Immunomodulatory therapy in non-infectious Uveitis: Current landscape, gaps, and future directions

Abstract

Non-infectious uveitis (NIU) is a potentially sight-threatening intraocular inflammatory condition that may arise idiopathically or in association with systemic immune-mediated diseases. While corticosteroids remain essential for rapid suppression of inflammation, their long-term use is limited by significant systemic and ocular side effects. Thus, immunomodulatory therapy (IMT)—including antimetabolites, calcineurin inhibitors, biologics, and emerging small molecules—has become central to achieving sustained control with a reduced corticosteroid burden in chronic cases.

Despite a range of therapeutic options, significant challenges persist. Safe, remission-inducing treatments remain elusive; tapering strategies are poorly standardized; and evidence for optimal combinations or long-term outcomes remains limited. Recent registries, such as Programme for Ocular Inflammation and Infection Translational Research (PROTON) and Treatment Exit Options for Uveitis (TOFU), are beginning to address the unmet need for structured treatment exit frameworks. Moreover, advances in imaging and artificial intelligence may soon enable real-time monitoring of disease status and risk stratification, although the development of large, well-annotated cohorts to be subject to such analysis remains a key hurdle.

This review summarizes the current role of IMT in the management of NIU, with an emphasis on corticosteroid-sparing strategies. We highlight the use of conventional immunosuppressants—including antimetabolites, calcineurin inhibitors, and alkylating agents—as well as newer biologic, smallmolecule, and interferon-based therapies. We outline where IMT fits within the broader treatment algorithm, discuss emerging evidence for earlier initiation, and explore future directions in targeted and personalized immunotherapy. We also explore future directions, including personalized approaches, biomarker-driven therapy, and the promise of AI-guided prediction models.