Background: CSF and plasma HIV-1 dynamics are independent in neurologic disease. In HAART-treated patients (pts), the effects of specific neurologic disorders and of different regimens on CSF HIV-1 replication were poorly investigated.

Methods: Retrospective analysis on pts with available paired CSF/plasma HIV-1 RNA measurements among those recruited in a neurologic Italian registry.

Results: Among 770 patients enrolled in IRINA from January 2000, 245 were included. General characteristics: median age 40 yrs (IQR 36–45); 192 (78%) male; 106 (43%) IVDU; and 75 (31%) previous AIDS. Median CD4 count, plasma, and CSF HIV-1 RNA were 68 cellx10⁹/l (IQR 21–136), 4.98 log₁₀c/ml (4.18–5.45) and 3.52 log₁₀c/ml (2.06–4.67). Prevalence: HIVE 24%; PML 15%; other leukoencephalopaties (OL) 18%; TE 12%; cryptococcal meningitis (CM) 11%; PCNSL 6%; TB 3%; and other diseases 9%. A high variability for mean values of CSF HIV-1 RNA in naïve (ANOVA; P = 0.001) as well as in experienced (P = 0.01), but not for plasma HIV-1 RNA and CD4 count was observed. Higher mean CSF HIV-1 RNA values were detected in TB (5.02 log₁₀c/ml), CM (4.49), and PCNSL (4.27), whereas lower values were observed in PML (3.13) and OL (2.95), with significant differences after Bonferroni’s adjustment for comparisons of CM vs PML (P = 0.004), CM vs OL (P = 0.0001), TB vs PML (P = 0.01) and TB vs OL (P = 0.003). 119 (49%) pts had a previous HAART exposure, and 78 (32%) were currently treated at neurologic diagnosis. A significant correlation between plasma and CSF HIV-1 RNA was observed in experienced (r = 0.414; P = 0.0001) but not in naïve pts (r = 0.165; P=0.08). Using a multivariable logistic regression model, taking efavirenz at neurologic diagnosis (OR 13.63; 95% CI, 1.59–116.65 vs not taking or naive) and having PML (3.29; 0.96–11.25 with HIVE as reference) were associated to CSF HIV-1 RNA < 500 c/ml, whereas CD4 count (OR 0.84 for each 50 cell increase; 0.73–0.97) and meningeal enhancement (0.19; 0.03–0.99) reduced probability of undetectable CSF viral load.

Conclusions: HIV-1 replication was strongly influenced by specific neurologic disorders even in HAART treated pts. Higher CSF HIV-1 burden was detected in inflammatory meningoencephalitis (TB and CM), probably by a macrophages activation, whereas PML and leukoencephalopathies other than HIVE developed despite lower HIV-1 replication in the compartment. Efavirenz, despite low levels in CSF, is highly effective in suppressing CSF HIV-1 load even adjusted for main covariates.