Use of OCT Retinal Thickness Deviation Map for Hydroxychloroquine Retinopathy Screening

Abstract

Purpose: To investigate the use of a retinal thickness deviation map generated from swept-source (SS) OCT images for hydroxychloroquine retinopathy screening. Design: Retrospective cohort study. Participants: This study included 1192 Korean patients with a history of hydroxychloroquine treatment: 881 patients (1723 eyes) in the discovery set and 311 patients (591 eyes) in the validation set. Patients were screened for retinal toxicity using SS OCT, fundus autofluorescence, and standard automated perimetry. Methods: According to the 2016 American Academy of Ophthalmology guidelines, hydroxychloroquine retinopathy was diagnosed by the presence of abnormalities on 1 objective structural tests alongside corre- sponding visual field defects. The 12 9-mm2 macular volume SS OCT scan was performed, and the retinal thickness deviation map was generated automatically using the built-in software. On this map, yellow (retinal thickness, <5% of the normative level) or red (<1% of the normative level) pixels were defined as abnormal. Abnormal findings were evaluated, and diagnostic criteria were developed based on the discovery set data; criteria were validated using the validation set data. Main Outcome Measures: The rate and patterns of abnormalities on the retinal thickness deviation map and sensitivity and specificity of the diagnostic criteria. Results: The retinal thickness deviation map showed the following abnormal patterns in eyes with hydrox- ychloroquine retinopathy: pericentral (36.0%) or parafoveal (6.1%) ring, mixed-ring (34.2%), central island (13.2%), and whole macular thinning (10.5%). The criterion of 5 contiguous red pixels showing 1 of the 5 characteristic patterns in both eyes yielded the greatest diagnostic performance (sensitivity and specificity of 98.2% and 89.1% and of 100% and 87.5% in the discovery and validation set data, respectively). Moreover, the area of abnormal pixels on the map was correlated significantly with the mean deviation (P < 0.001) and pattern standard deviation (P < 0.001) on the Humphrey 30-2 test in eyes with hydroxychloroquine retinopathy. Conclusions: The retinal thickness deviation map may facilitate the objective evaluation of hydroxychloroquine retinopathy because it does not require subjective, morphologic evaluation of the outer retinal layers. The map has the potential to enhance hydroxychloroquine retinopathy screening when used in conjunction with conventional screening methods.