Background:  HIV neurocognitive impairment (HNCI) is prevalent despite the use of combination ART. Protease inhibitors (PI) are potent ART, but may not penetrate into the central nervous system in therapeutic concentrations, which may allow ongoing replication and injury. Atazanavir (ATV) is a commonly used PI, 14% of which is unbound to plasma proteins and therefore available to penetrate into the central nervous system. Our study objective was to determine ATV penetration into cerebrospinal fluid (CSF).

Methods:  CHARTER is an ongoing, multi-center, observational study to determine the effects of potent ART on HIV-associated neurological disease. Single, random plasma and CSF samples were drawn within an hour of each other from subjects taking ATV/ritonavir between October 2003 and June 2004. Plasma samples were assayed by high-performance liquid chromatography (HPLC) and immunoassay with a detection limit of 45 ng/mL. CSF samples were assayed by immunoassay with a lower detection limit of 5 ng/mL. Data were analyzed with summary statistics and linear regression.

Results:  Paired samples were drawn from 26 participants (age 41±8 years; 84±16 kg; 3 females; 12 white, 11 black, 3 Hispanic) 12±7 hours post-dose. The median plasma and CSF ATV concentrations after a median 3 months of therapy (range 4 days to 26 months) were 1020 ng/mL (range below quantification limit [BQL] 4101) and 14 ng/mL (range BQL 39.9), respectively. Of 26 CSF samples, 17 (65%) were >11 ng/mL; the ATV wild type IC₅₀. The CSF/plasma concentration ratio was 0.0098 (n = 24), and the CSF/IC₅₀ ratio was 1.3. Two paired CSF and plasma samples were below detection; 5 additional patients with plasma concentrations <600 ng/mL had CSF samples below detection, as expected based on the observed CSF/plasma ratio. Plasma concentrations correlated with CSF concentrations (r² = 0.67). Plasma and CSF viral loads at time of sampling were 2.7±1.2 and 1.9±0.5 (mean ± SD) log₁₀ copies/mL, respectively.

Conclusions:  ATV concentrations in CSF are highly variable and are 100-fold lower than plasma concentrations, even with ritonavir-boosting. Observed CSF concentrations were less than the estimated free concentration in plasma (~140 ng/mL), suggesting active transport out of the CSF. Increasing plasma exposure to ATV may increase ATV CSF penetration. CSF concentrations of atazanavir do not consistently exceed the wild type IC₅₀ of ATV, and may not afford protection against viral replication in the CSF.