Plasma (1→3)-β-D-Glucan and suPAR levels correlate with Neurocognitive Performance in People Living with HIV on Antiretroviral Therapy: a CHARTER Analysis

Abstract

Background:

Despite antiretroviral therapy (ART), people living with HIV (PLWH) have higher rates of non-AIDS disorders, such as neurocognitive (NC) impairment (NCI) than the general population. (1–3)-b-D-glucan (BDG) is a fungal cell wall component which serves as a biomarker for gut barrier integrity failure and microbial and fungal translocation. The primary objective of this study was to determine whether higher plasma and cerebrospinal fluid (CSF) levels of BDG were associated with NCI in PLWH.

Methods:

Paired blood and CSF samples were collected cross-sectionally from 61 male adult PLWH on ART (95% virally suppressed) who underwent a detailed NC assessment as part of the prospective CHARTER study between 2005–2015. BDG and soluble urokinase plasminogen activator receptor (suPAR) were measured in frozen blood and CSF samples while soluble CD14 (sCD14), intestinal fatty acid binding protein (IFABP) and CD4/CD8 ratio were measured in blood only. Spearman’s rho correlation analysis assessed associations between BDG, other biomarkers and NC performance.

Results:

Median BDG levels were 18 pg/mL in plasma (range: 2–60 pg/mL) and 20 pg/mL in CSF (range: 0–830 pg/mL). Higher levels of plasma BDG were associated with worse NC performance (Spearman’s rho=−0.32; p=0.013) and with presence of NCI (p=0.027). A plasma BDG cut-off of >30pg/mL was 30% sensitive and 100% specific for NCI. After adjusting for age, higher plasma SuPAR levels were also associated with worse NC performance (p<0.01). No significant associations were observed between the remaining biomarkers and the NC variables.

Conclusion:

Plasma levels of BDG and age-adjusted suPAR, may be new biomarkers for detection of NCI in PLWH on suppressive ART.

Introduction

Although antiretroviral therapy (ART) has improved survival and morbidity, people living with HIV (PLWH) have higher rates of non-AIDS disorders, such as neurocognitive (NC) impairment (NCI), than the general population (1–3). While occurrence of severe forms of HIV-associated dementia has decreased with ART, milder forms of NC disorders are still estimated to affect one-third to one-half of PLWH (3). NCI has been associated with chronic immune dysfunction, which persists in some individuals despite long term suppressive ART (4). While the exact mechanism is incompletely understood and likely multifactorial, translocation of microbial and fungal products from the gastrointestinal tract into the systemic circulation is an important driver of immune dysfunction and persistent inflammation during suppressive ART (5,6). Microbial and fungal translocation may therefore play a role in the pathogenesis of NCI during HIV-infection (5,6), and previous studies reported associations between higher biomarkers of microbial translocations (e.g. sCD14, LPS) and worse NC performance (6, 7).

(1–3)-b-D-glucan (BDG) is a cell wall component of most fungal species and is also a common component of food products (8, 9). Among PLWH and without invasive fungal infection, higher levels of BDG in serum are associated with gut barrier integrity failure and luminal contents translocation, including microbes and/or endogenous fungal flora from the gastrointestinal tract into systemic circulation, and consequently with HIV-associated immunosuppression and inflammation (6,10–12). In line with this theory, our study team recently reported that BDG exhibited a strong negative correlation with the proportion of gut Lactobacillales in the distal gut microbiome (low proportion of gut Lactobacillales is an indicator of microbial translocation), and positive correlation with sCD14, the soluble form of the receptor for lipopolysaccharide (LPS) (10). BDG levels in blood and CSF may have significant prognostic potential, predicting mortality in patients with various underlying diseases (13). In fact, previous studies found that blood BDG levels correlated with cardiovascular non-AIDS comorbidities among individuals with chronic HIV infection (6,14). In a previous pilot study, BDG levels also correlated with neurocognitive performance among 14 PLWH (15).

The soluble urokinase plasminogen activator receptor (suPAR) is another biomarker that has been correlated with inflammation, non-AIDS events and AIDS-related mortality (14,16–18). The urokinase-type plasminogen activator system consists of a proteinase (uPA), a receptor (uPAR), and inhibitors. suPAR is the soluble form of uPAR and positively correlates with the activation level of the immune system (19–21). In blood, suPAR is a marker of monocyte activation and chronic inflammation (14,16–18). Interestingly, cerebrospinal fluid (CSF) suPAR levels are higher in PLWH with AIDS dementia complex than in neurologically asymptomatic adults (22,23).

