Abstract

Background:

Claudin-18.2 (CLDN18.2), a tight junction protein, has emerged as a promising molecular target in gastrointestinal cancers. However, its expression pattern and clinicopathological relevance in pancreatic ductal adenocarcinoma (PDAC) remain unclear.

Objective:

To assess the prevalence, tumor-specific, and clinicopathological associations of CLDN18.2 expression in PDAC, and to identify sources of heterogeneity to clarify its value as a therapeutic target.

Methods:

A systematic search across six major databases was conducted from inception to August 2025 in accordance with PRISMA guidelines. Included studies evaluated CLDN18.2 expression in adult PDAC using defined immunohistochemistry protocols. Random-effects models estimated pooled prevalence and odds ratios (ORs), with subgroup analyses exploring heterogeneity.

Results:

Sixteen studies including 2,025 patients with PDAC were included. The pooled prevalence of CLDN18.2 expression was 51.60% (95% CI: 40.93–62.19), with substantial heterogeneity (I2 = 95.4%). Expression varied significantly by antibody clone (p = 0.0006), but not by geographic regions. CLDN18.2 expression was highly specific to neoplastic tissue (OR = 102.40; 95% CI: 35.50–295.38). No significant associations were identified with sex, tumor location, T-stage, N-stage, or metastatic status. However, expression was significantly lower in poorly differentiated tumors (G3 vs. G1/G2: OR = 0.37; 95% CI: 0.20–0.70).

Conclusion:

CLDN18.2 is frequently expressed in PDAC and shows high tumor specificity, supporting its relevance as a therapeutic biomarker. Despite assay variability, even under stringent clinical trial criteria, a meaningful subset of patients may qualify for CLDN18.2-targeted therapy. These findings highlight the need for standardized testing and further clinical evaluation of CLDN18.2-directed strategies in PDAC.

Highlights

CLDN18.2 is a tumor-specific biomarker in pancreatic ductal adenocarcinoma, expressed in approximately 51.6% of cases.

Using stringent, trial-aligned immunohistochemical criteria, about 23% of PDAC cases may be eligible for CLDN18.2-targeted therapy.

Reported CLDN18.2 expression varies widely, likely due to differences in immunohistochemical methods, underscoring the need for standardized testing.

1 Introduction

Pancreatic ductal adenocarcinoma (PDAC), the most common form of pancreatic cancer, remains one of the deadliest malignancies worldwide, with over 511,000 new cases and a 5-year survival rate of just 13% (1–3). It has recently become the third leading cause of cancer-related death in the United States (1). These poor outcomes are largely due to late diagnosis of the disease (45.9% are diagnosed at stage IV) (4).

The Claudin family consists of 27 transmembrane tight junction proteins. In biological systems, Claudins have been shown to function as paracellular barriers, paracellular channels, and signaling hubs for cellular tumorigenesis (5). Altered expression of Claudins has been associated with many cancers (6), including Claudin-18 in PDAC (7). Claudin-18 has two isoforms (Claudin-18.1 and Claudin-18.2), which are normally expressed in pulmonary and gastric tissue, respectively (8). CLDN18.2 has been implicated in gastric, pancreatic, esophageal, ovarian, and lung tumors (9).

In 2024, the FDA approved zolbetuximab as the first CLDN18.2-targeted treatment for gastric and gastro-esophageal adenocarcinoma (10). Zolbetuximab is a chimeric immunoglobulin G1 monoclonal antibody that binds to CLDN18.2, leading to cellular apoptosis and inhibition of proliferation (11). It has been shown to prolong progression-free survival and overall survival in both the SPOTLIGHT and GLOW trials (10, 12). Currently, there are two ongoing studies assessing the use of zolbetuximab in metastatic pancreatic cancer (NCT06396091; and NCT03816163). Moreover, five additional therapies targeting CLDN18.2 are currently under clinical investigation (NCT07079228; NCT07025889; NCT06219941; NCT05862324; and NCT05482893).

Given the involvement of CLDN18.2 in PDAC and the ongoing investigations into therapies targeting this protein, our aim to systematically synthesize the evidence on CLDN18.2 expression in PDAC through a systematic review and meta-analysis, evaluating its association with clinicopathological features and its potential as a therapeutic target in future clinical applications.

