裂線網膜剥離修復後の早期黄斑再接着予測モデルOCT-RiSK

To develop and validate the OCT-RiSK (Optical Coherence Tomography-based Retinal Integrity and Surgical Knowledge) scoring model for predicting early macular anatomical outcomes following rhegmatogenous retinal detachment (RRD) repair.

This retrospective, single-center study included eyes with fovea-off primary RRD that underwent pars plana vitrectomy (PPV), scleral buckle (SB), or pneumatic retinopexy between January 2020 and December 2024 at a tertiary eye hospital in India. Preoperative OCT scans were analyzed for eleven predefined macular biomarkers, and 4-week postoperative OCT images were evaluated for macular reattachment. Suboptimal macular reattachment was defined as persistent macular subretinal fluid on 4-week OCT. An elastic-net logistic regression model incorporating OCT biomarkers and surgical technique was developed and validated to predict early macular attachment. Based on statistically significant predictors, a non-weighted OCT-RiSK score (+ 1 point per variable) was derived, stratifying eyes into low- and high-risk groups for suboptimal reattachment.

A total of 1,491 eyes (median age 55 years) were analyzed. Suboptimal anatomical recovery occurred in 23.9% (n = 357). Four OCT biomarkers-absence of posterior vitreous dots, intraretinal cysts, and outer retinal corrugations, and presence of photoreceptor loss-along with SB surgery were identified as independent predictors of suboptimal macular reattachment. The OCT-RiSK model achieved an overall accuracy of 84.3%, with higher OCT-RiSK scores (> 2) significantly associated with suboptimal outcomes (p < 0.001).

The OCT-RiSK scoring system provides a practical, data-driven framework for predicting early postoperative macular outcomes, supporting personalized surgical planning, patient counselling, and risk stratification in RRD management.