Relationship of plasma cytokines and clinical biomarkers to memory performance in HIV

Abstract

Chronic systemic immune activation and inflammatory processes have been linked to brain dysfunction in medically stable HIV-infected people. We investigated the association between verbal memory performance and plasma concentrations of 13 cytokines measured using multiplexed bead array immunoassay in 74 HIV-seropositive individuals and 50 HIV-seronegative controls. Memory performance was positively related to levels of IL-8 and IFN-γ, and negatively related to IL-10 and IL-18 and to hepatitis C infection. Memory performance was not significantly related to HIV disease markers. The results indicate the importance of systemic immune and inflammatory markers to neurocognitive function in chronic and stable HIV disease.

INTRODUCTION

The widespread availability and use of combination antiretroviral therapy (CART) over the past decade has resulted in dramatic reductions in mortality (Yeni, 2006) and improved cognitive status among many HIV-infected individuals (Cohen et al., 2001, Robinson-Papp et al., 2009). Yet, cognitive impairments remain prevalent affecting up to 50% of HIV-infected individuals in the United States (Heaton et al., 2011, Schouten et al., 2011), even among those with good HIV control on CART (Cysique and Brew, 2011, Garvey et al., 2011). The cognitive domains most commonly affected by HIV include psychomotor speed, attention, and executive functions (Grant, 2008). Memory, particularly impaired learning and retrieval of new information, is also affected (Woods et al., 2009). This pattern of memory performance is typically suggestive of disruption of subcortical memory and executive circuits rather than hippocampal involvement, which typically produces an amnestic pattern characterized by rapid forgetting as in Alzheimer’s disease (Dickerson and Eichenbaum, 2010). The reasons for persistence of cognitive dysfunction in the CART era are not well understood. One possibility relates to adverse cerebral neuronal impact of altered cytokine expression due to chronic infection-related inflammation and immune activation (Cohen et al., 2011, Langford and Masliah, 2001, Si et al., 2002, Sui et al., 2003, Woods et al., 2006).

The cerebral pathophysiology of HIV supports a link between inflammatory cytokines and cognitive dysfunction in HIV. HIV enters the brain soon after infection, triggering inflammatory responses associated with microglial activation and associated release of neurotoxic pro-inflammatory cytokines and chemokines (Gonzalez-Scarano and Martin-Garcia, 2005, Gisolf et al., 2000, Rostasy et al., 2000, Shah and Kumar, 2010, Shah et al., 2011a, Shah et al., 2011b, Silverstein et al., 2012), which are thought to contribute importantly to neuronal injury and protection in HIV (He et al., 1997, Kaul and Lipton, 1999, Merrill and Chen, 1991, Potula et al., 2004, Kim et al., 2004, Okamoto et al., 2003, Gorg et al., 2006, Krebs et al., 2000, Thompson et al., 2001, Minagar et al., 2002, Albright and Gonzalez-Scarano, 2004, Barber et al., 2004, Nukuna et al., 2004, Poluektova et al., 2004, Cartier et al., 2005, Sas et al., 2009, Guyon et al., 2008, Lewis et al., 2008, Chiodi, 2006, Sabri et al., 2003, Gray et al., 1996). Studies using magnetic resonance spectroscopy (MRS) provide evidence of adverse inflammatory impact on brain metabolism in individuals with HIV (Chang et al., 2004, Paul et al., 2008, Paul et al., 2007), particularly in basal ganglia and frontal brain regions (Cohen et al., 2010, Paul et al., 2008). These metabolic changes in frontal-subcortical circuits align with the cognitive profile of prominent attention and executive dysfunction in HIV and the subcortical pattern of memory difficulties characterized by encoding and retrieval impairments.

Our group (Cohen et al., 2011) recently demonstrated direct evidence that plasma cytokine levels are associated with attention and executive function in individuals with HIV infection. Further, our results showed these cytokines were retained in statistical models predicting dysfunction in these cognitive domains even when HIV clinical markers such as HIV RNA level and duration of illness were included in the statistical model. These results, viewed in the context of aforementioned MRS studies, raise the possibility that altered cytokine expression might also contribute to HIV-related memory impairment.

We addressed this possibility by examining the association between plasma cytokine levels and verbal memory functioning in individuals with and without HIV. We hypothesized that abnormal plasma cytokine levels are associated with reduced learning and memory performance even among HIV individuals with adequately reconstituted immune systems.