2 Materials and methods2.1 Study design

This study was conducted as a systematic review and meta-analysis of observational studies, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The review aimed to synthesize available evidence on CLDN18.2 expression in PDAC, its tumor specificity, and its association with clinicopathological characteristics.

2.2 Search strategy

We conducted a comprehensive and a systematic literature review in EBM Reviews, MEDLINE, and Embase (via Ovid), as well as in Web of Science, Scopus, and PubMed (inception to 08 of August 2025). Search terms included Claudin-18.2 and its variants (“Claudin 18,” “Claudin-18,” CLDN18.2, “Claudin 18 isoform 2,” “Claudin 18 variant 2”) and pancreatic cancer terms (“Pancreatic Ductal Adenocarcinoma,” PDAC, “Pancreatic cancer,” “Pancreatic carcinoma,” “Pancreatic neoplasm*,” “Pancreatobiliary adenocarcinoma”). Two reviewers independently screened titles, abstracts, and full texts based on predefined criteria. Discrepancies were resolved by discussion or a third reviewer. Reference lists of the included studies were reviewed manually to include relevant studies to the topic.

2.3 Eligibility criteria

We included studies that evaluated the expression of CLDN18.2 in adult patients with PDAC using immunohistochemistry (IHC). Eligible studies were required to report the number of CLDN18.2-positive cases using well defined IHC protocols. Studies were excluded if they involved non-PDAC pancreatic tumors, were inaccessible due to paywalls, or were case reports, abstracts, reviews, or non-original research. Additionally, studies not published in English were excluded.

2.4 Data extraction

Data were independently extracted by two reviewers using a standardized data extraction form. The extracted information included study characteristics (author, year, and country), sample size, number and percentage of CLDN18.2-positive cases, antibody clone used, staining protocols, positivity thresholds, and reported associations with clinicopathological features. Any discrepancies between reviewers were resolved through discussion or consultation with a third reviewer to ensure accuracy and consistency.

2.5 Interpretation of CLDN18.2 positivity

To harmonize definitions across studies, CLDN18.2 positivity was interpreted according to each study’s reported criteria. When quantitative thresholds (e.g., % positive cells, intensity score, or H-score) were provided, these were applied directly. In studies lacking explicit cut-offs, any membranous staining of tumor cells (≥1+) was considered positive, consistent with author reporting.

2.6 Quality assessment

The methodological quality of the included studies was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Studies Reporting Prevalence Data. The checklist includes eight domains: sample frame appropriateness, sampling method, sample size adequacy, description of participants and setting, data analysis coverage, validity of condition identification, standardization of measurement, and appropriateness of statistical analysis. Each domain was rated as “Yes,” “No,” or “Unclear.” Response rate was deemed not applicable in these retrospective tissue-based studies. Studies meeting ≥6 of 8 applicable criteria, were classified as high quality, 4–5 as moderate, and ≤3 as low quality. Two independent reviewers performed the assessment, and discrepancies were resolved through discussion.

2.7 Statistical analysis

Meta-analyses were performed using a random-effects model (DerSimonian and Laird method) to account for expected heterogeneity across studies due to differences in patient populations, antibody clones, staining protocols, and positivity thresholds. For the primary outcome, we calculated the pooled prevalence of CLDN18.2 expression in PDAC, expressed as a percentage with corresponding 95% confidence intervals (CIs). For all dichotomous outcomes, including comparisons of expression by sex, tumor location, tumor stage (T), nodal status (N), metastatic status (M), and tumor grade (differentiation), we computed odds ratios (ORs) with 95% CIs. Where applicable, subgroup meta-analyses were conducted based on geographic region (Asia, Europe, North America) and antibody clone used in IHC (e.g., 43–14A, EPR19202, ZMD395, HPA-018446). Statistical differences between subgroups were assessed using Q-statistics for subgroup difference. Heterogeneity across studies was quantified using the I2 statistic, with values of 25, 50, and 75% representing low, moderate, and high heterogeneity, respectively. A p-value of less than 0.10 for the Cochran Q test was considered indicative of significant heterogeneity. Forest plots were generated for all pooled analyses to visually represent the effect sizes and confidence intervals. To assess publication bias, we visually examined funnel plots of the primary outcome (CLDN18.2 prevalence in PDAC). All statistical analyses and data visualizations were performed using RStudio (Version 2023.06.1 Build 524) with R (Version 4.3.1). A p-value < 0.05 was considered a statistically significant.

