CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Controls had no EFV-consistent CNS AEs while receiving EFV for ≥96 weeks after entry. Neither group had neuropsychological signs or symptoms at entry. Genotypes were imputed from Illumina genome-wide assays. Tissue-specific RNA expression levels for about 10,000 genes were inferred from genotypes using PrediXcan (Nat Genet 2015; 47:1091-8). Associations were tested with multivariable logistic regression, adjusted for CYP2B6/CYP2A6 genotype, baseline age, sex and 2 ancestry principal components. Results: A total of 2863 participants were assigned to EFV-containing ART, of whom 2171 consented for genetic testing, 1798 had CYP2B6/CYP2A6 genotypes, and 1425 had imputed genome-wide genotypes and principal components. Of these participants, 820 met criteria for cases (n=167) or controls (n= 653). CYP2B6/CYP2A6 genotype was associated with CNS AEs (p=0.001). For candidate gene expression in 10 brain regions, the lowest P-value was for PGR (progesterone receptor) in hippocampus (p=0.012), and for all genes was for RCE1 (CAAX prenyl protease 2) in cerebellum (p=5.3x10 -4 ). For candidate gene (±100kB) SNPs, the lowest P-value was for rs12393326 in HTR2C (serotonin receptor 2C, p=6.7x10 -4 ), and for all SNPs was for rs7143465 in SLC8A3 (zinc transporter 10, p=2.1x10 -9 ). Conclusion: In ART-naïve patients randomly assigned to EFV-containing ART in ACTG protocols, grade ≥2 EFV CNS AEs were not associated with predicted neurotransmitter transporter/receptor gene expression levels in brain or with SNPs in these genes, after correcting for multiple testing. We confirmed an association of CYP2B6/CYP2A6 genotype with EFV CNS AEs. Variable susceptibility to EFV CNS AEs may not be explained by brain neurotransmitter transporter/receptor genomics. 403LB POORER NEUROCOGNITIVE PERFORMANCE ASSOCIATED WITH CSF HIV DNA DESPITE LONG-TERM ART Kevin Robertson 1 , Ronald Bosch 2 , Serena S. Spudich 3 , Rajesh T. Gandhi 4 , Joshua C. Cyktor 5 , Hanna Mar 2 , Bernard J. Macatangay 5 , Christina Lalama 2 , Charles Rinaldo 5 , Ann Collier 6 , Catherine Godfrey 7 , Joseph J. Eron 1 , Deborah McMahon 5 , John W. Mellors 5 1 University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, 2 Harvard University, Boston, MA, USA, 3 Yale University, New Haven, CT, USA, 4 Brigham and Women’s Hospital, Boston, MA, USA, 5 University of Pittsburgh, Pittsburgh, PA, USA, 6 University of Washington, Seattle, WA, USA, 7 NIH, Bethesda, MD, USA Background: We have found that almost half of individuals with suppressed viremia (<50 copies/ml) on long-term antiretroviral therapy (ART) have persistence of HIV-infected cells in cerebrospinal fluid (CSF), but the neurocognitive significance of this finding is unknown. We assessed neurocognitive performance in a large cohort of individuals on long-term ART who had CSF sampling. Methods: In ACTG A5321, participants underwent lumbar puncture, blood collection and neurocognitive assessments. Participants had sustained long- term viremia suppression after initiating ART during chronic HIV infection. Cell-associated (CA) HIV DNA, unspliced HIV mRNA, and cell-free HIV RNA were quantified. Markers of inflammation (IL-6, IP-10, neopterin, MCP-1, sCD14, and sCD163) in CSF and plasma were assessed. Neurocognition was measured with a 15-test battery covering Language, Attention, Executive, Learning, Memory, Speed of Processing, and Fine Motor domains, standardized into z and deficit scores to create the neurocognitive total z score and global deficit score (GDS) as the summary neurocognitive measures. Results: The 65 participants were 97%male, 75%white, had a median age 50 years and median duration of ART 8.6 years; median current and pre-ART CD4+ cells of 696/mm 3 and 292/mm 3 , respectively. All participants had at least a high school education. The median neurocognitive total z score was 0.2 (range -1.1, 1.5), and median GDS was 0.2 (0.0, 1.3). Detectable CSF HIV DNA (46%) was significantly associated with poorer neurocognitive total z-score (p = 0.044, Wilcoxon) and GDS (p = 0.005, figure 1). This association persisted after adjusting for pre-ART, current CD4 count and age. By contrast, soluble biomarkers of immune activation in CSF were not associated with neurocognitive performance. Conclusion: Poorer neurocognitive performance despite long-term effective ART is associated with the persistence of detectable HIV-infected cells in the CSF. We do not know whether this is due to the legacy effects of HIV infection on the CNS prior to ART initiation, or alternatively that persistent HIV in the CNS is driving neurocognitive injury. No association with inflammatory biomarkers suggests that current inflammation was not driving present function, but

