Clinicopathological and Imaging Distinction Between Ocular Adnexal MALT Lymphoma and IgG4-Related Ophthalmic Disease.

PURPOSE: To characterize clinicopathologic and imaging differences between ocular adnexal mucosa-associated lymphoid tissue lymphoma (OAML) and IgG4-related ophthalmic disease (IgG4-ROD) and to evaluate clinical and artificial intelligence-assisted approaches for differentiating these entities.

DESIGN: Retrospective, single-center observational cross-sectional study.

PARTICIPANTS: Four hundred patients with pathologically confirmed orbital lymphoproliferative disease, including OAML (n = 287) and IgG4-ROD (n = 113), treated at West China Hospital between 2010 and 2024.

METHODS: Clinical, serologic, imaging, histopathologic, immunophenotypic, and molecular features were systematically compared between groups. A multivariable logistic regression model based on routinely available clinical variables was developed for early differentiation. Artificial intelligence models were additionally trained using radiological CT/MRI images and digital pathology whole-slide images.

MAIN OUTCOME MEASURES: Distinct clinical and pathologic characteristics between OAML and IgG4-ROD and diagnostic discrimination assessed by area under the receiver operating characteristic curve (AUC).

RESULTS: OAML more frequently presented as unilateral orbital or conjunctival lesions and showed monoclonal features, including light-chain restriction and immunoglobulin gene rearrangements. In contrast, IgG4-ROD predominantly involved bilateral lacrimal glands and demonstrated fibro-inflammatory changes with abundant IgG4⁺ plasma cells. Serum IgG4 levels overlapped substantially between groups. The clinical model achieved an AUC of 0.865. Radiology- and pathology-based artificial intelligence models achieved AUCs of 0.933 and 0.946, respectively, and pathology-informed radiology training further improved discrimination (AUC 0.974).

CONCLUSIONS: Integrated clinical, imaging, and pathologic assessment provides practical distinctions between OAML and IgG4-ROD. Artificial intelligence-assisted imaging analysis may complement conventional diagnostic pathways and support differentiation in clinically ambiguous orbital lymphoproliferative disease.