ILM flap coverage versus conventional ILM peeling for idiopathic full-thickness macular hole: a retrospective cohort study

Abstract

Background Idiopathic full-thickness macular hole (iFTMH) is treated by pars plana vitrectomy with internal limiting membrane (ILM) peeling. Conventional peeling achieves high closure in small holes but may yield inferior closure morphology in larger lesions. The temporal inverted ILM flap technique may improve closure quality, yet observational data stratified by Type 1 versus Type 2 closure require careful handling of baseline imbalance. Methods We conducted a retrospective cohort study of 202 consecutive iFTMH eyes from 202 unique patients operated by a single team between January 2022 and December 2024. The strict primary-endpoint cohort comprised 198 eyes with 12-month OCT within a prespecified ± 4-week window (flap, n = 100; conventional peeling, n = 98). The primary endpoint was Type 1 closure at 12 months and the primary estimand was the average treatment effect in the overlap population (ATO). Propensity scores used pre-treatment covariates including hole dimensions, symptom duration, baseline BCVA, and calendar year. Common outcomes used g-computation with 1,000 bootstrap replicates (adjusted probabilities, risk differences, risk ratios); sparse-event outcomes used unweighted covariate-adjusted Firth-penalized regression as exploratory sensitivity analyses. Results The flap group had larger baseline holes, worse baseline BCVA, and longer symptom duration (all P < 0.001). After overlap weighting, all standardized mean differences were <0.001, reflecting exact mean balance of included covariates (weighted effective sample size 111.5 of 202 eyes). Type 1 closure occurred in 82/100 (82.0%) flap versus 44/98 (44.9%) peeling eyes. In the ATO population the adjusted absolute risk difference was +33.5 percentage points (95% bootstrap CI +14.5 to +51.1), the adjusted risk ratio was 1.68 (95% CI 1.24–2.39), and the adjusted odds ratio was 4.84 (95% CI 1.95–12.00; P = 0.001). Exploratory secondary analyses suggested greater BCVA improvement, shorter EZ and ELM defect lengths, and fewer secondary surgeries in the flap group. Conclusion In this retrospective single-center cohort with predominantly large iFTMH, temporal inverted ILM flap coverage was associated with a higher adjusted probability of Type 1 closure at 12 months than conventional ILM peeling within an overlap-population estimand. Exploratory secondary functional and microstructural findings should be interpreted as hypothesis-generating associations rather than confirmatory evidence.