Molecular imaging in gastric cancer: state-of-the-art techniques and emerging opportunities

Abstract

Background

Gastric cancer (GC) remains a major global health challenge with persistently poor survival rates, largely attributable to late-stage diagnosis and the inherent limitations of conventional imaging modalities, including CT, MRI, and FDG PET/CT, in detecting early lesions, identifying peritoneal metastases, and capturing tumor heterogeneity. There is a critical unmet need for non-invasive imaging approaches capable of visualizing GC-specific molecular biomarkers to enable more accurate diagnosis, risk stratification, and personalized treatment planning.

Main body

Molecular imaging integrates high-sensitivity tracer-based techniques with high-resolution anatomical imaging, allowing in vivo visualization of key biological processes and molecular targets in GC. This review provides a comprehensive overview of recent advances in molecular imaging platforms for GC, including PET, SPECT, molecular optical imaging, and molecular MRI. We summarize the biological rationale, imaging mechanisms, and emerging clinical relevance of representative molecular targets, such as HER2, PD-L1, CLDN18.2, FAP, VEGF, and Trop2. In addition, we discuss the development and application of diverse classes of imaging probes, including monoclonal antibodies, single-domain antibodies, affibodies, peptides, and nanomaterials, highlighting their roles in improving lesion detection, delineating intratumoral heterogeneity, guiding targeted and immune-based therapies, and supporting image-guided surgical and interventional procedures. Key challenges in clinical translation, including dynamic target expression, suboptimal pharmacokinetics, and the lack of standardized imaging protocols, are also critically examined.

Conclusions

Despite existing translational barriers, continued advances in probe engineering, chemical synthesis, and imaging instrumentation are rapidly expanding the clinical potential of molecular imaging in GC. These developments are expected to facilitate more precise diagnosis, treatment monitoring, and patient stratification, ultimately supporting the broader goals of precision oncology and personalized management of gastric cancer.

Introduction

Gastric cancer (GC) is a malignant neoplasm arising from the epithelial lining of the gastric mucosa, with the vast majority of cases classified as adenocarcinomas [1]. Less common histological subtypes include lymphomas and gastrointestinal stromal tumors. As the fifth most common malignancy and the third leading cause of cancer-related mortality worldwide, GC continues to carry a poor prognosis, with a five-year survival rate below 40% [2, 3]. This unfavorable outcome primarily stems from the nonspecific nature of early symptoms, such as epigastric discomfort and bloating, which often lead to delayed diagnosis.

Imaging plays a critical role in the diagnosis and clinical management of GC. Widely used modalities, including ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI), are vital for tumor localization, staging, assessment of resectability, and therapeutic planning. Their non-invasive nature and ability to provide detailed anatomical and functional information offer substantial advantages in clinical practice. However, these conventional techniques have notable limitations, particularly their limited sensitivity for detecting early-stage disease and insufficient accuracy in evaluating micrometastases or subtle therapeutic responses [4]. Consequently, histopathological confirmation through endoscopic biopsy remains the gold standard for definitive diagnosis. Because early GC is often asymptomatic, many patients are diagnosed at advanced stage, by which point the tumor has typically invaded locally or metastasized, leaving curative surgery feasible for only a minority. Therefore, improving early detection and advancing the development of more accurate molecular imaging modalities constitute critical priorities for future research and clinical translation.

Molecular imaging, a non-invasive approach that enables visualization of biological processes at cellular and molecular levels, offering a promising avenue for improving diagnostic precision. Nuclear medicine modalities, such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT), are particularly valuable due to their high sensitivity and specificity. Molecular imaging can also be integrated with optical imaging and conventional imaging techniques to enhance diagnostic accuracy and provide complementary information. Fluorine-18-fluorodeoxyglucose ([¹⁸F]FDG) PET/CT remains the most widely utilized molecular imaging modality for cancer, enabling visualization of tumor glucose metabolism. However, FDG uptake in GC is highly variable, often weak in certain histological subtypes, leading to false-negative results, and it can also produce false positives due to benign inflammatory conditions such as gastritis and peptic ulcers. These challenges underscore the need to shift from conventional metabolic imaging toward more precise, target-specific diagnostic strategies.

In this review, we summarize the contemporary diagnostic imaging landscape for GC and clarify key concepts in molecular imaging, emphasizing the biological foundations and signaling pathways underlying major biomarkers and therapeutic targets. We then provide a systematic and up-to-date evaluation of targeted molecular imaging strategies, spanning PET, SPECT, optical molecular imaging, and molecular MRI, with attention to diagnostic performance, clinical relevance, and emerging limitations reported over the past five years (Fig. 1). Finally, we critically examine the barriers that continue to impede clinical translation, including dynamic target expression, pharmacokinetic constraints of imaging probes, and methodological heterogeneity, and propose strategic directions to guide future development. Together, these discussions aim to establish a comprehensive framework for understanding the evolving role of molecular imaging in the personalized management of GC.

Fig. 1.

Molecular imaging modalities for gastric cancer. This review systematically summarizes recent advances in molecular imaging for gastric cancer, highlighting major technological platforms including PET, SPECT, molecular optical imaging, and molecular MRI. It focuses on the biological rationale, imaging mechanisms, and clinical relevance of key targets—including HER2, PD-L1, CLDN18.2, FAP, VEGF, and Trop2. PET, positron-emission tomography; SPECT, single-photon emission computed tomography; MRI, magnetic resonance imaging. Reproduced from ref [5]. Available under a CC-BY 4.0 license. Copyright 2021, Wen, Xia, Guo, Huang, Wang, Yang, Yang and Zhu. Reproduced from ref [6]. Available under a CC-BY 4.0. Copyright 2024, Ibrahim, Simó, Brown, Shmuel, Panikar, Benton, DeWeerd, Dehdashti, Park and Pereira. Reproduced from ref [7]. Copyright 2024, American Chemical Society

Diagnostic imaging of gastric cancer

Accurate diagnosis and staging form the foundation of GC management, directly informing therapeutic decisions ranging from endoscopic resection and surgical gastrectomy to neoadjuvant chemotherapy and palliative strategies. GC is diagnosed primarily through endoscopy, which allows precise tumor localization, macroscopic characterization, and acquisition of biopsy samples for histological confirmation. However, comprehensive staging relies heavily on medical imaging results.

CT is widely available, fast, and effective for evaluating gastric wall thickening, extragastric extension, and distant metastases, particularly in the liver and lungs. For preoperative staging, CT reliably detects serosal invasion with reported sensitivity up to 82.8–100% [8]. A systematic review of 10 studies [9] revealed considerable variability in CT’s ability to detect lymph node metastases, with median sensitivity and specificity of 80.0% and 77.8%, respectively. The main constraint stems from CT’s reliance on size-based criteria: non-enlarged metastatic nodes reduce sensitivity, whereas enlarged inflammatory nodes reduce specificity. Moreover, CT performs poorly in identifying small peritoneal implants or early peritoneal carcinomatosis (sensitivity ~ 23–76%) [8]. Such false negatives may underestimate disease burden and compromise surgical planning. Recent advances such as dual-energy CT (DECT) have improved lesion conspicuity through low-keV monochromatic reconstruction, reducing under-staging in some studies. However, its routine use remains limited by hardware availability and a lack of standardized reconstruction protocols [10, 11].

Magnetic resonance imaging (MRI), particularly multiparametric MRI (mpMRI) incorporating T2-weighted, diffusion-weighted (DWI), and dynamic contrast-enhanced (DCE) sequences, offers superior soft-tissue contrast and valuable functional information. In a prospective, paired single-center trial [11], Li et al. demonstrated that mpMRI achieved significantly higher accuracy than DECT for both T-staging (61–77% vs. 50–64%) and N-staging (54–68% vs. 51–58%). These findings are consistent with earlier smaller studies [12, 13] highlighting MRI’s advantage in depicting perigastric fat invasion and nodal morphology. Functional metrics such as the apparent diffusion coefficient (ADC) have also been correlated with tumor grade and chemotherapy response [14]. Nevertheless, mpMRI is limited by motion artefacts, longer acquisition times, and availability, and thus currently serves as a complementary, rather than primary, staging modality in most clinical centers.

[¹⁸F]FDG PET/CT has an established role in identifying distant metastatic disease and resolving indeterminate CT or MRI findings. However, its sensitivity in GC is highly variable and strongly influenced by tumor histology. Signet-ring cell carcinoma, mucinous tumors, and other diffuse-type cancers frequently show low FDG avidity due to low GLUT-1 expression or high mucin content, leading to false negatives for primary tumors and peritoneal metastases [15]. A meta-analysis of primary staging in GC reported pooled sensitivity and specificity of only 49% and 92% for nodal staging, and 56% and 97% for detecting distant metastases [16].

These limitations have driven interest in molecularly targeted imaging agents that investigate tumor biology beyond glycolytic activity. Unlike conventional CT, MRI, or [¹⁸F]FDG PET, which primarily reflect anatomical features or glucose metabolism, novel molecular tracers are designed to target specific biomarkers associated with GC pathogenesis—such as cell surface receptors, tumor angiogenesis, hypoxia, and immune microenvironment activity. Advances in radiochemistry and probe engineering have enabled the development of tracers that bind to targets including HER2, integrin αvβ3, FAP, among others. Early translational and preclinical studies suggest that these tracers provide improved lesion-to-background contrast and can detect disease sites that remain occult on FDG PET/CT. Moreover, by quantifying molecular expression and pharmacodynamic changes, targeted molecular imaging opens the possibility for non-invasive tumor characterization, treatment response prediction, and patient stratification.

Molecular imaging techniques in the diagnosis of gastric cancer

Molecular imaging uses specific probes to label biological targets, achieving precise detection of early disease-associated changes and providing a basis for personalized diagnosis and treatment [17]. Current targeted molecular imaging techniques for GC include nuclear medicine imaging, optical imaging, and molecular MRI.

Nuclear medicine imaging

Positron emission tomography (PET) imaging

PET is a functional imaging modality that employs molecular probes labeled with positron-emitting isotopes to evaluate tumor biology, including metabolism, receptor expression, and other biochemical processes. PET offers high sensitivity and enables quantitative evaluation of molecular pathways, such as glucose metabolism and amino acid transport. Unlike anatomical imaging techniques, PET characterizes physiological and metabolic alterations associated with disease, making it particularly valuable in oncology, neurology, and cardiology.

