[Pubmed] Disease severity as the dominant clinic-available marker of myasthenic crisis occurrence: a single-centre retro
Posté : 02 oct. 2026 12:00
Neurol Sci. 2026 Oct 2;47(10):830. doi: 10.1007/s10072-026-09448-9.
ABSTRACT
INTRODUCTION: Myasthenic crisis is the most feared acute complication of myasthenia gravis (MG), yet simple clinic-based approaches to risk stratification remain limited. We examined which routinely recorded clinical features were associated with crisis occurrence across the observed disease course and whether a parsimonious model could provide retrospective group-level stratification.
METHODS: This single-centre retrospective study included 86 adults with MG. Myasthenic crisis was defined as respiratory failure requiring mechanical ventilation. Disease severity was graded using the Myasthenia Gravis Foundation of America (MGFA) classification and grouped as mild, moderate or severe; MGFA class V was excluded to avoid circularity. Independent correlates were assessed by logistic regression and internally validated by cross-validation. The model was intended for retrospective group-level stratification, not prospective prediction of an individual future crisis.
RESULTS: Thirty-four patients (40%) had experienced a myasthenic crisis. Crisis frequency increased across severity groups: 11%, 26% and 73% in mild, moderate and severe disease, respectively (p < 0.001). Each severity grade increased the odds of crisis approximately seven-fold (OR 6.7, 95% CI 2.8-16.5), whereas more frequent outpatient visits were associated with lower odds (OR 0.73). The two-item model showed good discrimination (AUC 0.84; cross-validated AUC 0.79). Calibration-in-the-large was close to ideal, although the cross-validated calibration slope of 0.73 suggested moderate overfitting.
CONCLUSIONS: MGFA severity was the dominant clinic-available marker of crisis occurrence. Outpatient-visit frequency contributed less stable information and should be considered exploratory. These findings support retrospective group-level stratification but do not establish prospective prediction or a protective effect of follow-up.
PMID:42825837 | DOI:10.1007/s10072-026-09448-9
Source: https://pubmed.ncbi.nlm.nih.gov/4282583 ... 0&v=2.20.1