Our goal was to determine if systemic inflammatory and immunological processes influence learning and memory impairments, and to identify potential cytokine biomarkers that may be useful for assessing HIV-associated neurocognitive impairment. Understanding the role of cytokines in HIV-related cognitive dysfunction, particularly if the relationship can be detected using plasma cytokine levels would provide researchers and clinicians with an easily accessed and inexpensive biomarker for predicting and monitoring cognitive dysfunction in HV. This information could be very useful helping individual patients plan for their long-term care and in assessing the efficacy of new treatments to reduce cognitive decline in HIV.

METHODS

Participants

Participants were 124 adults including 74 HIV-infected (HIV+) and 50 HIV seronegative (HIV−) individuals. Participants were between 24 and 79 years of age (mean = 46.22 ± 11.65 years). There were 78 men and 46 women. Participants were recruited between 2007 and 2010 as part of an NIH-sponsored study of HIV-associated brain dysfunction at The Miriam Hospital / Brown University. The Institutional Review Boards at Miriam Hospital and Brown University approved the study, and informed consent was obtained from each participant prior to enrollment. Exclusion criteria included 1) history of head injury with loss of consciousness > 10 minutes; 2) neurological conditions including dementia unrelated to HIV, seizure disorder, stroke, and opportunistic infection of the brain; 3) severe psychiatric illness that may impact brain function, e.g., schizophrenia; and 4) substance use disorder within 6 months prior to neuroimaging.

HIV serostatus was documented by ELISA and confirmed by Western blot test. A significant number of participants including 24 HIV+ (32%) and 5 HIV− (10%) individuals had current hepatitis C infection (HCV+), defined as detectable serum HCV RNA by PCR. Viral load (HIV RNA by PCR) was classified as detectable or undetectable based on a lower limit of detection of 75 copies/ml. Most HIV+ participants (67.6%) had undetectable viral loads (n = 67), and 83.3% were on stable CART. Nadir CD4 count was available on 71 of the 74 HIV+ participants. Of these more than half (61%) reported nadir CD4 < 200 and average nadir CD4 was 174.7 ± 157.3 cells/μl indicating a history of significant immune suppression. Two of the individuals without nadir CD4 were also missing duration of HIV infection and we were unable to assess CART status. Duration of HIV infection (n=72) ranged from 1 to 26 years. Current CD4 was unavailable on 6 HIV participants including 3 with missing nadir CD4. Most (70.6%) HIV-infected subjects had reconstituted immune function as indicated by current CD4 counts above 350 cells/μl (n = 68). Demographic characteristics stratified by HIV and HCV status are presented in Table 1. Age and group sex composition did not differ significantly by HIV status (age, p = .969; sex, p = .479) or by current HCV status (age, p = .472; sex, p = .481). The HIV−/HCV− group had significantly more years of education compared to the HIV+/HIV− group (p = .001).

Table 1.

Demographic and clinical characteristics of study participants.

	HIV−		HIV+
	HCV−	HCV+	HCV−	HCV+
N	45	5	50	24
Age (years)	46.67 (14.69)	40.2 (10.03)	45 (10.16)	49.17 (7.39)
Education (years)	14.22 (3.46)	13.00 (2.35)	13.06 (2.14)	11.67 (1.66)
Sex (% male)	27 (60%)	3 (60%)	34 (68%)	14 (58%)
Ethnicity (% Caucasian)	35 (78%)	3 (60%)	32 (64%)	9 (38%)
Current CD4 (cells/µl)			528.83 (254.69)a	471.81 (338.04)b
Nadir CD4 (cells/µl)			187.15 (166.88)c	148.7 (134.91)d
Duration of HIV infection (years)			12.1 (6.45)e	16.5 (6.72)f
Undetectable plasma HIV RNA (%)			33 (73%)g	17 (77%)h
CART-treated (%)			42 (88%)i	21 (88%)j

Note: Continuous variables are reported as mean (SD). Proportions are reported as N (%). Analyses involving HIV clinical variables were conducted only in cases with complete data for all variables

^an = 47,

^bn = 21,

^cn = 48,

^dn = 23,

^en = 48,

^fn = 24,

^gn = 45,

^hn = 22,

ⁱn = 48,

^jn = 24.