The primary objective of this study was to determine whether higher plasma and CSF levels of BDG and suPAR were associated with NCI in PLWH.

Methods

In this cross-sectional analysis, we included paired plasma and CSF samples from a cohort of 61 PLWH on ART (~50% NCI and ~50% no NCI) who underwent a NC assessment as part of the prospective CNS HIV Antiretroviral Therapy Effects Research (CHARTER) study between 2005–2015 at the University of California San Diego (UCSD). The CHARTER study was funded in 2002 to explore the changing presentation of HIV neurological complications in the context of ART (3). The UCSD Human Research Protections Program approved the study protocol, consent and all study related procedures. All study participants provided voluntary, written informed consent before any study procedures were undertaken.

Neurocognitive assessment and Comorbid Conditions

Degree of NCI at the time of sample collection was measured using an established and sensitive method to determine neurocognitive functioning among PLWH (24–26). Briefly, individuals completed a comprehensive neuropsychological test battery consistent with Frascati recommendations for NeuroAIDS research (25). Raw NC test scores were converted to demographically-adjusted T-scores and used to derive a Global T-score (higher scores=better performance). Individual T-scores were also converted to deficit scores and averaged to derive a global deficit score (GDS) which was used to classify NCI (i.e., GDS≥0.5). The GDS has been shown to detect mild, HIV-associated cognitive impairment based on assessment of multiple cognitive domains (25). NCI in PLWH may not only be caused by HIV infection but also by comorbidities. Comorbidities in CHARTER participants were therefore classified with respect to whether they should be considered incidental, contributing, or confounding as causes of NCI, as described before (3,27).

Testing of BDG and suPAR and other biomarkers

As part of CHARTER study, blood samples were prospectively tested for HIV RNA, CD4 and CD8 T-cell counts. A subset of 31 CSF samples was also tested for uPAR levels (using the Quantikine Human uPAR ELISA, R&D Systems, Minneapolis, USA; assay marked for research use only). All samples analyzed as part of this study had been stored at −80°C within 90 minutes of collection.

61 paired Plasma and CSF samples were retrospectively evaluated for BDG levels using the Fungitell® assay at the Associates of Cape Cod, Inc., research laboratories (Associates of Cape Cod, Inc, East Falmouth, USA) and for suPAR levels using suPARnostic assay (ViroGates, Copenhagen, Denmark), a CE-marked in vitro diagnostic according to the manufacturer’s instructions. Blood plasma samples were tested for soluble cluster of differentiation 14 (sCD14) levels (a marker of monocyte activation) using an enzyme-linked immunosorbent assay (ELISA; R&D Systems Inc, Minneapolis, MN, USA), and intestinal fatty acid binding protein (IFABP, marker of gut epithelial dysfunction; R&D Systems Inc, Minneapolis, MN, USA) according to the manufacturer’s instructions.

Statistical analysis

For statistical analysis SPSS 25 (SPSS Inc., Chicago, IL, USA) was used. BDG, suPAR and other biomarkers as well as Global T-scores were used primarily as continuous variables. GDS was used as binary (cutoff ≥0.5, which defines NCI), or continuous variable. When using GDS as binary variable, biomarkers and biomarker combinations were evaluated using the Wilcoxon rank-sum test and receiver operating characteristic (ROC) curve analysis for determination of area under the curve (AUC) values. Correlation between levels of BDG, levels of other biomarkers, global T-scores, and GDS were calculated using Spearman rho correlation analysis due to the non-normal distributions of GDS scores and other biomarkers. We further calculated the coefficient of variability between the suPAR and the uPAR assays. Power calculation revealed that a sample size of 61 plasma and CSF samples provides at least 80% power (alpha = 0.05) to detect an effect size 0.73 when comparing BDG or suPAR between those with NCI versus others. Univariate and multivariable binary logistic regression analyses assessed predictors of NCI. Variables with a p-value <0.2 in univariate analysis were included in the multivariable model. Variables in the final model were selected with a stepwise forward procedure. Model discrimination was assessed by the goodness-of-fit Hosmer-Lemeshow statistics. Odds ratios (ORs) and adjusted odds ratios (aOR) including 95% confidence intervals (CIs) were calculated and a p-value of <0.05 was considered statistically significant.

