Secondary Logo

Journal Logo

Clinical Science

Social Isolation Is Linked to Inflammation in Aging People With HIV and Uninfected Individuals

Ellis, Ronald J. MD, PhDa; Iudicello, Jenny PhDb; Sun-Suslow, Ni PhDb; Grelotti, David MDb; Cherner, Mariana PhDb; Morgan, Erin PhDb; Letendre, Scott L. MD, Phdc; Heaton, Robert K. PhDb

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: April 15, 2021 - Volume 86 - Issue 5 - p 600-606
doi: 10.1097/QAI.0000000000002596
  • Free



Mounting evidence reveals that social support is beneficial for prevention and management of HIV disease.1–4Signs of poor social support, including social isolation and loneliness, are common in people living with HIV, particularly in older adults and ethnic minorities.5 Poor social support is associated with poor health outcomes among people with HIV (PWH), including faster progression to AIDS,6 functional impairment,7,8 poor health-related quality of life,9 poor retention in care,10 and higher rates of mortality.11 In some clinical populations, social isolation has been linked to systemic inflammation.11,12 However, the link between poor social support and systemic inflammation has not been studied among PWH, and therefore, it is not known whether the association may be stronger in them than in other populations. Even in the setting of viral suppression with antiretroviral therapy (ART), PWH have persistently elevated levels of inflammation biomarkers such as interleukin-6 (IL-6) and C-reactive protein (CRP). Increased inflammatory biomarkers are, in turn, associated with adverse health outcomes such as myocardial infarction and even death.13–17 Persistent inflammation in virally suppressed PWH also affects the central nervous system (CNS), where microglia and astrocytes are chronically activated,18,19 producing neurotoxic cytokines that can contribute to cognitive impairment, depression, and other adverse outcomes. To the best of our knowledge, the relationship between poor social support and neuroinflammation as indexed by cerebrospinal fluid (CSF) biomarkers has never been reported. Identification of poor social support as a driver of persistent inflammation in PWH would be important because it is potentially modifiable and could be improved with intervention. As such, we evaluated the hypothesis that poor social support would be associated with elevated levels of inflammatory biomarkers, both in blood and in CSF.



PWH and HIV− participants recruited from community sources were prospectively enrolled in cohort studies at the University of California, San Diego. Inclusion criteria included HIV-seropositive or -seronegative, consented to lumbar puncture and phlebotomy, and had a sufficient volume of CSF and plasma in storage at −80 °C to perform assays. Exclusions included abuse or dependence of methamphetamine, alcohol, and other substances within the past 18 months; significant CNS confounding conditions such as history of AIDS-defining opportunistic infection of the CNS; traumatic brain injury resulting in permanent neurological deficits; and major, active psychiatric disorders such as schizophrenia. All participants signed informed consent documents approved by the local Institutional Review Board.

Clinical Evaluations

The Medical Outcome Study Social Support Survey (MOS-SSS)20 consists of 19 items (item 2 to item 20) measuring the functional aspects of perceived social support and 1 additional item (item 1) assessing the number of close relatives and friends (item 1 was not included in data analysis). The instrument is easy to administer to chronically ill patients and demonstrates high convergent validity, discriminant validity, internal consistency, reliability, and 1-year stability. The MOS-SSS measures 4 domains of social support including tangible/instrumental support, emotional-informational support, positive social interactions, and affectionate support.21 An additional item queries whether the person has “Someone to do things with to help you get your mind off things.” Participants rated the MOS-SSS items using a five-point Likert rating scale ranging from (1) none of the time to (5) most of the time. The mean scores of the overall scale and 4 subscales were then transformed to a 100-point scale using the formula: transformed score = [(observed score − minimum possible score)/(maximum possible score − minimum possible score)]  × 100.20 A higher score indicates higher level of self-reported social support. Current mood was assessed using the Beck Depression Inventory (BDI)-II (1), a widely used 21-item self-report inventory measuring the severity of depressed mood.


