Given these similarities and the aging HIV population, there is a heightened need to distinguish vascular from non-vascular presentations of HAND, though the standard method of classifying HAND using Frascati criteria would not be able to distinguish the two phenotypically . Together these findings have generated increased interest in vascular contributions to NCI in ART-treated PWH . Although chronic systemic inflammation and CVD have been linked to HAND, far less is known regarding the relationship between HAND and vascular mechanisms such as endothelial dysfunction that underlie the relationship between HIV, inflammation and vascular disease. Endothelial dysfunction can be assessed by measuring circulating levels of endothelial activation biomarkers . Plasma levels ICAM-1 and VCAM-1 are elevated in treated PWH relative to people without HIV 😉 and associated with inflammation , though their links to NCI in ART-treated PWH are unclear. Thus, the present study sought to examine the relative contributions of biomarkers reflective of endothelial cell activation and dysfunction and biomarkers of inflammation to empirically-derived,cannabis drying rack domain-based neurocognitive deficit profiles in a cohort of virally-suppressed PWH and HIV-seronegative adults. We hypothesized that vascular and inflammatory biomarker levels would differ across NCI profiles in the entire sample, but expected these biomarkers to be independently and most strongly associated with NCI profile among PWH. Identification and validation of NCI profiles and their underlying biological substrates may help identify subtypes of HAND that will improve our understanding of its etiology in the ART era and inform diagnostic criteria that better detect earlier signs of HAND.
Participants were 84 HIV-seropositive and 126 HIVseronegative adults enrolled in the University of California San Diego’s Translational Methamphetamine AIDS Research Center . Study visits took place between 2009 and 2020 and all participants provided written informed consent to study procedures, which were approved by the UCSD Institutional Review Board. Exclusion criteria for the parent study, TMARC, were history of psychotic disorder, HCV co-infection, or presence of a neurological condition that would confound the interpretation of neuropsychological test results and their association with HIV disease. In order to focus the current analyses exclusively on HIV, the present study also excluded TMARC participants who: 1) met criteria for a current substance use disorder ; 2) reported illicit substance use in the past 90 days , and/or; 3) had a positive urine toxicology screen for drugs or positive breathalyzer test for alcohol on the day of testing. HIV + participants were also excluded from analysis if they had HIV RNA load >200 copies/ml. HIV serostatus was determined via self-report and confirmed with a finger stick point-of-care test . Participants completed a comprehensive and validated neuropsychological assessment that encompassed an estimate of premorbid verbal IQ and seven neurocognitive domains commonly impacted by HIV . The domains and individual tests were: verbal fluency , executive function , processing speed , learning and delayed recall , working memory , and motor skills . For participants with previous test battery exposure due to prior research visits, raw scores for each test were converted to practice effect adjusted scaled scores . Scaled scores were converted to demographically-corrected T-scores that adjusted for the effects of age, education, sex, and race/ethnicity, as appropriate . T-scores were averaged across the battery and within each domain to derive global and domain-specific T-scores.
T-scores were also converted to deficit scores that give differential weight to impaired, as opposed to normal scores, on a scale ranging from 0 to 5 . Deficit scores were used for cluster analysis in order to identify neurocognitive subgroups characterized by impairment severity, and dichotomous impairment classifications were calculated to aid cluster group interpretation. T-scores were selected over deficit scores for multiple linear regression analyses , because their normal distribution better conforms to the parametric assumptions of linear regression. Plasma for biomarker assays was collected using EDTA vacuum tubes via standard phlebotomy procedures. Soluble levels of inflammatory biomarkers , and biomarkers reflective of endothelial activation and/ or vascular permeability were measured using electrochemical luminescence immunoassays . All assays were performed in duplicate. Assays were repeated when the coefficient of variation was greater than 20%. In addition, 10% of all assays were repeated to assess operator and batch consistency. To reduce the influence of batch variation, raw biomarker data was normalized on a plateby-plate basis using median value across wells. Biomarkers were adjusted for batch variation and converted to z-scores to reduce skewness and facilitate interpretation. Individual biomarker z-scores were grouped and averaged to form separate composites z-scores for primary vascular and inflammatory processes .Participants also completed structured clinical interviews to determine medical and neuropsychiatric comorbidities. Medical comorbidities were determined based on the combination of self-report and/or specific drug treatment for the condition. Characterization of cardiometabolic risk included evaluation of hypertension, hyperlipidemia, diabetes mellitus, body mass index , and smoking history . Data for central adiposity, indexed via midwaist circumference , was also available on a subset of participants and is presented for descriptive purposes.