Results

The study cohort was comprised of 61 male and mostly white PLWH, with a median age of 48 years (range 27–63 years), that were without symptoms of opportunistic infections. Overall, 58/61 (95%) of participants had undetectable HIV RNA in blood plasma, and 59/61 (97%) undetectable HIV RNA in CSF at the time of sampling. Median CD4⁺ T-cell count was 589 cells/μL (IQR 407–776, range 69–1587 cells/μL), median CD4⁺/CD8⁺ T-cell ratio was 0.72 (IQR 0.44–1.01, range 0.11–2.00). Clinical characteristics are summarized in Table 1.

Table 1.

Demographics and Clinical Metrics of Study Participants

Characteristics	Total Subjects (N = 61)
Male sex, n (%)	61 (100)
Race, n (%)
White	32 (52)
Black	20 (33)
Hispanics	9 (15)
Age, years, median (range)	48 (27–62)
Undetectable HIV RNA plasma, n (%)	58 (95)
Undetectable HIV RNA CSF, n (%)	59 (97)
CD4/CD8 ratio, median (range)	0.72 (range 0.11–2.00).
CD4+ cell count1, cells/μl, median (range)	589 (69–1587)
Nadir CD4+ cell count1, cells/μl, median (range)	212 (3–816)
Neurocognitive Impairment (GDS>0.5), n (%)	33 (54)
Smoking Status, n (%)	48 (79)
Duration on current ART, years, median (range)	2 (0–9)
Duration on any ART, years, median (range)	5 (0–19)
ART regimens*
NNRTI-based, n (%)	23 (43)
PI-based, n (%)	25 (46)
Other regimens, n (%)	6 (11)

Legend: CSF: cerebrospinal fluid,

^*ART regiments available for 54 study participants,

NNRTI: non-nucleoside reverse transcriptase inhibitors, PI: Protease Inhibitors

Median BDG level was 18 pg/mL in plasma (range: 2–60 pg/mL) and 20 pg/mL in CSF (range: 0–830 pg/mL). Higher levels of plasma BDG were associated with lower global T Scores (Spearman’s rho=−0.32; p=0.013), and more severe cognitive impairment as measured by the GDS (Spearman r=0.35; p=0.006; Figure 1). No other significant associations were observed between the remaining biomarkers and the NC variables: CSF BDG and GDS (r=0.23), plasma suPAR and GDS (r=0.19), CSF suPAR and GDS (r=−0.022), sCD14 and GDS (r=−0.033), iFABP and GDS (r=0.057), CD4/CD8 ratio and GDS (r=−0.028). Interestingly, when plasma suPAR levels were adjusted for age they were significantly associated with lower global T scores (beta −0.317; p=0.014), and with higher GDS (beta 0.342; p=0.008).

Figure 1:

Scatterplots of correlation of plasma beta-D-glucan levels with Global Deficit Scores.

Comorbidities were classified as incidental in 32 participants, contributing in 19 participants, and confounding as causes of NCI in 10 participants. Correlation between plasma and CSF biomarkers and measures of NCI did not change after exclusion of participants with confounding comorbidities (plasma BDG level and global T-scores r=−0.33; plasma BDG and GDS r=0.31, other correlations not shown).

Individuals with NCI (i.e., GDS>0.5, n=33) had higher plasma BDG levels compared to unimpaired individuals (median 21 pg/mL, range 5–60 versus median 16 pg/mL, range 2–30; p=0.027; AUC=0.666, 95%CI 0.530–0.801) and there was also a trend towards higher CSF BDG levels in impaired individuals (p=0.083; AUC=0.631, 95%CI 0.486–0.776; Figure 2). Plasma BDG levels >30pg/mL were observed in 30% of participants with NCI but not a single participant without NCI. No significant difference was observed for other biomarkers, including suPAR, sCD14 and IFABP.

Figure 2:

Plasma and CSF beta-D-glucan levels in those with neurocognitive impairment (i.e., Global Deficit Scores>0.5) versus those without neurocognitive impairment.

In the univariate and multivariate model only plasma BDG levels remained a predictor of NCI, while all other biomarkers and variables evaluated did not (Table 2; not displayed plasma IFABP, and plasma sCD14, both p>0.8 in univariate analysis).

Table 2:

Univariate and Multivariable Binary Logistic Regression Models for Predicting Neurocognitive Impairment.