The most widely used PET tracer, [¹⁸F]FDG, enters highly metabolic cells via glucose transporters and is phosphorylated by hexokinase but cannot progress through glycolysis, unlike native glucose, resulting in intracellular accumulation. This property enables visualization of metabolically active tissues and underpins the use of [¹⁸F]FDG PET in detecting and characterizing malignancies, evaluating neurological disorders, and assessing myocardial viability.

In GC, numerous studies have demonstrated the value of [¹⁸F]FDG PET in detecting lymph node metastasis, and metabolic parameters such as maximum standardized uptake value (SUVmax) have prognostic relevance in patients with nodal involvement [18, 19]. However, no international guidelines mandate PET for GC stage, largely due to the variable and often limited uptake of FDG in several gastric tumor subtypes [20]. This substantially diminishes PET’s sensitivity and specificity for GC detection.

To address these limitations, novel PET radiotracers with enhanced target specificity have been introduced. ImmunoPET, which utilizes antibodies or peptides labeled with radioactive isotopes, enables precise targeting of tumor-associated biomarkers overexpressed on GC cells [21]. These probes significantly improve molecular specificity, reduce nonspecific background uptake, and enhance diagnostic accuracy.

Single photon emission computed tomography (SPECT) imaging

SPECT is a functional imaging modality that reconstructs three-dimensional tomographic images by detecting gamma rays emitted from radiolabeled compounds within the body [22]. Commonly used tracers, such as technetium-99 m, accumulate in target tissues to reveal metabolic activity or blood perfusion patterns. Due to limited spatial resolution and quantitative accuracy, SPECT is frequently combined with CT to integrate functional and anatomical information [23]. SPECT/CT has demonstrated clinical utility in detecting bone and lymph node metastases, monitoring disease recurrence, and evaluating therapeutic responses in GC [24, 25]. However, its overall imaging sensitivity remains lower than PET, limiting its utility in accurately detecting small or early GC lesions [26]. Future advances in SPECT for GC are anticipated to center on the development of molecular probes with high specificity for GC-associated antigens, thereby improving diagnostic accuracy and therapeutic monitoring.

Optical-based imaging

Optical molecular imaging utilizes fluorescent dyes or bioluminescent reporters to bind specific biological targets. The emitted visible or near-infrared (NIR) signals are captured by light detectors, enabling real-time visualization of biological processes [27, 28].

Fluorescence molecular imaging

Fluorescence molecular imaging (FMI) utilizes fluorescent probes to label biomolecules, such as tumor markers [29]. When excited by light of specific wavelengths, these probes emit detectable fluorescence, enabling real-time dynamic monitoring of biological activities. FMI has shown significant clinical utility in early lesion detection during endoscopy, intraoperative tumor margin delineation, and treatment efficacy monitoring [30–32]. Common fluorescent probes include Cy5.5, quantum dots (QDs), and indocyanine green (ICG). Emerging probes operating in the second near-infrared window (NIR-II, 900–1880 nm) exhibit enhanced tissue penetration and improved contrast-to-noise ratio. When combined with AI-based image analysis, NIR-II imaging holds promise for improved depth discrimination and quantitative accuracy [33, 34]. Specifically, unsupervised CycleGAN algorithms have been validated for wide-field cross-modality image translation to achieve high-definition NIR-IIb resolution from standard dyes, while supervised pix2pix models have been employed in light-sheet microscopy to computationally mitigate tissue scattering and extend imaging depth [35]. Although primarily validated in vascular and generic tumor models, the extrapolation of these AI-enhanced NIR-II techniques to gastric cancer holds significant potential for visualizing submucosal invasion. Future research is expected to emphasize multimodal probes capable of integrating multiple imaging modalities with integration of deep learning technology, as well as smart-responsive fluorescent probes that activate in response to specific physiological or pathological stimuli [36, 37].

Bioluminescence imaging

Bioluminescence imaging (BLI) is a non-invasive optical imaging modality based on endogenous light-emitting biochemical reactions [38]. With genetic engineering, luciferase genes can be stably introduced into target cells, including tumor cells [39]. Upon administration of the corresponding substrate, enzymatic reactions produce bioluminescent signals that are subsequently captured by highly sensitive charge-coupled device cameras [40, 41]. Although BLI has been explored for intraoperative guidance, its clinical translation is currently limited by relatively shallow tissue penetration, restricting its application primarily to small animal models. BLI remains a powerful tool in preclinical research for real-time monitoring of tumor growth, metastasis, and treatment response [42]. Significant technological and methodological advances will be required before BLI can be realistically considered for human imaging.

Molecular magnetic resonance imaging (MRI)

Molecular MRI integrates the high spatial resolution of conventional MRI with targeted molecular probes that bind specific biomarkers [43]. Upon binding, these probes alter MR signal properties by influencing the local magnetic environment. Unlike nuclear medicine modalities (PET/SPECT), which possess high sensitivity capable of detecting tracers in the picomolar to nanomolar range (10^{− 12}–10^{− 9} M), MRI is inherently limited by lower sensitivity, typically requiring contrast agent concentrations in the micromolar to millimolar range (10^{− 6}–10^{− 3} M) to generate detectable signal changes [44]. To address this limitation, molecular MRI probes often incorporate multiple contrast-generating components (e.g., nanoparticle-based systems) or utilize enzyme-activated signal amplification mechanisms. As a result, molecular MRI has become an important tool in biomedical research for early disease detection, precise monitoring of therapeutic responses, and evaluation of novel therapeutics [45].

Despite this preclinical promise, the clinical translatability of molecular MRI remains challenging compared to established ImmunoPET techniques. The high probe concentration required for MRI necessitates rigorous safety profiles to avoid toxicity, and accurate quantification of biomarker expression is often confounded by magnetic field inhomogeneities and baseline tissue signals. Consequently, molecular MRI is currently best positioned as a complementary preclinical tool for characterizing tumor heterogeneity as well as guiding interventions.

Molecular mechanisms of targets and biomarkers in gastric cancer

A detailed understanding of the molecular mechanisms underlying key targets and biomarkers in GC is critical for the development of targeted therapies and advanced molecular imaging strategies. In this section, we provide a concise overview of the principal molecular drivers currently utilized in therapeutic targeting and molecular imaging of GC.

Human epidermal growth factor receptor 2 (HER2)

HER2 is a transmembrane tyrosine kinase receptor that belongs to the erythroblastosis oncogene (ERBB) family. Approximately 13–22% of patients with GC exhibit HER2 overexpression or gene amplification, with higher prevalence observed in the intestinal subtype and cancers arising at the gastroesophageal junction [46, 47]. Under physiological conditions, HER2 regulates cellular proliferation, differentiation, and survival through multiple signal transduction pathways [48]. However, gene amplification or protein overexpression promotes homodimer formation and constitutive activation of downstream signaling cascades, including the phosphatidylinositol 3-kinase (PI3K)/protein kinase B (PKB/AKT) and rat sarcoma (RAS)/mitogen-activated protein kinase (MAPK) pathways [49]. This aberrant signaling enhances tumor cell proliferation, inhibits apoptosis, and increases invasive and metastatic potential [50].

Programmed cell death ligand 1 (PD-L1)

Immune checkpoint blockade has become a major therapeutic strategy for advanced GC. Programmed cell death protein 1 (PD-1), an immune checkpoint receptor expressed on T cells, B cells, and other immune populations, binds to its ligand PD-L1, which is expressed on GC cells and tumor-infiltrating immune cells such as macrophages [51]. Engagement of PD-L1 with PD-1 delivers inhibitory signals that suppress T-cell activation, cytokine production, and cytotoxicity, thereby facilitating tumor immune evasion [52, 53]. Moreover, PD-L1 expression in GC is dynamically regulated by inflammatory cytokines, including interferon (IFN)-α, IFN-β, and IFN-γ, as well as oncogenic signaling pathways, including PI3K and nuclear factor-κB (NF-κB) [54, 55]. These regulatory mechanisms collectively establish an immunosuppressive tumor microenvironment that supports tumor progression.

Claudin 18.2 (CLDN18.2)

CLDN18.2 is an isoform of the tight junction protein claudin 18, generated through alternative gene splicing [56]. It is predominantly localized at the tight junctions of gastric mucosal epithelial cells, where it maintains epithelial integrity and regulates paracellular permeability. CLDN18.2 is expressed in approximately 40–55% of GC cases, particularly within diffuse-type tumors as classified by the Lauren system [57–59]. CLDN18.2 influences several oncogenic pathways such as PD-1, wingless/integrated (Wnt), and extracellular signal-regulated kinase (ERK) signaling, thereby promoting tumor progression and metastasis [60–62]. While CLDN18.2 generally promotes cell proliferation, differentiation, and migration, it may exhibit tumor-suppressive functions in select gastric and lung cancer subtypes [63, 64]. This paradoxical role may be attributed to its regulatory effects on Wnt and Notch signaling pathways, as well as the downstream transcriptional coactivator Yes-associated protein (YAP) [64–66]. This context-dependent activity underscores its complex biological role in GC.

Fibroblast activation protein (FAP)

FAP is a type II transmembrane serine protease that is highly expressed in over 90% of cancer-associated fibroblasts (CAFs) in epithelial-derived tumors such as GC, while its expression in normal tissues is minimal [67]. FAP promotes epithelial-mesenchymal transition (EMT) via Wnt/β-catenin signaling and contributes to tumor invasion, angiogenesis, and metastasis [68, 69]. Furthermore, FAP has been shown to suppress T cell-mediated antitumor immunity, thereby supporting an immunosuppressive microenvironment conducive to tumor progression [70].

Fibroblast growth factor receptor (FGFR)

FGFRs are transmembrane receptor tyrosine kinases that regulate cell proliferation, differentiation, angiogenesis, and tissue repair under physiological conditions. In GC, FGFR signaling is frequently dysregulated due to gene amplification, chromosomal fusion events, or activating point mutations, leading to constitutive receptor activation. This aberrant activation contributes to tumor progression and malignancy.

FGFR dimerization, either homodimeric or heterodimeric, is stabilized by interactions with heparan sulfate proteoglycans (HSPGs), inducing autophosphorylation of FGFR intracellular domains and activating multiple downstream signaling pathways [71]. The adaptor protein fibroblast growth factor receptor substrate 2 alpha (FRS2α) interacts with phosphorylated tyrosine residues on the FGFR, promoting recruitment of the growth factor receptor-bound protein 2 (GRB2)–son of sevenless (SOS) complex [72]. This molecular interaction leads to the activation of both the RAS/MAPK and PI3K/AKT/mTOR signaling pathways, which drive tumor cell proliferation and survival [73].

