Factor Analysis of Proton MR Spectroscopic Imaging Data in HIV Infection: Metabolite-derived Factors Help Identify Infection and Dementia

Purpose: To develop a relevant pathophysiologic model of human immunodeficiency virus (HIV)-associated dementia by studying regional variations in metabolite levels measured with magnetic resonance (MR) spectroscopic imaging and their relationship to immunologic measures and cognitive dysfunction. Materials and Methods: This was a HIPAA-compliant, institutional review board-approved study involving written informed consent. Distributions of N-acetylaspartate (NAA), choline (Cho), and creatine (Cr) concentrations in 94 subjects (20 seronegative controls and 74 HIV-positive subjects; 34 of the HIV-positive subjects having HIV-associated dementia; 63 men, 31 women; mean age, 40 years) were determined with proton (hydrogen 1 [H-1]) MR spectroscopic imaging. HIV-positive subjects underwent neuropsychological testing and blood and cerebrospinal fluid (CSF) analysis. Factor analysis was utilized to determine associations between metabolites across regions. Analysis of variance and t tests were used to isolate differences between cohorts. Results: A Cho factor differentiated seronegative controls from HIV-infected cohorts, indicating elevated Cho levels across deep gray and white matter regions of HIV-positive individuals. An NAA factor differentiated those with dementia from those without and correlated best with psychomotor and executive function tests. A Cr factor indicated Cr elevations correlated with CSF monocyte chemoattractant protein-1 levels. NAA and Cr factor scores were strongly weighted to metabolite changes in white matter regions. Conclusion: These results highlight the importance of white matter involvement in HIV-associated dementia and support the current pathogenesis model of glial cell proliferation in HIV infection, denoted by regional Cho elevations, and neuronal dysfunction and/or death, denoted by NAA decreases, associated with dementia. Factor analysis of MR spectroscopic imaging data is a useful method for determining regional metabolic variations in HIV infection and its neuropsychological correlates. (C) RSNA, 2010