ABSTRACT
Claudin-18.2 is a promising therapeutic target for gastrointestinal cancer. However, its expression pattern in pancreatic ductal adenocarcinoma, especially the concordance between biopsy and resection specimens, is unknown. This study aimed to evaluate the consistency of claudin-18.2 positivity across different specimen types using the clinically validated antibody clone 43-14A employed in ongoing zolbetuximab trials. Immunohistochemical analysis for claudin-18 was conducted on 211 resected pancreatic cancer tissues, 133 matched preoperative biopsy samples, and 60 samples from recurrent lesions. The concordance between the biopsy and resection specimens was 92.5% using a 75% staining threshold. However, this high concordance likely reflects the large proportion of claudin-18.2-negative cases, as the biopsy sensitivity for detecting claudin-18.2-positive tumors was only 54.6%. This raises concerns about underdiagnosis and suggests that biopsy alone may miss patients eligible for zolbetuximab therapy. Receiver operating characteristic analysis showed that lowering the threshold to 20% in biopsy samples improved the sensitivity to 100%. However, patients meeting this threshold would still not qualify for therapy under the current trial criteria, highlighting a potential clinical dilemma. Claudin-18.2 expression was generally preserved in recurrent lesions (83.3% concordance with primary tumors), although reductions were noted in local recurrence and liver metastasis. These findings suggest that although biopsy-based assessments may be a practical initial tool, they should be interpreted with caution. Confirmatory studies on resection specimens using the same clinical trial protocol may be necessary to ensure the accurate identification of patients eligible for claudin-18.2-targeted therapy.
KEYWORDS: Biopsy, claudin-18.2, immunohistochemistry, pancreatic cancer, recurrence, tight junction, zolbetuximab
Introduction
Claudin-18.2 (CLDN18.2), a tight junction protein predominantly expressed in gastric tissues, has been identified as a promising therapeutic target for various gastrointestinal malignancies, including gastric and pancreatic cancers. The monoclonal antibody zolbetuximab, which specifically targets CLDN18.2, has demonstrated efficacy in improving the survival of patients with unresectable advanced gastric and gastroesophageal junction adenocarcinomas in the phase 3 clinical trials GLOW and SPOTLIGHT.1,2 These trials showed significantly prolonged median progression-free survival and overall survival in patients with CLDN18.2-positive cancer. Given these positive outcomes, zolbetuximab is anticipated to be a key component of future treatment protocols for pancreatic ductal adenocarcinoma (PDAC).
In an ongoing phase 2 clinical trial for metastatic pancreatic cancer (ClinicalTrials.gov ID: NCT03816163), patients treated with zolbetuximab are selected based on the positive expression of CLDN18.2, as determined by immunohistochemical staining. The criteria for CLDN18.2 positivity, consistent with those of the GLOW and SPOTLIGHT trials, require moderate-to-strong membrane staining of CLDN18.2 in at least 75% of the cancer cells.1–3 However, previous immunohistochemical analyses of resected gastric and pancreatic cancer specimens revealed heterogeneous CLDN18 staining intensity across tissue regions within the same samples.4–7 This variability raises concerns regarding the accuracy of CLDN18.2 expression assessments based on biopsy specimens, which may not accurately represent the staining pattern observed in the entire resected tumor.
Recent studies on gastric cancer tissues reported substantial heterogeneity in CLDN18.2 expression across tumor regions, indicating that the choice of biopsy site influences the assessment of CLDN18.2 expression.7,8 This finding suggests the need for similar studies on PDAC to ensure the reliability of biopsy-based evaluations, as discrepancies between biopsy and resection specimens may affect the accuracy of patient selection for zolbetuximab therapy. Furthermore, for patients with unresectable PDAC, which account for more than 75% of cases,9,10 therapeutic regimen will depend on the CLDN18.2 positivity in limited biopsy tissue, understanding the relationship between CLDN18.2 expression in biopsy and resection specimens is critical. The accurate assessment of biopsy specimens is essential to determine the eligibility for zolbetuximab treatment, making this area of research highly relevant for optimizing therapeutic outcomes.
Additionally, examining the concordance of CLDN18.2 expression between primary and recurrent PDAC tissues is crucial for determining the applicability of zolbetuximab in recurrent cases. Biopsies from metastatic PDAC lesions, typically identified using imaging modalities,11 are not always available. Thus, understanding the relationship between CLDN18.2 expression in primary and recurrent sites may provide valuable insights for clinical decision-making, making the findings of the present study particularly interesting for the management of patients with recurrent disease.
To date, no study has systematically compared the concordance of CLDN18.2 positivity between biopsy and resected PDAC specimens or between primary and recurrent PDAC specimens using the anti-CLDN18 antibody clone 43-14A, the same clone used in clinical trials.1,2 Assessing the concordance rates may optimize patient selection for zolbetuximab therapy, thus providing crucial data for improving cancer treatment strategies.
Materials and methods
Sample collection
Specimens from 222 consecutive patients with pancreatic cancer who underwent pancreatectomy between January 2011 and December 2020 were retrieved from the pathology files of Sapporo Medical University Hospital. Follow-up data were obtained from hospital records. The inclusion criteria for resection specimens were that pancreatic cancer was pathologically confirmed in these specimens. The pancreatic cancers included PDAC, acinar cell carcinoma, carcinoma derived from intraductal papillary mucinous neoplasia, adenosquamous carcinoma, and anaplastic carcinoma. Patients with pathological complete response after neoadjuvant therapy (n = 7), those diagnosed with pancreatic intraepithelial neoplasia (PanIN) only (n = 3), and those with specimens containing only a small number of cancer cells (n = 1) were excluded. Thus, specimens from 211 patients were analyzed (Figure 1).
Figure 1.
Summary of the specimens and histological types of pancreatic cancer in the present study. A total of 211 patients with pancreatic cancer were included in the study. The numbers in parentheses indicate the number of cases in each analysis. PDAC, pancreatic ductal adenocarcinoma. CLDN18, claudin-18; panin, pancreatic intraepithelial neoplasia.
Preoperative biopsy tissues were obtained using endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) in all cases. The inclusion criteria for preoperative biopsy specimens were as follows: (1) pancreatic cancer was pathologically confirmed in these specimens and (2) the specimens contained a minimum of 50 viable cancer cells for quantitative evaluation of immunostaining. Specimens diagnosed with suspected pancreatic cancer were excluded from the study. Biopsy specimens that met the inclusion criteria were obtained from 133 patients.
Specimens from recurrent lesions were obtained via resection or needle biopsy. The inclusion criteria for such samples were that tissue from recurrent lesions contained (1) pathologically confirmed metastatic carcinoma cells and (2) a minimum of 50 viable cancer cells for quantitative evaluation of immunostaining. Thus, samples from 60 recurrent lesions in 53 patients were collected, as seven patients had recurrent lesions in two different organs.