Cytokine Measurement

Plasma was separated from each participant’s blood sample and immediately frozen and stored at −80°C. Chemokine and cytokine levels were measured using an xMAP multiplexed bead array immunoassay with fluorescence intensity measured using the Luminex-100 system (Luminex Corp., Austin, TX). Samples were diluted 1:5 in reaction buffer according to the manufacturer's protocol. All samples were treated identically and all assays were performed simultaneously using the same platforms and reagents. The panel of cytokines used along with their proposed functions is listed in Table 2. All cytokine assay results fell within the linear dynamic range of the standards (fluorescent light units per milliliter=FLU/ml). However, the data were natural log transformed to reduce variance and outliers within the groups.

Table 2.

Descriptions of cytokines and chemokines measured in this study¹

Cytokine/Chemokine	Abbreviation	Function
Interferon gamma-soluble cytokine	IFN-γ	Produced by innate NK cells, acquired antigen-specific cytotoxic CD4+ and effector CD8+ T cells. Activates macrophages and critical for innate and adaptive immune responses to intracellular pathogens, tumor control, and inhibition of viral replication.
Interleukin-1-Beta	IL-1β	Produced by activated macrophages; mediates inflammatory responses, cell proliferation, apoptosis. Induces Cox-2 in CNS, causing inflammatory pain
Interleukin-6	IL-6	Secreted by T cells and macrophages; triggers inflammation, acute phase response, fever. Anti-inflammatory effects include inhibiting TNF-α and IL-1, and activating IL-1ra and IL-10.
Interleukin-8	IL-8	Made by macrophages and some epithelial and endothelial cells; Role in innate immune response. Major role in chemotaxis of neutrophils. Also mediates inflammatory response and angiogenesis.
Interleukin-10	IL-10	Produced by monocytes. Pleiotropic cytokine. As an anti-inflammatory cytokine, it inhibits macrophage and dendritic cell function, suppresses TNF-α. Acquires pro-inflammatory activity during immune response with IFN-α stimulation.
Interleukin-16	IL-16	Secreted by lymphocytes. Pleiotropic cytokine. Functions as a chemoattractant (CD4+ cells), modulates T cell activation, and inhibits HIV replication.
Interleukin-18	IL-18	Produced by macrophages and monocytes. Pro-inflammatory cytokine interacts with IL-12 to induce cell-mediated immune response with microbial infection and LPS, inducing severe inflammatory reactions. Stimulates NK and T cell release of IFN-γ, which activates macrophages. Inhibits IL4-dependent IgE, enhances B cell production.
Interferon-inducible protein-10	IP-10	Produced by various cell types including monocytes, endothelial cells, fibroblasts, keratinocytes. Induced by IFN-γ and TNF-α. Chemoattractant for activated T cells.
Monocyte chemoattractant protein-1	MCP-1	Expressed in monocytes, vascular endothelial cells, smooth muscle cells. CCL2 chemokine, induces monocyte attraction, and degranulation of basophils with histamine release. Induced by IL-1, TNF-α, PDGF, TGF-β, and LIF
Macrophage inflammatory protein-1-beta	MIP-1β	Produced by macrophages. CCL4 chemokine that generates local inflammatory responses, induces superoxide production by neutrophils. Chemotactic activity for lymphocytes, macrophages, NK cells, and monocytes with inflammation; down-regulates CCR5, inhibiting HIV-1 blocking.
Stromal cell-derived factor-1-alpha	SDF-1α	Expressed ubiquitously, except in blood cells. Small cytokine member of CXCL12 family of chemokines. Activates leukocytes due to strong chemotactic effects. Induced by pro-inflammatory stimuli, e.g. TNF-α and IL-1β.
Tumor Necrosis factor-alpha	TNF-α	Secreted by macrophages, monocytes, neutrophils, T cells, NK cells after stimulation with LPS. CD4+ cells secrete TNF-α. Also made by astrocytes, microglial cells, smooth muscle cells, and fibroblasts. Mediates systemic inflammation, inhibits viral replication, and inhibits tumorigenesis.
Tumor necrosis factor related apoptosis-inducing ligand	TRAIL	Expressed broadly in tissues. Cytokine induces proapoptotic caspase activity by up-regulating pro-apoptotic Bcl proteins. Causes apoptosis in hepatocytes, neural cells, and thymocytes

¹Reproduced from Cohen et al. (2011)

Bivariate correlations were used to determine if participant demographics were correlated with cytokine levels in the entire sample. Age did not correlate significantly with any of the 13 cytokines (maximum r = .112, p = .216). Education was significantly and negatively associated with IL-8 only (r = -.193, p = .032) but this did not survive correction for multiple comparisons.