3 Results

Following PRISMA guidelines, 393 records were identified. After duplicate removal, 326 were screened, with 131 excluded. Of 195 full-text articles assessed, 16 studies involving 2,025 PDAC patients were included in the final analysis (Figure 1). The key characteristics and findings of the included observational studies are summarized in Table 1.

PRISMA flow diagram of study selection showing the identification, screening, eligibility assessment, and inclusion of studies. Of 393 records identified across four databases, 326 remained after duplicate removal. Following title and abstract screening and full-text review, 16 studies met the eligibility criteria for qualitative and quantitative synthesis.

Study (author, year)CountryPatients with PDAC (n, % positive)CLDN18.2 in neoplastic vs. non-neoplastic tissuePositivity definitionAntibody clone; dilution; manufacturerKey findingsValentini et al., 2025 (22)Italy70 (35, 50%)75 (35) vs. 28 (0)H-score ≥ 5EPR19202; 1:200; AbcamOnly neoplastic cells expressed CLDN18.2. Associated with well/moderately differentiated tumors and N0 status.Kayikcioglu et al., 2023 (23)Turkey68 (37, 54.4%)NRAny membranous staining ≥1UnspecifiedAssociated with better overall survival.Wöll et al., 2014 (24)Germany174 (103, 59.2%)202 (109) vs. 24 (0)≥1% tumor cells stainedaGC182; NA; in-houseCorrelated with lymph node metastasis.Arseneau et al., 2025 (25)Canada120 (39, 32.5%)NR≥75% tumor cells, intensity ≥2+43–14A; pre-diluted; RocheAssociated with well-differentiated tumors and better survival.Park et al., 2023 (13)South Korea130 (41, 31.5%)NR≥80% tumor cells, intensity ≥2+34H14L15; 1:100; InvitrogenAssociated with well-differentiated tumors and N0 status.Wang et al., 2022 (15)China93 (46, 49.5%)93 (35) vs. 13 (0)H-score >150EPR19202; 1:500; AbcamHigher expression in PDAC than normal tissue. Linked with metastasis, stage, nerve invasion, poor survival in stage III/IV.Lyu et al., 2024 (18)Germany309 (94, 30.4%)NR≥75% tumor cells, intensity ≥2+LS-B16145; 1:200; LSBio & 43–14A; RocheAssociated with better differentiation and survival. Comparable clone performance.Zhang et al., 2022 (26)China302 (171, 56.6%)302 (171) vs. 10 (0)≥1% tumor cells stainedCatalog CLDN18.2Associated with better differentiation, female sex, non-smokers, less bile duct invasion/metastasis.Sanada et al., 2010 (27)Japan15 (6, 40%)NR≥10% tumor cells stainedClaudin 18; 1:200; ZymedNAKaranjawala et al., 2008 (7)USA166 (83, 50%)166 (159) vs. 105 (4)>80% tumor cells, intensity ≥2+ZMD395; 1:400; InvitrogenOverexpressed in neoplastic tissue. Associated with well-differentiated tumors.Soini et al., 2012 (28)Japan and Finland111 (78, 70.3%)18 (10) vs. 26 (0)>25% tumor cells stainedNo. 38–8,000; 1:100; InvitrogenAssociated with well-differentiated tumors.Isidro et al., 2022 (29)USA56 (37, 66.1%)NR>5% tumor cells, intensity ≥1+HPA-018446; 1:500; Sigma-AldrichNAYang et al., 2022 (30)Taiwan10 (10, 100%)All effusion samplesStaining intensity ≥1%HPA-018446; NA; Sigma-AldrichNATanaka et al., 2011 (31)Japan156 (109, 69.9%)NR≥1% tumor cells stainedZMD395; 1:1000; ZymedHigh expression in cancer tissue. Associated with well/moderately differentiated tumors.Kyuno et al., 2025 (32)Japan201 (20, 10%)211 (152) vs. 153 (3)≥75% tumor cells, intensity ≥2+43–14A; 1:5000; AbcamHigh expression in neoplastic tissue. Biopsy detected 54.6% of CLDN18.2 + tumors.Li et al., 2020 (33)Taiwan44 (30, 68.2%)NR>1 + (weak, <10% of cells)HPA-018446; 1:250; Sigma-AldrichCLDN18.2 has high sensitivity as a diagnostic tool for gastric and pancreatobiliary tumors.

