Using 24-h intraocular pressure-related patterns to identify open-angle glaucoma in thyroid eye disease.

Abstract

Our study aims to develop a diagnostic model using 24-h intraocular pressure (IOP) patterns to differentiate between open-angle glaucoma (OAG) and dysthyroid optic neuropathy (DON) in thyroid eye disease (TED) patients with glaucoma-like symptoms. TED patients with elevated IOP, abnormal optic disc, and/or visual fields were prospectively recruited. The subjects whose symptoms were relieved by DON first-line treatments were divided into the DON group, and the subjects with previous diagnosis of OAG before TED onset were divided into the OAG group. The 24-h IOP was monitored by Tono-Pen in a sitting position during awake time and in a supine position during sleep time. All subjects were divided into a training set and a testing set. The diagnostic models were generated from training set by using either IOP curve-derived parameters or principal component (PC) factors. The discrimination ability was tested in training set based on area under curve (AUC), and the calibration ability was verified in testing set by Hosmer-Lemeshow goodness-of-fit. The sensitivity and specificity were calculated by two-by-two table with the cutoff value determined by Youden's index. Thirty-two cases were recruited in each group. The 24-h IOP curves revealed a nocturnal pattern in both groups, with the acrophase moving slightly forward in the DON group (21:00pm-24:00pm) compared to the OAG group (22:00pm-3:00 am). Several IOP curve-derived parameters differed between the two groups, with larger amplitude during sleep time (P < 0.000) and longer duration of IOP ≥ 21mmHg at awake time (P = 0.004) in the DON group than the OAG group. However, the diagnostic model generated from IOP parameters showed poor reliability (P = 0.001) in calibration test and was rejected. The other model built on PC factors achieved good performance of discrimination (AUC = 0.943) and calibration (P = 0.139) with a sensitivity of 87.50% and a specificity of 95.83% at cutoff value of 0.538 to identify OAG cases. The diagnostic model facilitates discrimination between OAG and DON in TED patients based on 24-h IOP-related patterns. This work was registered on Chinese Clinical Trial Registry (ChiCTR1900025394).