To determine accuracy and relative risk (RR) of posttreatment optical coherence tomography (OCT) features in identifying complete or incomplete polypoidal regression in polypoidal choroidal vasculopathy (PCV).Validity analysis.Treatment-naive PCV eyes undergoing OCT and indocyanine green angiography (ICGA) at baseline and posttreatment were included. Two graders confirmed diagnosis and identified posttreatment complete or incomplete regression on ICGA. Two other graders classified OCT characteristics of pigment epithelial detachment (PED) (polypoidal lesion) based on 5 prespecified features: "A," no PED; "B," PED with internal homogeneous reflectivity with predominant "BUN" (blended retinal pigment epithelium with underlying structure) sign; "C," PED with internal homogeneous reflectivity with minimal "BUN"; "D," heterogeneous PED; and "E," PED with hyporeflectivity.Among 130 polypoidal lesions (65 pretreatment and 65 posttreatment) of 39 PCV eyes (39 patients; 54% female; mean age±SD: 64.6±8.2), all pretreatment lesions showed feature D on OCT. Posttreatment lesions with complete regression (31 lesions) showed OCT features A, B, C, D, and E in 32%, 45%, 13%, 10%, and 0%, respectively. Posttreatment lesions with incomplete regression (34 lesions) showed OCT features A, B, C, D, and E in 0%, 6%, 15%, 79%, and 0%, respectively. Presence of either feature A or B had highest accuracy (86%; 95% confidence interval: 75%-93%); 77% sensitivity; 94% specificity; RR 5.0 (3.5-7.1, P0.001) for complete regression. Presence of feature D had highest accuracy (85%; 95% confidence interval: 74%-92%); 79% sensitivity; 90% specificity; RR 4.6 (3.0-6.9, P0.001) for incomplete regression.Without ICGA, OCT features could provide high accuracy in identifying posttreatment complete or incomplete polypoidal regression in PCV.
