Fig. 4: Cerebrospinal fluid (CSF) proteins, pathways, and predicted upstream regulators associated with contrast-enhancing lesions (CEL) and their destructiveness to central nervous system (CNS) tissue.

a Associations of CSF biomarkers with the number of CEL (CEL#) were assessed by linear regression, with CSF biomarkers as dependent variables and CEL#, sex, and CEL#:sex interaction as independent variables. Coefficients of CEL# characterized biomarker associations. Regression lines and 95% confidence intervals (CI) are shown (green with gray shading). b Seven percent of high-signaling somamers were significantly associated with CEL#, with one-third positive and two-thirds negative. Positive associations were dominated by immune-derived proteins, whereas negative associations were enriched for central nervous system–derived proteins. c Volcano plot of 67 proteins significantly associated with CEL# (unadjusted p < 0.05). Gray line shows coefficient = 0. Two CST7 labels reflect distinct epitopes of the same protein. d Somamers associated with CEL# were analyzed by Ingenuity Pathway Analysis (IPA), identifying pathways, upstream regulators, and networks. e Volcano plot of CEL#-associated somamers affected by CD40 upstream regulators. f Selected upstream regulators identified by IPA, shown as bar plots of bias-adjusted z scores with unadjusted –log10 p-values. g CSF biomarker-predicted CEL# from an elastic net model showed significant increases in people with multiple sclerosis (pwMS) compared with healthy controls (HC), and in relapsing-remitting MS (RRMS) compared with primary progressive MS (PPMS) and secondary progressive MS (SPMS). The green rectangle shows HC mean ±1.5 standard deviation (SD); red % values indicate proportions above HC range. Significance was tested by two-sided one-way ANOVA with post-hoc t-tests. Boxplots display medians, quartiles, and whiskers (1.5×in interquartile range [IQR]). h Regression of biomarker-predicted CEL# and age showed CEL# decline with increasing age. Blue line with band shows regression with 95% CI; HC reference lines shown in green/black represent regression line with 95% CI/95% prediction interval. i Regression of neurofilament light chain (NEFL) levels and biomarker-predicted CEL# in 282 MS samples above the HC prediction interval was used to calculate NEFL–CEL residuals (dark red vertical lines), indicating axonal damage adjusted for CEL#. The regression line and 95% confidence interval are shown as a black line with a gray band. The regression model is characterized by the Pearson correlation coefficient (R), coefficient of variance (R²), and unadjusted p-value. j, k Scatter plots showing somamers associated with CEL# and NEFL–CEL residuals, highlighting biomarkers from neurons (green), monocytes (salmon), fibroblasts (gray), and endothelial cells (beige). The solid gray lines mark coefficients of 0. All p-values of regression coefficients were tested in a two-sided test. Created in BioRender. Kosa, P. (2025) https://BioRender.com/g03z1b4.