Neuropsychological Measures

All participants completed the Hopkins Verbal Learning Test –Revised (HVLT-R) (Benedict et al., 1998), a widely used standardized test with extensive norms and strong reliability and validity. The HVLT-R is a test of verbal memory involving immediate recall of a 12-item word list. In immediate recall phase, the examiner reads the aloud, after which the examinee calls out as many words as he or she can recall in any order. Three trials are administered. The delayed recall phase starts 20 minutes later. Examinees are asked to recall the word list. This free recall trial is followed by a forced- choice recognition recall trial. Scores used in this analysis are (1) immediate recall: the total number of recalled words summed across the three learning trails and (2) delayed recall: the number of words recalled after the 20-minute delay. The HVLT-R immediate and delayed memory scores have demonstrated sensitivity to HIV-related verbal memory impairment (Woods et al., 2005). The immediate and delayed recall raw scores were transformed into age-corrected T-scores using the norms provided in the published test manual. T-scores were used as dependent measures in subsequent analyses. T-scores have a mean of 50 and standard deviation of 10. For the purpose of the current analysis, higher T-scores indicate better performance.

Statistical Analysis

All statistical analysis was performed in R, version 2.12.1 (http://www.rproject.org) and SPSS version 20 (http://www-01.ibm.com/software/analytics/spss/). To examine HIV+/− group differences in plasma concentrations of the 13 cytokines, we used using multivariate ANCOVA covarying for current HCV status with follow-up univariate ANOVAs.

To examine the relationship between plasma cytokine concentrations and performance on the two memory measures (HVLT-R Immediate and Delayed recall T scores) , we used the Akaike Information Criterion (AIC) (Akaike, 1974, Burnham and Anderson, 2002) procedure. The procedure is a linear regression model-selection algorithm that balances model fit with complexity to identify a parsimonious set of independent variables that predict a dependent variable. Four model-selection procedures were run. The first two were run using the whole sample. Independent variables for these models were HIV and HCV status, and all 13 cytokines as independent variables. Dependent variables were HVLT-R Immediate Recall in the first model and Delayed Recall in the second. The next two model-selection procedures were run in the HIV+ group only. Independent variables were CD4 nadir (log transform), current CD4 count, HIV RNA level, duration of HIV infection, and CART status as a dichotomous variable (i.e., yes/no). Dependent measures were HVLT-R Immediate and Delayed Recall scores.

RESULTS

HIV effects on cytokine concentrations

Plasma cytokine levels by group stratified by HIV and HCV status are presented in Table 3 and graphically presented in Figure 1. We used MANCOVA to assess the impact of HIV status on cytokine concentration incorporating current HCV status as a covariate. Neither age nor education was included in this analysis due to the absence of significant associations between them and cytokine levels. Results showed a significant main effect of HIV (Pillai’s trace = 0.285, F(13, 109) = 3.340, p < .001, partial η² = .285) and HCV status (Pillai’s trace = 0.378, F(13, 109) = 5.100, p < .001, η² = .378) on cytokine concentrations. Follow-up univariate analyses of the effect of HIV status, controlled for HCV status, showed significant HIV group differences on 4 of the 13 cytokines. Compared to the HIV- group, the HIV+ group showed significantly higher levels of IL-8 (p = .003), IL-18 (p = .002), IP-10 (p < .001), and significantly lower SDF-1α (p = .025). Statistical trends occurred toward higher MIP-1β and lower IL-6 (p = .061) in the HIV+ group.

Table 3.

Plasma cytokine levels stratified by HIV and HCV status. Values are log transformed FLU/ml displayed as mean (SD); (95% confidence interval).

