Junki Kwon1, Kyung Rim Sung2, Seungbong Han3, Ye Ji Moon1, Joong Won Shin1. 1. Department of Ophthalmology, College of Medicine, University of Ulsan, Asan Medical Center, Seoul, South Korea. 2. Department of Ophthalmology, College of Medicine, University of Ulsan, Asan Medical Center, Seoul, South Korea. Electronic address: sungeye@gmail.com. 3. Department of Applied Statistics, Gachon University, Seongnam-Si, Gyeonggi-do, South Korea.
Abstract
PURPOSE: To classify eyes with primary angle closure (PAC) in terms of the features visualized using anterior segment optical coherence tomography (AS-OCT) and ultrasound biomicroscopy (UBM). DESIGN: Retrospective, observational study. PARTICIPANTS: A total of 73 eyes of 73 patients with PAC. METHODS: Participants' eyes that had undergone laser peripheral iridotomy (LPI) were imaged using AS-OCT and UBM under the same lighting conditions. Anterior chamber depth, anterior chamber width, iris cross-sectional area, peripheral iris thickness, iris curvature, lens vault (LV), and angle opening distance 500 μm from the scleral spur (SS) were determined using the AS-OCT image; trabecular-ciliary process angle (TCA), trabecular-ciliary process distance (TCPD), and ciliary body (CB) thickness 1 mm posterior to the SS were estimated on the UBM image using ImageJ software (Wayne Rasband, National Institutes of Health, Rockville, MD). Iris insertion, iris angulation, iris convexity, presence of ciliary sulcus, irido-angle contact, and CB orientation assessed on the UBM image were included. Partitioning around the medoids algorithm was used for cluster analysis based on the parameters obtained using AS-OCT and UBM. Axial length and pupil diameter were incorporated into statistical models. MAIN OUTCOME MEASURES: Clinical and anatomic characteristics were compared between the clusters, as classified using the partitioning around medoids algorithm method. RESULTS: Cluster analysis revealed that 2-group clustering produced the best results. The 2 clusters, which were defined in terms of parameters obtained using AS-OCT and UBM, showed differences in iris curvature (0.16±0.08 vs. 0.11±0.04 mm), TCA (91.0°±13.4° vs. 63.7°±6.2°), TCPD (0.99±0.22 vs. 0.78±0.16 mm), CB orientation (neutral/anterior, 35/13 vs. 0/25), and iris insertion (basal/middle/apical, 37/9/2 vs. 12/11/2). Pre-LPI intraocular pressure (IOP) (18.8±5.4 vs. 16.2±4.5 mmHg; P = 0.037) and percentage of IOP reduction after LPI (22.3%±17.9% vs. 8.3%±19.5%; P < 0.003) showed a significant difference between the 2 clusters. CONCLUSIONS: The most distinct difference between the 2 subgroups in the cluster analysis was TCA, suggesting that the position of the CB is important in subclassifying PAC. By using UBM, clinicians may obtain more clues about the mechanisms of PAC; in turn, they may learn to predict the IOP-lowering effects of LPI.
PURPOSE: To classify eyes with primary angle closure (PAC) in terms of the features visualized using anterior segment optical coherence tomography (AS-OCT) and ultrasound biomicroscopy (UBM). DESIGN: Retrospective, observational study. PARTICIPANTS: A total of 73 eyes of 73 patients with PAC. METHODS:Participants' eyes that had undergone laser peripheral iridotomy (LPI) were imaged using AS-OCT and UBM under the same lighting conditions. Anterior chamber depth, anterior chamber width, iris cross-sectional area, peripheral iris thickness, iris curvature, lens vault (LV), and angle opening distance 500 μm from the scleral spur (SS) were determined using the AS-OCT image; trabecular-ciliary process angle (TCA), trabecular-ciliary process distance (TCPD), and ciliary body (CB) thickness 1 mm posterior to the SS were estimated on the UBM image using ImageJ software (Wayne Rasband, National Institutes of Health, Rockville, MD). Iris insertion, iris angulation, iris convexity, presence of ciliary sulcus, irido-angle contact, and CB orientation assessed on the UBM image were included. Partitioning around the medoids algorithm was used for cluster analysis based on the parameters obtained using AS-OCT and UBM. Axial length and pupil diameter were incorporated into statistical models. MAIN OUTCOME MEASURES: Clinical and anatomic characteristics were compared between the clusters, as classified using the partitioning around medoids algorithm method. RESULTS: Cluster analysis revealed that 2-group clustering produced the best results. The 2 clusters, which were defined in terms of parameters obtained using AS-OCT and UBM, showed differences in iris curvature (0.16±0.08 vs. 0.11±0.04 mm), TCA (91.0°±13.4° vs. 63.7°±6.2°), TCPD (0.99±0.22 vs. 0.78±0.16 mm), CB orientation (neutral/anterior, 35/13 vs. 0/25), and iris insertion (basal/middle/apical, 37/9/2 vs. 12/11/2). Pre-LPI intraocular pressure (IOP) (18.8±5.4 vs. 16.2±4.5 mmHg; P = 0.037) and percentage of IOP reduction after LPI (22.3%±17.9% vs. 8.3%±19.5%; P < 0.003) showed a significant difference between the 2 clusters. CONCLUSIONS: The most distinct difference between the 2 subgroups in the cluster analysis was TCA, suggesting that the position of the CB is important in subclassifying PAC. By using UBM, clinicians may obtain more clues about the mechanisms of PAC; in turn, they may learn to predict the IOP-lowering effects of LPI.