Background:  HIV-1 can replicate in neurological compartment independently from plasma. The neuroactive role of antiretroviral drugs and protective role of various drug classes has to be conclusively defined.

Methods:  To analyze the effect of drugs and classes of ART on HIV-1 RNA load in cerebrospinal fluid (CSF), HIV-infected neurologic patients enrolled in the Italian Registry of Investigative NeuroAIDS (IRINA) was investigated. HIV RNA in CSF were considered undetectable if below the detection limit. Zidovudine (AZT), stavudine (D4T), lamivudine (3TC), abacavir (ABV), nevirapine (NVP), efavirenz (EFV), and indinavir (IDV) were considered to be neuroactive drugs. Logistic regression was used for multivariate adjustment of covariates.

Results:  Among the 363 patients included, 50.1% were experienced and 33.1% were taking ART at diagnosis of neurological disorders. Median CD4 count, plasma, and CSF HIV-1 RNA were 71 cellx10⁹/L (IQR 22 to 162), 4.98 log₁₀ copies/mL (3.81 to 5.44) and 3.63 log₁₀ copies/mL (2.17 to 4.83), respectively. In CSF of 16.5% of patients HIV RNA was undetectable despite the neurological picture. Mean CSF HIV RNA was significantly higher in naïve (4.3 log₁₀ copies/mL) than in experienced (3.2 log₁₀ copies/mL) (p <0.001), and in patients taking ART (2.9 log₁₀ copies/mL) than in patients not taking (4.2 log₁₀ copies/mL) (p <0.001). A linear correlation between CSF HIV RNA and number of neuroactive drugs included in the HAART regimen was found (r = 0.17; p <0.001). By multivariable logistic regression, assuming 2 (OR 6.07; 95%CI 2.87 to 12.85) or 3 or more (OR 7.07; 3.46 to 14.47) neuroactive drugs significantly increases probability of undetectable CSF HIV-1 RNA. Compared with no treatment, exposure to a regimen containing non-nucleoside reverse transcriptase inhibitor (NNRTI) (OR 12.46; 3.28 to 47.41) or boosted protease inhibitor (PI) (OR 5.64; 1.31 to 24.25), and including EFV (OR 5.59; 1.29 to 24.27) or NVP (OR 5.18;1.15 to 23.44) in the regimen, all independently predicted an undetectable CSF HIV-1 RNA, whereas higher burden of plasma HIV-1 RNA (OR 0.70;0.51 to 0.96) and occurrence of neurocognitive impairment (OR 0.38; 0.16 to 0.88) were associated with a reduced probability of CSF viral suppression. When PI exposure was used as reference group, only NNRTI (OR 5.38; 1.52 to 19.00), but not boosted PI (OR 1.47; 0.15 to 14.09) had a protective role over HIV-1 CSF replication.

Conclusions:  Exposure to NNRTI and, in lower measure, to boosted PI increases probability of CSF HIV-1 suppression during therapy. Including a higher number of neuroactive drugs in the HAART regimen results in a more effective control of CSF HIV-1 replication in neurologically impaired patients.