations. Therefore, the absence of an alteration on plasma testing should not necessarily be interpreted as definitive exclusion of a tissue-detectable actionable alteration.

Fifth, resistance mechanisms are not equally validated across targets. Some mechanisms, such as secondary kinase-domain alterations or bypass signaling, have clinical or translational evidence, whereas others remain based primarily on laboratory or early clinical observations. The rapidly evolving molecular landscape also means that resistance mechanisms identified in earlier treatment eras may not fully reflect current treatment sequences or combination strategies.

Sixth, the review includes rapidly evolving and investigational treatment platforms, including next-generation KRAS inhibitors, ADCs, bispecific approaches, and novel targeted-delivery systems. Evidence for these approaches may change substantially as ongoing trials mature, and early response signals may not ultimately translate into durable clinical benefit.

Finally, regulatory approvals, companion-diagnostic requirements, and treatment availability are jurisdiction-specific and subject to change. Clinical evidence, regulatory status, guideline recommendations, and practical access were therefore considered distinct dimensions. The review should consequently be interpreted in the context of the applicable contemporary regulatory and clinical practice environment rather than as a universal treatment algorithm. These limitations highlight the need for prospective biomarker-standardized trials, longer follow-up, harmonized diagnostic methods, and real-world evaluation of treatment effectiveness and resistance across molecularly defined GI cancer populations.

CONCLUSION

Molecularly targeted therapies have shifted GI oncology toward biomarker-driven treatment, with established benefits in selected molecular subgroups across colorectal, gastric, biliary, pancreatic, and GISTs. The strongest evidence supports therapies directed against validated targets, including EGFR, VEGF, HER2, BRAF, FGFR2, IDH1, KIT, PDGFRA, and BRCA-associated DNA-repair defects. Future progress will depend on improving biomarker standardization, overcoming acquired resistance, expanding access to molecular testing and targeted treatments, and evaluating rational combinations and emerging therapeutic platforms.