Predicting worsening risk in MGFA class I, II and III myasthenia gravis patients: development and validation of a predictive nomogram.

Background: Myasthenia gravis (MG), a neuromuscular junction autoimmune disorder, causes skeletal muscle weakness. MG worsening frequently occurs during the disease course, severely impairing quality of life and elevating myasthenic crisis risk. Existing predictive models remain scarce. This study developed a predictive model for MG worsening to facilitate early risk stratification and personalized care.

Research design & methods: Retrospective analysis included 437 the Myasthenia Gravis Foundation of America (MGFA) class I - III myasthenia gravis patients from December 2019 to September 2024. Sociodemographic, clinical variables and worsening status were analyzed. Predictors were identified via univariate analysis, the Least Absolute Shrinkage and Selection Operator (LASSO) regression, and multivariate logistic regression. Model performance was assessed using receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis.

Results: Patients were randomized into training (n = 305) and validation (n = 132) cohorts. Worsening rates were comparable (26.52% vs. 31.15%, p = 0.331). Six predictors emerged: age, MGFA classification, thymectomy history, chills, fatigue, and emotional disturbances (ED). The nomogram demonstrated strong discrimination (AUC: 0.82 training, 0.83 validation) and calibration (Hosmer-Lemeshow p > 0.05). Decision curve analysis confirmed clinical utility at 10-70% probability thresholds.

Conclusion: This nomogram integrates accessible clinical variables to stratify MG worsening risk, enabling early intervention. Validation through multicenter prospective studies is warranted to optimize generalizability.