Functional Outcomes and Ocular Safety of Subretinal AAV-RPGR Gene Therapy for RPGR-Associated X-Linked Retinitis Pigmentosa: A Systematic Review and Meta-Analysis

Background/Objectives: We aimed to provide an updated synthesis of the efficacy and ocular safety of subretinal adeno-associated viral (AAV) RPGR gene therapy for retinitis pigmentosa GTPase regulator (RPGR)-associated X-linked retinitis pigmentosa (RPGR-XLRP) and to assess the certainty of the available evidence. Methods: We performed a PROSPERO-registered (CRD420251163589), PRISMA 2020 systematic review and meta-analysis using database and trial-registry searches. The primary microperimetry synthesis was restricted to mesopic Macular Integrity Assessment (MAIA) 68-loci acquisitions; continuous-outcome syntheses were conducted at the cohort level, using observed paired-eye contrasts when available; dose levels were interpreted within clinical programs; and above-maximum-tolerated-dose and peripheral-injection cohorts were reported separately. Random-effects analyses used DerSimonian–Laird models with Hartung–Knapp–Sidik–Jonkman small-sample correction. Risk of bias and certainty were assessed with design-matched tools and Grading of Recommendations Assessment, Development and Evaluation (GRADE). Results: The quantitative synthesis included four trials reported in five publications (128 participants; AAV8, AAV5, and AAV2tYF platforms). Mesopic retinal sensitivity numerically favored treatment, but the pooled differences were not statistically significant at six months (mean difference [MD] +1.11 dB; 95% confidence interval [CI], −0.88 to +3.09) or twelve months (MD +0.97 dB; 95% CI, −3.99 to +5.94). Best-corrected visual acuity showed no statistically significant pooled difference (MD +0.94 letters; 95% CI, −4.97 to +6.85). The six-month ≥10-letter low-luminance responder estimate favored treatment, but was nonsignificant and attenuated by twelve months; no primary pooled efficacy outcome reached significance under small-sample-adjusted inference. Certainty was very low for functional outcomes and low for safety outcomes. Uncontrolled treated-eye proportions were 80.8% for intraocular inflammation, 17.0% for ocular serious adverse events, and 4.3% for central-injection retinal detachment. Conclusions: Current evidence does not establish a clinically actionable functional benefit, although a modest true effect cannot be excluded. Clinically relevant ocular safety signals were observed, but the estimates were uncontrolled and program dependent. Adequately powered randomized trials with standardized functional endpoints, optimized subretinal delivery strategies, and careful within-program dose selection are needed.