Article Text

Download PDFPDF
  1. Sui Wong1–3,
  2. Aviva Petrie4,
  3. Gordon Plant1
  1. 1Moorfields Eye Hospital
  2. 2St Thomas' Hospital
  3. 3National Hospital for Neurology and Neurosurgery
  4. 4UCL Eastman Dental Institute


Background There is no prognostic test to predict generalized myasthenia gravis (GMG) risk in patients with ocular disease. We aimed to create a predictive risk score.

Method Multi-centre retrospective cohort study: OMG for minimum 3 months, untreated with immunosuppression for minimum 2 years or until GMG onset.

Results 101 (57F) patients included. Median follow-up 8.4 y(2–42) from onset. 31 developed GMG at median 1.31 y (3.5 mo–20.2 y); 19 within 2 y.

3 significant predictors from univariable logistic regression analysis (p<0.10); adjusted ORs in multivariable model (Chi-square p=0.001): seropositivity 5.63 (1.42,22.33); presence of ≥1 comorbidity including autoimmune disorders 5.56 (0.66,46.62); thymic hyperplasia 6.66 (0.45,98.01).

Prognostic score derived from the coefficients of the logistic model: sum of the points(one point for presence of each of the above predictive factors); classified ‘low risk’ if ≤1, ‘high risk’ if ≥2. Predicted probabilities: 0.07 (SD 0.03) for ‘low risk’, 0.39(SD 0.09) for ‘high risk’. Negative predictive value 91%(79,98), positive predictive value 38% (23,54), area under ROC curve 0.74 (0.64,0.85).

Conclusions We present one of the few natural history studies on OMG and create the first predictive score to prognosticate the risk of GMG. This approach of risk stratification moves us towards addressing the question of modifying GMG risk in high-risk patients.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.