does not rule out prior inflammation as the underlying cause of neuronal injury. These findings underscore the importance of the CSF compartment as a reservoir with clinical significance that needs additional evaluation.

404 DETERMINANTS OF COGNITIVE FUNCTION DIFFER IN A EUROPEAN AND A KOREAN COHORT Davide De Francesco 1 , Alan Winston 2 , Jun Yong Choi 3 , Rosan van Zoest 4 , Jonathan Underwood 2 , Judith Schouten 4 , Nam Su Ku 3 , Woo Joo Kim 5 , Peter Reiss 4 , Caroline Sabin 1 1 University College London, London, UK, 2 Imperial College London, London, UK, 3 Yonsei University, Seoul, South Korea, 4 Amsterdam Institute for Global Health and Development, Amsterdam, Netherlands, 5 Korea University, Seoul, South Korea Background: HIV-associated cognitive impairment (CI) remains relevant in people living with HIV (PLWH) treated with antiretroviral therapy. However, risk factors for CI may differ in populations of PLWH of different ethnicity. We compared the prevalence and determinants of CI in a Northern European and a Korean cohort of PLWH, and assessed the ability of individual cognitive tests to discriminate between those with and without CI. Methods: Cognitive performances were assessed using a comparable battery covering 6 domains in 134 PLWH aged ≥45 years in the EU-funded COBRA study (Netherlands, UK), and 194 PLWH aged ≥18 years from the NeuroAIDS Project (Korea). Cognitive scores were standardized using population-specific normative scores and averaged into an overall score. Determinants of cognitive function were evaluated using linear regression. Factors that were associated with cognitive function in univariate analyses were selected for inclusion in a multivariable model. The discriminative ability of individual cognitive tests to detect CI, as defined by an overall score ≤0.5 standard deviations (SD) below the mean, was assessed using the area under the receiver operating characteristic curve (AUROC). Results: The 134 COBRA PLWH (mean age: 57 yrs, 93%male, 88%white ethnicity, 100% on cART) had a higher CD4 (mean (SD): 646 (214) cells/µL) and lower rate of anaemia (8.3%with haemoglobin ≤13 g/dL) compared to the 194 Korean PLWH (45 yrs, 94%male, 90% on cART, mean (SD) CD4: 481 (236) cells/ µL, 19.1%with anaemia). The prevalence of CI was 18.8% in COBRA PLWH and 18.0% in Korean PLWH (p=0.86). In COBRA, being of African descent was the main determinant of cognitive function (p<0.01) whereas in the Korean cohort anaemia (other than years of education) was the main risk factor (p=0.1, Table). The discriminative ability of CI screening was highest for tests of attention (AUROC of 0.81 to 0.84) and executive function (0.80-0.88) in COBRA PLWH and for tests of processing speed (0.73-0.80) and motor skills (AUROC=0.78) in Korean PLWH. Conclusion: Two cohorts of PLWH from different geographic regions show similar CI rates when assessed using similar cognitive tests. However, determinants of cognitive performance in the two cohorts differ considerably with ethnicity and anaemia being important determinants in one but not the other cohort. These findings suggest that differences in ethnicity and other diseases should be taken into consideration when comparing CI rates in different geographic regions.

Poster Abstracts

CROI 2018 141