FGFR signaling also contributes to tumor invasion and metastasis through the activation of the phospholipase C gamma (PLCγ)/diacylglycerol (DAG)/protein kinase C (PKC) and Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathways [74, 75]. In addition, FGFR signaling plays a pivotal role in remodeling the tumor microenvironment, promoting angiogenesis, and inducing the EMT in cancer cells, further enhancing tumor aggressiveness and metastatic capacity [76].

Vascular endothelial growth factor (VEGF)

VEGF is a key pro-angiogenic cytokine highly upregulated in GC and strongly correlated with tumor invasion, metastasis, and reduced patient survival [77]. Its oncogenic effects are mediated primarily through VEGF receptor-2 (VEGFR-2), which activates downstream signaling cascades such as the PI3K/AKT and RAS/MAPK pathways [78, 79]. These pathways stimulate endothelial cell proliferation, migration, and neovascularization, ultimately supplying tumors with essential nutrients and oxygen to support their growth [80]. In addition, VEGF enhances vascular permeability, facilitates the extravasation of plasma proteins, and contributes to the formation of a supportive tumor microenvironment that enables cancer cell migration and invasion [81]. Moreover, VEGF modulates immunity by inhibiting dendritic cell maturation and promoting regulatory T cell (Tregs) expansion, thereby suppressing anti-tumor immune responses and contributing to immune evasion [82, 83].

Trophoblast cell surface antigen 2 (Trop2)

Trop2 is a transmembrane glycoprotein that belongs to the tumor-associated calcium signal transducer (TACSTD) family and is highly expressed in many epithelial-derived tumors, including GC [84]. Its overexpression is closely linked to aggressive tumor behavior.

Trop2 synergistically drives the growth and metastasis of GC through several interconnected mechanisms. It can downregulate the epithelial marker E-cadherin and upregulate the mesenchymal marker N-cadherin, facilitating EMT and enhancing migratory and invasive capacity of tumor cells [84, 85]. Additionally, Trop2 can directly interact with β-catenin to activate the Wnt/β-catenin signaling pathway, further driving EMT and metastatic behavior [86].

Trop2 can also regulate calcium-dependent signaling through the phosphatidylinositol-4,5-bisphosphate (PIP2)/1,4,5-triphosphate inositol (IP3) pathway, resulting in intracellular calcium release and subsequent activation of the MAPK cascade to promote proliferation. Moreover, Trop2 can modulate other key signaling pathways, including PI3K/Akt and JAK/STAT, to inhibit apoptosis and enhance survival [86, 87]. Its ability to influence growth factor signaling, including interactions with insulin-like growth factor 1 (IGF-1) and receptor tyrosine kinase (RTK) class III members such as ErbB3, further amplifies its oncogenic potential [84].

Molecular imaging of gastric cancer

Targeting HER2

The most extensively studied ligands for targeting HER2 are monoclonal antibodies (mAbs), especially trastuzumabs. HER2-targeted PET imaging using radiolabeled tracers such as [⁸⁹Zr]Zr-trastuzumab, [¹²⁴I]I-trastuzumab, and [⁶⁴Cu]Cu-DOTA-trastuzumab offers a non-invasive method to assess HER2 status in GC [88, 89]. This approach overcomes key limitations of tissue biopsy, particularly tumor heterogeneity and sampling error, by visualizing HER2 expression across all metastatic sites. In diagnosis and staging, HER2-PET demonstrates high specificity for identifying HER2-positive lesions, often outperforming conventional imaging and guiding biopsy site selection. Therapeutically, HER2-PET is crucial for selecting patients for HER2-directed treatments, ensuring that only those with confirmed, accessible HER2-positive disease receive targeted therapies.

Based on trastuzumab, Wen et al. [5] developed a HER2-targeted multimodal nanoprobe, Her-PEG-dMNPs, by conjugating trastuzumab to biocompatible dopamine-melanin nanoparticles (dMNPs) labeled with ¹²⁴I (for PET) and Mn²⁺ (for MRI). This platform enabled chelator-free labeling and improved tumor targeting by enhancing the stability and pharmacokinetics of trastuzumabs. As a result, the probe demonstrated high specificity for HER2-expressing cells in vitro (cellular uptake of 3.61% ± 0.74% at 2 h) and strong tumor retention in GC patient-derived xenograft (PDX) models as evidenced by micro-PET/CT (Fig. 2A) and PET/MRI imaging (Fig. 2B) (SUVmax 1.63 ± 0.07 vs. 0.90 ± 0.04 for the non-targeted probe at 24 h). Reduced cardiac toxicity was also reported. These results support the nanoprobe’s potential for HER2-targeted theranostics in GC.

Fig. 2.

HER2 is an effective target for molecular imaging of GC. (A) Micro-PET/CT imaging of [124I]I-Her-PEG-dMNPs and [124I]I-PEG-dMNPs in a gastric cancer PDX model. (B) T1-weighted MR axial images of the GC PDX model of Mn-Her-PEG-dMNPs. Reproduced from ref [5]. Available under a CC-BY 4.0 license. Copyright 2021, Wen, Xia, Guo, Huang, Wang, Yang, Yang and Zhu. (C) Micro-PET/CT imaging of [18 F]F-ZHER2:342 in NCI-N87 (a) and in SGC7901 (1 h p.i.) models (c) acquired at different time points after administration. Blocking images of the NCI N87 models were also obtained at 1 h p.i. (b). (D) The uptake data were described as %ID/g. (E) In vitro cell uptake and specificity assays. Reproduced from ref [90]. Available under a CC-BY 4.0. Copyright 2019, Pan, Yang, Xu, Bai, Pan, Yang, Wang, Guan and Yang

Although mAb-based probes show excellent diagnostic and therapeutic performance, their large molecular weight leads to prolonged blood circulation times, necessitating delayed imaging, and poor tumor penetration, especially in dense tumors. To address these limitations, researchers have resorted to antibody fragments, nanobodies, and affibodies. Affibodies, engineered small scaffold proteins (∼7 kDa) derived from staphylococcal protein A, offer several advantages for HER2-targeted imaging in GC. Their compact size enables rapid tumor penetration and fast systemic clearance, achieving high tumor-to-background ratios (TBR) within hours while maintaining nanomolar affinity for HER2 [90–93]. Preclinical evaluation demonstrated that ¹⁸F-labeled HER2 affibody ([¹⁸F]AlF-NOTA-HER2) accumulated significantly more in HER2-positive NCI-N87 tumors (1.46 ± 0.65%ID/g at 60 min p.i.) than in HER2-negative MKN74 tumors (0.16 ± 0.01%ID/g, p = 0.016). Notably, the probe exhibits an exceptionally low hepatic uptake (< 0.6%ID/g at 1 h), ideal for gastric delineation compared with mAbs, however demonstrating intense renal retention (> 15%ID/g) [92].

To further overcome the limitations of nonspecific uptake and stability, researchers have focused on structural optimization through linker engineering and novel chelator design. Pharmacokinetic profiles can be optimized by incorporating hydrophilic linkers such as GGGRDN, which can reduce hepatic and intestinal uptake to < 2% ID/g (Fig. 2C-E) [90]. This markedly reduces abdominal background and enhances tumor-to-organ contrast and detection sensitivity of gastric and other abdominal lesions. Parallel advancements in chelator design have further improved probe performance. Substituting NOTA with RESCA significantly enhanced renal clearance and reduced nonspecific uptake. The resulting tracer, [¹⁸F]AlF-RESCA-HER2-BCH, demonstrated significantly lower renal accumulation (e.g., 5.16 ± 0.22% ID/g vs. 158.73 ± 5.44% ID/g in mice at 2 h) and higher TBR in metastatic lesions, highlighting its potential for both diagnostic and therapeutic applications [93]. In another study [94], TE1PA was introduced as a copper chelator for trastuzumab-based probes, demonstrating superior stability compared to DOTA/NOTA and offers a promising translational pathway for personalized copper theranostics in HER2-positive GC.

In terms of clinical translation, a notable clinical study using [68Ga]Ga-NOTA-MAL-MZHER2 [91] in advanced GC patients confirmed the feasibility of affibody-based HER2 imaging, achieving optimal imaging quality 2 h post-injection. Importantly, anti-HER2 therapy did not affect probe binding, as affibodies target non-overlapping epitopes, offering a distinct advantage for monitoring treatment response.

Targeting CLDN18.2

CLDN18.2 is a tight junction protein normally restricted to gastric mucosa. During oncogenic transformation, however, it becomes aberrantly exposed on the surface of cancer cells. Its high-frequency and tumor-specific expression in gastrointestinal malignancies, particularly gastric and pancreatic adenocarcinomas, makes it an attractive therapeutic target [95]. This recognition led to the development of targeted therapies, most notably the mAb zolbetuximab [96, 97].

Building on zolbetuximab and its analogs, researchers have explored radiolabeled mAbs for molecular imaging of GC. ¹²³I and ¹²⁵I labeled zolbetuximab exhibited excellent specific binding ability to CLDN18.2-expressing tumor cells and enabled clear visualization of tumors using SPECT/CT (Fig. 3A), thus emerging as a promising radiotracer for identifying potential patients eligible for CLDN18.2-targeted therapy [7, 98]. For PET/CT, Zhao et al. [99] were the first to apply 5C9, a highly specific anti-CLDN18.2 antibody generated via hybridoma technology, for in vivo imaging of GC. In parallel, they developed the NIR-II fluorescent probe FD1080-5C9, facilitating real-time visualization during tumor resection in orthotopic models. Together, these tools established a “diagnosis and image-guided surgery” platform with substantial translational potential for CLDN18.2-positive gastrointestinal cancers.

Fig. 3.

