Visual Outcomes of Autologous RPE-Based Transplantation for Massive Submacular Hemorrhage in Neovascular AMD and PCV: A Systematic Review and Meta-Analysis

Massive submacular hemorrhage (SMH) secondary to age-related macular degeneration (AMD) or polypoidal choroidal vasculopathy (PCV) has a poor visual prognosis, and the role of autologous retinal pigment epithelium (RPE)-based transplantation remains uncertain. To synthesize visual, anatomical, and safety outcomes after autologous RPE monolayer or RPE complex transplantation for AMD/PCV-related massive SMH, with BCVA at 12 ± 6 months as the primary continuous outcome. The protocol was registered in PROSPERO (CRD420251164359). We searched seven English- and Chinese-language databases from inception through June 1, 2026. The primary outcome was within-eye BCVA change, defined as postoperative minus preoperative logMAR; negative values indicated improvement. For each eye, the primary analysis used the measurement closest to 12 months within an inclusive 6–18-month window. Continuous outcomes were synthesized using random-effects models fitted by restricted maximum likelihood, and single-arm binary outcomes using logistic-normal generalized linear mixed models. Certainty was assessed using GRADE. Thirteen studies involving 140 eyes were included. Eight cohorts comprising 54 eyes contributed to the primary 12 ± 6-month analysis. The pooled mean BCVA change was − 0.276 logMAR (95% CI, − 0.534 to − 0.019; P = 0.0355; I2 = 63.2%), with a 95% prediction interval of − 0.894 to 0.341 logMAR. The 12 ± 3-month analysis and the joint exclusion of Karasu et al. 2019, MacLaren et al. 2007, and the Chen et al. 2022 RPE-monolayer cohort were compatible with no mean change. All-available best and last available BCVA analyses remained favorable but showed substantial heterogeneity. Graft perfusion was confirmed in 96.7% of eyes (95% CI, 74.4–99.7%); 34.6% achieved BCVA better than 20/200, and 2.9% achieved 20/40 or better. Retinal detachment occurred in 16.5%, postoperative subretinal hemorrhage in 18.2%, epiretinal membrane in 16.7%, postoperative neovascularization in 11.8%, and reoperation in 28.2%. Certainty was very low. The fixed-effect sensitivity analysis of the same primary dataset yielded a pooled mean change of − 0.254 logMAR (95% CI, − 0.403 to − 0.105; P = 0.0008), supporting the direction of the random-effects estimate but with a narrower confidence interval. Autologous RPE-based transplantation may offer salvage value for highly selected eyes with AMD/PCV-related SMH. However, the primary prediction interval crossed no mean change, sensitivity results depended on follow-up definition and cohort inclusion, and complications were frequent. Heterogeneous, non-randomized evidence of very low certainty cannot establish comparative effectiveness. PROSPERO: CRD420251164359. A large bleed under the center of the retina can cause severe and permanent sight loss in age-related macular degeneration and polypoidal choroidal vasculopathy. Some surgeons remove the blood and damaged tissue and transplant the patient’s own retinal support tissue. We reviewed the published evidence for this complex salvage operation. We included 13 studies involving 140 eyes. Vision improved on average, but results varied widely among studies. Blood flow returned to most transplanted tissue, yet only about one-third of eyes achieved vision better than 20/200. Retinal detachment, repeat bleeding, membrane formation, new abnormal blood vessels, and further surgery were not uncommon. The operation may help selected eyes with otherwise poor prospects, but the available studies were small, non-randomized, and very uncertain. They do not show that transplantation works better than other treatments. Larger prospective studies with consistent outcome reporting are needed.