Our selection of plasma biomarkers was informed by previous studies linking them and other inflammatory factors to lack of social support and stress: soluble tumor necrosis factor alpha type II (sTNFRII),22,23 d-dimer, interleukin-6 (IL-6),24,25 CRP,26,27 monocyte chemoattractant protein type I (MCP-1),28 soluble CD14 (sCD14),29 IP-10,30 uPAR, IL-8,31 and VEGF.32,33 Biomarkers were measured in plasma and CSF using standard immunoassays (Fig 1).


Additional Clinical and Laboratory Assessments

Comprehensive neuromedical assessments were performed. These assessments included vital signs, neurological and physical examination, collection of medical history including antiretroviral regimen, and collection of blood and CSF. HIV serostatus was documented by enzyme-linked immunosorbent assay and confirmed by Western blot. Routine clinical assays, such as blood CD4+ T-cell count and CSF total protein, were measured in the Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory at the University of California, San Diego Medical Center. HIV RNA levels were measured in CSF and plasma by real-time polymerase chain reaction with a lower quantification limit of 50 copies/mL (Abbott Diagnostics, Des Plaines, IL).

Statistical Analyses

Demographics, medical history, and HIV disease characteristics were summarized using mean values and SDs, medians and interquartile ranges, or counts and percentages as appropriate. Demographic data were compared between PWH and HIV− individuals using the independent-samples t test for continuous variables and the Fisher exact test for categorical variables. Log10 transformations were applied to biomarker measures to improve symmetry and normality of distributions. Exploratory factor analyses with oblique Varimax rotation were used to reduce the dimensionality of the biomarkers, grouping them into related categories. All biomarkers studied were allowed to enter the factor analysis. Factors with eigenvalues ≥ 1 were included. We then examined Pearson correlations between overall social support variable and the plasma and CSF inflammation biomarker factors. A Bonferroni correction for multiple comparisons was applied (P = 0.05/the number of factors). Secondary analyses are presented for each of the social support subcomponents. Separate multivariable regressions were used to assess the interaction effects between HIV serostatus and factor levels. Demographics (ie, age and ethnicity) were included in multivariable models and retained as covariates if P value was less than 0.2 in backward model selection. The results were considered statistically significant at the 5% alpha level. JMP Pro version 14 was used for all analyses.


As shown in Table 1, participants were 69 PWH and 87 HIV−, 31.4% female with a mean (SD) age of 44.9 (14.5) years. Approximately half (51.9%) were non-Hispanic white, 27.6% Hispanic, 14.1% black, and 6.4% other ethnicities. Among PWH, 55 (80.9%) were virally suppressed. Table 1 compares PWH and HIV− participants on demographic and disease variables and on social support variables. The groups were well-matched on age and ethnicity, but the proportion of women in the HIV− group was much larger. Social support was worse in PWH than HIV−; this was marginal for overall social support, but significant for affectionate (P = 0.0078) and positive support (P = 0.0138). Overall social support was poorer with increasing age among HIV− (r = −0.268, P = 0.0122), but not PWH (r = −0.00996, P = 0.935). Ethnicity and sex were not significantly associated with overall social support in either group.

