Evidence map›Paper›PMID 42407198›Full record

ArticleESMO open2026

Artificial intelligence-based prediction of claudin 18.2 expression and immune phenotype from routine histology to guide treatment decisions in patients with gastric cancer.

H-D Kim, S Shin, W Hwang, J Shin, T Lee, J Hyung, J Park, S Pereira, C-Y Ock, A Puccini and 3 more

Abstract read
In one paragraph

Article in ESMO open, 2026. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Not yet cited in PubMed.

0numbers the graph read from it
0cells of the map it votes in
0citing papers in PubMed
–field-weighted citation impact
1 · What the graph read from it

What it found

Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.

The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.

2 · The registry

The trial behind it

Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.

Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.

3 · Its place in the literature

Who cites it

0 citing papers in PubMed.

No citing paper in PubMed yet.

4 · The record

Corrections and comments

PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.

5 · Who and what money

Authors and funding

13 authors.

H-D KimDepartment of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea. Electronic address: https://twitter.com/HyungDonKim86.
S ShinLunit, Seoul, Republic of Korea.
W HwangLunit, Seoul, Republic of Korea.
J ShinDepartment of Pathology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea.
T LeeLunit, Seoul, Republic of Korea.
J HyungDepartment of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea.
J ParkLunit, Seoul, Republic of Korea.
S PereiraLunit, Seoul, Republic of Korea.
C-Y OckLunit, Seoul, Republic of Korea.
A PucciniDepartment of Biomedical Sciences, Humanitas University, Pieve Emanuele, Milan, Italy; Medical Oncology and Hematology Unit, Humanitas Cancer Center, IRCCS Humanitas Research Hospital, Rozzano, Milan, Italy.
G CarloniLunit, Seoul, Republic of Korea. Electronic address: gianluca.carloni@lunit.io.
Y S ParkDepartment of Pathology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea. Electronic address: youngspark@amc.seoul.kr.
M-H RyuDepartment of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea. Electronic address: miniryu@amc.seoul.kr.

Funding

No grant is acknowledged in the PubMed record.

6 · The paper itself

Abstract

backgroundFirst-line treatment of gastric cancer is evolving with the integration of immune checkpoint inhibitors (ICIs) and targeted agents, complicating biomarker stratification. Claudin 18.2 (CLDN18.2) is an established target for zolbetuximab; however, immunohistochemistry (IHC) is limited by tissue requirements, cost, and turnaround time. Artificial intelligence (AI) analysis of hematoxylin and eosin (H&E)-stained slides may provide a scalable alternative. We developed and validated an AI model to predict CLDN18.2 expression from H&E slides and evaluated its clinical utility integrated with AI-derived immune phenotyping. PATIENTS AND

methodsThis retrospective study included three independent cohorts of patients with gastric cancer. The development cohort comprised 622 patients (497 for training and 125 for tuning). The internal validation cohort included 378 patients treated with first-line nivolumab plus chemotherapy or chemotherapy alone. The external validation cohort included 98 patients from diverse ethnic backgrounds. Whole-slide H&E-stained images were analyzed using a Vision Transformer-based AI model to predict CLDN18.2 expression. A separate AI model classified the immune microenvironment as inflamed or noninflamed. Primary outcomes were predictive performance metrics, including area under the receiver operating characteristic curve (AUROC). Secondary outcomes included progression-free survival (PFS) and overall survival (OS), stratified by AI-predicted CLDN18.2 status and immune phenotype.

resultsCLDN18.2 positivity by IHC was 42.9% (development), 36.8% (internal validation), and 25.5% (external validation). The AI model yielded AUROC values of 0.752 (internal validation) and 0.856 (external validation). In the internal validation cohort, patients with AI-predicted CLDN18.2-negative/inflamed tumors exhibited improved outcomes with nivolumab plus chemotherapy versus chemotherapy alone [PFS: hazard ratio (HR) 0.35, 95% confidence interval (CI) 0.15-0.82; OS: HR 0.40, 95% CI 0.18-0.89]. Patients with CLDN18.2-positive/noninflamed tumors showed no benefit from nivolumab plus chemotherapy.

conclusionsAn AI model using routine histology predicted CLDN18.2 expression and immune phenotype in gastric cancer, identifying subgroups with differential benefit from ICI-based chemotherapy.

Indexed as

Artificial IntelligenceClaudinsStomach NeoplasmsAgedBiomarkers, TumorFemaleHumansImmunohistochemistryMaleMiddle AgedPhenotypeRetrospective StudiesTumor MicroenvironmentBiomarkers, TumorClaudinsCLDN18 protein, humanartificial intelligenceclaudin 18.2digital pathologygastric cancerimmune checkpoint inhibitorsimmune phenotype

Identifiers

PMID42407198
PMCPMC13355801

What Socratic holds

Textmetadata
LicenceCC BY-NC-ND
Read underepoch 390

Registered trials

None linked

Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the Socratic graph.