Model may predict treatment outcomes in ocular myasthenia gravis

A simple model that uses five clinical factors could help doctors predict which people with ocular myasthenia gravis (OMG) are most likely to achieve minimal symptoms or better within one year, according to a study in China.

“Based on prognostic predictions, early efforts can be made to develop therapeutic interventions that are more specific to individual patient disease progression and clinical needs,” the researchers wrote.

The study, “Identification of Ocular and Systemic Predictors and Development of a Prediction Model in Ocular Myasthenia Gravis,” was published in Translational Vision Science & Technology.

Recommended Reading

June 2, 2026 News by Marisa Wexler, MS

Myasthenia Gravis Awareness Month features events spanning the globe

Researchers focused on 1-year outcomes

OMG is a type of myasthenia gravis that affects only the muscles controlling the eyes and eyelids, causing drooping eyelids and double vision. Predicting which patients will achieve minimal manifestation status (MMS) — a state in which symptoms are absent or minimal after treatment — or better is important for guiding treatment decisions.

However, “most studies on the prognosis of patients with OMG have focused on [progressing] to generalized myasthenia gravis (GMG),” the researchers wrote. GMG is a form of the disease that affects several muscle groups in addition to the eyes.

Few studies have evaluated “treatment-induced remission status in OMG, including complete stable remission (CSR) or MMS,” they added.

With this in mind, a team of researchers in China aimed to develop a tool that combines clinical factors to predict which OMG patients are most likely to reach MMS or better within one year. To identify the most important predictors, they retrospectively reviewed medical records from 216 people with OMG who were seen at their hospital between 2020 and 2023, ultimately including 126 in the study.

All were followed for at least a year, and patients who had GMG at diagnosis or progressed to GMG within the first three months after diagnosis were excluded. A total of 61 patients (48.4%) achieved MMS or better after one year of follow-up.

Significant differences between participants achieving MMS or better (MMS group) and those who didn’t (non-MMS group) were detected in several factors. The researchers then selected five key factors for the prediction model.

Five factors shaped the prediction model

Specifically, the MMS group had significantly shorter disease duration compared with the non-MMS group (median of about 2.5 months vs. one year), and showed a smaller ocular deviation angle, a measure of how much the eyes are misaligned.

Patients achieving MMS or better were also significantly less likely to test positive for self-reactive antibodies against the AChR protein — the most common target of MG-driving antibodies (11.5% vs. 41.5%) — and to have abnormal findings on a thyroid ultrasound (52.5% vs. 72.3%). Problems with the thyroid gland often occur alongside MG.

Corticosteroid therapy was more complicated. In the raw group comparison, fewer patients in the MMS group were receiving corticosteroids than in the non-MMS group (21.3% vs. 44.6%). However, after the researchers analyzed the factors together, corticosteroid therapy was still selected as one of the five variables in the prediction model.

These five clinical factors were combined into a prediction model and presented as a nomogram, a graphical tool that allows doctors to estimate the probability of a specific outcome for an individual patient. The researchers also reported designing an online version where doctors can enter a patient’s clinical data and receive an immediate prediction.

The model’s ability to distinguish between OMG patients who would and would not achieve MMS was assessed using the area under the receiver operating characteristic curve (AUC). The AUC typically ranges from 0.5, indicating no predictive ability, to one, indicating perfect prediction.

The new model achieved an AUC of 0.889 in a training cohort and 0.813 in an internal validation cohort, showing strong internal predictive performance.

The researchers reported that the model’s sensitivity was 82.3% and its specificity was 91.4%, meaning it showed good ability to distinguish patients who did and did not achieve MMS.

Model showed strong internal performance

Additional testing showed that the model’s predictions closely matched actual patient outcomes. Decision curve analysis, which is used to check for clinical usefulness, suggested that the model could provide meaningful benefits in clinical decision-making.

“Using this model … is convenient and can be widely extended to neurological, as well as ophthalmic [eye care], healthcare settings,” the researchers wrote. “Based on prognostic predictions, early efforts can be made to develop therapeutic interventions that are more specific to individual patient disease progression and clinical needs.”

Among the study’s limitations, the researchers noted the small sample size, the short follow-up period, lack of external validation, and the fact that they only assessed anti-AChR antibody status, which “may overlook the potential prognostic significance of other antibody profiles.”

“Future validation in larger OMG cohorts is essential to enhance the model’s robustness and clinical applicability,” the team concluded.