Abstract
Purpose
Janus kinase (JAK) inhibitors have entered clinical testing for non-infectious ocular inflammation, but trial evidence is unsynthesized. We evaluated efficacy and safety within a pre-specified tiered framework.Methods
We searched five databases, three registries, and grey literature to July 4, 2026. Eligible studies were pre-assigned to Stratum A (systemic agents for intraocular or scleral inflammation; primary) or Stratum B (topical, ocular surface disease); bias was assessed with RoB 2 and ROBINS-I. Single-arm proportions were pooled with a beta-binomial model; with I2 above the pre-specified 75%, Synthesis Without Meta-analysis (SWiM) was primary.Results
Seven studies (four randomized by design; 465 participants) were included: 138 in Stratum A (systemic), 327 in the single Stratum B trial. In Stratum A, disease control ranged from 33.3% to 100%; five of six showed a favorable direction. The pooled estimate, 76.2% (95% CI 55.8-96.5), fell to 57.6% at 24 weeks or excluding two 100%-response studies; a prediction interval (1.0-87.1; I2 = 91.5%) confirmed no single value is reliable. In HUMBOLDT, the only placebo-controlled trial of a systemic agent, filgotinib reduced treatment failure (RR 0.55, 95% CI 0.34-0.92; NNT 3.3; p = 0.006). Serious adverse events occurred in 5/140 systemically exposed patients (3.6%, 95% CI 1.2-8.1), with no thromboembolic, major cardiovascular, or malignant events.Conclusion
Systemic JAK inhibitors show a consistent favorable direction in refractory intraocular inflammation, most reliably filgotinib, but heterogeneity and very low certainty preclude pooling. They may serve as third- or fourth-line agents pending Phase 3 data.