A Prediction Model for Detecting Dysthyroid Optic Neuropathy Based on Clinical Factors and Imaging Markers of the Optic Nerve and Cerebrospinal Fluid in the Optic Nerve Sheath.

OBJECTIVE

This study aimed to develop and test a model for predicting dysthyroid optic neuropathy (DON) based on clinical factors and imaging markers of the optic nerve and cerebrospinal fluid (CSF) in the optic nerve sheath.

METHODS

This retrospective study included patients with thyroid-associated ophthalmopathy (TAO) without DON and patients with TAO accompanied by DON at our hospital. The imaging markers of the optic nerve and CSF in the optic nerve sheath were measured on the water-fat images of each patient and, together with clinical factors, were screened by Least absolute shrinkage and selection operator. Subsequently, we constructed a prediction model using multivariate logistic regression. The accuracy of the model was verified using receiver operating characteristic curve analysis.

RESULTS

In total, 80 orbits from 44 DON patients and 90 orbits from 45 TAO patients were included in our study. Two variables (optic nerve subarachnoid space and the volume of the CSF in the optic nerve sheath) were found to be independent predictive factors and were included in the prediction model. In the development cohort, the mean area under the curve (AUC) was 0.994, with a sensitivity of 0.944, specificity of 0.967, and accuracy of 0.901. Moreover, in the validation cohort, the AUC was 0.960, the sensitivity was 0.889, the specificity was 0.893, and the accuracy was 0.890.

CONCLUSIONS

A combined model was developed using imaging data of the optic nerve and CSF in the optic nerve sheath, serving as a noninvasive potential tool to predict DON.