Overview and characteristics of included observational studies on CLDN18.2 expression in pancreatic ductal adenocarcinoma.

PDAC, Pancreatic ductal adenocarcinoma; NR, Not reported; NA, Not available; N0, No regional lymph node metastasis. Percent positivity reflects the proportion of PDAC cases expressing CLDN18.2 per study-specific criteria. Antibody information includes clone name, dilution, and manufacturer, where available. Staining thresholds vary across studies, contributing to inter-study heterogeneity.

The methodological quality of the 16 included studies was evaluated using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Prevalence Studies. Overall, 7 studies (43.8%) were rated as high quality, and 9 studies (56.2%) as moderate quality. None were rated as low quality. The most common limitation was the lack of sample size justification and unclear or non-random sampling methods. However, all studies used valid and standardized IHC methods to assess CLDN18.2 expression. Details of the quality assessment are presented in Table 2.

Study (author, year)Q1: sampling frameQ2: sampling methodQ3: sample sizeQ4: subjects and setting describedQ5: valid method for conditionQ6: standardized measurementQ7: appropriate analysisQ8: adequate response rate“Yes” totalOverall qualityValentini et al., 2025 (22)YesYesNoYesYesYesYesYes7HighKayikcioglu et al., 2023 (23)YesYesNoYesYesYesYesYes7HighWöll et al., 2014 (24)UnclearUnclearNoYesYesYesYesYes5ModerateArseneau et al., 2025 (25)YesYesNoYesYesYesYesYes7HighPark et al., 2023 (13)YesUnclearNoYesYesYesYesYes6HighWang et al., 2022 (15)UnclearUnclearNoYesYesYesYesYes5ModerateLyu et al., 2024 (18)YesYesNoYesYesYesYesYes7HighZhang et al., 2022 (26)YesYesNoYesYesYesYesYes7HighSanada et al., 2010 (27)UnclearUnclearNoYesYesYesYesUnclear5ModerateKaranjawala et al., 2008 (7)UnclearUnclearNoYesYesYesYesYes5ModerateSoini et al., 2012 (28)UnclearUnclearNoYesYesYesYesYes5ModerateIsidro et al., 2022 (29)UnclearUnclearNoYesYesYesYesYes5ModerateYang et al., 2022 (30)UnclearNoNoYesYesYesYesYes5ModerateTanaka et al., 2011 (31)UnclearUnclearNoYesYesYesYesYes5ModerateKyuno et al., 2025 (32)YesYesNoYesYesYesYesYes7HighLi et al., 2020 (33)UnclearUnclearNoYesYesYesYesYes5Moderate

Quality assessment of included studies based on the JBI checklist for prevalence studies.

This table summarizes the quality assessment of included studies using the Joanna Briggs Institute (JBI) Checklist for Prevalence Studies, which evaluates eight domains (Q1–Q8) related to sampling, measurement, and analysis. Each item was rated as “Yes,” “No,” or “Unclear.” Studies were categorized as high quality (6–8 “Yes” responses), moderate quality (4–5 “Yes”), or low quality (≤3 “Yes”).

The pooled prevalence of CLDN18.2 expression in PDAC was 51.60% (95% CI: 40.93–62.19%) under a random-effects model, with substantial heterogeneity observed across studies (I2 = 95.4%, p < 0.0001; Figure 2). Subgroup analyses were conducted to explore sources of heterogeneity in CLDN18.2 expression among patients with PDAC. When stratified by geographic region, the pooled expression rates were relatively consistent across continents but demonstrated notable heterogeneity within each subgroup. Studies conducted in Asia reported a pooled expression rate of 52.67% (95% CI: 32.87–72.07%), which was slightly higher than those.