CLDN18.2 is an effective target for molecular imaging of GC. (A) Small-animal SPECT/CT imaging of tumor-bearing nude mice injected with [125I]I-zolbetuximab. Reproduced from ref [7]. Copyright 2024, American Chemical Society. (B) Section images of [89Zr]Zr-DFO-TST001 of tumor uptake 48 h p.i. were compared to section images of [18 F]FDG in BGC823 CLDN18.2 mice 1 h. (C) Tumor/heart at each point. Reproduced from ref [100]. Available under a CC-BY 4.0. Copyright 2023, Elsevier. (D) Representative fluorescence images of intravenous injection of SPIO@1D5 or SPIO@IgG in MFCCLDN18.2 and MFC subcutaneous xenograft tumor models at different time points. (E) Magnetic Particle Imaging (MPI) showing signal retention over time. (F) Quantitative comparison of normalized MPI signal of the nanoparticles. Reproduced from ref [107]. Available under a CC-BY 4.0. Copyright 2023, Wiley-VCH Verlag GmbH & Co. KGaA

Other mAbs have also been evaluated, including osemitamab (TST001), a second-generation humanized antibody with higher affinity and enhanced Fc receptor binding compared to zolbetuximab. When radiolabeled with ⁸⁹Zr, osemitamab exhibited excellent specificity and rapid tumor accumulation in CLDN18.2-positive GC, with a 2.51-fold higher cellular uptake in CLDN18.2-positive versus negative cells and a high tumor-to-muscle ratio (14.95 ± 1.63 at 96 h)(Fig. 3B-C) [100]. When labeled with ¹⁷⁷Lu [101], it also demonstrated potent antitumor efficacy. To optimize ligand structure and improve imaging performance, Li et al. [102] developed a novel scFv-Fc fusion protein (SF106) targeting CLDN18.2, radiolabeled with ¹²⁴I for PET imaging and ¹⁷⁷Lu for therapeutic purposes. The scFv-Fc construct combines the antigen-binding region of a single-chain variable fragment (scFv) with the Fc domain of IgG, thus preserving high affinity and bivalent binding while offering improved tumor penetration and pharmacokinetics compared with full-length mAbs.

Therapeutically, Wu et al. [98] demonstrated that a single 2 MBq dose of [¹³¹I]I-IMAB362 markedly suppressed tumor growth and reduced [¹⁸F]FDG uptake, with no observable toxicity as evidenced by stable body weight and normal histology. Advancing this concept, Liu et al. [103] developed an innovative radiolabeled antibody-drug conjugate (RADC), [¹³¹I]I-HLX58-Der, combining the fully human anti-CLDN18.2 monoclonal antibody HLX58, the radiotherapeutic isotope ¹³¹I, and the topoisomerase I inhibitor Deruxtecan. This novel RADC exhibited high tumor-specific accumulation (5.83 ± 0.41%ID/g in CLDN18.2-positive tumors) and potent anti-tumor efficacy, reducing tumor volume by 12.15-fold compared to controls and outperforming the non-radiolabeled HLX58-Der. This study pioneers the RADC paradigm, addressing the modest efficacy of conventional CLDN18.2-targeted therapies and advancing precision oncology toward synergistic, image-guided radio-chemotherapy.

In addition to antibody-based approaches discussed above, smaller ligands have increasingly been pursued. The first radiolabeled CLDN18.2-targeted nanobody was pioneered by Wei et al. [104], who evaluated ⁶⁸Ga-,¹⁸F-, and ⁶⁴Cu-labeled hu19V3, all of which showed excellent PET imaging capability and strong target specificity. In terms of tissue penetration, these nanobody-based tracers demonstrate rapid tumor accessibility, achieving high-contrast imaging of both primary tumors and disseminated peritoneal metastases within 1 h post-injection. This contrasts sharply with full-length antibodies, which are hindered by their large size and the “binding-site barrier”. Expanding this approach to SPECT/CT imaging, Bai et al. [105] developed a site-specifically labeled nanobody, [^99mTc]Tc-PHG102, which demonstrated high binding affinity (Kd = 24.01 nM) and rapid tumor accumulation (4.63 ± 0.81%ID/cc at 0.5 h) in xenografts. Polypeptides have also been explored as molecular imaging ligands, offering an alternative to nanobodies, which often exhibit excessive renal accumulation. For example, Wang et al. [106] successfully identified a CLDN18.2-specific peptide, T37, using phage display technology. The radiolabeled probe, [⁶⁸Ga]Ga-DOTA-T37, exhibited high specificity and affinity for CLDN18.2-positive tumors (SUVmax = 0.38 ± 0.01 vs. 0.24 ± 0.004). Importantly, the peptide probe exhibits rapid renal clearance, with kidney uptake dropping from 9.35 ± 2.03%ID/g at 5 min to 1.73 ± 0.10%ID/g at 2 h post-injection. This contrasts with the nanobody-based tracer hu19V3, which retain high renal radioactivity (> 10%ID/g) for prolonged periods, potentially obscuring adjacent abdominal lesions.

In the realm of multimodal imaging, Wang et al. [107] presented a novel CLDN18.2-targeted nanoplatform, SPIO@1D5-ICG, that integrates fluorescence and magnetic particle imaging (MPI) for spatiotemporally controlled magnetic hyperthermia (MHT) in GC. They developed 1D5, a new mAb with six-fold higher affinity than IMAB362 (zolbetuximab), and conjugated it to superparamagnetic iron oxide nanoparticles capable of dual-modal imaging (FMI/MPI) and localized heating (Fig. 3D). Guided by MPI, the nanoplatform achieved precise tumor targeting and optimal MHT timing (Fig. 3E-F), effectively inducing immunogenic cell death and reversing the immunosuppressive microenvironment. When combined with anti-PD-1 therapy, the strategy synergistically enhanced dendritic cell recruitment, CD8⁺ T-cell activation, and M1 macrophage polarization, resulting in robust antitumor immunity and significant tumor regression in both allograft and PDX models. This work provides a promising theranostic approach for CLDN18.2-positive GC patients who respond poorly to conventional immunotherapies.

Clinical translation of CLDN18.2 probes has progressed rapidly. A first-in-human study in 17 patients (including 12 with GC) using [¹²⁴I]I-18B10(10 L) confirmed high specificity and favorable dosimetry, enabling noninvasive lesion detection of treatment response, and highlighting ovarian metastases as imaging “hot spots” [108]. Similarly, first-in-human studies by Qi et al. ([⁶⁸Ga]Ga-NC-BCH, n = 11) [109] and Wang et al. ([⁶⁸Ga]Ga-PMD22, n = 16) [110] confirmed excellent safety, high tumor specificity, and strong correlation between PET uptake (SUVmax) and immunohistochemical CLDN18.2 expression in gastrointestinal cancers (r = 0.848, P < 0.01 for PMD22). Notably, [⁶⁸Ga]Ga-NC-BCH outperformed [¹⁸F]FDG PET in detecting lymph node and peritoneal metastases, while [⁶⁸Ga]Ga-PMD22 showed superiority in identifying CLDN18.2-positive signet ring cell and other low-metabolic cancers. Complementing these PET tracers, a recent SPECT/CT study [105] (n = 8) using the nanobody [^99mTc]Tc-PHG102 demonstrated clear primary tumor visualization with a favorable effective dose (0.148–0.218 mSv), offering a cost-effective alternative for patient screening.

PD-L1

PD-L1 has long been an important therapeutic target across multiple malignancies, including GC, colorectal cancer, renal cell carcinoma, and hepatocellular carcinoma, with clinically approved mAbs including atezolizumab, durvalumab, and avelumab. In recent years, efforts have been devoted to developing PET imaging approaches targeting PD-L1, aiming to overcome the inherent limitations of immunohistochemical analysis, which can only provide static, spatially restricted snapshots of PD-L1 expression.

Early work focused on repurposing existing therapeutic mAbs as imaging ligands. ¹¹¹In-labeled atezolizumab was first used for SPECT/CT imaging of PD-L1-high tumors, serving as a valuable tool for patient stratification, therapeutic monitoring, and intraoperative guidance within the framework of immune checkpoint blockade [111, 112]. In 2018 [113], [⁸⁹Zr]Zr-atezolizumab PET/CT was evaluated across multiple tumor types as a noninvasive method for assessing PD-L1 expression. Pretreatment uptake of [⁸⁹Zr]Zr-atezolizumab correlated more strongly with clinical response, progression-free survival, and overall survival than standard PD-L1 immunohistochemistry or RNA sequencing biomarkers. These results highlight the unique ability of immunoPET to capture the spatial and temporal heterogeneity of PD-L1 expression throughout the body.

Despite increasing clinical interest in PD-L1-targeted antibody drug conjugates and immunoPET tracers, consistent antibody binding remains challenging due to the complex spatiotemporal dynamics of PD-L1 expression, especially its dynamic N-linked glycosylation, which can mask epitopes and hinder antibody access [114, 115]. In 2024, Ibrahim et al. [6] investigated PD-L1 glycosylation in GC, demonstrating that PD-L1 expression in the NCI-N87 cell line is both time-dependent and spatially heterogeneous following IFN-γ stimulation. Western blotting, immunohistochemistry, and PET imaging confirmed that PD-L1 exists predominantly in a glycosylated form (~ 50 kDa) in GC cells and PDXs. Treatment with PNGase F, which removes N-linked glycans, shifted the protein to ~ 35 kDa, verifying N-linked glycosylation. Importantly, deglycosylation enhanced avelumab binding (Fig. 4D-E), suggesting that glycan structures sterically hinder antibody-epitope interactions. The glycosylation poses a critical challenge to imaging reliability, as steric hindrance can mask epitopes and lead to false-negative results when using standard monoclonal antibodies, underscoring the need for optimized radiotracer design, such as deglycosylation-compatible strategies or smaller ligands, to improve imaging accuracy and predictive value.

Fig. 4.

PD-L1 is an effective target for molecular imaging of GC. (A) Representative coronal PET images of two mice bearing bilateral PD-L1(+) L2987 (purple arrows) and PD-L1(-) HT-29 (yellow arrows) tumors at 90–120 min post [18 F]BMS-986,229 administration. The left image shows tracer alone; the right image shows co-administration of 2 mg/kg BMS-986,189 (PD-L1 peptide inhibitor). (B) Representative transverse PET images of mice bearing bilateral PD-L1(+) L2987 (purple arrows) and PD-L1(-) HT-29 (yellow arrows) tumors at 90–120 min after ligand administration. (i) shows [18 F]BMS-986,229 alone (ii) shows [18 F]BMS-986,229 with a 10 mg/kg BMS-936,559 (PD-L1 mAb) blocking dose (iii) shows [18 F]BMS-986,229 with a 2 mg/kg BMS-986,189 (PD-L1 peptide inhibitor) dose. (C) Representative time-activity curves, [18 F]BMS-986,229 baseline, [18 F]BMS-986,229 with a 2 mg/kg BMS-986,189 blocking dose. Reproduced from ref [121]. (D, E) Fluorescence intensity of fluorescently labeled avelumab in NCIN87 cells and gastric PDXs with and without PNGase F treatment. Reproduced from ref [6]. Available under a CC-BY 4.0. Copyright 2024, Ibrahim, Simó, Brown, Shmuel, Panikar, Benton, DeWeerd, Dehdashti, Park and Pereira

Beyond mAbs, low-molecular-weight probes have also been developed for PD-L1 imaging. WL12, a 15-amino acid cyclic peptide with high affinity for PD-L1, has served as a widely used scaffold for PET tracer development. Variants labeled with ¹⁸F, ⁶⁸Ga, and ⁶⁴Cu have all been reported [116–118]. Other peptide-based tracers include derivatives of the natural peptide TPP-1 labeled with ¹⁸F and ⁶⁴Cu, as well as the small peptide-based PET imaging agent, [¹⁸F]DK222 [119]. In 2022, Sun et al. [120] fused an albumin-binding moiety, IPB, with the PD-L1-targeting peptide, RK-10, thus prolonging blood circulation while preserving the pharmacokinetic benefits of a low molecular weight ligand. Although these probes have not yet been directly evaluated in GC models, they hold potential for imaging PD-L1-high tumors, including GC. In 2023, Donnelly et al. [121] reported the macrocyclic peptide-based radiotracer [¹⁸F]BMS-986,229, designed specifically for PET imaging of PD-L1 expression. The tracer was synthesized efficiently using click chemistry, showed picomolar affinity for PD-L1, and exhibited high target specificity, with an 8:1 binding ratio in vitro and five-fold higher uptake in PD-L1-positive tumors and non-human primate spleens (Fig. 4A-C). In 2024, a pilot clinical study confirmed that [¹⁸F]BMS-986,229 PET is a safe, feasible, and noninvasive imaging method for assessing PD-L1 expression in patients with gastroesophageal cancer [122].

