To explore whether in pediatric multiple sclerosis (MS) patients treated with interferon-b (IFN-b) increases in the number of MRI lesions, clinical relapses or disability worsening over the first year of treatment or combination of them could be able to predict treatment failure, and to identify a score of treatment response specific for pediatric patients.

A cohort of 334 pediatric MS patients treated with any IFN-b preparations was analyzed. MRI and clinical assessments during first year of treatment, and clinical follow-up of at least 2 additional years were collected. From this cohort a training (233 patients) and a validation set (101 patients) were obtained to estimate the diagnostic accuracy of the score. The association of 1-year MRI or clinical variables with the risk of treatment failure (defined as treatment switch for inefficacy) was evaluated using multivariate Cox regression models.

The presence of at least three new T2 or gadolinium enhancing (Gd+) lesions (hazard ratio [HR]=2.40; p=0.003) detected on 1-year MRI scan and the occurrence of at least 2 relapses (HR=2.27; p=0.001) over the first year of IFN-b treatment resulted as independent predictors of treatment failure. According to these results, risk levels were grouped into 3 classes: group 1 (<2 relapses, <3 T2/Gd+ lesions), group 2 (≥2 relapses or ≥T2/Gd+ lesions), group 3 (≥2 relapses and 1 or ≥3 T2/Gd+ lesions). In the validation set, group 2 patients had an intermediate risk of treatment failure (HR=5.26, p<0.001) and group 3 patients had the highest risk of treatment failure (HR=18.94, p=0.007). We found that the score had moderate sensitivity (52.26%) and very high specificity (91.9%).

We obtained a simple score specific for pediatric MS patients able to predict IFN-b treatment failure, supporting the clinicians in optimizing treatment strategies.