The Composite International Diagnostic Interview ; was administered to determine DSM-IV diagnoses of current and lifetime Major Depressive Disorder and substance use disorders . The Beck Depression Inventory-II assessed for frequency and severity of depressive symptoms in the past two weeks . JMP Pro 16.0.0 was used for statistical analyses . HIV serostatus differences on demographic, cardiometabolic, neuropsychiatric, biomarker, and neurocognitive variables were examined using analysis of variance , Wilcoxon/Kruskal-Wallis tests, and Chi-square statistics as appropriate. The relationship of vascular and inflammation composite z-scores with neurocognitive profiles was examined via a two-step approach that employed a person-centered, data-reduction method followed by a confirmatory multiple regression analysis. First, a data-driven cluster analysis of domain-specific deficit scores was performed to empirically-derive homogenous sub-populations of neurocognitive deficit profiles. We applied the JMP hierarchical agglomerative clustering function and the Ward method based on sum of squares to determine the distance between the clusters. The number of clusters is selected visually based on graphical representations and by examining the distance between clusters. Since this approach is primarily descriptive and determination of the optimal number of clusters also considers sample size, parsimony, and meaningful interpretation, we also conducted discriminant function analyses to evaluate the ability of domain deficit scores to discriminate the clusters derived from the hierarchical clustering analyses. These methods are consistent with those utilized in prior studies . To maximize sample size, the cluster analysis was conducted on 247 participants, which included the 210 individuals in the present study and 37 additional TMARC participants without biomarker data. These additional 37 participants had comparable demographic and clinical backgrounds to the 210 participants with biomarker data. Cluster group differences in vascular and inflammation z-scores were then examined in the full study sample and stratified by HIV serostatus. Cohen’s d statistics are presented as estimates of effect size for pairwise comparisons. Next, after interpreting the cluster analysis,vertical grow system we selected a subset of neurocognitive domains for multiple regression analysis based on the domains most impacted in the cluster group with the highest elevations in biomarker composite z-scores. In addition to limiting the number of domains, we controlled for multiple comparisons by only conducting multiple regressions for biomarker-domain relationships that exhibited significant univariable associations with the False Discovery Rate set at 5% . The multiple regression analyses tested the interaction of HIV serostatus and biomarker composite z-scores on domain T-scores, adjusting for clinical factors that significantly differed by HIV serostatus or neurocognitive subgroups identified in the cluster analysis. The following variables were identified as having met our criterion for covariates: hypertension, hyperlipidemia, smoking status , lifetime MDD, and BDI-II. Demographic variables were not included as covariates in neurocognitive analyses given that neurocognitive T-scores already adjusted for these factors. However, regression analyses also controlled for premorbid verbal intelligence in order to account for differences in educational quality that may not be captured by test score adjustments for years of formal education .