Targeting FAP

FAP is a type II transmembrane serine protease that is highly expressed on CAFs within the tumor stroma of over 90% of epithelial carcinomas, while exhibiting minimal expression in normal adult tissues [67]. To exploit this target, a novel class of quinoline-based radiotracers, fibroblast activation protein inhibitors (FAPIs), was developed. These low-molecular-weight ligands, including FAPI-04 and FAPI-46, can be radiolabeled with various diagnostics or therapeutics isotopes and have rapidly gained prominence in molecular imaging.

FAPI-04 is the earliest, most extensively studied, and most widely used FAPI probe in clinical practice. Since 2021, multiple studies have investigated [⁶⁸Ga]Ga-DOTA-FAPI-04 for molecular imaging of GC (Fig. 5B) [123, 124]. Lin et al. [123] reported that [⁶⁸Ga]Ga-DOTA-FAPI-04 significantly outperformed [¹⁸F]FDG in detecting peritoneal (159 vs. 47 lesions, P < 0.001) and bone metastases (64 vs. 55, P = 0.003). Similarly, Zhang et al. [124] evaluated 25 patients for staging and restaging, finding that FAPI PET/CT exhibited higher sensitivity for primary tumors (94.7% vs. 68.4%, P < 0.01) and resulted in management changes in 28% of patients. A 2023 systematic review and meta-analysis [125] demonstrated that [⁶⁸Ga]Ga-FAPI PET, predominantly using FAPI-04, exhibits superior diagnostic performance compared with [¹⁸F]FDG PET in GC, with pooled sensitivity and specificity of 0.84 and 0.91 versus 0.46 and 0.88 for FDG. With higher sensitivity, specificity, SUVmax, and TBR, [⁶⁸Ga]Ga-FAPI provides enhanced visualization of tumors with low FDG avidity, such as signet ring cell carcinoma. Although heterogeneity across studies and regional bias remain limitations, these findings support the potential of [⁶⁸Ga]Ga-FAPI as a future standard in GC imaging, especially when combined with [¹⁸F]FDG for comprehensive staging. Qin et al. [126] further explored [⁶⁸Ga]Ga-FAPI PET/MR, suggesting that it offers improved staging accuracy and treatment response evaluation compared with PET/CT, owing to the superior soft-tissue contrast of multiparametric MRI. Although physiological uptake in reproductive organs remains a challenge, PET/MRI integration enhances diagnostic confidence, particularly in evaluating peritoneal and small-volume disease.

Fig. 5.

FAP is an effective target for molecular imaging of GC. (A) Micro-PET evaluation of the dual-targeting probe [18 F]F-FAPI-42-RGD versus the monospecific [18 F]F-FAPI-42 in MKN-45 xenografts, showing enhanced tumor uptake. Reproduced from ref [129]. Copyright 2025, e-Century Publishing Corporation. (B) Comparison of [18 F]F-FAPI-74 and [18 F]FDG PET/CT imaging in a patient with advanced gastric cancer. Reproduced from ref [124]. Available under a CC-BY 4.0 license. Copyright 2022, Zhang, Wang, Xu, Ding, Li, Liu, Huang, Liu, Du, Zhao, Chen and Qiu. (C) Representative comparison between [68Ga]Ga-FAPI-04 and [18 F]FDG PET/CT in a patient with advanced gastric cancer. Reproduced from ref [128]. Available under a CC-BY 4.0 license. Copyright 2024, the Society of Nuclear Medicine and Molecular Imaging

Beyond FAPI-04, Pang et al. [127] compared [⁶⁸Ga]Ga-DOTA-FAPI-46 and [⁶⁸Ga]Ga-DOTA-FAPI-2286 in a prospective cohort of 46 patients (including 19 with head-to-head comparison). [68Ga]Ga-FAP-2286 exhibited lower physiological uptake than [⁶⁸Ga]Ga-FAPI-46 in muscle, salivary glands, thyroid, and pancreas but higher uptake in the kidneys, liver, and heart. These differences underscore the importance of choosing the optimal tracer to balance tumor contrast with background signal depending on the clinical scenario. Notably, expanding the utility to SPECT/CT, Raeisi et al. [25] recently reported the first case of utilizing [^99mTc]Tc-FAPI-46 to successfully detect peritoneal carcinomatosis in gastric adenocarcinoma that was missed by contrast-enhanced CT, serving as a more cost-effective method to substitute PET/CT in certain clinical scenarios.

In addition to ⁶⁸Ga tracers, ¹⁸F-labeled FAPI have been increasingly adopted due to the production advantages of ¹⁸F and their potential for broader clinical use. A prospective study by Xu et al. [128] showed that [¹⁸F]F-FAPI-74 PET/CT outperformed [¹⁸F]FDG PET/CT in detecting primary tumors, local recurrences, lymph node involvement, and distant metastases in patients with gastric, hepatic, and pancreatic cancers (Fig. 5C). Notably in 2025, Zhao et al. [129] introduced [¹⁸F]F-FAPI-42-RGD, a novel heterodimeric PET tracer designed to simultaneously target FAP (via FAPI-42) and integrin αv​β3 (via cyclic RGD peptide). Across multiple GC xenograft models, [¹⁸F]F-FAPI-42-RGD demonstrated significantly higher tumor uptake and retention compared to the monospecific tracer [¹⁸F]F-FAPI-42, showing approximately four-fold higher uptake at 4 h p.i., along with excellent radiochemical stability and favorable TBR (Fig. 5A). These findings highlight the potential of dual-targeting strategies to overcome tumor heterogeneity and improve diagnostic accuracy in GC.

Despite its high sensitivity, FAP expression is not exclusive to malignancy. Significant FAPI uptake has been observed in benign processes involving active tissue remodeling, such as wound healing, arthritis, and IgG4-related disease, as well as in chronic inflammation [130, 131]. Therefore, careful interpretation of FAPI PET/CT images in the context of clinical history is essential to avoid potential false-positive diagnoses.

Others

Targeting VEGF

VEGF plays a pivotal role in the pathogenesis and progression of GC by stimulating angiogenesis, a process essential for tumor growth, invasion, and metastasis. Given its critical function, VEGF has emerged as a key therapeutic target in GC, with anti-angiogenic agents such as ramucirumab demonstrating clinical efficacy and improving survival in patients with advanced disease [132]. Zhu et al. [133] first reported a novel VEGFR1 specific 12-mer peptide, F56, and radiolabeled it with ¹²⁵I to enable non-invasive visualization of GC tumors via SPECT/CT imaging (Fig. 6A). The probe exhibited strong tumor targeting (tumor uptake: 2.1 ± 0.2% ID/g), favorable biodistribution, and receptor-specific binding, as evidenced by effective blocking with unlabeled F56. In 2020, the same group improved the imaging properties of the probe by labeling F56 with ¹¹¹In [134], which showed higher tumor uptake (4.9 ± 0.6% ID/g at 24 h) and lower hepatic accumulation (< 1.2% ID/g) than the previous ¹²⁵I-labeled version. These improvements resulted in clearer micro-SPECT images and improved tumor-to-liver contrast. For PET imaging, the group subsequently developed ⁶⁴Cu-DOTA-F56 [135], achieving high radiochemical purity and preserved VEGFR1 affinity. The tracer provided distinct VEGFR1-specific tumor visualization in BGC823 xenografts, with markedly higher tumor uptake (4.7–8.0% ID/g from 2 to 24 h) than earlier SPECT tracers and superior contrast compared with [¹⁸F]FDG or free ⁶⁴Cu.

Fig. 6.

VEGF, Trop2 and HER3 are effective targets for molecular imaging of GC. (A) Micro-SPECT/CT validation of the VEGFR1-targeted probe [125I]I–F56 in BGC-823 xenografts, showing specific tumor uptake blocked by excess peptide. Reproduced from ref [133]. Copyright 2017, American Chemical Society. (B) ImmunoPET imaging of [89Zr]Zr-DFO-Trodelvy in GC model. (C) Comparative biodistribution of [89Zr]Zr-DFO-Trodelvy in tumor-bearing mice, showing significantly higher accumulation in NCI-N87 tumors compared to MDST8, Blocking, and IgG groups. Reproduced from ref [136]. Copyright 2025, Springer Nature. (D) Coronal [68Ga]Ga-HER3P1 PET and overlaid PET and T1-weighted MR (PET/MRI) images of tumor-bearing mice at baseline and 4 days after treatment with afatinib demonstrates significantly increased PET uptake in NCI-N87 tumors on day 4. Reproduced from ref [139]. Available under a CC-BY 4.0 license. Copyright 2024, the Society of Nuclear Medicine and Molecular Imaging

Trop2

In GC, high Trop2 expression promotes tumor growth, invasion, and metastasis through amplification of key oncogenic signaling pathways, such as the ERK/MAPK cascade. As a result, Trop2 has emerged not only as a robust prognostic biomarker but also as a highly promising therapeutic target, particularly for the development of antibody-drug conjugates, exemplified by Trodelvy. In 2025, Huang et al. [136] pioneered the use of immunoPET for Trop2 imaging in GC by radiolabeling Trodelvy with ⁸⁹Zr. The radiotracer exhibited high radiochemical purity (≥ 99%) and high-affinity binding (Kd ~ 3.5–9.4 nM), with peak tumor uptake of ~ 14.3 ± 2.1% ID/g at 48 h in Trop2-high xenografts (Fig. 6B). Uptake was markedly reduced by co-injection of unlabeled Trodelvy, confirming Trop2-specific targeting. Although hepatic accumulation (Fig. 6C) and delayed imaging kinetics posed challenges, this work established [⁸⁹Zr]Zr-DFO-Trodelvy as a transformative tool for noninvasive Trop2 quantification with the potential to improve patient selection for therapies based on antibody-drug conjugates. Building on these findings, the same team [137] engineered an antibody fragment-based tracer, [⁶⁴Cu]Cu-NOTA–Trodelvy-F(ab’)₂, designed to overcome the prolonged circulation time and suboptimal imaging window associated with full-length antibodies. In GC xenografts, this tracer demonstrated rapid and selective tumor accumulation, with peak uptake at 12 h post-injection (10.83 ± 2.76%ID/g in NCI-N87) and significantly enhanced tumor-to-heart ratios (3.53 ± 0.22) compared with intact Trodelvy (0.83 ± 0.34). Importantly, the fragment-based construct enabled same-day, high-contrast imaging, minimizing background signal and improving clinical feasibility.

