The study was approved by the University of Rochester Research Subjects Review Board. Written informed consent was obtained from all study participants before enrollment into the study as per the Declaration of Helsinki.

This is a longitudinal observational study. Only cross-sectional data from the entry study visit are reported here. The longitudinal data analysis is in progress and has not been included in this report. Inclusion and exclusion criteria and all the study procedures as well as sample size justification are described in detail in our previous report (Murray et al., 2020). Briefly, inclusion criteria for PWH included, stable cART for at least 6 months prior to screening and age ≥ 18. We excluded individuals with symptomatic cardiovascular disease (CVD) such as angina, myocardial infarction, stroke or other peripheral atherosclerotic disease and uncontrolled vascular risk factors, such as diabetes mellitus and hypertension. We also excluded those with severe premorbid or comorbid psychiatric disorders, schizophrenia, bipolar disorder, and active depression. Brain infections other than HIV-1 and space-occupying brain lesions were also exclusionary. Individuals with dementia from any cause were not enrolled. For safety reasons, those with metallic implants were excluded. The PWoH population differed from PWH only in HIV status. A total of 108, PWH on cART and 111, age, sex, and Reynold’s cardiovascular risk score-matched uninfected individuals (PWoH) were enrolled.

All cognitive testing was performed by staff trained and supervised by a clinical neuropsychologist. We administered a comprehensive neuropsychological battery that included tests of Attention and Working Memory (CalCAP), Processing Speed (Symbol Digit Modalities Test and Stroop Color Naming), Executive Function (Trail making Test Parts B, Stroop Interference task), Fine Motor Skills (Grooved Pegboard, left and right hand); Verbal and visual learning (Rey Auditory Verbal Learning Test AVLT (Trials 1-5), Rey Complex Figure Test Immediate Recall), verbal and visual memory (Rey Auditory Verbal Learning Test RAVLT Delayed Recall, Rey Complex Figure Test Delayed Recall) and language (verbal fluency tests) at each visit. Premorbid intellectual functioning ability was estimated via WRAT-4 Reading at the baseline visit only.

Raw scores from each test were converted to z-scores using test manual norms. We created cognitive domain scores by averaging the z-scores of each test within each domain. A total cognitive score was calculated by summing the six cognitive domain z-scores (Attention and Working Memory, Processing Speed, Executive Function, Fine Motor Skill, Verbal and Visual Learning, Verbal and Visual Memory, and Language). We used Frascati criteria (Antinori et al., 2007) to determine HAND diagnoses for each participant.

Imaging was conducted on a Siemens 3T MAGNETOM PrismaFit whole-body MRI scanner (Erlangen, Germany; software version VE11c) equipped with a 64-channel phased array head coil. Detailed study protocols have already been reported by our group (Murray et al., 2020). Briefly, the following imaging was acquired: high-resolution 3D T1-weighted (T1w) anatomical images using a magnetization prepared rapid gradient echo (3D MPRAGE) sequence (inversion time -TI=926 ms; repetition time/echo time -TR/TE=1840/2.45 ms; image resolution = 1 × 1 × 1 mm³); 2D T2-weighted (T2w) images using a turbo spin echo sequence (TR/TE = 6000/100 ms; image resolution = 0.5 × 0.5 × 5 mm³); 3D fluid-attenuated inversion recovery (FLAIR) image (TI = 1800 ms; TR/TE = 5000/100 ms; resolution = 1 × 1 × 1 mm³); and 3D T2*-weighted images using a multi-echo gradient echo (MEGE) sequence with bi-polar readout (TR/TE 1st echo = 84/5.43 ms; number of echoes = 8; image resolution = 0.94 × 0.94 × 2.0 mm³). The T1w, T2w, T2*w and FLAIR images were used to evaluate CSVD burden by an experienced neuroradiologist using the Fazekas score (CSVD+ if Fazekas score ≥1) (Fazekas et al., 1987). Fazekas score is a qualitative neuroimaging scale used to assess the severity of WMH in the brain. Finally, VolBrain (Manjón and Coupé, 2016) was used to quantitatively measure the total WMH burden using the T1w and FLAIR images.