TABLE 1. - Demographics, Social Support, and Disease Variables by HIV Status for Participants in the Plasma Factor Analysis
N 87 69
Age yr (mean, SD) 44.6, 15.9 45.2, 12.6 0.79
Sex female (N, %) 45, 51.7 4, 5.8% <0.0001
Ethnicity non-Hispanic white (N, %) 44, 50.6 37, 53.6% 0.705
Hispanic 20, 23.0% 23, 33.3%
Black 15, 17.2% 7, 10.1%
Other 8, 9.20% 2, 2.90%
CD4 nadir cells/µL (median; IQR) 297; 156, 434
CD4 current cells/µL (median; IQR) 640; 500, 842
Virally suppressed (N, %) 55, 80.9%
Overall social support (mean, SD) 3.94, 0.95 3.66, 0.96 0.0673
Tangible (mean, SD) 3.53, 1.27 3.30, 1.27 0.282
Overall (mean, SD) 3.99, 1.25 3.45, 1.25 0.0673
Tangible (mean, SD) 4.18, 1.01 3.78, 1.01 0.282
Affectionate (mean, SD) 4.08, 1.05 3.77, 0.99 0.0078
Positive social (mean, SD) 4.02, 0.91 3.86, 0.94 0.0138
Additional item (mean, SD) 3.94, 0.95 3.66, 0.96 0.0598
Educational/inform (mean, SD) 3.53, 1.27 3.30, 1.27 0.278

Plasma Factor Analysis

The plasma factor analysis (N = 156) yielded 4 factors that accounted for 68.8% of the variance, with CRP, IL-6, and D-dimer loading onto factor 1; sTNFR-II, IP-10, and uPAR loading onto factor 2; MCP-1, IL-8, and VEGF loading onto factor 3; and sCD14 loading onto factor 4. Levels of plasma factors 2–4 were significantly higher in PWH than HIV− (factor 2 mean [sd], 0.152 [0.896] vs −0.210 [0.850]; P = 0.007; factor 3, 0.0242 [0.840] vs. −0.424 [0.898]; P = 0.001; factor 4, 0.0101 [0.607] vs. −0.466 [1.0455]; P = 0.0005). Factor 1 levels did not differ significantly between the 2 groups (0.413 [0.101] vs −0.160 [0.093]; P = 0.0665).

Table 2 shows correlations of the overall support index and subscales with the plasma biomarker factors. Worse overall social support and each of the subscales were associated with higher levels of factor 3. None of the other factors was associated with social support. None of the social support variables interacted with HIV serostatus in predicting plasma inflammation factors, indicating that associations between inflammation and social support were not significantly different for the 2 groups. In a multivariable model, age and its interaction with overall social support also was nonsignificant.

TABLE 2. - Demographics, Social Support, and Disease Variables by HIV Status for Participants in the CSF Factor Analysis
N 34 43
Age yr (mean, SD) 45.4, 14.2 44.2, 13.7 0.703
Sex female (N, %) 17, 50.0% 3, 6.98% <0.0001
Ethnicity non-Hispanic white (N, %) 18, 52.9% 16, 47.1% 0.418
Hispanic 9, 26.4% 17, 39.5%
Black 5, 14.7% 3, 6.98%
Other 1, 2.94% 2, 4.65%
CD4 nadir cells/µL (median; IQR) 297; 156, 434
CD4 current cells/µL (median; IQR) 640; 500, 842
Virally suppressed (N, %) 55, 80.9%
Overall social support (mean, SD) 3.94, 0.95 3.66, 0.96 0.0673
Tangible (mean, SD) 3.53, 1.27 3.30, 1.27 0.282
Overall (mean, SD) 3.99, 1.25 3.45, 1.25 0.0673
Tangible (mean, SD) 4.18, 1.01 3.78, 1.01 0.282
Affectionate (mean, SD) 4.08, 1.05 3.77, 0.99 0.0078
Positive social (mean, SD) 4.02, 0.91 3.86, 0.94 0.0138
Additional item (mean, SD) 3.94, 0.95 3.66, 0.96 0.0598
Educational/inform (mean, SD) 3.53, 1.27 3.30, 1.27 0.278