	HIV−		HIV+
Cytokine	HCV− n = 45	HCV+ n = 5	HCV− n = 50	HCV+ n = 24
IFN-γ	1.57 (0.78) (1.33 – 1.80)	1.58 (0.15) 1.39 – 1.77	1.44 (0.51) 1.30 – 1.59	1.37 (0.53) 1.15 – 1.59
IL-1β	1.85 (0.48) (1.71 – 1.99)	2.07 (0.46) (1.50 – 2.64)	1.99 (0.49) (1.85 – 2.13)	2.00 (0.52) (1.79 – 2.22)
IL-6	2.13 (0.62) (1.94 – 2.32)	2.37 (0.5) (1.76 – 2.99)	2.19 (0.69) (1.99 – 2.38)	2.34 (0.64) (2.07 – 2.61)
IL-8	2.79 (0.45) (2.65 – 2.92)	3.13 (0.5) (2.50 – 3.75)	3.11 (0.55) (2.95 – 3.27)	3.33 (0.62)* (3.07 – 3.59)
IL-10	2.86 (0.37) (2.75 – 2.97)	3.3 (0.48) (2.70 – 3.90)	2.87 (0.26) (2.80 – 2.95)	3.03 (0.4) (2.86 – 3.20)
IL-16	4.2 (0.54) (4.04 – 4.36)	4.14 (0.77) (3.19 – 5.10)	3.99 (0.47) (3.86 – 4.12)	4.1 (0.51) (3.88 – 4.31)
IL-18	3.96 (0.65) (3.76 – 4.15)	4.22 (0.65) (3.41 – 5.03)	4.44 (0.86) (4.20 – 4.69)	4.62 (0.91)* (4.23 – 5.00)
IP-10	5.04 (0.83) (4.79 – 5.29)	5.97 (1.29) (4.37 – 7.57)	5.61 (1.05) (5.31 – 5.91)	7.21 (0.82)** (6.87 – 7.56)
MCP-1	4.85 (0.70) (4.64 – 5.06)	5.41 (0.59) (4.67 – 6.15)	4.95 (0.73) (4.74 – 5.16)	4.66 (0.75) (4.34 – 4.98)
MIP-1β	3.65 (0.82) (3.41 – 3.90)	4.37 (1.11) (2.99 – 5.75)	3.93 (0.78) (3.71 – 4.15)	4.8 (0.89)* (4.42 – 5.17)
SDF-1α	2.99 (0.6) (2.81 – 3.17)	2.86 (0.49) (2.25 – 3.46)	2.77 (0.38) (2.66 – 2.87)	2.76 (0.37) (2.61 – 2.92)
TNF-α	1.81 (0.36) (1.70 – 1.92)	2.02 (0.33) (1.61 – 2.44)	1.88 (0.43) (1.75 – 2.00)	1.94 (0.41) (1.76 – 2.11)
TRAIL	3.43 (0.50) (3.28 – 3.58)	3.63 (0.33) (3.21 – 4.03)	3.5 (0.46) (3.37 – 3.62)	3.58 (0.55) (3.35 – 3.81)

^*mean value is greater than 1 standard deviation discrepant from mean value of HIV−/HCV− group.

^**mean value is greater than 2 standard deviations discrepant from mean value of HIV−/HCV− group.

Figure 1.

Mean cytokine levels, expressed as log FLU/ml, in the three clinical groups (light grey), displayed in comparison to the 95% confidence intervals of cytokine levels in the control group (HIV−/HCV−; dark grey).

Cytokines and cognitive function

Table 4 shows results of a linear regression model-selection algorithm run in the entire sample and in the HIV+ group only to identify parsimonious models predicting cognitive status as described in the Methods section. We did not include age or sex because the groups did not differ on these variables; moreover, memory T-scores already include a correction for age. We did not include education in these models because correlations between education and cytokine levels were non-significant after correction for multiple tests. Accordingly, education is unlikely to moderate an association between cytokines and memory.

Table 4.

Regression coefficients and associated p-values for clinical and cytokine markers included in the final model selected by minimizing AIC for each HVLT-R Immediate and Delayed Recall scores in the whole group and in the HIV+ group only. Rows indicate the predictors under consideration, and each final regression model is represented by one column of the table. Statistically significant covariates are highlighted in bold. Adjusted R² values of the final models are included in the final rows.

	Whole Group		HIV+ Only
	HVLT-R Immediate Recall	HVLT-R Delayed Recall	HVLT-R Immediate Recall	HVLT-R Delayed Recall
HIV+ status		−6.622 (0.015)
Current CD4
HIV RNA level			−3.815 (0.148)	−7.031 (0.057)
CART+				−8.021 (0.084)
HIV duration
CD4 nadir
Current HCV+		−6.515 (0.034)	−6.019 (0.020)	−10.194 (0.007)
IFN-γ				10.506 (0.005)
IL1-β
IL6
IL8		5.353 (0.048)	5.274 (0.027)	−8.444 (0.008)
IL10			−10.699 (0.036)	−8.177 (0.118)
IL16
IL18				−3.314 (0.050)
IP10	−2.640 (0.002)			2.040 (0.186)
MCP1	3.658 (0.011)
MIP1-β		2.900 (0.134)
SDF1-α
TNF-α			6.323 (0.093)
TRAIL
ICV
Model Adjusted R2	0.086	0.132	0.213	0.383

Note: Cell values are unstandardized regression coefficient (p-value).

