Comparison of the performance of clinical classification systems for myasthenia gravis relapse after thymoma...

To characterize clinically significant myasthenia gravis (MG) relapse/exacerbation after thymoma resection and compare classification systems for risk stratification.

This multicenter retrospective cohort included 257 patients with thymoma-associated MG who underwent extended thymectomy at 3 centers from July 2009 through November 2025. The end point was new MG symptoms after postoperative complete stable remission, pharmacologic remission, or minimal manifestation status plus verifiable MG-related treatment escalation or rescue therapy. Death without MG relapse/exacerbation was treated as a competing event. Cumulative incidence functions were estimated with Aalen-Johansen and compared with Gray tests; predictors were assessed with Fine-Gray models.

During median follow-up of 41.6 months, 69 relapse/exacerbation events and 21 competing deaths occurred. Cumulative incidence was 20.4% at 3 years and 29.0% at 5 years; 68% and 90% of events occurred within 3 and 5 years, respectively. Among 52 relapse/exacerbation events in patients who underwent surgery in 2018 or later with retrievable treatment-status data, 28 (53.8%) occurred during ongoing immunosuppression and 24 (46.2%) after withdrawal. The a/b phenotype-based grouping of the Myasthenia Gravis Foundation of America classification (1/A/B) showed the strongest separation (3-year cumulative incidences, 2.5%, 12.3%, and 56.4%; 5-year cumulative incidences, 7.6%, 29.6%, and 56.4% for groups 1, A, and B). Groups A and B were independently associated with relapse/exacerbation (subdistribution hazard ratio, 2.65; 95% CI, 1.18-5.91; and 10.55; 95% CI, 4.55-24.47).

Clinically significant MG relapse/exacerbation after thymoma resection clustered in the early postoperative years. The a/b phenotype-based grouping of the Myasthenia Gravis Foundation of America classification (1/A/B) provided the most informative risk stratification.