Background: Clostridium perfringens epsilon toxin (ETX) is a potent neurotoxin that crosses the blood-brain barrier (BBB), and has been suggested to be involved in demyelinating CNS disorders. Here, we investigated the prevalence of C. perfringens and anti-ETX immunoglobulin G (IgG) and immunoglobulin M (IgM) antibody levels in stool and sera of patients with multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) compared to healthy controls.
Methods: Between September 2019 and June 2020, 43 stool and 43 sera specimens were collected from MS (n = 33) and NMOSD (n = 10) patients, and 49 samples from healthy subjects. Stool samples were cultured for the presence of C. perfringens and PCR was used to detect etx-encoding gene. The level of serum IgG and IgM against ETX was examined using competitive ELISA. The predictive value of sex, age, ETX IgG and IgM levels, and C. perfringens was assessed using multivariable prediction models, with all models trained on the full set of variables. Independent contributions were evaluated separately using logistic regression (LR) analysis.
Results: Totally, 11 C. perfringens were isolated from MS (n = 4), NMOSD (n = 2), and healthy (n = 5) subjects. The etx gene was detected in 1/4 (25%) isolates from MS and 2/2 (100%) from NMOSD patients, while none of the isolates from healthy subjects carried the gene. The analysis of serum IgG and IgM level against ETX showed no significant difference between patients and healthy subjects. Among the prediction models used, LR and Naïve Bayes (NB), trained on all predictor variables, exhibited the highest statistical accuracy for MS diagnosis, with AUC = 0.71 and AUC = 0.93, respectively. Among individual parameters, sex (AUC = 0.63) and IgG level (AUC = 0.57) were the most reliable predictors. The nomogram prediction demonstrated that the probability of patients' risk of developing MS and NMOSD can be predicted using these predictors.
Conclusions: Our findings indicate a potential association between ETX-producing C. perfringens strains and MS/NMOSD diseases, as its presence was detected in a subset of patient samples. However, no significant differences in anti-ETX antibody levels were observed between control and patients. Our prediction models highlighted key demographic and serological factors that may contribute to MS diagnosis. These results highlight the need for further research to meticulously decipher the biological interplay of C. perfringens and its ETX toxin with the development of MS and NMOSD.
Keywords: Clostridium perfringens; ETX; Epsilon toxin; Multiple sclerosis; NMOSD; Neuromyelitis optica; Prediction models.
© 2025. The Author(s).