Variables for Predicting Neurocognitive Impairment (n=61)	OR	95% CI	P-value	aOR	95% CI	p value
	Univariate Model			Multivariable Model
Plasma BDG (per pg/mL)	1.066	1.010 – 1.124	0.020	1.066	1.010 – 1.124	0.020
CSF BDG (per pg/mL)	1.003	0.966 – 1.010	0.383
Plasma SuPAR	1.165	0.862 – 1.574	0.321
Current CD4	0.999	0.997 – 1.001	0.290
Current CD4/CD8 ratio	1.587	0.402 – 6.262	0.509
Nadir CD4 count	0.997	0.994 – 1.000	0.083	-	-	n.s.
Months of ART exposure	1.004	0.995 – 1.013	0.341
Age (per year)	1.041	0.977 – 1.110	0.213
Higher education category	1.046	0.865 – 1.264	0.644
Current Smoker	0.627	0.195 – 2.019	0.434

Legend: * Chi square 6.731; p=0.566 Hosmer Lemeshow; Forward Wald Binary Logistic Regression

Abbreviation: OR=odds ratio; aOR=adjusted odds ratio, CSF: Cerebrospinal fluid, n.s.: not significant

Plasma levels of BDG correlated significantly with plasma suPAR levels (rho=0.31, p=0.016), but not with other biomarkers.

CSF suPAR levels determined as part of this study correlated significantly with CSF uPAR levels determined at the time of sample collection as part of the CHARTER study with an assay marked for research use only (Spearman rho=0.795; p<0.001; n=31). The coefficient of variation between the two assays (log10 values used for the research only assay) was mean 0.76, SD 0.23.

Discussion

Microbial imbalances in the gut likely play an important role in the pathogenesis of NCI (28–30). Translocation of bacterial and fungal products is one of many drivers of systemic inflammation and might contribute to HIV disease progression (31,32), as well as to NCI (33–35). As part of this study we measured novel biomarkers of microbial translocation and immune activation in blood and CSF in a cohort of 61 PLWH on ART who underwent a detailed battery of NC assessments. We also compared performance for these novel biomarkers with established markers of monocyte activation (sCD14), immune dysfunction (CD4/CD8 ratio) and gut leakage (IFAB).

We found that in our cohort of chronically infected PLWH on suppressive ART, median levels of BDG in blood plasma (18 pg/ml) were similar to a previous study with a similar cohort of PLWH where the level of BDG was 15 pg/ml (15). Higher BDG levels (median 66 pg/ml) were reported from another study where PLWH began ART very early during infection and also had shorter follow up times (10). Overall, the levels in our cohort were within the range of BDG levels previously reported for healthy subjects (36, 37), indicating that blood BDG levels may not consistently differentiate between those with and those without HIV infection. Nevertheless, an increasing number of studies supports BDG as a biomarker for microbial translocation and immune activation among those with HIV infection (6, 10–12, 14, 15, 35). In contrast, the levels of BDG in CSF in our cohort were higher as compared to the cohort that received early ART (20 pg/ml compared to 5pg/ml) (10). Subsequently, we found that elevated levels of BDG in blood plasma (but not in CSF) were significantly associated with worse NC performance in chronically-infected PLWH on suppressive ART. This is in line with a previous smaller study from our group, showing a similar association in a group of 14 PLWH who started ART during early HIV-infection (15). Also, higher plasma suPAR levels were significantly associated with worse NC performance, but only after adjusting for age. Interestingly, in our cohort none of the other measured biomarkers (i.e. sCD14, IFABP, CD4/CD8 ratio) were associated with NCI, which is in contrast with other studies (38). These discrepancies might be because of differences in study populations, stage of HIV disease and treatment status. Other limitations in our current study include the limited sample size and the cross-sectional design which may have biased our analysis and limited our ability to detect associations between NC performance and some of our other measured biomarkers.

Despite the mentioned limitations, our study confirms previous findings that elevated plasma levels of BDG may be an indicator of gut barrier integrity failure and an independent biomarker associated with NC performance in PLWH on suppressive ART. The same may be true for plasma suPAR levels, once adjusted for age.

Identification of biomarkers (or combination of biomarkers) that are sensitive to HIV-associated NC performance are essential avenues of future research. While our findings might have clinical implications, future research is needed to validate our findings in an independent cohort. Incorporation of independent measurements of HIV-associated brain dysfunction (e.g., neuroimaging or neuropathological data) may provide further support of associations between BDG, suPAR and NC performance. Future research is also needed to determine associations between biomarkers and the profile and pattern of neurocognitive performance, which is not feasible using dichotomous impairment outcome variables. Finally, longitudinal studies are needed to determine the prognostic and clinical utility of biomarkers such as BDG and suPAR in identifying those who may be at risk for subsequent neurocognitive decline.