Objective: To identify the sources of HIV virions in CSF by modeling treatment-associated HIV dynamics.
Background: We postulated a model in which cell-free CSF virions originate from two major sources, namely, systemic non-CNS and CNS tissues, the latter including brain parenchyma and meninges. The model predicted that with initiation of antiretroviral therapy, the acute-phase decline in CSF HIV RNA levels would be controlled by the kinetics of the dominant virion source (systemic versus CNS). Based on prior observations, we hypothesized that the dominant source of CSF virions would shift from systemic to CNS in more advanced disease.
Methods: Three patient groups were studied: Group 1 (n = 5): nondemented, with early HIV disease (CD4+ lymphocytes > or = 400/microL) or pleocytosis (CSF leukocytes > or = 4/microL); Group 2 (n = 5): nondemented, with advanced HIV disease (CD4+ < 400/microL) and no pleocytosis; Group 3 (n = 2): patients with HIV-associated dementia (HAD). All patients began a new, highly active antiretroviral treatment regimen and underwent serial lumbar punctures and phlebotomies.
Results: For patients in Group 2, the rate of decline in CSF HIV RNA was slower than in plasma (p < 0.00001). For Group 1, the rate of decline in CSF was not different from plasma (p > 0.25). Patients with HAD showed high CSF HIV RNA after 5 to 6 weeks of treatment despite a 100-fold decrease in plasma HIV RNA.
Conclusions: CSF and plasma HIV dynamics became increasingly independent in advanced HIV disease, and the compartmental discrepancy was largest in HAD. Our findings suggest that viral replication in CNS tissues may constitute a major, independent source of CSF HIV RNA. In patients with HAD, brain parenchyma itself may be the principal CNS tissue source, and CNS-targeted treatment strategies may be required to eradicate this infection.