Pooled prevalence of CLDN18.2 expression in pancreatic ductal adenocarcinoma (PDAC) shown as a forest plot summarizing 16 studies comprising 2025 patients. The overall pooled prevalence was 51.60% (95% CI, 40.93–62.19) with substantial heterogeneity (I2 = 95.4%, p < 0.0001).

from Europe (48.19, 95% CI: 31.96–64.61%) and North America (48.90, 95% CI: 31.97–65.96%). Despite these variations, no statistically significant difference in CLDN18.2 expression was observed across regions (p = 0.9401), and high levels of heterogeneity persisted within each subgroup (Asia: I2 = 97.0%, Europe: I2 = 93.3%, North America: I2 = 89.6%; Figure 3).

Subgroup analysis of CLDN18.2 expression by geographic region shown as a forest plot stratified by study location. Pooled prevalence estimates were highest in Asia (52.67%), followed by North America (48.90%), and Europe (48.19%), with no statistically significant differences between subgroups (p = 0.9401).

A separate subgroup analysis was performed based on the antibody clone used for IHC detection of CLDN18.2. Considerable differences in expression rates were observed across antibody clones. Studies using the 43–14A clone reported the lowest pooled expression rate at 23.27% (95% CI: 10.01–39.94%), while those employing the HPA-018446 clone reported the highest rate at 77.63% (95% CI: 57.45–93.07%). Intermediate expression rates were observed with the EPR19202 clone (49.69, 95% CI: 41.98–57.41%) and the ZMD395 clone (60.12, 95% CI: 40.24–78.42%). The test for subgroup differences revealed a statistically significant difference in CLDN18.2 expression depending on the antibody clone used (p = 0.0006), as shown in Figure 4. These findings emphasize the importance of methodological consistency and antibody selection in the assessment of CLDN18.2 expression across studies.

Subgroup analysis of CLDN18.2 expression by antibody clone shown as a forest plot comparing pooled prevalence estimates across assays. Expression was lowest with clone 43–14A (23.27%) and highest with HPA-018446 (77.63%), with statistically significant differences observed between clones (p = 0.0006), indicating assay-related variability.

Seven studies compared CLDN18.2 expression between male and female patients. The pooled odds ratio (OR) slightly favored higher expression in females (OR = 1.22, 95% CI: 0.81–1.84), though this was not statistically significant (Figure 5).

Association between CLDN18.2 expression and patient sex shown as a forest plot of odds ratios comparing male and female patients. No statistically significant association was observed (OR, 1.22; 95% CI, 0.81–1.84), with moderate heterogeneity (I2 = 60.6%).

In six studies that compared CLDN18.2 expression between neoplastic and non-neoplastic pancreatic tissues, expression was significantly higher in neoplastic tissues (OR = 102.40, 95% CI: 35.50–295.38, I2 = 35.7%, p = 0.1555), indicating tumor-specific expression (Figure 6). Studies such as Karanjawala et al. and Kyuno et al. demonstrated near-complete absence of CLDN18.2 in normal tissue.

CLDN18.2 expression in neoplastic versus non-neoplastic pancreatic tissue shown as a forest plot comparing odds ratios across seven studies. CLDN18.2 was significantly overexpressed in neoplastic tissue compared with adjacent non-tumor tissue (OR, 102.40; 95% CI, 35.50–295.38), supporting tumor-specific expression.

CLDN18.2 expression did not significantly differ between early (T1–T2) and advanced (T3–T4) stages (OR = 1.12, 95% CI: 0.84–1.49; I2 = 0%; Figure 7a). Similarly, no statistically significant difference was observed between CLDN18.2 expression and nodal status (OR = 1.20, 95% CI: 0.65–2.21; I2 = 73.7%; Figure 7b). Three studies evaluated the correlation between CLDN18.2 expression and M stage. The pooled analysis suggested higher expression in metastatic disease (OR = 1.36, 95% CI