Prediction and Diagnosis of HIV Dementia

Abstract & Commentary

By Michael T. Lin, MD, Assistant Professor of Neurology, NewYork-Presbyterian Hospital, Weill Cornell Medical Center. Dr. Lin reports no financial relationships relevant to this field of study.

Synopsis: Several CSF biomarkers, sphingomyelin, sphingomyelinase, ceramide, 4-hydroxynonenal, and vitamin E, are associated with the development of HIV dementia in patients being treated with highly active antiretroviral therapy (HAART).

Source: Bandaru VV, et al. Associative and predictive biomarkers of dementia in HIV-1-infected patients. Neurology. 2007,68:1481-1487.

Highly active anti-retroviral therapy (HAART) has changed the nature of HIV dementia: the degree of cognitive impairment is milder, the course may improve as well as worsen, and traditional biomarkers such as CSF viral load or levels of monocyte chemotactic protein 1 (MCP-1) are less likely to be associated with dementia. The objective of this study was to identify new CSF biomarkers that are associated with or predictive of HIV dementia in the setting of HAART.

The study involved 48 patients from the North Eastern AIDS Dementia cohort, assessed cognitively at 0, 6, and 12 months. CSF was drawn at the 6 month time point, and analyzed for biochemical markers. Biochemical markers correlating with cognitive change over the preceding 6 months (between 0 and 6 months) were said to be biomarkers associated with cognitive change; biochemical markers correlating with cognitive change over the subsequent 6 months (between 6 and 12 months) were said to be biomarkers predictive of cognitive change. Based on previous work showing accumulation of sphingomyelin, ceramide, sterol, and lipid peroxidation products in brains of HIV dementia patients, these were the primary biomarkers the authors investigated.

Considering first the cognitive changes occurring before the CSF draw (between the 0 and 6 month time points), the authors divided the cohort into 3 groups: those who were not demented at both time points ("nondemented"), those who were initially demented and remained stably so ("inactive dementia"), and those who were initially not demented but became demented ("active dementia"). Those with inactive dementia over the preceding 6 months had 3-5 fold elevations of CSF sphingomyelin compared to nondemented or active dementia subjects. Those with active dementia over the preceding 6 months had increased CSF ceramide levels and sphingomyelinase activity, increased levels of lipid peroxidation products (4-hydroxynonenal), and decreased levels of the lipid antioxidant vitamin E.

Considering, next, the cognitive changes occurring after the CSF draw (between the 6- and 12-month time points), the authors divided the cohort into 3 groups, based on whether scores improved, remained stable, or worsened. The group that subsequently worsened had increased CSF levels of triglyceride C52 and vitamin E. Sphingomyelin, ceramide, and hydroxynonenal levels were not predictive of subsequent worsening.


These are interesting findings, consistent with the known biology of HIV. In tissue culture, the HIV-1 coat protein gp120 and the trans acting protein Tat can increase sphingomyelin levels in 6 hours, followed by activation of neutral sphingomyelinase (Haughey et al. 2004; Jana et al. 2004). In response to inflammatory mediators, sphingomyelinase cleaves sphingomyelin to produce ceramide, which induces apoptosis and increases oxidant stress. Additionally, cleavage of sphingomyelin by sphingomyelinase disrupts the association between cholesterol and sphingomyelin. The cholesterol released can then accumulate in lipid droplets containing cholesterol esters and triglycerides. This sequence could explain the increases in sphingomyelin in stably demented subjects; the increased sphingomyelinase, ceramide, and lipid peroxidation seen in actively demented subjects; and the increased triglyceride C52 seen in subjects who subsequently worsened. The authors interpret the elevated vitamin E levels initially seen in subjects who subsequently worsened as a secondary compensatory response to ongoing pathology. When the pathology overwhelms defensive systems, decreased vitamin E levels would then be found in actively dementing subjects.

Several potential follow-up studies should be considered. Although the design of this study cleverly minimized the number of CSF draws, true biomarkers should be shown to vary over time in parallel with the process being followed. The results of this study should thus be confirmed with serial determinations of biomarkers in parallel with cognitive assessments. Additionally, with at least one key biomarker of active pathogenesis, ceramide, the magnitude of difference between groups, although statistically significant, was small, and it is not clear that such a biomarker could usefully discriminate between groups. The sensitivity, specificity, and ability to discriminate between groups should be characterized. Finally, the pathogenetic sequence identified could lead to new potential therapy of HIV dementia, targeting the HIV-induced perturbations of sphingomyelin/ceramide metabolism.


Haughey NJ, et al. Perturbation of sphingolipid metabolism and ceramide production in HIV-dementia. Ann Neurol. 2004;55:257-267.

Jana A, Pahan K. Human immunodeficiency virus type 1 gp120 induces apoptosis in human primary neurons through redox-regulated activation of neutral sphingomyelinase. J Neurosci. 2004;24:9531-9540.