Myasthenia gravis is an antibody-mediated autoimmune disease of the neuromuscular junction, in which autoantibodies against the acetylcholine receptor and other postsynaptic membrane proteins disrupt signaling between nerves and muscle. The hallmark is fluctuating skeletal muscle weakness and fatigability, and in severe cases respiratory muscle involvement can progress to life-threatening myasthenic crisis. Although corticosteroids, immunosuppressants, intravenous immunoglobulin, and targeted biologics have improved survival and quality of life, the disease remains clinically heterogeneous, with wide variation in age at onset, affected muscle groups, treatment response, and long-term outcome. Some patients never achieve sustained remission despite standard treatment, and repeated relapses, disease progression, and myasthenic crisis remain common—making biomarkers that could aid risk stratification a persistent clinical goal.

The study enrolled 468 anti-AChR antibody-positive patients hospitalized at a single center in Shijiazhuang, China, between December 2023 and April 2025. Fifty-three patients were excluded by the study criteria and 25 were lost to follow-up, leaving 390 patients in the final analysis. All were scheduled to begin immunosuppressive therapy, and critically, blood samples were drawn before any immunosuppressive treatment was started, so the measured protein levels reflected the untreated disease state. A control group of 100 healthy individuals undergoing routine physical examinations was frequency-matched by age and sex. Serum Sema4D and ANGPTL4 concentrations were measured by enzyme-linked immunosorbent assay, with all samples tested in duplicate and assay coefficients of variation below 10 percent; laboratory technicians were blinded to clinical data.

The comparison with healthy controls was striking. Median serum Sema4D was 43.52 ng/mL in patients versus 33.52 ng/mL in controls, and median ANGPTL4 was 17.15 ng/mL versus 12.20 ng/mL, with both differences highly significant (p < 0.001). Age, sex, and body mass index were comparable between the groups, suggesting these general characteristics did not confound the biomarker comparison.

After one year of follow-up, patients were classified by Myasthenia Gravis Foundation of America post-intervention status. Good prognosis was defined as achieving minimal manifestations or better—encompassing complete stable remission, pharmacologic remission, or minimal manifestations—while poor prognosis meant failing to reach that threshold. Among the 390 patients, 308 (79.0 percent) achieved a good prognosis and 82 (21.0 percent) a poor prognosis. The poor-prognosis group had significantly higher proportions of MGFA class IV–V disease (52.44 percent versus 17.86 percent), baseline dysphagia (40.24 percent versus 21.75 percent), and thymoma (24.39 percent versus 11.04 percent), along with higher quality-of-life and nutritional-risk scores. Both biomarkers were also significantly elevated in the poor-prognosis group.

The correlations extended across several clinical dimensions. Serum levels of both proteins were significantly higher in patients with MGFA class IV–V disease than in those with class I–III disease, and higher in patients with thymoma than in those without. Spearman rank correlation analysis showed that Sema4D and ANGPTL4 levels were positively correlated with the MG-QoL15r quality-of-life score (r = 0.378 and r = 0.370, respectively, both p < 0.001) and with the CONUT nutritional status score (r = 0.355 and r = 0.369). The two proteins also correlated with each other (r = 0.431, p < 0.001), a pattern the authors describe as an observational finding that requires further validation, since no mechanistic indicators were collected that could establish a direct molecular interaction.

Multivariable logistic regression, which included MGFA class, dysphagia, thymoma, quality-of-life score, and nutritional score alongside the two biomarkers, identified all of these as independent risk factors for poor prognosis. Each unit increase in Sema4D carried an odds ratio of 1.051 for poor prognosis, and each unit increase in ANGPTL4 an odds ratio of 1.085, after adjustment for the clinical variables. Notably, the biomarker associations persisted after adjustment for thymoma, indicating that their elevation was not solely attributable to thymoma-associated immune dysregulation. Variance inflation factors below 2 for all predictors suggested multicollinearity was not a problem.

The predictive performance analysis offered a quantitative picture of how much the proteins add. Alone, Sema4D yielded an area under the ROC curve of 0.770 (95% CI 0.707–0.832) and ANGPTL4 an AUC of 0.759 (95% CI 0.698–0.820) for predicting poor prognosis. Combined, the two markers reached an AUC of 0.841 (95% CI 0.790–0.893), with sensitivity of 78.05 percent and specificity of 79.22 percent. When the biomarkers were added to a baseline clinical model built from MGFA class IV–V, dysphagia, thymoma, MG-QoL15r score, and CONUT score—which alone achieved an AUC of 0.874—the combined model rose to 0.907 (95% CI 0.875–0.939), a statistically significant improvement (p = 0.0002 by DeLong test). Scatter-plot analysis showed that patients with poor outcomes were disproportionately found in the region where both biomarkers exceeded their optimal cutoffs: 40.2 percent of the poor-prognosis group versus 7.5 percent of the good-prognosis group had both Sema4D above 48.24 ng/mL and ANGPTL4 above 20.35 ng/mL.

The authors caution that the study has several limitations. It was a single-center prospective study with a 12-month follow-up, so the findings require validation in larger multicenter cohorts with longer observation. Only anti-AChR antibody-positive patients were enrolled, meaning the conclusions may not generalize to anti-MuSK-positive, anti-LRP4-positive, or seronegative myasthenia gravis. Disease-specific quantitative measures such as MG-ADL, QMG, and MGC scores were not systematically collected, and immunotherapy regimens during follow-up were individualized by clinicians, introducing treatment heterogeneity. The CONUT nutritional score may also partly reflect disease severity rather than act as a fully independent prognostic factor, and the observational design cannot establish causal relationships between the biomarkers and outcome. Serial serum measurements during treatment were not performed, so the markers’ potential as indicators of treatment response remains unclear. The specific biological pathways through which Sema4D and ANGPTL4 might act in myasthenia gravis also remain to be determined by future basic research.

Within those constraints, the study adds to a growing literature linking these proteins to autoimmune disease. Sema4D is a transmembrane glycoprotein expressed on T cells, B cells, and monocytes that regulates lymphocyte activation and inflammatory responses, and it has been found at elevated levels in rheumatoid arthritis, multiple sclerosis, and psoriasis. ANGPTL4 participates in lipid metabolism as well as inflammatory cell recruitment and tissue remodeling, and its dysregulated expression has been reported in rheumatoid arthritis and systemic lupus erythematosus. Prior work also found ANGPTL4 gene transcripts upregulated in extraocular muscle fibroblasts from patients with refractory ocular myasthenia gravis after exposure to patient serum. The present findings suggest that combined measurement of the two proteins may help evaluate prognostic risk in anti-AChR antibody-positive myasthenia gravis, though the authors emphasize that their clinical utility requires further investigation before any practical application.

Subject of Research: Serum biomarkers Sema4D and ANGPTL4 and their association with disease severity and prognosis in myasthenia gravis

Article Title: Association of serum Sema4D and ANGPTL4 levels with disease severity and prognosis in myasthenia gravis patients

Article References: Li, W., Zhang, M., Wang, Z., Gong, C., Sun, L., Feng, W., Qi, G., Yang, Y., & Yu, F. (2026). Association of serum Sema4D and ANGPTL4 levels with disease severity and prognosis in myasthenia gravis patients. Frontiers in Neurology, 17, Article 1965303. https://doi.org/10.3389/fneur.2026.1965303

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1965303

Keywords: myasthenia gravis, Sema4D, ANGPTL4, biomarkers, autoimmune disease, neuromuscular junction, prognosis, disease severity, thymoma, ELISA, prospective cohort study, neurology