All specimens were fixed using formalin, and alcohol-fixed specimens were not included. Cytological specimens were not used in this study.
The 211 cases were staged according to the UICC staging system (8th edition).12 Tumor resectability status was classified according to the NCCN and Japanese criteria.13,14 Preoperative CEA and CA19–9 levels were measured within four weeks prior to surgery. Postoperative surveillance was performed monthly during the first postoperative year and every one to three months thereafter. Recurrence was diagnosed based on radiological findings and confirmed through biopsy if possible.
Immunohistochemical analysis of specimens
Immunohistochemistry was performed using the anti-CLDN18 antibody clone 43-14A (1:5000; Abcam, ab314690, Cambridge, UK), which was previously used to determine CLDN18.2 positivity in gastric cancer.3 The primary antibody against CLDN18 (43-14A) used in the current study targets the conserved C-terminal region of CLDN18 proteins; therefore, it cannot distinguish CLDN18.1 from CLDN18.2. CLDN18.1 is expressed in normal lung and lung tumors, while CLDN18.2 is highly expressed in tumors including gastric and pancreatic cancer.15,16 Immunostaining using this primary antibody has been adopted as the criterion for the administration of zolbetuximab (a monoclonal antibody against CLDN18.2) in gastric cancer.1,2 Therefore, we used an anti-CLDN18 antibody (43-14A) as a criterion for CLDN18.2 positivity in pancreatic cancer.
We observed all cases in whole tissue sections and did not use tissue microarrays to evaluate the heterogeneity of CLDN18 staining within tumors. In the present study, tissue sections were deparaffinized in xylene, rehydrated using a graded series of ethanol and phosphate-buffered saline. After antigen retrieval by microwave heating (95°C for 30 min) in 10 mmol/L Tris/1 mmol/L EDTA buffer, the sections were incubated with 3% H2O2 for 10 min to block endogenous peroxidase activity, followed by 4% skim milk as a blocking buffer for 30 min, and then with antibodies diluted in Dako REAL™ Antibody Diluent (Agilent, Santa Clara, CA, USA) overnight at 4°C. The sections were then incubated with the Dako REAL™ EnVision™ HRP rabbit/mouse for 30 min at room temperature, and color was developed using the Dako REAL™ EnVision™ DAB plus CHROMOGEN and substrate buffer for 10 min, according to the manufacturer’s instructions. The slides were then counterstained with hematoxylin.
As negative controls, adjacent non-neoplastic regions were examined as normal tissues. As a positive control, normal gastric mucosal tissue was stained simultaneously with pancreatic cancer tissue (Supplemental Figure S1(a)). The staining intensity of the surgical specimens was scored as follows: 0, no reactivity at the cellular membrane; 1+, weak reactivity; 2+, moderate reactivity; and 3+, strong reactivity, as in gastric cancer (Supplemental Figure S1(b)).8,17 The assessment was based on the VENTANA CLDN18 (43-14A) RxDx Assay Interpretation Guide for Gastric Adenocarcinoma including GEJ, and the evaluation of expression intensity was also based on this guide. The 1+/2+/3+ evaluation criteria are as follows: 1+ (weak intensity) is staining with a light brown hue, and the membrane staining may appear to be partial or circumferential, 2+ (moderate intensity) is staining with a chocolate brown hue, and the membrane staining may appear to be partial or circumferential, and 3+ (strong intensity) is staining with a dark brown to black hue, saturated DAB, and the membrane staining is thickened and may be partial or circumferential. The proportion of the stained area was scored between 0 and 100 in increments of 5. The histological scores (H-scores) of pancreatic cancer, low-grade PanIN, and normal ducts were obtained by multiplying the proportion and intensity scores, therefore ranging between 0 and 300. Because all high-grade PanINs were difficult to differentiate from the intraductal spread of cancer, the H-score of high-grade PanIN alone was not analyzed. CLDN18 staining intensity did not correlate with the year of the surgical tissue collection (Supplemental Figure S1(c)). CLDN18.2 positivity was defined as ≥ 75% of areas with moderate-to-strong staining intensity; the same criterion was employed in previous clinical trials.1–3
Two pathologists (DK and MO) were blinded to the clinical data when evaluating the slides. Cohen’s Kappa coefficient was 0.942 (95%CI: 0.885–0.999) (Supplemental table S1). Discordant cases were discussed, and a consensus on the staining intensity and area was reached.
Although the antibody clone 43-14A was used in this study, we did not employ the VENTANA CLDN18 (43-14A) RxDx Assay on the Benchmark Ultra platform because of limitations in access and cost, particularly as the study was initiated prior to May 2024, when the clinical kit became available in Japan. Instead, we validated our in-house staining protocol using gastric mucosa and gastric cancer tissue samples, as shown in Supplemental Figure S1(a). Furthermore, we compared CLDN18 staining in a tissue microarray (PA242f purchased from TissueArray.Com LLC (Derwood, MD, United States)) containing 20 pancreatic cancer samples (including two CLDN18-positive cases and 18 negative cases, along with four normal tissues and one adrenal tumor) using both our in-house method and the standard VENTANA assay. The staining patterns and intensity scores were consistent across all samples, with no CLDN18.2-positive cases detected by either method, supporting the comparability of our approach (Supplemental Figure S1(d)). While the overall staining contrast and negative-positive differentiation appeared superior with the VENTANA assay, our method yielded reproducible and aligned results, justifying its use for large-scale retrospective analyses.
Statistical analysis
The measured values are presented as mean ± standard deviation. Data were analyzed and compared using Fisher’s exact test and the Mann – Whitney U test. The correlation between two variables was tested using Spearman’s correlation coefficient. The difference between two paired data was tested using the Wilcoxon signed-rank sum test. Survival rates were calculated using the Kaplan – Meier method and compared using the log-rank test and a Cox proportional hazard model to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). To detect CLDN18.2 true-positive cases using biopsy specimens, receiver operating characteristic (ROC) analysis was conducted to establish a suitable cutoff proportion for the moderate-to-strong staining area of CLDN18 in biopsy samples. Youden Index is used to identify the point on an ROC curve where both sensitivity and specificity reach their peak values. This point represents the highest vertical distance from the diagonal line, indicating no discrimination on the ROC curve.18 Statistical significance was set at p < 0.05. All statistical analyses were performed using BellCurve for Excel version 4.08 and EZR software version 1.68.19
Results
Expression profiles of claudin-18 in surgical specimens of pancreatic cancer, panin, and normal ducts
We examined the expression and distribution of CLDN18 in pancreatic cancer, PanIN, and normal pancreatic ducts of resected specimens using immunohistochemical staining with the anti-CLDN18 antibody 43-14A. Strong CLDN18 expression was detected in PanIN, moderate expression in pancreatic cancer, and almost no expression in normal ducts (Figure 2(a-c)). CLDN18 was located on the plasma membrane of all tissue types. In the continuous area between low- and high-grade PanIN, CLDN18 staining delignated the low- and high-grade areas with a clear boundary (Figure 2(d)). In addition, moderate-to-strong levels of CLDN18 staining were observed in acinar-to-ductal metaplasia (Figure 2(e)). This metaplasia was excluded from the evaluation of the staining intensity and area. Even within the same tissue, the CLDN18 staining properties of cancer cells were heterogeneous (Figure 2(f)).
