An Immune Model Incorporating NK Cells and IL-12 to Predict the Efficacy of the Initial Treatment Cycle of Efgartigimod in Generalized Myasthenia Gravis.

BACKGROUND

Efgartigimod (EFG), a neonatal Fc receptor (FcRn) antagonist for generalized myasthenia gravis (gMG), exhibits variable treatment responses. This study aimed to investigate the associations of baseline clinical characteristics and multidimensional immune profiles with the clinical response to EFG in patients with acetylcholine receptor antibody-positive (AChR+) gMG, and develop a predictive model.

METHODS

This multicenter retrospective observational study enrolled 35 AChR+ gMG patients who received at least one cycle of EFG. Responders were defined as those achieving a Myasthenia Gravis Activities of Daily Living (MG-ADL) score reduction of ≥ 2 points that was sustained for ≥ 1 month. After univariable screening, three models (Clinical, Immune, Integrated) were constructed using a complete-case modeling subset of 26 patients via multivariable logistic regression, least absolute shrinkage and selection operator (LASSO), and random forest algorithms. The optimal model was selected via leave-one-out cross-validation (LOOCV) and translated into a clinical risk score and an online calculator.

RESULTS

Among 35 patients, 25 responded (68.6%). Non-responders had significantly higher baseline NK cell counts (p = 0.013) and showed a trend toward lower IL-12p70 levels (p = 0.095). Both NK cell (OR = 0.985) and IL-12p70 (OR = 11.657) were independent predictors of EFG response. The Immune Model (NK cells + IL-12p70) demonstrated robust discrimination (AUC = 0.861), outperforming the Clinical Model (AUC = 0.625, p = 0.021) and the Integrated Model (AUC = 0.847, p = 0.782). It showed good calibration and clinical utility. Bootstrap validation confirmed robustness (corrected AUC = 0.872, optimism=0.017). A derived clinical risk score stratified patients into high (100%), moderate (80.0%), and low (33.3%) response probability groups. An online prediction calculator was developed.

CONCLUSION

Baseline NK cell counts and IL-12p70 levels may predict EFG response in AChR+ gMG. The dual-biomarker Immune Model demonstrates robust performance, and has been translated into an online tool but yet requires validation in prospective, larger cohorts.