Approximately, 40 ml of whole blood was collected in Acid Citrate Dextrose (ACD) vacutainers, incubated at room temperature with slow shaking, and processed within 2 hours of collection. Plasma was isolated by centrifugation at 1000 X g for 10 minutes at room temperature. Plasma was aliquoted and cryopreserved at -80 °C and used to perform ELISA analyses mentioned below. 1ml of whole blood was processed by flow cytometry to measure different monocyte subsets and expression of monocyte activation markers such as c-c chemokine receptor 2 (CCR2), CD40, p-selectin glycoprotein ligand (PSGL-1), tumor necrosis factor receptor 2 (TNFR2) and tissue factor (TF) as described previously (Murray et al., 2020). Please note that we are missing data from whole blood flow cytometry analyses and ELISA analyses from 4 participants in PWoH group and 4 participants in PWH group due to instrument failure or inability to obtain blood samples.

Within two hours of the blood draw, whole blood was fixed with paraformaldehyde (PFA) followed by RBC lysis using ACK lysis buffer. The cells were then washed and stained with titrated amounts of antibodies against anti-CD14 PE (BD Biosciences # 555398; 10 μl), anti-CD16 PE Cy7 (BD Biosciences 557744; 3 μl), anti-PSGL-1 FITC (R&D Systems # FAB9961G; 1.5 μl), anti-CD40 FITC BD Biosciences. # 555598; 10 μl), anti-CCR2 FITC (R&D Systems # FAB151G; 1.5 μl), anti-TF FITC (Miltenyi Biotech Inc # REA949, 10μl) and anti-TNFR2 FITC (Miltenyi Biotech #130-107-743; 1μl) for 30 minutes at room temperature in dark. The cells were washed and acquired using Accuri C6 flow cytometer. 75,000 gated leukocytes were acquired based on forward and side scatter per tube. Data was analyzed using Flow Jo software version 10.4.2. Florescence minus one (FMO) controls such as unstained cells, cells stained with anti-CD14 and anti-CD16 were used to gate on three subsets of monocytes, classical monocytes (CD14+ CD16-), intermediate monocytes (CD14+CD16+; IMMs) and non-classical monocytes (CD14lo CD16+, NCMs). Expression of CCR2, CD40, PSGL-1, TNFR2 and TF was measured on all monocyte subsets. In depth information about the gating strategy used for data analysis is described in detail in our previous report using the same cohort (Murray et al., 2020).

Plasma samples were used to conduct ELISA assays for ICAM (# DY720-05, R and D systems, MN, USA), VCAM (# DY809, R and D systems, MN, USA), sCD14 (# DY383, R and D systems, MN, USA), CD163 (# DY1607, R and D systems, MN, USA), Osteoprotegerin (# DY805, R and D systems, MN, USA) and lipoprotein associated phospholipase A2 (LpPLA2) (# ab235643, Abcam, Cambridge, UK) as per manufacturer’s instructions.

The descriptive characteristics of study participants are shown in Table 1 . Percentages and frequencies for discrete variables as well as the means and standard deviations for continuous variables were reported. Two-way analysis of variance (ANOVA) was performed to study the effects of HIV infection, CSVD status, and their interaction, to the cognitive status of subjects. HIV status was included because it significantly impacts cognitive performance, as confirmed by our study ( Table 2 ). CSVD status was included because CSVD is associated with brain structural changes that might lead to reduced cognitive performance. Multiple linear regression models were used to examine linear associations between a continuous outcome variable and several continuous and/or categorical covariates. For example, we investigated the association between NCMs (a continuous regressor) and total cognitive scores (a continuous outcome variable), controlling for HIV infection (binary), CSVD status (binary), and their interactions, and summarized the results in Table 3 . The D’Agostino & Pearson test was performed to analyze the normality of the data. Based on the normality results, Two-sample Welch t-test (parametric) or Mann-Whitney U test (nonparametric) was used to perform two group comparisons (PWoH with CSVD group and PWH with CSVD group). Correlation analyses were done using Spearman’s rank correlation test. A p-value of less than 0.05 was considered significant. Error bars represent mean ± standard deviation. All statistical analyses were performed with GraphPad Prism 9.4.1 (GraphPad Software, Inc., San Diego, CA) and the R programming language 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria).

Anonymized data not published within this article will be made available by request from any qualified investigator.

As shown in Table 1 , the PWoH and PWH cohorts were comparable for age and sex distribution as well as Reynold’s cardiovascular risk score (RRS). PWH had significantly lower cognitive performance and a higher percentage met criteria for mild neurocognitive disorder (MND) than PWoH. PWH were more likely to have diabetes or hepatitis C compared to PWoH (10% vs. 1%). 81 of the 111 PWH and 68 of the 108 PWoH had CSVD (as defined by Fazekas score ≥ 1). Supplementary Figure 1A shows the age distribution of the study participants when they were divided based on HIV and CSVD status. CSVD- groups (PWoH 44.08 years ± 15.31 years and PWH 48.13 years ± 13.09 years) were younger compared to the CSVD+ groups (PWoH 54.75 years ± 16.17 years and PWH 54.62 years ± 10.05 years). Similar distributions were seen for RRS ( Supplementary Figure 1B ). RRS in CSVD- PWoH was 4.025 ± 4.44 and in PWH without CSVD was 4.233 ± 3.99. The RRS was higher in those with CSVD. In PWoH with CSVD it was 7.831 ± 7.173 and in PWH with CSVD it was 7.654 ± 6.751. These differences in the distribution of CSVD and RRS reflect their known association with age.