Potential Confounds—Plasma Analysis

Increasing age was related to higher levels of factor 1 in both PWH (r = 0.411, P = 0.0004) and HIV− participants (r = 0.293, P = 0.006). Factors 2–4 were not significantly related to age. Among HIV− participants, women had higher factor 1 levels than men (P = 0.0002), but lower factor 3 levels than men (P = 0.0041). Among PWH, there were no sex differences. None of the factors was related to ethnicity. Older individuals had poorer overall social support among HIV− (r = −0.268, P = 0.0122), but not PWH (r = −0.0001, P = 0.935). Lack of plasma viral suppression (19.1% of PWH) was related to higher levels of plasma factor 2 (P = 0.0012) and factor 4 (P = 0.0078), but not significantly related to the other factors. Higher current CD4 was associated with higher levels of factor 1 (r = 0.223; P = 0.0250) and lower levels of factor 4 (P = 0.202; r = 0.0427). Lower nadir CD4 predicted higher levels of factor 2 (r = −0.286; P = 0.0034). Current and nadir CD4 were not related to the other factors. In a multivariable model predicting factor 3 levels from overall social support and age, social support remained significant (P = 0.0372), while age was not. Sex was not significant in any multivariable models. Because a subtype of major depressive disorder has been associated with inflammation, we also modeled the potential contribution of depressive mood to the plasma inflammation factors. In a multivariable model predicting plasma factor 3, including overall social support, BDI-II score, and their interaction, neither the main effect of BDI-II nor the interaction term was significant, and their inclusion did not alter the statistical significance of social support (P = 0.0193).

CSF Factor Analysis

A subset of 43 PWH and 34 HIV− had CSF available for analysis. The CSF factor analysis (N = 86) yielded 3 factors accounting for 63.2% of the variance, with uPAR, sCD14, and sTNFR-II loading onto factor 1; MCP-1 and IL-6 loading onto factor 2, and D-dimer and CRP loading onto factor 3. Levels of factor 1 were significantly higher in PWH than HIV− (P = 0.0194).

Table 3 shows the correlation matrix for the social support variables and the CSF factors for PWH and HIV− together. Poorer social support was significantly associated with higher levels of factor 2 for the overall support index and for all but one of the support subscales. Factors 1 and 3 were not significantly associated with any of the social support measures. In a multivariable model, overall social support was significantly associated with CSF factor 2, but HIV serostatus and the interaction were not.

TABLE 3. - Correlations of the Overall Support Index and its Component Scores With Plasma Inflammation Factors
Variable By Variable r 5%ile 95%ile P
Overall support Factor 1 CRP IL-6 D-dimer −0.063 −0.218 0.095 0.434
Index Factor 2 sTNFR-II IP-10 uPAR −0.053 −0.209 0.105 0.508
Factor 3 MCP-1 IL-8 VEGF −0.186 −0.333 −0.030 0.020
Factor 4 sCD14 −0.096 −0.249 0.062 0.235
Tangible Factor 1 CRP IL-6 D-dimer −0.072 −0.226 0.086 0.373
Support Factor 2 sTNFR-II IP-10 uPAR 0.043 −0.115 0.199 0.593
Factor 3 MCP-1 IL-8 VEGF −0.167 −0.316 −0.010 0.037
Factor 4 sCD14 −0.034 −0.191 0.123 0.670
Affectionate Factor 1 CRP IL-6 D-dimer −0.056 −0.211 0.102 0.490
Support Factor 2 sTNFR-II IP-10 uPAR −0.141 −0.292 0.016 0.078
Factor 3 MCP-1 IL-8 VEGF −0.180 −0.328 −0.023 0.025
Factor 4 sCD14 −0.120 −0.272 0.038 0.136
Positive social Factor 1 CRP IL-6 D-dimer −0.052 −0.207 0.106 0.520
Support Factor 2 sTNFR-II IP-10 uPAR −0.090 −0.243 0.069 0.266
Factor 3 MCP-1 IL-8 VEGF −0.193 −0.340 −0.037 0.016
Factor 4 sCD14 −0.050 −0.206 0.108 0.535
Additional item Factor 1 CRP IL-6 D-dimer −0.096 −0.249 0.063 0.235
Factor 2 sTNFR-II IP-10 uPAR −0.080 −0.235 0.078 0.318
Factor 3 MCP-1 IL-8 VEGF −0.204 −0.350 −0.049 0.011
Factor 4 sCD14 −0.058 −0.213 0.100 0.470
Emotional/ Factor 1 CRP IL-6 D-dimer −0.043 −0.199 0.115 0.595
Informational Factor 2 sTNFR-II IP-10 uPAR −0.040 −0.196 0.118 0.618
Support Factor 3 MCP-1 IL-8 VEGF −0.144 −0.294 0.014 0.074
Factor 4 sCD14 −0.122 −0.274 0.036 0.128
Corrected for multiple comparisons, significant if P ≤ 0.0125. Bold indicates significant correlation.