Figure 2.
Representative images of CLDN18 immunohistochemical stainings. The left panels show images of hematoxylin and eosin stainings, and the right panels show CLDN18 immunohistochemistry images of PDAC (a), low-grade PanIN (b), normal pancreatic duct (c), low- and high-grade PanIN (d), acinar-to-ductal metaplasia in the pancreas (e), and low-power fields of PDAC (f). Scale bar: 200 µm. CLDN18, claudin-18; PDAC, pancreatic ductal adenocarcinoma; panin, pancreatic intraepithelial neoplasia.
The intensity of CLDN18 expression was analyzed using immunohistochemistry in 211 resected pancreatic cancer tissues. The H-score of CLDN18 was significantly higher in PanIN than in pancreatic cancer, and CLDN18 was not expressed in almost all normal pancreatic ducts (p < 0.001 between tissue types; Figure 3(a)). CLDN18 expression was not detected in adenosquamous (n = 3), anaplastic (n = 1), or acinar cell carcinomas (n = 2). One case of invasive intraductal papillary carcinoma showed 20% of moderate-to-strong CLDN18 staining, and three cases did not show any moderate-to-strong CLDN18 staining. In pancreatic cancer, positive cases for any staining, including slight staining for CLDN18, were found in 72.0% (152/211); 97.8% (88/90) for PanINs and 3.3% (5/153) for normal pancreatic ducts. The cases showing CLDN18.2 positivity, that is, moderate to severe staining of 75% or more, were 9.5% (20/211), 73.3% (66/90), and 0.0% (0/153) for pancreatic cancer, PanINs, and normal pancreatic ducts, respectively (Table 1). The intensity of CLDN18 staining was heterogeneous within the same pancreatic cancer specimen (Figure 3(b)). CLDN18 was not expressed in 59 cases (28.0%), and 56 cases (26.5%) showed such heterogeneity that both no-to-weak and moderate-to-strong expression regions represented more than 25% in the same specimen. In the SPOTLIGHT and GLOW studies, CLDN18.2 positivity was defined as ≥ 75% of tumor cells demonstrating moderate-to-strong membranous CLDN18 staining.1–3 In the current study, the prevalence of CLDN18.2 positivity was 9.5% (20/211) among all screened patients (Figure 3(c).
Figure 3.
CLDN18 expression in pancreatic cancer. (a) H-score of CLDN18 in normal ducts, panin, and pancreatic cancer. Statistical significance was determined using the Mann – Whitney U test. *p < 0.001. (b) overall expression intensity of CLDN18 in pancreatic cancer tissues. (c) distribution of CLDN18 staining in all screened patients. The numbers above each bar indicate the proportion of samples within each bin. The dotted line indicates the cutoff for CLDN18.2 positivity (≥75% of tumor cells with moderate-to-strong membranous CLDN18 staining). (d) Kaplan – Meier curves for overall survival (left) and recurrence-free survival (right) of patients with pancreatic cancer grouped by CLDN18.2 positivity. CLDN18, claudin-18; H-score, histological score; panin, pancreatic intraepithelial neoplasia.
Table 1.
Percentage of each tissue types expressed CLDN18.
| Total number |
Any staining | 2+/3+ staining ≧75% |
|
|---|---|---|---|
| Pancreatic cancer | 211 | 152 (72.0%) | 20 (9.5%) |
| PanIN | 90 | 88 (97.8%) | 66 (73.3%) |
| Normal duct | 153 | 5 (3.3%) | 0 (0.0%) |
2+/3+ staining indicates moderate-to-strong membrane staining for CLDN18.
Using information from our hospital patient database, we performed a survival analysis of patients with pancreatic cancer regarding CLDN18.2 positivity. No significant correlation was observed between CLDN18.2 positivity and any clinicopathological feature (Table 2). CLDN18.2 positivity was not associated with recurrence-free survival (HR: 1.05, 95% CI: 0.63–1.77) and overall survival (HR: 1.04, 95% CI: 0.59–1.86; Figure 3(d)).
Table 2.
Prevalence of CLDN18.2 positivity by demographic characteristics among all screened patients.
| Variables | Cldn18 negative |
Cldn18 positive |
p-value |
|---|---|---|---|
| Overall | 191 (90.5) | 20 (9.5) | |
| Age | 0.819 | ||
| ≤70 | 101 (47.9) | 10 (4.7) | |
| >70 | 90 (42.7) | 10 (4.7) | |
| Sex | 0.640 | ||
| Female | 91 (43.1) | 11 (5.2) | |
| Male | 100 (47.4) | 9 (4.3) | |
| ASA-PS classification system | 0.440 | ||
| 0–2 | 155 (73.5) | 15 (7.1) | |
| 3 | 19 (9.0) | 3 (1.4) | |
| BMI | 0.765 | ||
| ≤25 | 139 (65.9) | 14 (6.6) | |
| > 25 | 35 (16.6) | 4 (1.9) | |
| Tumor location | 0.083 | ||
| Head | 133 (63.0) | 10 (4.7) | |
| Body/Tail | 58 (27.5) | 10 (4.7) | |
| Resectability | 0.811 | ||
| R | 133 (63.0) | 63 (29.9) | |
| BR/UR | 14 (6.6) | 8 (3.8) | |
| Neoadjuvant treatment | |||
| No | 103 (48.8) | 13 (6.2) | |
| Yes | 88 (41.7) | 7 (3.3) | |
| Poorly differentiated carcinoma | 0.317 | ||
| Negative | 129 (61.1) | 16 (7.6) | |
| Positive | 62 (29.4) | 4 (1.9) | |
| T-stage | 0.500 | ||
| pT1 | 52 (24.6) | 4 (1.9) | |
| pT2 | 95 (45.0) | 13 (6.2) | |
| pT3 | 44 (20.9) | 3 (1.4) | |
| pT4 | 0 (.0) | 0 (.0) | |
| N-stage | 1 | ||
| pN0 | 82 (38.9) | 8 (3.8) | |
| pN1 | 109 (51.7) | 12 (5.7) | |
| TNM-stage | 0.596 | ||
| pStage I | 58 (27.5) | 6 (2.8) | |
| pStage II | 63 (29.9) | 4 (1.9) | |
| pStage III | 27 (12.8) | 4 (1.9) | |
| pStage IV | 15 (7.1) | 2 (.9) | |
| Pre-operative CEA | 0.576 | ||
| Within normal | 146 (69.2) | 17 (8.1) | |
| Above normal (> 5 U/mL) |
45 (21.3) | 3 (1.4) | |
| Pre-operative CA19–9 | 1 | ||
| Within normal | 99 (46.9) | 10 (4.7) | |
| Above normal (> 37 U/mL) |
92 (43.6) | 10 (4.7) | |
Presented as n (%). ASA-PS, American Society of Anesthesiologists Physical Status; CEA, Carcinoembryonic Antigen; CA 19‑9, Carbohydrate Antigen 19‑9.
