Background:  Relatively few data are available on the long-term effect of HAART on the severity of neurocognitive impairment in patients with HIV-dementia. HAART improves neurocognitive impairment, but predictors of reversible neurocognitive impairment are unknown.

Methods:  A group of 116 with HIV-related neurocognitive impairment underwent 1 to 6 serial neuropsychological batteries, neurological examination, and brain imaging studies. Patients received HAART for a mean of 32 (range 4 to 107) months. Neuropsychological performance was transformed into z-scores so that negative values indicated below-average performances. The change in mean z-score was evaluated by paired t tests. According to the results of serial neuropsychological assessments, patients were considered as having either reversible or persistent neurocognitive impairment. A multiple logistic regression model was performed to assess factors associated with reversible neurocognitive impairment.

Results:  Neuropsychological z-scores showed significant (p <0.05) improvements after 6, 12, 18, 24, 36, and 48 months of HAART. The improvements were more prominent in measures exploring mental flexibility and fine motor functioning and less prominent in measures exploring memory and concentration and speed of mental processing. Overall, a reversible neurocognitive impairment was observed in 35 of 116 (30.2%) patients. Age, CDC stage, risk category, baseline CD4⁺ count and viral load, virological response to HAART, changes in viral load and CD4⁺ count, use of CNS-penetrating drugs did not differ between patients with persistent or reversible neurocognitive impairment. By contrast, patients with reversible neurocognitive impairment were more educated (12.4 vs 10.1 years; p = 0.004), predominantly male (94.3 vs 75.3%; p = 0.017), had a better baseline mean NPZ8 global score (–1.47 vs –2.29; p = 0.026), and a greater improvement in mean NPZ8 score between baseline and first follow-up (+0.84 vs +0.18; p = 0.044). In multivariable analyses, only baseline NPZ8 scores (OR 5.5; 95%CI 2.0 to 5.0; p = 0.001) and changes between baseline and first follow-up NPZ8 scores (OR 6.9; 95%CI 2.4 to 20.1; p <0.001) remained significantly associated with reversible neurocognitive impairment.

Conclusions:  Although HAART was associated with improvements in cognitive functions, the impairment persisted in nearly 2 of 3 of cases, despite 3 years of HAART. Less severe impairment and prompt improvement in cognitive performance were independently associated with reversible neurocognitive impairment. Our data indicate that HAART should be initiated as soon as neurocognitive impairment is diagnosed to avoid a potentially irreversible neurological damage. Additional treatment strategies are needed in patients with persistent neurocognitive impairment.