In addition to the Fazekas score, WMH lesion volume was quantified via VolBrain. The PWH group had higher WMH lesion volume (0.4274 cm³± 0.7476 cm3) as compared to the PWoH group (0.1794 cm³ ± 0.3408 cm³) Data is shown in Supplementary Figure 1C (p=0.0047, 95% CI for difference between means: 0.0772 to 0.4187). Even among individuals that had CSVD (as assessed by Fazekas score), PWH with CSVD group showed higher WMH lesion volume (0.4989 cm³ ± 0.8014 cm³) as compared to PWoH without CSVD group (0.2412 cm³ ± 0.3888 cm³). Data is shown in Supplementary Figure 1D (p=0.0156, 95% CI for difference between means: 0.0496 to 0.4657). It should be noted that values from individuals that had no WMH lesion (0 cm³) were not included in the analysis (n=26 PWH and n=16 PWoH had 0 values and were excluded). Further, data from four PWoH study participants were omitted as outliers (as determined by values more than three standard deviation from the mean).

Table 2 shows the results of two-way ANOVA to measure the effects of HIV status, CSVD status, and their interactions on cognitive scores. Being HIV positive was associated with lower cognitive scores (p=0.0146, estimate: -2.7664, 95% CI: -4.9804 to -0.5525), and this effect was most prominent and statistically significant for the total z-scores and two sub-domains (Learning and Memory). CSVD status also trended toward a negative correlation with cognitive performance especially with Executive Function ( Supplementary Table 1 ). Further it was worth noting that none of the interactions of HIV and CSVD status are significant, suggesting that the effects of these two factors to cognitive scores are mostly additive. Supplementary Figure 1E shows the total cognitive score when the cohort is categorized based on HIV and CSVD status into four groups. PWH with CSVD (-2.110 ± 4.126, n=77) had significantly reduced score as compared to PWoH without CSVD (1.781 ± 4.381, n=38, p=0.001) as well as PWoH with CSVD (1.115 ± 5.014, n=66, p=0.002). However, as shown in Supplementary Figure 1A , it should be noted that PWH with CSVD are a significantly older age group as compared to PWoH without CSVD and the effect of age on cognition may have significant impact on this comparison.

The next set of analyses were performed only in participants with CSVD (PWoH with CSVD and PWH with CSVD). It is noteworthy that these two groups have similar age and RRS distribution ( Supplementary Figure 1 ), indicating that the relative contribution of these covariates on CSVD status and cognitive impairment between both the groups is comparable. The results show that there was a significant decrease in total cognitive scores in PWH with CSVD group (-2.110 ± 4.126) as compared to PWoH with CSVD group (1.115 ± 5.014). Data is shown in Figure 1A for total cognitive score (p<0.0001, 95% CI for mean difference: -4.761 to -1.690). With the exception of attention/working memory domain, all the individual cognitive domains showed similar results ( Figures 1A–H ).

A two-way ANOVA of HIV status and CSVD and their interaction on monocyte phenotype showed that HIV status was positively associated with levels of NCMs (p=0.0473, estimate: 3.2130, 95% CI: 0.0387 to 6.3874) and there was no significant interaction between HIV status and CSVD status. Supplementary Figure 1F shows NCM levels when the cohort is categorized based on HIV and CSVD status into four groups. PWH with CSVD (14.13 ± 5.625, n=68) had significantly reduced score as compared to PWoH without CSVD (10.92 ± 4.207, n=31, p=0.0006) as well as PWoH with CSVD (11.4 ± 5.915 n=50, p=0.0012). Comparison between the two age matched CSVD+ groups exhibited that NCM levels were significantly higher in PWH with CSVD (n=68, median: 15.12 ± 6.929) as compared with PWoH without CSVD (n=50, median: 10.49 ± 5.915, p<0.0001, 95% CI of difference in median: 2.368 to 6.560, Figure 2A ). No significant differences were observed in levels of intermediate ( Supplementary Figure 2A ) or classical monocyte subsets ( Supplementary Figure 2B ). Among the different monocyte activation markers measured, NCMs in PWH with CSVD group expressed higher levels of tissue factor (mean: 37.20 ± 23.42), as compared to PWoH with CSVD group (mean: 27.97 ± 19.88, Figure 2B , p=0.0233, 95% CI for mean difference: 1.276 to 17.17). No significant differences were observed in the expression of other monocyte activation markers in either NCM or other monocyte subsets.

