CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

407 VACS INDEX PREDICTS SYMPTOMATIC HIV-ASSOCIATED NEUROCOGNITIVE DISORDERS IN UGANDA

408 NEUROPSYCHOLOGICAL PHENOTYPES IN THE MACS

Samantha A. Molsberry 1 , Yu Cheng 2 , Lawrence Kingsley 2 , Lisa Jacobson 3 , Andrew Levine 4 , Eileen Martin 5 , Eric N. Miller 4 , Cynthia Munro 3 , Ann B. Ragin 6 , Ned Sacktor 3 , James T. Becker 2 1 Harvard University, Cambridge, MA, USA, 2 University of Pittsburgh, Pittsburgh, PA, USA, 3 The Johns Hopkins University, Baltimore, MD, USA, 4 University of California Los Angeles, Los Angeles, CA, USA, 5 Rush University Medical Center, Chicago, IL, USA, 6 Northwestern University, Chicago, IL, USA Background: HIV infection can affect brain function; prior to the development of combination anti-retroviral therapies (cART), HIV infected individuals often performed poorly on tests of cognition and, in many cases, became severely demented. In the cART era mild forms of HIV-associated neurocognitive disorder (HAND) remain prevalent.This study’s objective was to identify neuropsychological subgroups within the Multicenter AIDS Cohort Study (MACS) based on the participant-based latent structure of cognitive function and to identify factors associated with subgroups Methods: Using MACS Neuropsychological substudy participants’ domain scores from the time of their first neurocognitive classification [HIV+: n=1531; HIV-: n=1370], we applied a cluster variable selection algorithm to identify the optimal subset of neuropsychological domains that have cluster information. We then conducted a latent profile analysis based upon scores from the identified domains. Exploratory analyses were conducted to identify factors associated with cluster membership. For those factors found to be associated with cluster membership, post-hoc analyses were conducted to determine the drivers of the observed results. Results: Cluster variable selection identified all domains as containing cluster information except for the working memory domain. A three-cluster solution with variable distribution, variable volume, and equal shape and orientation (VVE) was identified as the best fit for the data. Profile 1 performed below average on all domains; Profile 2 performed at average on executive functioning, motor, and speed and below average on learning and memory; Profile 3 performed at or above average across all domains (See figure). Approximately 20% of men identified most closely with Profile 1, 35%with Profile 2, and 45%with Profile 3. Demographic, cognitive, and social factors but not HIV-related factors were found to be associated with profile membership. Post-hoc analyses of these associations demonstrated that they were generally driven by differences between Profile 1 and the other two profiles. Conclusion: There is an identifiable pattern of neuropsychological domain scores among MACS members and this pattern is determined by all domains except for the working memory domain. Neither HIV nor HIV-related biomarkers were related with cluster membership, which is consistent with other recent findings that patterns of neuropsychological test performance do not map directly onto HIV serostatus.

Deanna Saylor 1 , Gertrude Nakigozi 2 , Noeline Nakasujja 3 , Kevin Robertson 4 , Ronald H. Gray 5 , Maria Wawer 5 , Ned Sacktor 1 1 Johns Hopkins Hospital, Baltimore, MD, USA, 2 Rakai Health Sciences Program, Kalisizo, Uganda, 3 Makerere University College of Health Sciences, Kampala, Uganda, 4 University of North Carolina Chapel Hill, Chapel Hill, NC, USA, 5 The Johns Hopkins University, Baltimore, MD, USA Background: The Veterans Aging Cohort Study (VACS) Index, a clinical score originally developed in HIV+military veterans to predict hospitalizations, mortality, and other adverse health outcomes, was recently found to correlate with HIV-associated neurocognitive disorder (HAND) in cohorts from the United States, Canada and Italy, but this relationship varies by ethnicity. It has never been studied in Sub-Saharan Africa. We investigated the relationship between HAND stage and VACS Index score among HIV+ patients in rural Uganda. Methods: 141 HIV+ adults from the Rakai Community Cohort Study underwent a detailed neurological history and examination, a comprehensive neurocognitive battery, functional status assessments, and peripheral blood draw for CD4 count, viral load, complete blood count, and comprehensive metabolic panel. HAND stage was determined using Frascatti criteria and local normative data derived from 400 HIV- adults in Rakai, Uganda. HAND diagnoses were then classified into symptomatic (minor neurocognitive disorder, HIV- associated dementia) and asymptomatic (normal cognition, asymptomatic neurocognitive impairment) HAND. Because prior community-based studies in Rakai have found very low rates of hepatitis C infection, VACS Index scores were calculated assuming all participants were hepatitis C negative. Scores between HAND groups were compared using t tests. Results: Participants were 48%male (n=68) with mean age of 37 (SD 9) years. 140 participants (99%) were taking antiretroviral therapy. Median CD4 count was 363 cells/µL (Interquartile Range 245-490), and 77% (n=109) had undetectable HIV plasma viral loads. Those with symptomatic HAND (n=51) had significantly higher VACS Index scores (19.4 (SD 14.9)) than those without symptomatic HAND (n=90; 14.9 (SD 9.1); p=0.04) (Figure). For every one quartile increase in VACS Index score, odds of symptomatic HAND increased by 42% [OR 1.42, 95% CI (1.03, 1.96), p=0.03]. This difference was most strongly associated with the VACS component scores for age (2.4 vs 0.4, p=0.008) and FIB4 score (2.1 vs 1.0, p=0.02). Other component scores were not significantly different between those with and without symptomatic HAND. Conclusion: VACS Index scores were significantly higher among HIV+ adults with symptomatic HAND than those with normal cognition or asymptomatic neurocognitive impairment in rural Uganda. This suggests that the VACS Index may be a useful indicator of symptomatic HAND in African populations in addition to Western populations.

Poster Abstracts

CROI 2018 143