Comparison between preoperative biopsied and resection specimens
CLDN18 expression was compared in 133 cases in which both preoperative biopsy and resection specimens of the primary tumor were available for analysis. The proportions of samples with moderate-to-strong membrane staining were correlated between biopsy and resection specimens (Spearman’s rank correlation coefficient: 0.500, p < 0.001; Figure 4(a)). The concordance rate of CLDN18.2 positivity between these specimens was 92.5% (123/133; p < 0.001, Fisher’s test) at a 75% cutoff value (Table 3). The sensitivity and specificity were 54.6% and 95.9%, respectively. The concordance of CLDN18.2 positivity did not correlate with any clinicopathological features (Table 4), and the presence or absence of neoadjuvant treatment did not correlate with the difference in the proportion of moderate-to-strong CLDN18 staining areas between biopsy and resection specimens (p = 0.760, Mann – Whitney U test).
Figure 4.
Correlation of CLDN18.2 positivity between biopsy and resection specimens. (a) bubble chart showing the proportions of moderate-to-strong membranous staining for CLDN18 in biopsy and resection specimens. The bubble size indicates the number of cases. Dotted lines indicate 75%. Orange bubbles represent cases in which the CLDN18.2 positivity of a biopsy specimen did not match that of the corresponding resection specimen. (b and c) Representative images of cases in which the positivity of biopsy and resection specimens matched (B) and did not match (c). The left panels show images of hematoxylin and eosin staining, and the right panels show CLDN18 immunohistochemistry images of PDAC. (d) ROC curve of the proportion of moderate-to-strong CLDN18 staining of biopsy specimens that is indicative for the CLDN18.2 positivity of resection specimens. CLDN18.2, claudin-18.2; PDAC, pancreatic ductal adenocarcinoma; ROC, receiver operating characteristic.
Table 3.
Summary of CLDN18.2 positivity comparisons across biopsy, resection, and recurrence specimens under different cutoff conditions.
| Comparison | Group | Resection ≧75% | Resection < 75% | Total | Concordance | Sensitivity | Specificity |
|---|---|---|---|---|---|---|---|
| 1. Biopsy vs Resection | Biopsy ≧75% | 6 | 5 | 11 | |||
| (cutoff: 75%/75%) | Biopsy < 75% | 5 | 117 | 122 | |||
| Total | 11 | 122 | 133 | 92.5% | 54.6% | 95.9% | |
| 2. Biopsy vs Resection | Biopsy ≧20% | 11 | 44 | 55 | |||
| (cutoff: 20%/75%) | Biopsy < 20% | 0 | 78 | 78 | |||
| |
Total |
11 |
122 |
133 |
66.9% |
100.0% |
63.9% |
| Comparison |
Group |
Recurrence ≧75% |
Recurrence < 75% |
Total |
|
|
|
| 3. Primary vs Recurrence | Primary ≧75% | 3 | 6 | 9 | |||
| (cutoff: 75%/75%) | Primary < 75% | 4 | 47 | 51 | |||
| Total | 7 | 53 | 60 | 83.3% | 42.9% | 88.7% |
Comparison 1–3 compares different cutoff proportions of moderate-to-strong membrane staining for CLDN18 and sample types.
Table 4.
Prevalence of unmatched CLDN18.2 positivity between biopsy and resected specimens.
| Variables | unmatched | matched | p-value |
|---|---|---|---|
| Overall | 10 (7.5) | 123 (92.5) | |
| Age | 0.520 | ||
| ≤70 | 4 (3.0) | 65 (48.9) | |
| > 70 | 6 (4.5) | 58 (43.6) | |
| Sex | 0.744 | ||
| Female | 6 (4.5) | 61 (45.9) | |
| Male | 4 (3.0) | 62 (46.6) | |
| ASA-PS classification system | 0.610 | ||
| 0–2 | 10 (7.5) | 97 (72.9) | |
| 3 | 0 (.0) | 15 (11.3) | |
| BMI | 0.106 | ||
| ≤25 | 6 (4.5) | 92 (69.2) | |
| > 25 | 4 (3.0) | 20 (15.0) | |
| Tumor location | 0.725 | ||
| Head | 6 (4.5) | 85 (63.9) | |
| Body/Tail | 4 (3.0) | 38 (28.6) | |
| Resectability | 1 | ||
| R | 7 (5.3) | 82 (61.7) | |
| BR/UR | 3 (2.3) | 41 (30.8) | |
| Neoadjuvant treatment | 1 | ||
| No | 5 (3.8) | 56 (42.1) | |
| Yes | 5 (3.8) | 67 (50.4) | |
| Evans grade | 0.559 | ||
| 1 | 3 (2.3) | 18 (13.5) | |
| 2a/2b | 2 (1.5) | 37 (27.8) | |
| 3 | 0 (.0) | 5 (3.8) | |
| Poorly differentiated carcinoma | 1 | ||
| Negative | 7 (5.3) | 84 (63.2) | |
| Positive | 3 (2.3) | 39 (29.3) | |
| T-stage | 0.753 | ||
| pT1 | 3 (2.3) | 31 (23.3) | |
| pT2 | 4 (3.0) | 63 (47.4) | |
| pT3 | 3 (2.3) | 29 (21.8) | |
| pT4 | 0 (.0) | 0 (.0) | |
| N-stage | 0.519 | ||
| pN0 | 5 (3.8) | 48 (36.1) | |
| pN1 | 5 (3.8) | 75 (56.4) | |
| TNM-stage | 0.959 | ||
| pStage I | 3 (2.3) | 34 (25.6) | |
| pStage II | 4 (3.0) | 44 (33.1) | |
| pStage III | 2 (1.5) | 20 (15.0) | |
| pStage IV | 0 (.0) | 11 (8.3) | |
| Pre-operative CEA | 0.690 | ||
| Within normal | 9 (6.8) | 95 (71.4) | |
| Above normal (> 5 U/mL) |
1 (.8) | 28 (21.1) | |
| Pre-operative CA19–9 | 0.753 | ||
| Within normal | 6 (4.5) | 66 (49.6) | |
| Above normal (> 37 U/mL) |
4 (3.0) | 57 (42.9) | |
Presented as n (%). ASA-PS, American Society of Anesthesiologists Physical Status; CEA, Carcinoembryonic Antigen; CA 19‑9, Carbohydrate Antigen 19‑9.
