Background:  Cognitive impairment can occur or persist during ART. Explanations include co-morbidities, neurotoxic ART, persistent neuroinflammation, or persistent HIV replication. This analysis assessed whether low levels of HIV in CSF were associated with inter-individual differences in ART and neuropsychological (NP) performance.

Methods:  We selected 300 participants from the CHARTER cohort because they were taking ART, had HIV RNA levels <50 copies/mL (Roche Amplicor) in cerebrospinal fluid (CSF) and blood, and had at least 2 mL of specimen in storage at –70 C. CSF specimens were assayed with a modified version of the NucliSens EasyQ (bioMerieux) assay capable of qualitatively detecting HIV at 2 copies/mL. Those who had detectable HIV in CSF were identified and 100 matched plasma specimens (obtained within 1 hour of CSF) were also assayed. Penetration of ART into the CNS was estimated by CNS Penetration-Effectiveness (CPE) ranks. Global NP performance was summarized by a validated method. Analyzes were performed using univariate (Fisher’s exact test, t test) and multivariate (logistic regression) methods.

Results:  Participants were mostly non-white (54%), middle-aged (mean 45 years), men (76%) with AIDS (74%) who were HCV seronegative (74%). Median duration of ART was 16 months. Median CD4 count was 535/µL; 122 (41%) had detectable HIV in CSF. Detectable HIV was associated with worse CPE scores (mean 1.5 vs 1.7, d = 0.25, p = 0.03) and HCV seronegativity (81% vs 69%, p = 0.03). Multivariate analysis confirmed that HIV in CSF was associated with both worse CPE scores and HCV seronegativity. Among the 100 matched plasma specimens selected for assay, 26% had undetectable HIV (i.e., undetectable in plasma with detectable HIV in CSF). These individuals had worse NP performance (mean Global Rating 5.3 vs 4.1, d = 0.71, p = 0.006, see the figure) than those who had detectable HIV in both fluids.

Conclusions:  ART-treated individuals frequently (41%) have low levels of HIV in CSF which is associated with less penetrant ART. At least a quarter of these individuals appear to have HIV that persists solely within the central nervous system (detectable HIV in CSF but not in blood) and these individuals have much worse cognitive performance than those who appear to have a systemic source of HIV. We conclude that people living with HIV may have cognitive impairment as a result of ART that is incompletely effective in the nervous system. A more sensitive HIV assay may be needed to monitor CSF in treated individuals.