Background: Though common in HIV infection, CSF pleocytosis is not well characterized. This study aimed to further define the relationship among 4 salient interacting variables: CSF WBCs, CSF, and plasma HIV concentrations, and blood CD4 counts.

Methods: Prospective study of a convenience sample using 2 approaches: cross-sectional analysis of 100 clinically diverse HIV-1-infected subjects, and longitudinal study of a subset undergoing serial lumbar punctures after initiating (n=29) or interrupting (n=9) antiretroviral therapy. CSF HIV was measured in the cell-free fraction.

Results: Cross-sectionally, CSF HIV correlated strongly with both the CSF WBC counts and plasma viral loads (ρ= 0.53 and 0.63, respectively, p <0.001), and subjects with the highest CSF HIV concentrations had both elevated plasma virus and CSF lymphocytic pleocytosis. The lowest quartile of subjects (CD4 counts <50 cells/μL) had both low prevalence of CSF pleocytosis and higher plasma:CSF HIV ratios (p = 0.005 and 0.0005, Kruskal-Wallis) than other quartiles. The CSF:plasma ratios of acute-phase viral decay after starting treatment correlated with the baseline blood CD4 count (ρ=0.805, p <0.001). However, all subjects observed for >1 month had favorable CSF responses, approaching or reaching the limit of viral detection at last observation, including 5 with persistent, drug-resistant plasma viremia. CSF pleocytosis resolved in all of the treated subjects. Conversely, the majority interrupting treatment developed pleocytosis, at times robust but always asymptomatic. Highest CSF HIV developed in those with the greatest WBC responses.

Conclusions: These results show that CSF pleocytosis is intimately related to the local HIV concentration either as a response, contributing cause, or both. Slower treatment-induced acute viral decay in CSF compared to plasma in those with low CD4 counts indicates a change in the character of CSF infection with immunological progression. However, the favorable virological outcomes in this series suggests that high CSF drug penetration may not always be critical for CNS treatment, and CSF responses in the face of persistent viremia may indicate sparing of an important target organ despite virological failure in blood. The fall in CSF lymphocytes with treatment and their rise with its interruption, along with parallel changes in CSF HIV, suggest that treatment of systemic infection may importantly reduce the linked entry of lymphocytes and virus into the CNS.