The sensitivity of biopsied specimens was not high at 54.6% (6/11); five unmatched cases showed negative results in the biopsied specimens but CLDN18.2 positivity in the resection specimens (Figure 4(b,c)). To reduce the risk of false-negative results in the biopsy specimens, ROC analysis was performed to determine the optimal threshold for moderate-to-strong membranous CLDN18 staining predictive of CLDN18.2 positivity in resected tissues (Figure 4(d)). The area under the curve was 0.866 (95% CI: 0.774–0.958, p < 0.001), and a 20% staining threshold showed the highest Youden index (0.639). At this cutoff, the sensitivity for detecting CLDN18.2-positive tumors in resection specimens increased to 100% (11/11). However, this lower threshold does not correspond to the current therapeutic criteria for zolbetuximab eligibility, which requires ≥ 75% of tumor cells with moderate-to-strong membranous staining. Thus, while the 20% cutoff may serve as a useful screening tool to guide further testing, it cannot be used directly to define treatment eligibility.
Comparison between primary and recurrent specimens
CLDN18 expression was compared in 53 patients and 60 samples in which both primary tumors and recurrent tissues were available for analysis. The recurrent sites of the tissues used in this analysis were local sites (remnant pancreas or tumor beds; n = 15), the liver (n = 14), lungs (n = 18), peritoneum (n = 6), and others (para-aortic lymph nodes and tumor cells in pleural effusion). The proportions of moderate-to-strong membrane stainings were correlated between primary and recurrent specimens (Spearman’s rank correlation coefficient: 0.396, p = 0.002; Figure 5(a)). The concordance rate of CLDN18.2 positivity between these specimens was 83.3% (50/60; p = 0.062, Fisher’s test) at a 75% cutoff (Table 3). The sensitivity and specificity were 42.9% and 88.7%, respectively. The concordance of CLDN18.2 positivity did not correlate with any clinicopathological features, including the presence or absence of adjuvant treatment (Table 5). In many cases of local recurrence and liver metastasis, the proportions of moderate-to-strong CLDN18 stainings were reduced in recurrent lesions compared to those in the corresponding primary lesions (p = 0.029 for local recurrence, p = 0.175 for liver metastasis, Wilcoxon signed-rank sum test; Figure 5(b-d)). The overall survival rate was not correlated with CLDN18.2 positivity of recurrent specimens (HR: 1.144, 95% CI: 0.439–2.984) and the concordance between primary and recurrent specimens (HR: 1.358, 95% CI: 0.477–3.870).
Figure 5.
Correlation of CLDN18.2 positivity between primary and recurrent specimens. (a) bubble chart showing the proportions of moderate-to-strong membranous staining for CLDN18 in primary and recurrent specimens. The bubble size indicates the number of cases. Dotted lines indicate 75%. Orange bubbles represent cases in which the CLDN18.2 positivity of a primary specimen did not match that of the corresponding recurrent specimen. (b) proportion of moderate-to-strong membrane staining for CLDN18 in primary and recurrent specimens. Statistical significance was determined using the Wilcoxon signed-rank sum test. *p = 0.029. The heatmap shows the difference in the proportion of moderate-to-strong staining areas between primary and recurrent lesions (proportion of recurrent lesions minus the proportion of primary lesions). (c and d) Representative images of cases in which the positivity of the primary and recurrent specimens did not match. The left panels show images of hematoxylin and eosin stainings, and the right panels show CLDN18 immunohistochemistry images. (c) CLDN18-negative staining in primary PDAC (upper panels) and positive staining in lung metastasis (lower panels). (D) CLDN18-positive staining in primary PDAC (upper panels) and negative staining in liver metastasis (lower panels). CLDN18.2, claudin-18.2; PDAC, pancreatic ductal adenocarcinoma.
Table 5.
Prevalence of unmatched CLDN18.2 positivity between primary and recurrent specimens.
| Variables | unmatched | matched | p-value |
|---|---|---|---|
| Overall | 10 (16.7) | 50 (83.3) | |
| Age | 0.263 | ||
| ≤70 | 5 (8.3) | 36 (60.0) | |
| > 70 | 5 (8.3) | 14 (23.3) | |
| Sex | 0.737 | ||
| Female | 4 (6.7) | 24 (40.0) | |
| Male | 6 (10.0) | 26 (43.3) | |
| ASA-PS classification system | 0.190 | ||
| 0–2 | 7 (11.7) | 42 (70.0) | |
| 3 | 2 (3.3) | 3 (5.0) | |
| BMI | 1 | ||
| ≤25 | 8 (13.3) | 38 (63.3) | |
| > 25 | 1 (1.7) | 7 (11.7) | |
| Tumor location | 1 | ||
| Head | 6 (10.0) | 29 (48.3) | |
| Body/Tail | 4 (6.7) | 21 (35.0) | |
| Resectability | 0.073 | ||
| R | 3 (5.0) | 33 (55.0) | |
| BR/UR | 7 (11.7) | 17 (28.3) | |
| Poorly differentiated carcinoma | 0.148 | ||
| Negative | 9 (15.0) | 32 (53.3) | |
| Positive | 1 (1.7) | 18 (30.0) | |
| T-stage | 0.356 | ||
| pT1 | 1 (1.7) | 15 (25.0) | |
| pT2 | 7 (11.7) | 23 (38.3) | |
| pT3 | 2 (3.3) | 12 (20.0) | |
| pT4 | 0 (.0) | 0 (.0) | |
| N-stage | 0.305 | ||
| pN0 | 6 (10.0) | 20 (33.3) | |
| pN1 | 4 (6.7) | 30 (50.0) | |
| TNM-stage | 0.664 | ||
| pStage I | 5 (8.3) | 17 (28.3) | |
| pStage II | 1 (1.7) | 12 (20.0) | |
| pStage III | 1 (1.7) | 9 (15.0) | |
| pStage IV | 2 (3.3) | 6 (10.0) | |
| Post-operative CEA | 0.409 | ||
| Within normal | 7 (11.7) | 40 (66.7) | |
| Above normal (> 5 U/mL) |
3 (5.0) | 9 (15.0) | |
| Post-operative CA19–9 | 0.577 | ||
| Within normal | 10 (16.7) | 43 (71.7) | |
| Above normal (> 37 U/mL) |
0 (.0) | 6 (10.0) | |
| Adjuvant treatment | 0.675 | ||
| No | 3 (5.0) | 10 (16.7) | |
| Yes | 7 (11.7) | 40 (66.7) | |
| Early recurrence ( < 1y) | 1 | ||
| No | 5 (8.3) | 30 (50.0) | |
| Yes | 3 (5.0) | 15 (25.0) | |
Presented as n (%). ASA-PS, American Society of Anesthesiologists Physical Status; CEA, Carcinoembryonic Antigen; CA 19‑9, Carbohydrate Antigen 19‑9.
