Early diagnostic prediction model for inflammatory and ischemic ocular motor cranial nerve palsy.

OBJECTIVE: To develop an early diagnostic prediction model to differentiate cavernous sinus inflammation (inflammation group) from microvascular ischemic ocular motor cranial nerve (CN) palsy (ischemic group) at early presentation. METHODS: A total of 66 and 117 patients within 2 weeks of symptom onset were enrolled in the inflammation and ischemic groups. Twenty-two potential predictors were evaluated; predictors that remained significant after Benjamini-Hochberg adjustment (false discovery rate 0.05) were entered into a multivariable logistic regression model. Discrimination was assessed by the area under the receiver operating characteristic curve (AUC), and calibration by the Hosmer -Lemeshow goodness-of-fit test. RESULTS: Significant predictors included vascular risk factor scores (VRFs), ocular motor nerve palsy scale (OMNPS) score, aggregate index of systemic inflammation (AISI), history of diabetes, and cavernous sinus thickness. The AUC was 0.899 (95% confidence interval, 0.838 to 0.939). At the optimal probability cutoff(0.674), sensitivity and specificity were 72.3% and 89.7%, respectively. The Hosmer-Lemeshow test yielded χ2 = 4.262, P = 0.833. CONCLUSIONS: The logistic regression–based model showed good discrimination and calibration for differentiating cavernous sinus inflammation from microvascular ischemic ocular motor CN palsy during early presentation, and may assist early clinical decision-making.