Glial and neuroaxonal biomarkers in a multiple sclerosis (MS) cohort.

We present normally distributed variables using means and standard deviations and non-normally distributed variables using medians along with their ranges. Spearman's correlation coefficients were used to analyse correlations between the CSF's biomarker concentrations, and both unadjusted and Bonferroni adjusted p-values are reported. Orthogonal projection to latent structure discriminant analysis (OPLS-DA) was also used to find differences in terms of CSF metabolites between the relapsing and remitting patients [CIS, RR, PR versus PP, SP patients]. The OPLS-DA algorithm finds the projection direction, score vector, that gives the largest covariance between the variables and the pre-defined classes (i.e. relapsing and remitting phases) and that maximizes the separation between the classes. The variables that are found to have an influence on the projection (VIP: Variable importance on the projection plots) and that contribute to discriminate between the classes are summarized in a VIP bar-plot. The higher the VIP bar, the more influential is the variable on the model. We used linear regression to examine the association of the biomarkers with EDSS-change, :adjusting for age. A global validation test was used in each model to examine if the assumptions are met. All statistical analyses were performed in R. RESULTS: 87 MS patients (aged 41.1±11.96) enrolled in the study and 21 controls (aged 44.17±12.8). The female/male ratio was 8 females/12 males in the controls and...
Source: Hellenic Journal of Nuclear Medicine - Category: Nuclear Medicine Tags: Hell J Nucl Med Source Type: research