The present study empirically-derived profiles of neurocognitive deficits in virally suppressed PWH and PWoH, and examined biomarker evidence of systemic inflammation and vascular processes as underlying risk factors contributing to these profiles. Using hierarchical cluster analyses, three distinct neurocognitive subgroups emerged: one with no to minimal deficits across all domains , one with relatively isolated and prominent deficits in learning and memory , and one with global deficits across domains but, relative to the other two groups, this group had disproportionate impairment in executive functions, processing speed, working memory, and motor skills . Notably, Dysexecutive/Slow membership was more prevalent among PWH, whereas the prevalence of Learning/Memory membership was similar across PWH and PWoH. PWH exhibited elevations in vascular biomarkers, which in turn showed a selective association with Dysexecutive/Slow group membership among PWH. This association between circulating vascular biomarkers and HIV-related neurocognitive performance was reproduced in linear regression models that controlled for traditional cardiovascular risk factors , psychiatric comorbidities , and HIV disease and treatment characteristics. Furthermore, the link between the vascular biomarker composite and the HIV-related Dysexecutive/Slow phenotype was robust to the inflammation composite, which was also elevated in PWH relative to PWoH but only weakly associated, regardless of HIV status, with worse motor performance. Collectively, these findings support evidence of ongoing inflammation and associated vascular mechanisms in the context of viral suppression , and highlight the potential unique role of endothelial dysfunction and vascular permeability as mechanisms underlying a cluster of deficits characterized by executive dysfunction, cognitive slowing, and poor working memory in virally suppressed PWH. The deficit profiles observed in our study as well as the proportions of PWH with each deficit profile are consistent with those observed in another recent study by our group that was conducted in an independent sample of older PWH and PWoH using latent class analyses . Interestingly, similar numbers and patterns of deficit profiles have been observed in other clinical samples of PWoH with similar heterogeneity of neurocognitive deficits and targeted brain regions ; mild cognitive impairment. While the number and pattern of deficit profiles may be similar across clinical populations, the risk factors underlying each profile may differ depending on the clinical population of interest and characteristics of the sample. In the current study, medical and psychiatric risk factors associated with the multidomain Dysexecutive/Slow deficit profile included HIV disease, hypertension, hyperlipidemia, increased central adiposity, and LT MDD, and with regard to HIV disease and treatment characteristics, risk factors included longer duration of HIV disease and greater cumulative lifetime exposure to ART. The increased cardiovascular disease risk coupled with greater ART exposure in the Dysexecutive/Slow group is salient given that HIV treatment, particularly PI-based regimens , has been linked to cardiometabolic dysfunction , which is consistently associated with NCI in other studies of ART-treated PWH , including a multi-domain deficit profile . The elevated central adiposity, but not BMI, in the Dysexecutive/Slow group is also consistent with a prior study demonstrating central adiposity as a stronger predictor of neurocognitive impairment than BMI in PWH . A primary objective of this study was to determine and differentiate the contributions of inflammation and vascular biomarkers to neurocognitive deficit profiles and performance in virally suppressed PWH. The inflammatory and vascular composites were strongly correlated and both were elevated in our sample of virally suppressed PWH relative to PWoH, consistent with prior studies . Significant elevations with regard to individual biomarkers were found for VCAM-1 and chemokines CXCL10/IP-10 and CCL2/MCP-1 . In concert with the higher cardiometabolic disease burden observed in the Dysexecutive/Slow group, the vascular composite was uniquely associated with the domains of executive function, working memory, and processing speed, which further supports a potential “vascular” profile of neurocognitive deficits in PWH. While deficits in executive function, working memory, and processing speed are frequent in HAND , these domains are also commonly affected in vascular cognitive impairment , and rely on the prefrontal cortex and its extensive connections with other cortical and subcortical brain regions. As such, these domains are highly vulnerable to white matter damage in PWH , consistent with imaging studies conducted in other clinical and non-clinical samples . Although our data do not directly address the molecular cross-talk between viral, immune, and vascular processes, our results provide preliminary insight into specific vascular mechanisms in ART-treated PWH that may lead to this particular profile of neurocognitive deficits. The three soluble biomarkers included in our vascular composite, VCAM-1, ICAM-1, and uPAR, were selected as indicators of endothelial dysfunction that have shown previous associations with HIV as well as inflammation, atherosclerosis and CVD .