Background: Recently, the incidence of AIDS dementia complex (ADC) has been found to have halved as a consequence of the introduction of highly active antiretroviral therapy (HAART) but it is unknown whether this can be maintained with long-term HAART.  Moreover, it is not clear what factors are important in ADC development in patients who are taking HAART and who have plasma HIV viral suppression.

Methods: HIV-infected patients who were well and without clinical evidence of ADC were enrolled in a 4-year prospective study to determine the  factors that govern involvement of the brain by HIV.  All had a nadir CD4 cell count <200/mL and all were taking HAART.  Plasma HIV viral load was  < 400 copies/mL.  At entry all were assessed neurologically and neuropsychologically with a standard battery.  In addition, all had quantitative assessment of eye movements, MRI brain scans, cerebrospinal fluid (CSF) analyses, and FDG PET brain scans.  Each PET scan was co-registered with the MRI brain scan and dynamic data acquisition with blood sampling was used to supply quantitative data.  Normative data were obtained from age-matched HIV-negative neurologically normal subjects.  Patients were subsequently assessed neurologically every 4 months until ADC developed or 4 years had passed.  Patients reaching either of these endpoints were then re-assessed using the same investigations that were employed at entry into the study.

Results: 20 patients entered the study.  The mean age was 40.4 years; all were male except one.  One-third had evidence of minor neuropsychlogical dysfunction (psychomotor slowing) and abnormal saccadic eye movements.  20% had mild CSF abnormalities (minor elevations of  β-2 microglobulin and HIV viral load).  One third had marked (<2 SDs) reduction in global (both cortex and basal ganglia) glucose metabolism.  In 15% there was a mild (between 1 and 2 SDs) relative increase in basal ganglia glucose metabolism.  At the end of the study 1 patient had died, 1 had developed progressive multifocal leukoencephalopathy and 6 had developed ADC.  Factors associated with ADC were psychomotor slowing, elevated CSF HIV viral load, elevated CSF β-2 microglobulin concentrations and basal ganglia hypermetabolism as demonstrated by FDG PET scanning.

Conclusions: Despite sytemically effective HAART, central nervous system abnormalities persist in a sizeable proportion of patients.  Moreover, these patients are at increased risk of developing ADC.