A clinical predictive score for postoperative myasthenic crisis

Objective: Myasthenia gravis (MG) is an autoimmune disease mostly caused by autoantibodies against acetylcholine receptor associated with thymus abnormalities. Thymectomy has been proven to be an efficacious treatment for patients with MG, but postoperative myasthenic crisis often occurs and is a major complication. We aimed to develop and validate a simple scoring system based on clinical characteristics in the preoperative status to predict the risk of postoperative myasthenic crisis. Methods: We studied 393 patients with MG who underwent thymectomy at 6 tertiary centers in Japan (275 patients for derivation and 118 for validation). Clinical characteristics, such as gender, age at onset and operation, body mass index, disease duration, MG subtype, severity, symptoms, preoperative therapy, operative data, and laboratory data, were reviewed retrospectively. A multivariate logistic regression with LASSO penalties was used to determine the factors associated with postoperative myasthenic crisis, and a score was assigned. Finally, the predictive score was evaluated using bootstrapping technique in the derivation and validation groups. Results: Multivariate logistic regression identified 3 clinical factors for predicting postoperative myasthenic crisis risk: (1) vital capacity < 80%, (2) disease duration < 3 months, and (3) bulbar symptoms immediately before thymectomy. The postoperative myasthenic crisis predictive score, ranging from 0 to 6 points, had areas under the curve of 0.84 (0.66–0.96) in the derivation group and 0.80 (0.62–0.95) in the validation group. Interpretation: A simple scoring system based on 3 preoperative clinical characteristics can predict the possibility of postoperative myasthenic crisis. Ann Neurol 2017;82:841–849.