Whole group

In the entire sample, predictor variables included the 13 cytokines, HIV status, and HCV status, and education.

HVLT-R Immediate Recall was positively associated MCP-1 (p = .011) and negatively associated with IP-10 (p = .002).

HVLT-R Delayed Recall was positively associated with IL-8 (p = .048) and negatively associated with both HIV and current HCV infections (p = .015, p = .034, respectively).

HIV+ only

Predictor variables in the regression models for the HIV+ group only included the 13 cytokines, HCV status, the following HIV clinical variables: CD4 nadir (log transformed), current CD4 count, HIV RNA load (detectable vs. not detectable), HIV duration, and CART status (yes/no).

HVLT-R Immediate Recall was positively associated with IL-8 (p = .027), and negatively associated with IL-10 (p = .036) and HCV infection (p = .020). There was a statistical trend toward positive association with TNF-α (p = .093).

HVLT-R Delayed Recall was positively associated with IFN-γ (p = .005) and IL-8 (p = .008) and negatively associated with IL-18 (p = .050) and current HCV infection (p = .007). There were trends toward negative associations with detectable HIV RNA level (p = .057) and CART status (p = .084).

DISCUSSION

This study examined the associations between verbal learning and memory and a panel of plasma cytokine measures and clinical markers in the context of HIV infection. We used a model-selection procedure to identify which cytokines are most relevant to learning and memory functions. Our main finding is that a number of cytokines are significantly associated with verbal memory ability in patients with HIV. Higher levels of IL-8 and IFN-γ were associated with better performance on memory measures. In contrast, higher IL-10 and IL-18 levels, and HCV infection were related to poorer memory performance.

There was an additional positive statistical trend between TNF-α and delayed recall. In contrast, traditional clinical markers of HIV status including CD4 nadir and CD4 count were not retained in the final statistical models. There were statistical trends toward poorer delayed verbal recall among individuals with detectable HIV RNA viral load and among those on CART. We also found an additional adverse impact of HCV status on verbal memory in this group. Age was not significantly associated with cytokines. Education was also not significantly associated with cytokines after correction for multiple correlations.

In the combined group of HIV+ and HIV− individuals, immediate verbal memory was associated with MCP-1 and IP-10 and delayed recall was associated with IL-8. HIV infection negatively impacted delayed verbal memory only. Current HCV infection impacted delayed, but not immediate verbal memory.

In this HIV+ group, none of the clinical HIV variables were significantly associated with memory performance. Of these clinical variables, only HIV RNA level and CART status showed trend-level associations with delayed verbal memory. These results provide evidence that a group of inflammatory cytokines measured in plasma predict memory performance in a group of HIV patients with well-reconstituted immune systems. The results also indicate that cytokines remain significant predictors of memory status even when standard clinical HIV markers were considered simultaneously in the statistical model.

These results extend prior work in this area and advance understanding of the impact of cytokine concentrations on neurocognitive functioning in HIV-infected individuals (Cohen et al., 2011). Our prior paper reported significant associations between a subset of the same 13 cytokines and performances on tests of attention and executive function in HIV-infected individuals. The total number of cytokines showing significant associations in that analysis was greater than the number significantly associated with memory in the current analysis suggesting fewer cytokines impact memory function in HIV than impact attention and executive function in HIV. The prior study found the associations were more common for the interleukins. The current results align with this finding in that three of the four cytokines significantly associated with memory function in the HIV+ group were interleukins. At least one of the interleukins (IL-1β) has been shown to mediate hippocampal-based memory function although the precise mechanism of mediation is not understood (Huang and Sheng, 2010) nor is it clear that this effect of IL-1β extends to other interleukins.

Our results show robust associations between HCV infection and verbal memory. This is not unexpected given the prior work of Huckans et al. (2009). Nearly all of those in our sample with HCV infection were also infected with HIV (45 co-infected, 5 mono-infected with HCV), which raises some question about the independence of the effect of HCV on memory in the absence of HIV. Unfortunately, our sample of people with HCV who were not also infected with HIV is too small to investigate this statistically.

