Clinical and soft-tissue computed tomographic predictors of dysthyroid optic neuropathy: refinement of the constellation of findings at presentation

• Published in: Archives of ophthalmology (1960) Vol. 129; no. 10; p. 1332
• Main Authors: Weis, Ezekiel, Heran, Manraj K S, Jhamb, Ashu, Chan, Andy K, Chiu, Jack P, Hurley, Michael C, Rootman, Jack
• Format: Journal Article
• Language: English
• Published: United States 01.10.2011
• Subjects: Area Under Curve; False Positive Reactions; Female; Graves Ophthalmopathy - diagnostic imaging; Humans; Male; Middle Aged; Oculomotor Muscles - diagnostic imaging; Optic Nerve Diseases - diagnostic imaging; Predictive Value of Tests; Reproducibility of Results; Retrospective Studies; Risk Factors; ROC Curve; Sensitivity and Specificity; Tomography, X-Ray Computed; Visual Acuity - physiology
• ISSN: 1538-3601
• DOI: 10.1001/archophthalmol.2011.276

To evaluate the ability to predict the presence of dysthyroid optic neuropathy (DON) using computed tomography assessment of soft-tissue and clinical features. A retrospective consecutive case series of patients with thyroid-related orbitopathy. One hundred eighty-nine orbits from 99 patients were evaluated. Statistically significant clinical predictors of DON on univariate analysis included a difference in intraocular pressure from primary gaze to upgaze (P = .02), the presence of lagophthalmos (P = .04), and inflammation as measured by the VISA (vision, inflammation, strabismus, appearance/exposure) inflammatory scale (P = .004). Dysthyroid optic neuropathy was inversely related to the marginal reflex distance (P = .01), levator function (P = .02), total ductions (P = .003), and interpalpebral fissure (P = .04). Statistically significant radiologic predictors determined on univariate analysis included apical crowding (P < .001), presence of enlarged tendons (P = .004), increasing total rectus diameter (P = .02), and presence of small, low densities within the recti muscles (P = .04). Multivariate analysis found only total ductions (P = .02) and marginal reflex distance (P = .04) determined on clinical examination and apical crowding shown on computed tomography (P = .003) to be significantly associated with DON. Receiver operating characteristic curves were used to evaluate the ability of the clinical and radiologic assessment, as well as the combination of these assessments, to predict DON. All 3 models were strong predictors of DON, with no statistically significant differences in the area under the receiver operating characteristic curve among them (P = .14). Total ductions, marginal reflex distance, and apical crowding observed on computed tomography scans are able to predict the presence of DON with high sensitivity, specificity, positive predictive value, and negative predictive value. Eyelid ptosis is a novel predictor of DON.