Discussion
This study demonstrated a significant concordance in CLDN18.2 positivity between preoperative biopsy and resected PDAC specimens (92.5%) and between primary and recurrent lesions (83.3%). However, these results should be interpreted with caution. Although the apparent concordance between biopsy and surgical samples is high, this is largely due to the high frequency of CLDN18.2-negative findings in both. Notably, the sensitivity of biopsies in detecting CLDN18.2-positive tumors was only 54.6%, implying that nearly half of the positive cases went undetected preoperatively, which is a substantial concern for zolbetuximab treatment planning.
Consequently, the observed agreement does not necessarily reflect a robust predictive validity. Our results suggest that relying solely on biopsy assessments may fail to identify all potential candidates for CLDN18.2-targeted therapy. Clinically, this could mean that patients who might benefit from zolbetuximab are inadvertently excluded if only the biopsy data are considered. These findings support a more critical appraisal of biopsy-based results and highlight the importance of re-testing resected tissue when possible. Furthermore, although ROC analysis suggested that a 20% cutoff for moderate-to-strong membranous staining in biopsy samples yielded 100% sensitivity, the clinical utility of this threshold is limited. Current therapeutic eligibility requires ≥ 75% staining; therefore, individuals within the 20–74% range would still not qualify despite being potentially positive. This discrepancy creates a diagnostic gray zone, highlighting the risk of misclassification if the biopsy criteria are modified without concurrent adjustment of the treatment protocols. As such, while the 20% threshold may serve as a useful prescreening indicator, it must be applied within a framework of clearly defined clinical follow-ups. These issues underscore the importance of integrating diagnostic thresholds and therapeutic strategies.
The concordance of CLDN18.2 positivity between primary and recurrent pancreatic cancer lesions was slightly lower (83.3%) than that observed between biopsy and resection specimens. The concordance rates of CLDN18.2 positivity between primary and recurrent lesions differ depending on the tissue type and primary site of the cancer. For gastric cancer, the concordance rate between the primary tumor and its metastases ranged from 74.8% to 81.5%,5,7,8 whereas for ovarian cancer, it was 58% in mucinous tubo-ovarian carcinoma depending on the histological subtype.20 Although this concordance is relatively high, it should be interpreted with caution. In some cases, particularly those with local recurrence or liver metastasis, the proportion of CLDN18.2-positive cells was reduced in recurrent lesions compared to that in primary tumors. Similarly, decreased concordance rates in liver metastases have been observed in gastric cancer.5 This observation may reflect clonal selection or phenotypic shifts associated with tumor progression, therapeutic interventions, or changes in the tumor microenvironment during metastasis. Despite these changes, our findings suggest that CLDN18.2 expression in primary tumors may, in most cases, serve as a surrogate for its expression in recurrent sites, which is clinically valuable given the limited feasibility of re-biopsy in recurrent pancreatic cancer. These results support the potential utility of initial CLDN18.2 assessment in guiding treatment decisions for both primary and recurrent diseases, although confirmatory studies with larger cohorts and more metastatic samples are warranted.
Importantly, the potential clinical and medico-legal implications of misclassifying CLDN18.2-positive patients should not be overlooked. If biopsy-based evaluation is used as the sole determinant of treatment eligibility, patients with true CLDN18.2-positive tumors may be excluded from zolbetuximab therapy due to false-negative results. In such cases, failure to offer appropriate targeted therapy may result in suboptimal clinical outcomes and could pose liability risks in clinical practice. To mitigate these risks, institutions may consider implementing standardized protocols that recommend retesting resected specimens, when available, particularly in patients who test negative on initial biopsy but are otherwise candidates for targeted treatment. These safeguards are essential to ensure both diagnostic accuracy and ethical responsibility in patient care.
Our study reported a CLDN18.2 positivity rate of only 9.5%, using the analytically validated 43–14A clone and the same scoring threshold used in zolbetuximab trials.1,2 Previous studies on CLDN18.2 expression in PDAC have reported widely varying positivity rates, ranging from approximately 30% to over 90%, depending on the antibody clone, scoring criteria, and tissue processing methods used.6,21–26 Among these studies, only one used the same primary antibody clone (43-14A) and an automated staining device as the clinical trial protocol for pancreatic cancer.21 They reported a CLDN18.2 positivity rate of 32.5%. This difference from our results may reflect the use of the clinical VENTANA assay, which offers higher contrast and sensitivity than our in-house protocol, despite concordant validation results. Ethnic and geographic variations may also be a factor, as a previous study was conducted in Western populations, whereas our cohort was exclusively Japanese. Methodological differences, including the use of whole tissue sections versus tissue microarrays and fixation protocols, may further contribute to these discrepancies. Although our in-house staining protocol was validated and showed concordance with the VENTANA assay in a subset of cases, we acknowledge that the lower staining contrast of our method may have contributed to the underestimation of CLDN18.2 positivity, particularly in borderline cases. This technical factor should be considered when interpreting absolute positivity rates. Although the overall positive rate was low for CLDN18.2, this is the only study to systematically and quantitatively compare CLDN18.2 expression across biopsy, resection, and recurrent tumor specimens. These methodological strengths enhance the clinical relevance and generalizability of our findings to the general population of the study.
While our findings showed no significant association between CLDN18.2 positivity and overall survival in pancreatic cancer, Arseneau et al. presented differing results, noting enhanced survival involving CLDN18 expression.21 Several explanations may underlie these differences in the results. Notably, our larger cohort size (211 vs. 120 patients) may offer greater statistical confidence. Moreover, differences in patient populations, including geographic and ethnic factors, may have led to variations in tumor biology and clinical outcomes of the studies. Arseneau et al. observed a significant link between CLDN18 positivity and better-differentiated (grade 1) tumors, which was not present in our cohort, suggesting distinct biological behavior. Furthermore, methodological differences may have influenced the results; Arseneau et al. based their conclusions on survival improvement at a specific median survival point (23 months), whereas our findings were consistently negative using conventional Kaplan-Meier and Cox regression analyses. Together, these factors likely contribute to the observed disparities between studies and highlight the importance of carefully interpreting the prognostic relevance of CLDN18.2 expression.