Two-way ANOVA analysis for the entire cohort, using soluble markers of monocyte and endothelial activation, showed that HIV status was positively associated with sCD14 (p=0.0070, estimate:0.1577, 95% CI: 0.0436 to 0.2718) while CSVD status did not show any significant association. Results from two group comparisons (PWH with CSVD vs. PWoH with CSVD) showed that PWH with CSVD participants had significantly increased levels of sCD14 [median (log₁₀ values): 6.175 ± 0.2321)] and ICAM levels [median (log₁₀ values): 4.612 ± 0.1823)] than the PWoH with CSVD group (sCD14 median: 6.00 ± 0.2409, ICAM median: 4.490 ± 0.2056). Data is shown in Figure 2C for sCD14 (p=0.0033, 95% CI of median difference: 0.04 to 0.1877) and Figure 2D for ICAM (p=0.0015, 95% CI of median difference: 0.0420 to 0.1814). Comparisons for the two groups with Osteoprotegerin, CD163, LpPLA2 and VCAM showed no statistically significant differences ( Supplementary Figures 2C–F ).

Linear regression analysis for the entire cohort showed a significant association of sCD14 with ICAM, (p<0.0001, estimate: 0.5163, 95% CI: 0.3550 to 0.6777) and with HIV status (p=0.0453, estimate:0.0632, 95% CI: 0.0013 to 0.1251). This result was further confirmed in participants with CSVD. We found that Spearman correlation between sCD14 and ICAM levels was significant (p<0.0001, r_s=0.4708, 95% CI: 0.3263 to 0.5938, Figure 2E ).

All three subsets of monocytes and as well as soluble markers of endothelial and monocyte activation were assessed against cognitive scores (total and individual domain scores) using multiple regression analysis. Results showed that HIV status had the strongest association with total cognitive performance, followed by osteoprotegerin and NCMs ( Table 3 ). Two group comparisons within CSVD+ groups showed that NCM levels and total cognitive score were inversely correlated with each other ( Figure 3A , p=0.0043, r_s = -0.2631, 95% CI: -0.4295 to -0.0794). This correlation was even stronger in PWH with CSVD group ( Figure 3B , p=0.0032, r_s = -0.3579, 95% CI: -0.5572 to -0.1197), while there was no significant correlation in PWoH with CSVD group ( Figure 3C , p=0.5025, r_s = 0.09705, 95% CI: -0.1945 to 0.3728). Table 4 shows significant correlation coefficients for NCMs and individual cognitive domains in combined CSVD+ groups and PWH with CSVD group alone. The correlations are stronger in most domains for the PWH with CSVD group compared to the combined CSVD groups.

Among other markers, osteoprotegerin showed mild negative correlation with total cognitive score only when data from PWoH with CSVD and PWH with CSVD groups were pooled together ( Figure 3D , p=0.0409, r_s=-0.1743, 95% CI: -0.3361 to -0.0023). Further, we correlated the blood markers and the total cognitive score with absolute WMH volumes, however, no significant correlations between these measures were found (data not shown). It should be noted that Fazekas scores for most of the participants with CSVD were between 1 and 2 indicating that the volume of WMH was relatively small.

Lastly, we conducted two group analysis to investigate the effect of virus associated covariates, such as CD4 T cell counts, viral loads and protease inhibitor usage on levels of NCM as well as total cognitive score, since these factors have been known to be associated with inflammation and cognition. Results indicated that there was no significant difference in the levels of NCM or the total cognitive score when PWH were categorized into two groups either based on detectable vs. undetectable viral load ( Supplementary Figure 3A , p= 0.4246, Supplementary Figure 3B , p=0.5447 respectively) or CD4 T cell counts below 500 cells/mm³ or equal to/more than 500 cells/mm³ ( Supplementary Figure 3C , p= 0.5626, Supplementary Figure 3D , p= 0.83 respectively) or PWH with or without protease inhibitor treatment ( Supplementary Figure 3E , p= 0.5737, Supplementary Figure 3F , p= 0.6764). These results suggest that the contribution of these variables on NCM levels and cognition may be minimal in our cohort.