Targeting human epidermal growth factor receptor 3 (HER3)

HER3, a member of the EGFR tyrosine kinase receptor family, is characterized by its impaired kinase activity and reliance on heterodimerization with other receptors (e.g., HER2) to initiate oncogenic signaling. Once activated, HER3 triggers key downstream pathways such as PI3K/AKT, promoting tumor cell survival, proliferation, and metastasis. In GC, HER3 is frequently overexpressed or aberrantly activated has often been associated with resistance to targeted therapies and chemotherapy, making it a marker of poor prognosis [138]. In 2022, a study [139] evaluated the potential of HER3-targeted molecular imaging by radiolabeling an oligopeptide with ⁶⁸Ga to develop the tracer [⁶⁸Ga]Ga-NOTA-HER3P1 for PET/MRI. The authors demonstrated that quantitative changes in HER3-PET signal could predict early therapeutic response to pan-RTKI therapy in GC. In afatinib-sensitive NCI-N87 xenografts, the SUVmean of [⁶⁸Ga]Ga-NOTA-HER3P1 PET/MRI increased from 1.6 ± 0.6 at baseline to 3.8 ± 0.7 at day 4 (Fig. 6D), reflecting HER3 upregulation in response to therapy, as confirmed by immunofluorescence and Western blot analyses. In contrast, HER3-PET signal in afatinib-resistant SNU16 tumors remained essentially unchanged, consistent with continued tumor proliferation and lack of HER3 modulation. These findings indicate that HER3 may serve as a noninvasive biomarker for early treatment response and resistance stratification in GC, with broader implications for guiding personalized therapy.

Targeting transglutaminase 2 (TGM2)

TGM2 is a multifunctional enzyme that catalyzes protein cross-linking and is implicated in various cellular processes including apoptosis, cell adhesion, and extracellular matrix stabilization. In GC, TGM2 is frequently overexpressed and contributes to tumor progression, metastasis, and therapy resistance by promoting epithelial-mesenchymal transition (EMT), angiogenesis, and pro-survival signaling pathways. Its specific upregulation on tumor-associated vascular endothelium in GC models provides an opportunity for selective molecular targeting [140]. In 2021, Yin et al. [141] developed a novel dimeric peptide probe, [⁶⁸Ga]Ga-DOTA-KEK-(GX1)₂, for PET imaging of GC. The dimerization strategy significantly enhanced binding affinity and specificity toward tumor vasculature. Compared with the monomeric GX1 peptide, the dimeric probe showed markedly higher in vitro binding (p < 0.001) and significantly improved in vivo tumor uptake (5.2 ± 0.6%ID/g at 60 min) along with superior TBR (Fig. 7A-B). A distinctive feature of this study was the incorporation of Cerenkov imaging as a complementary optical modality to validate in vivo targeting. Cerenkov imaging detects the faint light emitted by β⁺ particles from radionuclides such as ⁶⁸Ga when they travel faster than the speed of light in tissue. The imaging results demonstrated clear tumor accumulation of [⁶⁸Ga]Ga-DOTA-KEK-(GX1)₂, with a significantly higher signal-to-background ratio (SBR) compared to the monomeric control. These results corroborated PET/CT findings and provided independent confirmation of the probe’s targeting specificity, while highlighting the potential of Cerenkov imaging as a more accessible method for pre-clinical evaluation of positron-emitting radiopharmaceuticals.

Fig. 7.

TGM2, Integrin αvβ3 and CEACAM6 are effective targets for molecular imaging of GC. (A) Nano PET/CT imaging of tumor-bearing nude mice of equal radioactive [68Ga]Ga-DOTA-KEK-(GX1)2, [68Ga]Ga-DOTA-GX1, and [68Ga]Ga-DOTA-URP. (B) The SUV of the tumor site. The SUV of the [68Ga]Ga-DOTA-KEK-(GX1)2 group was higher than that of the [68Ga]Ga-DOTA-GX1 group. Reproduced from ref [141]. Available under a CC-BY 4.0 license. Copyright 2021, Yin, Xin, Zhang, Hui, Chai, Hu, Xu, Wang, Nie, Zhou, Wang, Lu, Yao, Chen and Wu. (C) In vivo T1-weighted MRI of Mn3O4@PEG-RGD NPs at different time points. Reproduced from ref [143]. Available under a CC-BY 4.0 license. Copyright 2020, Li K, Li P, Wang and Han. (D) Longitudinal fluorescence imaging visualizing CEACAM6 expression in tumor xenografts over 7 days. Reproduced from ref [146]. Available under a CC-BY 4.0 license. Copyright 2021, Li K, Li P, Wang and Han

Targeting integrin αvβ3

Integrin αvβ₃ is a transmembrane adhesion receptor significantly upregulated in various cancers, including GC. It facilitates cancer cell survival, proliferation, migration, and invasion by interacting with components of the extracellular matrix and activating pro-survival signaling pathways such as PI3K-AKT [142]. In 2020, Li et al. [143] developed Mn₃O₄@PEG-RGD nanoparticles as a targeted MRI contrast agent for postoperative GC monitoring. The 7.3 nm Mn₃O₄ core provided T₁-weighted contrast enhancement, while the RGD peptide conferred specific binding to αvβ₃ integrin overexpressed in gastric tumors. In tumor-bearing mice, the nanoparticles provided rapid tumor accumulation (peaking at 1 h post-injection), enhanced MRI contrast, and reduced off-target retention (Fig. 7C). The biocompatibility, stability, and low cost of the platform support its potential for long-term surveillance of GC recurrence.

Beyond MRI, αvβ₃ has also been leveraged in fluorescence molecular imaging. Luo et al. [144] introduced M₁, a bis-pyrene-based nanoprobe functionalized with RGD and designed for early gastric cancer (EGC) detection using blue laser endoscopy (BLE). M₁ emits bright yellow-green fluorescence when excited by BLE (410–450 nm), facilitating real-time visualization of angiogenesis associated with early neoplastic lesions. Validated across cellular models, subcutaneous tumor xenografts, primary rabbit EGC models, and human clinical specimens, M₁ demonstrated robust tumor accumulation and high imaging specificity. This work represents the first molecular probe tailored for BLE and highlights its potential to reduce diagnostic misses and improve clinical outcomes in EGC. Integrin αvβ₃ targeting has also been incorporated into PET imaging, Zhao et al. [129] developed [¹⁸F]F-FAPI-42-RGD, a dual-targeting PET tracer that simultaneously binds FAP and integrin αvβ3. By capturing both stromal and tumor-cell-associated features, this heterodimeric tracer improved tumor uptake and contrast in GC models, highlighting the promise of multi-target imaging strategies for addressing tumor heterogeneity and enhancing diagnostic accuracy.

Targeting carcinoembryonic antigen-related cell adhesion molecule 6 (CEACAM6)

CEACAM6 is a glycosylphosphatidylinositol (GPI)-anchored cell surface protein involved in cell adhesion and intracellular signaling. In GC, CEACAM6 functions as an oncoprotein by promoting key hallmarks of cancer: it enhances tumor cell proliferation, inhibits anoikis, and facilitates invasion and metastasis [145]. Elevated CEACAM6 expression has been associated with aggressive tumor behavior and poor clinical outcomes. An et al. [146] developed fluorescently labeled and near-infrared-labeled anti-CEACAM6 probes using Alexa Fluor488 and IRDye800CW. These probes exhibited high labeling purity (> 95%) and demonstrated significantly stronger fluorescence signals, approximately 2- to 3-fold higher, in CEACAM6-positive GC cell lines compared to negative controls (Fig. 7D). These probes enabled precise tracking of primary gastric tumors and metastatic lesions in mouse models. Importantly, the clinical relevance of the probes was demonstrated through their successful application in wide-field and micro-fluorescence endoscopy, where they successfully identified dysplastic lesions in human gastric specimens. These findings indicate that anti-CEACAM6 fluorescent probes hold considerable promise as a next-generation endoscopic diagnostic tool for EGC, with the potential to enhance detection accuracy and reduce missed lesions.

Targeting caveolin-1 (CAV1)

CAV1 exhibits a context-dependent dual role in cancer biology. It can act as a tumor suppressor by negatively regulating oncogenic pathways such as EGFR and PI3K/AKT, yet it can also promote metastasis and chemoresistance by enhancing invasion, EMT, and cell survival under stress. Previous studies have shown that elevated CAV1 expression in GC drives heterogeneous HER2 distribution on the tumor cell membrane, thereby limiting the efficacy of HER2-targeted antibody therapies [147]. Building on this concept, Surendra Panikar et al. [148] established CAV1 as a theranostic target in HER2-positive GC by developing two complementary imaging agents: a ⁸⁹Zr-labeled anti-CAV1 antibody for immunoPET and an ICG-conjugated anti-CAV1 antibody for optical imaging. Using both subcutaneous and orthotopic HER2-positive/CAV1-high GC xenograft models, the authors demonstrated that the anti-CAV1 immunoPET tracer specifically accumulated in CAV1-expressing tumors, achieving uptake of 9.29 ± 2.74% ID/g at 72 h, compared with only 0.97 ± 0.29% ID/g in non-tumor stomach tissue. This work supports the potential of CAV1 as a complementary biomarker to HER2, with utility in patient stratification, prediction of response to antibody-based therapies, and enhanced precision imaging in GC.

