Antoine Guéguen1, Benoit Hijazi2, Kevin Zuber3, Romain Deschamps4, Cedric Lamirel5, Catherine Vignal-Clermont5, Olivier Gout4. 1. Department of Neurology, Foundation Hospital A. de Rothschild, 29, Rue Manin, 75019, Paris, France. agueguen@for.paris. 2. Department of Ophthalmology, Regional University Hospital Center Morvan, Brest, France. 3. Department of Clinical Research and Statistics, Foundation Hospital A. de Rothschild, Paris, France. 4. Department of Neurology, Foundation Hospital A. de Rothschild, 29, Rue Manin, 75019, Paris, France. 5. Department of Neuro-Ophthalmology, Foundation Hospital A. de Rothschild, Paris, France.
Abstract
BACKGROUND: Generalization of ocular myasthenia gravis (OMG) represents a pejorative evolution, and no validated generalization-prevention strategy exists. The study aimed to determine the percentage of patients with OMG generalization and identify factors predictive of it to establish a prediction score. METHODS: This retrospective, observational study included 151 patients diagnosed with OMG after an initial work-up in our institution. The outcome measure was time to MG generalization. The explanatory variables were age at onset (> 55 years), sex, first-year anti-acetylcholine-receptor antibody-positivity, repetitive nerve stimulation showing electromyogram decrement and corticosteroid use. Kaplan-Meier estimations of the probability of risk of generalization, and descriptive and multivariate Cox model analyses were computed. A nomogram combining explanatory variables was used to establish a score to predict the probability of OMG generalization. RESULTS: Among 183 patients' charts identified, 151 had confirmed OMG. Their median follow-up was 5.7 years. Estimations (95% CI) of OMG-generalization risk at 1, 3 and 10 years post-symptom onset, respectively, were: 13.0% (7.3-18.2), 25.1% (17.5-32.0) and 37.8% (27.2-45.2). The p-value-based multivariate analysis associated generalization with female sex, electromyogram decrement and first-year anti-acetylcholine-receptor antibody positivity, and Akaike information criterion-based analysis retained those three parameters and corticosteroid use. A nomogram was built and validated with an optimism-corrected C-statistic of 0.68, and calibration plots showed good fit. CONCLUSIONS: Our population's percentage of OMG generalization is in line with recent publications. Using the identified prognostic factors, the nomogram provided a score to predict the probable risk of generalization in our cohort.
BACKGROUND: Generalization of ocular myasthenia gravis (OMG) represents a pejorative evolution, and no validated generalization-prevention strategy exists. The study aimed to determine the percentage of patients with OMG generalization and identify factors predictive of it to establish a prediction score. METHODS: This retrospective, observational study included 151 patients diagnosed with OMG after an initial work-up in our institution. The outcome measure was time to MG generalization. The explanatory variables were age at onset (> 55 years), sex, first-year anti-acetylcholine-receptor antibody-positivity, repetitive nerve stimulation showing electromyogram decrement and corticosteroid use. Kaplan-Meier estimations of the probability of risk of generalization, and descriptive and multivariate Cox model analyses were computed. A nomogram combining explanatory variables was used to establish a score to predict the probability of OMG generalization. RESULTS: Among 183 patients' charts identified, 151 had confirmed OMG. Their median follow-up was 5.7 years. Estimations (95% CI) of OMG-generalization risk at 1, 3 and 10 years post-symptom onset, respectively, were: 13.0% (7.3-18.2), 25.1% (17.5-32.0) and 37.8% (27.2-45.2). The p-value-based multivariate analysis associated generalization with female sex, electromyogram decrement and first-year anti-acetylcholine-receptor antibody positivity, and Akaike information criterion-based analysis retained those three parameters and corticosteroid use. A nomogram was built and validated with an optimism-corrected C-statistic of 0.68, and calibration plots showed good fit. CONCLUSIONS: Our population's percentage of OMG generalization is in line with recent publications. Using the identified prognostic factors, the nomogram provided a score to predict the probable risk of generalization in our cohort.