Background
Inflammation in people with HIV (PWH) involves distinct CNS and systemic immune processes. Plasma biomarkers may miss CNS-specific inflammation due to limited molecular exchange across the blood-brain barrier (BBB). We examined how BBB integrity and viral suppression influence within- and cross-compartment cerebrospinal fluid (CSF)-plasma biomarker relationships.
Methods
Paired CSF and plasma samples from 563 CHARTER participants were analyzed using Olink’s Target 96 Inflammation panel. BBB permeability was estimated from CSF:plasma protein ratios. Pearson correlation, regression with interactions, and network analysis were used to evaluate within- and cross-compartment biomarker correlations and the effects of BBB permeability and viral suppression.
Results
Intra-CSF (r=0.56–0.90) and intra-plasma (r=0.51–0.79) correlations were moderate to strong, while cross-compartment (i.e., CSF-plasma) correlations were weaker and more variable (r=0.18–0.62). For instance, CD5 and CD6 correlated strongly within CSF (r=0.90) and plasma (r=0.76) but only modestly across compartments (r=0.35–0.40). BBB permeability significantly influenced cross-compartment relationships, with CD8A, IL-12B, TNFRSF9, and TNFB showing stronger associations at higher permeability (interaction ps<0.03). Eight proteins, including IL-12B, CCL3, CXCL9, CXCL10, CD8A, and TRAIL, had stronger CSF–plasma correlations in unsuppressed versus suppressed participants (FDR-adjusted p<0.05). Random forest models classified high BBB permeability with an AUC-ROC of 0.79 ± 0.034 using CSF biomarker signatures, highlighting IL-10RB, CCL23, and SCF as key contributors. Network analysis of highly connected biomarkers identified central nodes such as TNF and IFNG, and KEGG pathway enrichment highlighted cytokine–cytokine receptor and Toll-like receptor signaling as key cross-compartment mechanisms (Figure).
Conclusions
BBB integrity and viral suppression shape biomarker compartmentalization in HIV. Plasma markers best reflected CNS biology when the BBB was disrupted or viremia was present, but were least informative with intact barrier and suppression. Network and pathway analyses underscored that systemic inflammatory and vascular signaling pathways underlie CSF–plasma coordination, providing mechanistic insight into how BBB disruption and viral replication modulate biomarker exchange. Integrating BBB assessment with network and machine learning methods may improve biomarker interpretation, diagnostic precision, and therapeutic monitoring.
Table or Figure