Targeting AXL

AXL is a receptor tyrosine kinase belonging to the TAM family and is activated by its ligand GAS6. In GC, AXL is often overexpressed and plays a critical role in driving tumor progression and metastasis. It is particularly notable for mediating resistance to therapies targeting other receptors such as EGFR and HER2, making it an attractive diagnostic and therapeutic target. In 2024, Zhang et al. [149] developed a novel AXL-targeted affibody molecule, ZAXL:239, designed to inhibit GAS6/AXL signaling while enabling GC-specific molecular imaging. In vitro, ZAXL:239 effectively suppressed tumor cell proliferation and induced apoptosis by blocking the GAS6/AXL axis and downregulating key downstream pathways, including PI3K/AKT and MEK/ERK/c-myc. In vivo, ZAXL:239 enabled rapid, AXL-specific near-infrared imaging and significantly suppressed tumor growth without observable toxicity. In AXL-positive GC xenograft models, the probe achieved a tumor-to-skin fluorescence ratio of 5.8 ± 1.2 at 4 h post-injection and reduced tumor volume by 62% relative to controls over 21 day treatment period. These results provide compelling evidence that ZAXL:239 is a promising AXL-targeted affibody with dual functionality as a tumor-specific molecular imaging agent and a potent therapeutic candidate for GC.

Targeting cadherin-17 (CDH17)

CDH17 is a calcium-dependent cell adhesion protein predominantly expressed in the normal intestinal epithelium. In GC, CDH17 is frequently overexpressed and is recognized as a highly specific diagnostic marker for intestinal-type and metastatic gastric carcinomas [150]. In 2025, Mao et al. [151] developed an anti-CDH17 nanobody derivative (VHH-ABD) labeled with ¹⁷⁷Lu for radioimmunotherapy of GC. Nanobodies offer key advantages such as small molecular size, high binding affinity, and excellent tissue penetration; however their rapid renal clearance often limits their efficacy in radioimmunotherapy. To address this problem, the authors fused the nanobody to an albumin-binding domain (ABD), thereby prolonging systemic circulation and increasing tumor accumulation. As expected, the ABD fusion significantly extended the nanobody’s circulation half-life (t_1/2 = 2.11 h vs. 0.31 h for the unmodified VHH) and enhanced tumor uptake. In MKN-45 xenograft models, a single administration of [¹⁷⁷Lu]Lu-VHH-ABD (11.1 MBq) resulted in complete suppression of tumor growth over a 14-day period. SPECT/CT imaging confirmed selective localization to CDH17-expressing tumors, while histological analysis demonstrated reduced cellular proliferation (as indicated by decreased Ki67 expression) and significant disruption of tumor architecture.

Targeting cellular mesenchymal-epithelial transition factor (c-Met)

The c-Met receptor tyrosine kinase, upon binding its ligand hepatocyte growth factor (HGF), activates multiple downstream signaling pathways essential for cell proliferation, survival, motility, and invasion [152]. A recent study by Guillou et al. [153] reported the design, synthesis, and evaluation of three novel heptadentate chelates based on a desferrioxamine B (DFO) scaffold, each functionalized with a tripeptide unit and a photoactivatable aryl azide moiety. Density functional theory calculations predicted enhanced thermodynamic stability (logβ ≈ 48.6) for the corresponding Zr⁴⁺ complexes compared to conventional hexadentate DFO. Using a one-pot photoradiochemical strategy, the authors achieved direct conjugation of the heptadentate chelators to onartuzumab, a clinical monoclonal antibody targeting c-Met, without requiring pre-purification. In vivo PET imaging and biodistribution studies in tumor-bearing mice demonstrated tumor-specific targeting and significantly reduced bone uptake, a common indicator of ⁸⁹Zr complex instability. These findings highlight the potential of heptadentate DFO analogs for enhancing the in vivo stability, specificity, and overall imaging performance of ⁸⁹Zr-based immuno-PET tracers, particularly for targets such as c-Met.

Targeting triggering receptor expressed on myeloid cells 2 (TREM2)

TREM2 is a transmembrane receptor belonging to the immunoglobulin superfamily and is primarily expressed on tumor-associated macrophages. It is significantly upregulated within the tumor microenvironment of various cancers, including gastric carcinoma, while maintaining low expression in normal tissues. A study by Shi et al. [154] developed and evaluated four immuno-PET tracers targeting human TREM2 for the diagnosis of gastric carcinoma. Among these candidates, [¹²⁴I]I-5-F(ab′)2, derived from a novel high-affinity monoclonal antibody (5-mAb), demonstrated superior performance. The tracer exhibited strong binding affinity, rapid blood clearance (blood activity decreased to < 2% ID/g by 48 h), and high TBR contrast (12.3 ± 2.4). Compared to full-length antibodies, the F(ab′)₂ fragment format reduced nonspecific accumulation and enabled earlier high-contrast tumor visualization. Importantly, the tracer showed strong diagnostic value even in tumors with low [¹⁸F]FDG uptake, an important clinical advantage for GC that exhibit heterogeneous or minimal glycolytic activity.

Integrative summary of imaging strategies

To synthesize the diverse imaging strategies discussed, it is important to recognize that current gastric cancer molecular imaging probes differ substantially in biological specificity, imaging performance, and translational maturity. Targeted probes against cell-surface proteins such as HER2, CLDN18.2, and EGFR provide improved molecular selectivity and enable patient stratification, although their performance can be affected by intratumoral heterogeneity and variable expression levels. Stromal and microenvironment-directed agents (e.g., FAP-targeted tracers) demonstrate high tumor-to-background contrast and promising theranostic potential, yet limited by uptake in inflammation. Emerging probes targeting immune markers or angiogenic pathways offer opportunities for therapy monitoring but are still largely preclinical. Regarding probe classes, a critical trade-off exists between tumor retention and imaging kinetics. Full-length antibodies provide high binding capacity but are limited by slow blood clearance and significant hepatic background, necessitating delayed imaging. Conversely, low-molecular-weight agents—such as nanobodies, affibodies, and peptides—offer rapid tissue penetration and high contrast within hours, yet their clinical application for abdominal imaging requires optimization to mitigate high renal retention which may obscure local metastases.

Regarding the theranostic outlook, a clear distinction must be made between approaches ready for clinical translation and those requiring further exploration. Radionuclide therapies targeting established biomarkers, particularly CLDN18.2 and HER2, represent the most immediate path to clinical implementation. Utilizing clinically validated isotopes such as ¹⁷⁷Lu or ¹³¹I conjugated to antibodies or nanobodies (e.g., [¹⁷⁷Lu]Lu-DOTA-SF106 or [¹³¹I]I-IMAB362) leverages existing regulatory frameworks and supply chains established by successful PSMA-targeted therapies. In contrast, complex nanoplatforms integrating magnetic hyperthermia or photothermal therapy (e.g., SPIO-based systems) remain largely exploratory. Despite their potential for synergistic multimodal treatment, these ‘all-in-one’ nanoprobes face significant hurdles in large-scale manufacturing, reproducibility, and long-term toxicity assessment, positioning them as next-generation solutions rather than immediate clinical candidates.

Challenges and future perspectives

Despite growing recognition of its potential, the widespread clinical adoption of molecular imaging in GC faces several important challenges. First, spatiotemporal heterogeneity of target expression remains a major obstacle. Biomarkers such as HER2, PD-L1, CLDN18.2, and FAP exhibit dynamic variations not only within primary tumors but also across metastatic sites. These changes are influenced by tumor microenvironmental factors, treatment exposure, and disease evolution. As a result, imaging based on a single target may yield false-negative findings or underestimate expression levels, limiting diagnostic accuracy and therapeutic decision-making.

Second, unresolved pharmacokinetic limitations and nonspecific uptake continue to affect probe performance. For example, nanobodies show relatively high renal accumulation; monoclonal antibodies exhibit prolonged circulation and delayed optimal imaging windows; and certain small-molecule tracers exhibit elevated background uptake. Such properties can compromise signal-to-noise ratios, particularly in abdominal imaging. Structural optimization strategies, including incorporation of albumin-binding domains, hydrophilic linker designs, and construction of dual-target heterodimers, may improve pharmacokinetics and enhance tumor specificity.

Third, the field lacks standardized evaluation framework across imaging centers, scanner platforms, and probe types. Variability in acquisition parameters, quantitative methods, and diagnostic thresholds limits reproducibility and poses a barrier to conducting large-scale, multicenter clinical trials. A prime example lies in [⁶⁸Ga]Ga-FAPI-04 imaging, where acquisition parameters vary widely across institutions. Current clinical studies report uptake times ranging from 10 min to over 1 h post-injection. Given the rapid pharmacokinetics of FAPI probes, even a slight discrepancy in scan initiation can drastically alter the TBR and SUV, rendering data pooling for meta-analyses unreliable. Similarly, variations in injected dose (e.g., fixed dose vs. weight-based dosing) further complicate quantitative comparisons. Another challenge persists in hybrid modalities like PET/MRI. Unlike PET/CT, which uses CT-based density for attenuation correction, PET/MRI relies on MR-based tissue segmentation. In the context of gastric cancer, the stomach’s continuous peristalsis and respiratory motion can lead to misregistration between the MR attenuation map and the PET emission data, resulting in inaccurate SUV measurements. Establishing consistent imaging protocols and quantitative benchmarks is essential for translating molecular imaging into routine clinical practice.

Looking ahead, several avenues are expected to drive future advances in GC molecular imaging. Multitarget and multimodal approaches—such as FAP-RGD heterodimers, PET–MRI hybrid systems, and combined PET–fluorescence imaging for intraoperative navigation—may enhance the detection of micro-metastases and low-metabolic tumor subtypes. Extensive integration of imaging and therapy, such as dual-functional nanoplatforms, radioimmunotherapy, and imaging-guided magnetic or photothermal treatments, holds promise for establishing a seamless personalized diagnosis–therapy–monitoring continuum. As synthetic chemistry, protein engineering, and imaging instrumentation continue to advance, molecular imaging is expected to evolve from a mere visualization tool into a central decision-making technology within the precision medicine framework for GC.