HIV RNA viral load and CART treatment were the only HIV clinical variables to enter any AIC model. This occurred only for delayed verbal recall. The association was in the expected direction for HIV RNA level but was in the opposite direction for CART. This CART finding is not entirely unexpected in that others have reported decreased cognitive function in the presence of CART (Heaton et al., 2011, Robertson et al., 2010, Schouten et al., 2011).

Our results show that the association between plasma cytokine concentrations and memory functioning occurs partly independently of traditional clinical markers of HIV-infection status and demographic factors. Statistical analyses in the HIV+ group retained cytokines in the most parsimonious models predicting performance on a widely used verbal memory measure even when HCV status and HIV clinical factors were included as potential predictors. This result aligns with previous recent reports showing that cognitive impairment was not associated with HIV clinical markers in patients with good HIV control (Cysique and Brew, 2011, Garvey et al., 2011). Our result, considered with those of Cysique et al. and Garvey et al. raise the possibility that, compared to plasma cytokine levels, traditional clinical HIV markers may be less robust predictors of neurocognitive status in HIV patients with well-reconstituted immune status.

Interestingly, cytokines were associated with immediate verbal recall in the entire sample (HIV+ and HIV− groups combined), even when HIV status itself was not retained in the final statistical model. This latter finding tentatively suggests that cytokines influence memory performance even in the absence of HIV. Alternatively, it is possible that the association between cytokines and immediate verbal recall is a function of our specific sample of HIV− individuals. Clearly the influence of cytokines on memory independent of HIV infection requires additional study in a larger sample.

Cytokine concentrations differed as both a function of HIV and HCV status suggesting that both HIV and HCV infection affects systemic inflammatory cytokine production. Alterations of serum and CSF cytokine concentrations (e.g., MCP-1, IL-6, TNF-α) among people infected with HIV is well documented in past studies (Schoeniger-Skinner et al., 2007, Persidsky et al., 1997, Lokensgard et al., 1997, Laurenzi et al., 1990), with elevated concentrations of specific concentrations associated with increased neurovirulence (Na et al., 2011), pro-inflammatory response in the brain (Alcendor et al., 2012, Feuerstein et al., 1994, Sawada et al., 2006), and reduced cognitive functioning (Cohen et al., 2011). Although they have received less attention, cytokine disturbances also occur secondary to HCV (Nishitsuji et al., 2013, Zhang et al., 2011, Lemmers et al., 2009, Hung et al., 2009, Huang et al., 1999). The current findings suggest that HIV and HCV may differentially affect certain cytokines, but also have similar effects on others. People with HIV had elevated concentrations of IL-8, IL-18, IP-10, and reduced concentrations of SDF-1α. HCV co-infection was associated with over a two standard deviation increase in IL-10 and one standard deviation increases in IL-8, IL-18, and MCP-1. The confidence intervals provided in Table 3 provide reference ranges for evaluating the degree to which cytokines concentrations within individual patients infected with HIV with or without HCV co-infection are abnormal relative to healthy people without either HIV or HCV. These results are not intended for clinical application at this time. This will eventually require validation studies to determine the consistency of observed cytokine concentrations measured across different laboratories and accounting for use of equipment and reagents produced by different vendors. The current preliminary findings show that a cytokine panel such as the one employed in the current study can provide information about the concentrations of particular cytokines relative to normative values, which can eventually be clinically useful in detecting individuals with significant systemic inflammation secondary to HIV and HCV.

In sum, our results extend current understanding of the role of cytokines on cognitive function in individuals with HIV infection. When considered along with our previous findings, our results suggest that among HIV patients with good immune function, plasma cytokine concentration adds incrementally to the prediction of cognitive function, over and above the contribution of clinical HIV markers. These preliminary results are not intended for clinical application at this time. Rather, they suggest that plasma cytokine levels, which can be easily accessed and measured reasonably inexpensively may, in the future, be helpful for treatment planning; specifically for identifying patients at risk of memory problems. They may also be helpful for selecting participants for new interventions aimed at minimizing or reversing the impact of HIV on cognitive function. However, such applications would require replication of these results in other samples of HIV-infected patients using cross-sectional and longitudinal designs to establish reliability and validity.