CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Spatially independent effects of age and past drug use were also seen (p<0.05 FWE-corrected), with age broadly affecting numerous frontal and temporal cortical regions (Figure 1). There were no significant interactions between HIV serostatus and chronological age on any of the GMv measures. Conclusion: Regional GMv is subject to additive effects of serostatus, age and drug use. HIV infection is independently associated with basal ganglia volume reductions equivalent to 20-30 years of typical aging. Serostatus effects were regionally specific when age and drug use were controlled, suggesting that basal ganglia GMv measured using computational neuroanatomy methods may be a useful biomarker to follow HIV treatment effects. White matter reductions in treated HIV infection may reflect independent effects of age, age-related comorbidities, and past or current drug use. References (1) O’Connor EE, Jaillard A, Renard F, Zeffiro TA. Reliability of White Matter Microstructural Changes in HIV Infection: Meta-Analysis and Confirmation. AJNR. 2017 (2) O’Connor EE, Zeffiro TA, Zeffiro TA. Brain Structural Changes Following HIV Infection: Meta- analysis. AJNR. (in press) Talia M. Nir 1 , Hei Y. Lam 1 , Chris R. Ching 1 , Neda Jahanshad 1 , Jaroslaw Harezlak 2 , Kenia Martínez 3 , Tong Zhu 4 , Giovanni Schifitto 5 , Ronald Cohen 6 , Paul M. Thompson 1 , Bradford A. Navia 7 1 University of Southern California, Los Angeles, CA, USA, 2 Indiana University, Bloomington, IN, USA, 3 General University Hospital Gregorio Marañón, Madrid, Spain, 4 University of North Carolina Chapel Hill, Chapel Hill, NC, USA, 5 University of Rochester, Rochester, NY, USA, 6 University of Florida, Gainesville, FL, USA, 7 Tufts University, Boston, MA, USA Background: Neurological injury can persist in individuals with chronic HIV infection, despite viral suppression with combined antiretroviral therapy (cART). Brain metabolite disturbances in treated HIV+ individuals, linked to inflammatory processes and neuronal loss, have been shown to predict neurocognitive impairment (NCI). We hypothesized that using multimodal markers of brain injury-diffusion weighted imaging (DWI) measures of white matter (WM) microstructure, in addition to magnetic resonance spectroscopy (MRS)-would better predict NCI and AIDS Dementia Complex (ADC) decline in chronically HIV-infected adults on stable cART. Methods: 50 adults underwent MRS and DWI brain scans (mean age: 48.1+/- 7.6 yrs; sex: 32M/18F). Levels of N-acetylaspartate (NAA), myo-inositol (MI), choline (Cho), glutamate/glutamine (Glx), and creatine (Cr) were measured in frontal gray matter, frontal WM, and the basal ganglia (BG). Using ENIGMA’s DTI protocol (http://enigma.usc.edu), fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), and axial diffusivity (AD) measures were extracted from the full WM. Baseline (bl) NCI was defined as an ADC ≥1 (n=8). Logistic regressions tested whether absolute metabolite concentrations, WMmeasures, and standard plasma measures of disease severity (nadir/current CD4+ count and viral load) individually predicted bl-ADC status or, relative to those that remained stable (n=24), longitudinal decline in ADC (n=8), controlling for sex and age (and time interval in longitudinal analyses). To compare models, McFadden and Nagelkerke pseudo R², Akaike information criterion (AIC), and area under the ROC curve (AUC) were obtained for each predictor separately, and the model including all significant predictors. Results: Compared to neuroasymptomatic individuals and those that remained stable, individuals with bl-ADC and those that declined showed significantly lower BG NAA, a neuronal integrity marker, and higher WM diffusivity (Table 1). Only bl-ADC was associated with lower current CD4+ count. Of the DWI measures, AD, which may reflect axonal injury, best predicted bl-ADC and decline. Inclusion of BG NAA, CD4+ count, and AD measures in the model improved bl-ADC prediction, while further inclusion of MD and RD improved prediction of decline. Conclusion: Chronically HIV-infected individuals on cART are at risk for brain injury and progressive NCI. Multimodal approaches may help us better understand HIV-associated neuropathology and help identify those at risk for NCI.

432 MULTIMODAL BRAIN IMAGING PREDICTS NEUROCOGNITIVE IMPAIRMENT IN CHRONIC HIV

Poster Abstracts

433 LESION LOAD IS ASSOCIATED WITH NEUROCOGNITIVE DEFICITS IN HIV- INFECTED INDIVIDUALS Manori De Alwis 1 , Matthew K. Schindler 1 , Anuradha Ganesan 2 , Seung Hyun Won 3 , Robert Deiss 4 , Xiuping Chu 3 , Daniel B. Hawley 4 , Brian K. Agan 3 , Stanley Rapoport 1 , Daniel S. Reich 1 , Joseph Snow 1 , Edmund C. Tramont 1 , Bryan Smith 1 , Govind Nair 1 , Avindra Nath 1 1 NIH, Bethesda, MD, USA, 2 Walter Reed National Military Medical Center, Bethesda, MD, USA, 3 Uniformed Services University of the Health Sciences, Bethesda, MD, USA, 4 Naval Medical Center San Diego, San Diego, CA, USA Background: Comorbid infections and use of illicit drugs so often accompany HIV infection and confound radiological interpretation. These confounders could be addressed by using suitable control groups, thereby parsing out any effects of HIV infection. Methods: Three cohorts were used: (i) subjects with HIV recruited at the National Institutes of Health (NIH+, n=115, 34%male, mean and SD of age 51±8 years), (ii) subjects with HIV from the Department of Defense with fewer comorbidities and less drug use (DOD+, n=83, 99%m, 43±11 yrs), and (iii) control subjects without HIV matched to the NIH+ group for socioeconomic and demographic factors (NIH-, n=44, 45%m, 49±9 yrs). Assessment by 3T MRI was done on 1mm isotropic 3D-FLAIR VISTA sequence. MD and MKS rated subjects as having mild, moderate, or severe lesion number, lesion volume, amount of confluent lesions, diffuse white matter (WM) abnormality, and ventricular enlargement. Chi-square and t-tests were used. Results: NIH+ and NIH- did not differ in socioeconomic and demographic factors other than in prevalence of drug use (38% in NIH- and 86% in NIH+, p<0.05). NIH+ subjects had more lesions (p=0.005, Fig 1A), more confluent lesions (p=0.003), and larger ventricles (p=0.02, Fig 1B) than NIH- subjects. DOD+ group was younger (p<0.001) and had lower drug use (4%, p<0.0001) than both NIH groups. DOD+ also had lower lesion numbers, lesion volumes, confluent lesions, and WM abnormality than NIH- and NIH+ groups (p=0.016 and p<0.01, respectively). In addition, DOD+ had lesser ventricular enlargement than NIH+ (p=0.0062), but not the NIH- (p=0.97). Within the NIH+ group, those with more lesions were more likely to have HIV Associated Neurocognitive Disorder (HAND, p=0.04), and those with larger ventricles had lower neurocognitive T-scores, a measure indicating lower cognitive performance (p=0.027). In the DOD+ group, too, higher lesion number was associated with lower T-scores (p<0.04). Conclusion: Both DOD+ and NIH- groups had less WM lesions and WM abnormality than NIH+, with DOD+ exhibiting the lowest lesion burden.

CROI 2018 153