This study has several limitations. First, the retrospective nature and single-center design may limit the generalizability of our findings. Second, although we used the same antibody clone (43-14A) as in clinical trials, our CLDN18 immunostaining was not performed using the VENTANA CLDN18 (43-14A) RxDx Assay on the BenchMark ULTRA platform (VMSI/Roche), which is the clinical standard for gastric cancer. This assay became available in Japan in May 2024, after our study was already underway. Moreover, the high cost of the system renders its use infeasible for large-scale research involving over 200 cases. To address this limitation, we validated our in-house staining protocol by comparing the staining patterns and scores with those obtained using the clinical-standard method in a subset of tissue microarray samples. This comparison showed high concordance with no discrepancies in positivity or staining distribution. However, we acknowledge that the VENTANA system provided a slightly superior contrast between positive and negative regions, which may have influenced the overall CLDN18.2 positivity rate. Therefore, we note this as a potential limitation when interpreting the absolute frequency of CLDN18.2-positive cases in our study. Third, the relatively low proportion of CLDN18.2-positive cases may have contributed to an overestimation of concordance, owing to the high number of true-negative pairs. Forth, the number of recurrent specimens, particularly from metastatic sites, was limited. Finally, the influence of treatment history on CLDN18.2 expression could not be fully evaluated. These limitations might be related to patient cohort bias, including race and treatment modality, which may change the percentage of CLDN18.2- positive patients. In this case, the concordance rate of CLDN18.2 positivity among biopsy, resection, and recurrent specimens and the cutoff values in biopsied tissues would also change. Future multicenter studies involving larger cohorts, especially with more CLDN18.2-positive cases and diverse metastatic sites, are warranted to validate and refine our findings. In addition, longitudinal monitoring of CLDN18.2 expression before and after treatment may help to clarify its temporal stability and predictive value. Despite these limitations, the present study provides a foundation for incorporating CLDN18.2 testing into the diagnostic and therapeutic workflows for pancreatic cancer and highlights its potential role as a reliable biomarker in both localized and recurrent disease contexts.
Conclusion
This study demonstrated that CLDN18.2 expression in biopsy specimens of patients with PDAC showed partial concordance with resected tissues and, to a lesser extent, with recurrent lesions. Although the concordance rate between the biopsy and resection specimens was numerically high, this finding was largely driven by a high proportion of CLDN18.2-negative cases. The sensitivity of biopsy in detecting CLDN18.2-positive tumors was limited, suggesting that biopsy-based evaluation alone may lead to underdiagnosis and exclusion of eligible patients from zolbetuximab therapy. While ROC analysis indicates that a lower threshold could improve sensitivity in biopsy screening, this approach is not currently aligned with the therapeutic eligibility criteria and may introduce diagnostic ambiguity.
These findings highlight the potential and limitations of CLDN18.2 assessment in small biopsy samples. From a clinical standpoint, confirmatory testing of resected specimens, when available, may be warranted, especially in cases with negative biopsy results but with high clinical suspicion. Incorporating CLDN18.2 testing into the diagnostic workflow for PDAC must be accompanied by careful interpretation and institutionally defined guidelines to ensure accurate and ethically responsible patient selection for targeted therapy. Further prospective multicenter studies are needed to refine the diagnostic thresholds and validate the reliability of biopsy-based assessments in broader clinical contexts.
Given the dismal prognosis of pancreatic cancer and the emerging potential of zolbetuximab to improve patient survival, we hope that the findings of this study will help facilitate appropriate patient selection and ultimately contribute to better clinical outcomes for individuals affected by this devastating disease.
Supplementary Material
Acknowledgments
We would like to thank Yui Kawami and Fuminori Daimon for technical assistance with the experiments, Reona Okumura and Hinae Asano MD-PhD students at Sapporo Medical University for sampling and statistical analysis, and Editage (http://www.editage.com) for editing and reviewing this manuscript for English language.
Funding Statement
This work was supported by JSPS KAKENHI [grant numbers JP21K08715 and JP24K11826], the Pancreas Research Foundation of Japan, and THE SUHARA MEMORIAL FOUNDATION. Japan Society for the Promotion of Science
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/21688370.2025.2535047
Author contributions
Conceptualization, D.K.; methodology, D.K; validation, D.K., K.Y., M.O.; formal analysis, D.K., K.Y.; investigation, D.K., K.Y., A.S., Y.O., M.O.; resources, D.K., T.I., M.I.; data curation, D.K., K.Y.; writing – original draft preparation, D.K.; writing – review and editing, D.K.; visualization, D.K., K.Y.; supervision, D.K.; project administration, D.K.; funding acquisition, D.K., M.O. All authors have read and agreed to the published version of the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Ethics approval
This study was approved by the Institutional Review Board of Sapporo Medical University (IRB study number 292–68). The study was approved on August 25, 2017. The need for informed consent was waived by the Institutional Review Board of Sapporo Medical University because of the retrospective nature of the study.