Furthermore, AI-assisted quantitative analysis is emerging as a promising adjunct to molecular imaging; however, its application in gastric cancer remains largely exploratory. Radiomic analyses of [¹⁸F]FDG PET/CT can capture intra-tumoral heterogeneity and improve prediction of lymph node metastasis, histologic characteristics, and prognosis compared with conventional parameters such as SUVmax alone [155]. Applications to newer tracers, including FAPI-based imaging, are still exploratory, with current literature largely confined to feasibility observations rather than validated predictive models [156]. Future work must address key barriers—including annotation bias, limited multicenter validation, model generalizability, and evolving regulatory requirements—to enable reliable integration of AI with gastric cancer molecular imaging.

Conclusion

Molecular imaging is steadily reshaping the landscape of precision management in GC (Table 1). From early diagnosis and staging to prediction of treatment response and intraoperative guidance, probes targeting HER2, PD-L1, CLDN18.2, FAP, VEGF, Trop2 and other biomarkers are helping overcome the sensitivity and specificity limitations of conventional imaging modalities (Table 2). These agents enhance the detection of micrometastases, enable detailed characterization of tumor heterogeneity, and support individualized treatment planning. A diverse array of imaging probes—including monoclonal antibodies, single-domain antibodies, affibodies, peptides, and nanoparticle-based platforms—has demonstrated high specificity, excellent contrast generation, and favorable safety profiles, forming a solid foundation for future personalized GC care.

Table 1.

Established molecular targets in gastric cancer: comparison of imaging modalities and representative probes

Target	Year	Author	Ligand	Ligand type	Molecular Imaging	Sensitivity and Performance	Clinical Applicability	Limitations
FAP	2022	Lin et al. [123]	[68Ga]Ga-DOTA-FAPI-04	Small molecule	ImmunoPET	High. Often superior to FDG for detecting primary tumors, lymph node and peritoneal metastases, especially in low-metabolic subtypes.	Clinical research phase. An important complement to FDG PET for staging and restaging.	False positive due to inflammatory uptake.
	2022	Zhang et al. [124]	[68Ga]Ga-DOTA-FAPI-04	Small molecule
	2022	Qin et al. [126]	[68Ga]Ga-DOTA-FAPI-04	Small molecule
	2023	Pang et al. [127]	[68Ga]Ga-FAP-2286	Small molecule
	2024	Xu et al. [128]	[18F]F-FAPI-74	Small molecule
	2025	Zhao et al. [129]	[18F]F-FAPI-42-RGD	Small molecule
	2025	Raeisi et al. [25]	[99mTc]Tc-FAPI-46	Small molecule	SPECT	Moderate.	Clinical case.	Inferior image quality and quantitative accuracy compared to PET.
HER-2	2018	Guo et al. [88]	[64Cu]Cu-NOTA-Trastuzumab	mAb	ImmunoPET	High sensitivity. Enables whole-body imaging to overcome heterogeneity; small ligands (e.g., affibodies) achieve high tumor-to-background ratios within hours.	Under clinical translation. Used for patient selection targeted therapy and response monitoring.	High Physiological Liver Uptake and long circulation time of mAb-based probes.
	2019	Pan et al. [90]	[18F]Al-NOTA-MALCys-GGGRDN(M0)-ZHER2:342	Affibody
	2020	Guo et al. [89]	[124I]I-trastuzumab	mAb
	2020	Zhou et al. [91]	[68Ga]Ga-NOTA-MAL-MZHER2 affibody	Affibody
	2021	Wen et al. [5]	([124I]I/[64Cu]Cu)-HerPEG-dMNPs	mAb
	2022	Han et al. [92]	[18F]AlF-NOTA-HER2 affibody	Affibody
	2023	Liu et al. [93]	[18F]AlF-RESCA-HER2-BCH	Affibody
	2025	Pineau et al. [94]	[67Cu]Cu-Trastuzumab-p-SCN-BnTE1PA	mAb
	2021	Wen et al. [5]	Mn-HerPEG-dMNPs	mAb	Molecular MRI	Moderate sensitivity. Provides fusion of high-resolution anatomy and target information; useful for surgical planning.	Preclinical. Serves as a complement to PET, offering multiparametric data.	Novel technology with a long translation path; limited device availability.
PD-L1	2015	Heskamp et al. [111]	[111In]In-PD-L1.3.1	mAb	SPECT	Moderate. Can be used for detection of PD-L1 expression and patient stratification.	Clinical research phase.	Inferior image quality and quantitative accuracy compared to PET.
	2016	Chatterjee et al. [112]	[111In]PD-L1-mAb	mAb
	2018	Bensch et al. [113]	[89Zr]Zr-atezolizumab	mAb	ImmunoPET	High. Enables non-invasive assessment of whole-body expression heterogeneity and prediction of immunotherapy response.	Clinical research phase. Multiple probes under clinical validation.	Dynamic expression influenced by glycosylation; high background for mAb probes.
	2022	Sun et al. [120]	Al[18F]-NOTA-IPB-PDL1P	Polypeptide
	2023	Donnelly et al. [121]	[18F]BMS-986,229	Polypeptide
	2024	Ibrahim et al. [6]	[89Zr]-avelumab	mAb
	2024	Cytryn et al. [122]	[18F]-BMS-986,229	Polypeptide
CLDN18.2	2022	Zhao et al. [99]	[124I]-5C9	mAb	ImmunoPET	High. Superior detection rate for low-metabolic tumors (e.g., signet-ring cell carcinoma) compared to FDG.	Under clinical translation. Multiple probes in clinical studies for patient stratification and selection for targeted therapy (e.g., TST001).	High physiological uptake in the gastric mucosa; heterogeneous target expression
	2022	Wei et al. [104]	[68Ga]Ga-NOTA-hu19V3, [64Cu]Cu-NOTA-hu19V3, [18F]F-hu19V3	Nanobody
	2023	Chen et al. [100]	[89Zr]Zr-DFO-TST001	mAb
	2023	Wang et al. [108]	[124I]I-18B10(10 L)	mAb
	2023	Wang et al. [106]	[68Ga]Ga-DOTA-T37	Polypeptide
	2024	Qi et al. [109]	[68Ga]Ga-NC-BCH	Nanobody
	2024	Wang et al. [110]	[68Ga]Ga-PMD22	Nanobody
	2024	Li et al. [102]	[124I]I-SF106, [177Lu]Lu-DOTA-SF106	Antibody derivatives
	2024	Wang et al. [7]	[124I]I-zolbetuximab	mAb
	2024	Wu et al. [98]	[123I]I-IMAB362	mAb	SPECT	Moderate.	Early clinical. A cost-effective alternative for patient screening.	Inferior image quality and quantitative accuracy compared to PET.
	2024	Bai et al. [105]	[99mTc]Tc-PHG102	nanobody
	2025	Liu et al. [103]	[131I]I-HLX58-Der	mAb
	2025	Wang et al. [107]	SPIO@1D5-ICG	mAb
	2022	Zhao et al. [99]	Cy5.5-5C9, FD1080-5C9	mAb	NIR	High. Real-time surgical navigation.	Preclinical.	Poor penetration depth
Integrin αvβ3	2020	Li et al. [145]	Mn3O4@PEG-RGD NPs	Polypeptide	Molecular MRI	Moderate. Provides T1-weighted contrast enhancement and targeted accumulation, useful for postoperative monitoring.	Preclinical.	Lower molecular sensitivity inherent to MRI
	2023	Luo et al. [146]	BP-FFVLK-(PEG)-RGD, M1	Polypeptide	BLE	High. Real-time visualization of early lesions.	Preclinical.	Poor penetration depth.
	2025	Zhao et al. [131]	[18F]F-FAPI-42-RGD	Small molecule	PET/CT	High. Dual-targeting strategy improves tumor uptake and addresses heterogeneity.	Preclinical.	Heterodimer probe design and synthesis are more complex.

Table 2.

Emerging molecular targets in gastric cancer

Target	Year	Author	Ligand	Ligand type	Molecular Imaging
CDH17	2025	Mao et al. [151]	[177Lu]Lu-anti-CDH17 VHH, [177Lu]Lu-anti-CDH17 VHH-ABD	Nanobody	RIT
HER-3	2022	Esfahani et al. [139]	[68Ga]-NOTA-HER3P1	Polypeptide/Oligopeptide	ImmunoPET
VEGFR1	2020	Yu et al. [134]	[111In]In-DOTA-F56	Polypeptide	SPECT
CEACAM6	2021	An et al. [146]	CEACAM6-mAb-Alexa Fluor 488, CEACAM6-mAb-IRDye 800CW	mAb	Confocal Microscopic, NIRF
AXL	2024	Zhang et al. [149]	ZAXL:239	Affibody	NIRF
TGM2	2021	Yin et al. [141]	[68Ga]-DOTA-KEK-(GX1)2	Polypeptide	ImmunoPET, Cerenkov Imaging
TREM2	2023	Shi et al. [154]	[124I]I-anti-TREM2-mAb, [124I]I-anti-TREM2-F(ab′)2, [124I]I-5-mAb, [124I]I-5-F(ab′)2	mAb, Antibody fragment	ImmunoPET
c-Met	2022	Guillou et al. [153]	[89Zr]Zr-2-onartuzumab	mAb	ImmunoPET
Trop2	2025	Huang et al. [136]	[89Zr]Zr-DFO-Trodelvy	mAb	ImmunoPET
Caveolin-1	2023	Surendra Panikar et al. [148]	[89Zr]Zr-DFO-Anti-CAV1 Antibody	mAb	ImmunoPET, NIR

However, broad clinical translation remains constrained by challenges related to target stability, probe pharmacokinetics, imaging protocol standardization, and validation of clinical utility. Continued progress in multimodal hybrid imaging, dual-targeting ligand design, AI-driven image analysis, and integrated theranostic platforms, is expected to drive molecular imaging toward more comprehensive and actionable clinical applications. Ultimately, the strategic integration of molecular imaging into GC management promises to translate into meaningful improvements in diagnostic accuracy, therapeutic precision, and patient survival and quality of life.

Author Contributions

Kexin Lan: Conceptualization, Writing – original draft, Writing – review & editing. Yongshun Liu: Writing – original draft, Writing – review & editing. Ruobing Li: Writing – review & editing. Zhaonan Sun: Supervision, Writing – review & editing. Wenpeng Huang: Conceptualization, Supervision, Writing – review & editing.

Funding

Scientific Research Seed Fund of Peking University First Hospital (2025SF053).

Data availability

Data will be made available on request.

Declarations

Ethics declaration

This review does not report any new studies involving human participants or animals. All studies discussed herein were previously published and were conducted in accordance with relevant institutional and national ethical standards, as well as the Declaration of Helsinki and its later amendments.

Competing interests

All authors declare no conflict of interest.