Background:  HIV neurocognitive impairment (NCI) remains highly prevalent despite the use of ART. The predictive ability of several biomarkers may be reduced in treated populations. The objective of this project was to identify biomarkers that were associated with neuropsychological performance before and during ART.

Methods:  We enrolled 29 subjects in studies of ART-associated changes in neuropsychological performance and HIV RNA in plasma and cerebrospinal fluid (CSF). Neuropsychological performance was summarized as a continuous measure, the global deficit score. Analyses were performed in 2 stages. First, 13 biomarkers were measured in CSF from 19 subjects prior to and during (median follow-up, 15 weeks) a change in ART (HIV RNA, total protein, MCP-1, BDNF, FGF-1, RANTES, uPAR, sFAS, s100β, IP-10, 8-isoprostane, neopterin, sVCAM-1). Based on initial findings, 7 biomarkers were measured in an additional 10 subjects (total protein, IP-10, sFAS, uPAR, s100b, MCP-1, HIV RNA). All biomarkers were measured using commercial immunoassays. Data were transformed when appropriate and analyzed using descriptive statistics and routine parametric and non-parametric univariate methods.

Results:  Subjects were mostly middle-aged (median 40.4 years), white (62%), men (79%). At baseline, biomarker- neuropsychological relationships were modified by ART use. Among the 16 subjects who were not taking ART, worse performance correlated with higher levels of total protein (R² = 0.63, p = 0.009), IP-10 (R²= 0.54, p = 0.04), sFas (R² = 0.54, p = 0.04), and possibly uPAR (R² = 0.43, p = 0.09). At follow-up, greater neuropsychological improvements correlated with reductions in total protein (R² = 0.52, p = 0.03), IP-10 (R² = 0.51, p = 0.03), and neopterin (R² = 0.52, p = 0.03). BM-NP relationships were strongly modified by HIV suppression during ART. Among the 20 total subjects who achieved HIV suppression, worse performance correlated with higher MCP-1 (R² = 0.58, p = 0.02) and total protein levels (R² = 0.54, p = 0.02). In this group, the sensitivity of total protein levels ≥39 mg/dL to predict impairment was 82% and specificity was 75%.

Conclusions:  Biomarker-neuropsycholical relationships varied with the use, duration, and effectiveness of ART. In this pilot project, higher levels of total protein, a clinically available assay, were associated with worse performance in all analyses, possibly because it provides a pooled measure of many neuropathogenic proteins. Total protein may be a useful marker of NCI in a clinically important population, ART-treated individuals, and may indicate the need for intensified or adjunctive therapy.