References
- 1.Shitara K, Lordick F, Bang Y-J, Enzinger P, Ilson D, Shah MA, Van Cutsem E, Xu R-H, Aprile G, Xu J, et al. Zolbetuximab plus mFOLFOX6 in patients with CLDN18.2-positive, HER2-negative, untreated, locally advanced unresectable or metastatic gastric or gastro-oesophageal junction adenocarcinoma (SPOTLIGHT): a multicentre, randomised, double-blind, phase 3 trial. Lancet. 2023;401(10389):1–17. doi: 10.1016/S0140-6736(23)00620-7. [DOI] [PubMed] [Google Scholar]
- 2.Shah MA, Shitara K, Ajani JABang Y-JEnzinger PIlson DLordick FVan Cutsem EGallego Plazas JHuang J, et al. Zolbetuximab plus CAPOX in CLDN18.2-positive gastric or gastroesophageal junction adenocarcinoma: the randomized, phase 3 GLOW trial. Nat Med. 2023;29(8):2133–2141. doi: 10.1038/s41591-023-02465-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Shitara K, Xu R-H, Ajani JA, Moran D, Guerrero A, Li R, Pavese J, Matsangou M, Bhattacharya P, Ueno Y, et al. Global prevalence of claudin 18 isoform 2 in tumors of patients with locally advanced unresectable or metastatic gastric or gastroesophageal junction adenocarcinoma. Gastric Cancer. 2024;27(5):1058–1068. doi: 10.1007/s10120-024-01518-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kim T-Y, Kwak Y, Nam SK, Han D, Oh D-Y, Im S-A, Lee HS.. Clinicopathological analysis of claudin 18.2 focusing on intratumoral heterogeneity and survival in patients with metastatic or unresectable gastric cancer. ESMO Open. 2024;9(12):104000. doi: 10.1016/j.esmoop.2024.104000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Choi E, Shin J, Ryu M-H, Kim H-D, Park YS. Heterogeneity of claudin 18.2 expression in metastatic gastric cancer. Sci Rep. 2024;14(1):17648. doi: 10.1038/s41598-024-68411-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wang X, Zhang C-S, Dong X-Y, Hu Y, Duan B-J, Bai J, Wu Y-Y, Fan L, Liao X-H, Kang Y, et al. Claudin 18.2 is a potential therapeutic target for zolbetuximab in pancreatic ductal adenocarcinoma. World J Gastrointest Oncol. 2022;14(7):1252–1264. doi: 10.4251/wjgo.v14.i7.1252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ogawa H, Abe H, Yagi K, Seto Y, Ushiku T. Claudin-18 status and its correlation with HER2 and PD-L1 expression in gastric cancer with peritoneal dissemination. Gastric Cancer. 2024;27(4):802–810. doi: 10.1007/s10120-024-01505-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Pellino A, Brignola S, Riello E, Niero M, Murgioni S, Guido M, Nappo F, Businello G, Sbaraglia M, Bergamo F, et al. Association of CLDN18 protein expression with clinicopathological features and prognosis in advanced gastric and gastroesophageal junction adenocarcinomas. J Personalized Med. 2021;11(11):1095. doi: 10.3390/jpm11111095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sui K, Okabayashi T, Shima Y, Morita S, Iwata J, Sumiyoshi T, Saisaka Y, Hata Y, Noda Y, Matsumoto M, et al. Clinical effects of chemoradiotherapy in pursuit of optimal treatment of locally advanced unresectable pancreatic cancer. The Br J Radiol. 2017;90(1075):20170165. doi: 10.1259/bjr.20170165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Maulat C, Canivet C, Touraine C, Gourgou S, Napoleon B, Palazzo L, Flori N, Piessen G, Guibert P, Truant S, et al. A new score to predict the resectability of pancreatic adenocarcinoma: the BACAP score. Cancers (Basel). 2020;12(4):783. doi: 10.3390/cancers12040783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kovač JD, Mayer P, Hackert T, Klauss M. The time to and type of pancreatic cancer recurrence after surgical resection: is prediction possible? Academic Radiol. 2019;26(6):775–781. doi: 10.1016/j.acra.2018.07.025. [DOI] [PubMed] [Google Scholar]
- 12.Brierley JD, Gospodarowicz MK, Wittekind C. TNM classification of malignant tumours. Hoboken, NJ, USA: John Wiley & Sons; 2017. [Google Scholar]
- 13.Tempero MA, Malafa MP, Al-Hawary M, Behrman SW, Benson AB, Cardin DB, Chiorean EG, Chung V, Czito B, Del Chiaro M, et al. Pancreatic adenocarcinoma, version 2.2021, NCCN clinical practice guidelines in oncology. J Nat Comp Cancer Network: JNCCN. 2021;19(4):439–457. doi: 10.6004/jnccn.2021.0017. [DOI] [PubMed] [Google Scholar]
- 14.Japan Pancreas Society ed . General rules for the study of pancreatic cancer. 8th ed. Tokyo, Japan: Kanehara & Co, Ltd; 2023. [Google Scholar]
- 15.Hashimoto I, Oshima T. Claudins and gastric cancer: an overview. Cancers (Basel). 2022;14(2):290. doi: 10.3390/cancers14020290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Singh P, Toom S, Huang Y. Anti-claudin 18.2 antibody as new targeted therapy for advanced gastric cancer. J Hematol Oncol. 2017;10(1):105. doi: 10.1186/s13045-017-0473-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Rohde C, Yamaguchi R, Mukhina S, Sahin U, Itoh K, Ö T. Comparison of Claudin 18.2 expression in primary tumors and lymph node metastases in Japanese patients with gastric adenocarcinoma. Jpn J Clin Oncol. 2019;49(9):870–876. doi: 10.1093/jjco/hyz068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Habibzadeh F, Habibzadeh P, Yadollahie M. On determining the most appropriate test cut-off value: the case of tests with continuous results. Biochem Med (Zagreb). 2016;26(3):297–307. doi: 10.11613/BM.2016.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kanda Y. Investigation of the freely available easy-to-use software “EZR” for medical statistics. Bone Marrow Transplant. 2013;48(3):452–458. doi: 10.1038/bmt.2012.244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wagner P, Gass P, Pöschke P, Eckstein M, Gloßner L, Hartmann A, Beckmann MW, Fasching PA, Ruebner M, Emons J, et al. Spatial expression of claudin 18.2 in matched primaries and metastases of tubo-ovarian carcinoma of all subtypes. Virchows Arch: An Int J Pathol. 2024;485(1):63–74. doi: 10.1007/s00428-024-03756-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Arseneau RJ, Kempster E, Bekkers C, Samson T, Gala-Lopez BL, Ramjeesingh R, Boudreau JE, Arnason T. Claudin 18 (43–14A clone) expression in pancreatic ductal adenocarcinoma: assessment of a potential clinical biomarker for zolbetuximab therapy. Transl Oncol. 2025;55(102362):102362. doi: 10.1016/j.tranon.2025.102362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Park S, Shin K, Kim I-H, Hong T, Kim Y, Suh J, Lee M. Clinicopathological features and prognosis of resected pancreatic ductal adenocarcinoma patients with claudin-18 overexpression. J Clin Med Res. 2023;12(16):5394. doi: 10.3390/jcm12165394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tanaka M, Shibahara J, Fukushima N, Shinozaki A, Umeda M, Ishikawa S, Kokudo N, Fukayama M. Claudin-18 is an early-stage marker of pancreatic carcinogenesis. J Histochem Cytochem. 2011;59(10):942–952. doi: 10.1369/0022155411420569. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Valentini AM, Arborea G, Grassi I, Savino M, Labarile N, Donghia R, Iacovazzi P, Vallarelli S, Ostuni C, Lotesoriere C, et al. Claudin 18.2: an attractive marker in pancreatic ductal adenocarcinoma. Oncol Lett. 2025;29(3):140. doi: 10.3892/ol.2025.14886. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zhang Z, Liu X, Zhou L, Zhang M, Liang Z. Investigation of clinical application of claudin 18 isoform 2 in pancreatic ductal adenocarcinoma: a retrospective analysis of 302 Chinese patients. Histol Histopathol. 2022;37(10):1031–1040. doi: 10.14670/HH-18-477. [DOI] [PubMed] [Google Scholar]
- 26.Lyu SI, Fretter C, Simon AG, Spielmann S-M, Damanakis AI, Zhao Y, Bruns CJ, Schmidt T, Popp FC, Waldschmidt D, et al. Extent and clinical significance of the therapy-relevant tight junction protein Claudin 18.2 in pancreatic ductal adenocarcinoma - real-world evidence. Transl Oncol. 2024;47(102044):102044. doi: 10.1016/j.tranon.2024.102044. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.