To evaluate the relative importance of CSF versus plasma inflammation with respect to social support, we performed a multivariable regression predicting the overall support index from plasma factor 3 and CSF factor 2. CSF factor 2 retained significance in the model (P = 0.00646), while plasma factor 3 did not (P = 0.203). The interaction term was nonsignificant (Table 4).

TABLE 4. - Correlations of the Overall Support Index and its Component Scores With CSF Inflammation Factors
Variable By Variable r 5%ile 95ile% P
Overall support Factor 1 uPAR sCD14 sTNFR-II −0.051 −0.271 0.176 0.644
Index Factor 2 MCP-1 IL-6 −0.294 −0.486 −0.075 0.006
Factor 3 D-Dimer CRP 0.140 −0.087 0.353 0.200
Tangible support Factor 1 uPAR sCD14 sTNFR-II 0.076 −0.151 0.295 0.486
Factor 2 MCP-1 IL-6 −0.196 −0.402 0.029 0.070
Factor 3 D-Dimer CRP 0.181 −0.045 0.389 0.096
Affectionate Factor 1 uPAR sCD14 sTNFR-II −0.172 −0.382 0.054 0.112
Support Factor 2 MCP-1 IL-6 −0.256 −0.454 −0.034 0.017
Factor 3 D-Dimer CRP 0.080 −0.147 0.298 0.467
Positive social Factor 1 uPAR sCD14 sTNFR-II −0.176 −0.384 0.051 0.106
Support Factor 2 MCP-1 IL-6 −0.290 −0.482 −0.070 0.007
Factor 3 D-Dimer CRP 0.183 −0.043 0.391 0.092
Additional item Factor 1 uPAR sCD14 sTNFR-II −0.157 −0.368 0.070 0.149
Factor 2 MCP-1 IL-6 −0.268 −0.464 −0.047 0.013
Factor 3 D-Dimer CRP 0.174 −0.052 0.383 0.108
Emotional/ Factor 1 uPAR sCD14 sTNFR-II 0.007 −0.217 0.231 0.949
Informational Factor 2 MCP-1 IL-6 −0.302 −0.492 −0.083 0.005
Factor 3 D-Dimer CRP 0.079 −0.148 0.298 0.469
Corrected for multiple comparisons, significant if P ≤ 0.0125. Bold indicates significant correlation.

Potential confounds—CSF analysis

In the combined PWH and HIV− sample, older individuals had higher levels of CSF factor 1 (r = 0.461, P < 0.0001); age was not related to the other CSF factors. Sex and ethnicity were not significantly related to any of the factors. Older age was associated with poorer overall social support (r = −0.168, P = 0.0355). None of the other demographic variables was significantly related to overall social support. In a multivariable model predicting factor 2 levels from age and overall social support, the latter remained significant (P = 0.00392), whereas age was not significant (P = 0.0843). Again, in a multivariable model, there was no interaction between HIV serostatus and overall social support in predicting factor 2 levels. Because of the large difference in the proportions of women between PWH and HIV−, we included sex in a multivariable model predicting factor 2 levels from serostatus, sex, and overall social support. The interaction term (sex X social support) was nonsignificant, and overall social support remained significant in the model (P = 0.0118), while sex was not (P = 0.463). PWH with lower current CD4 counts had higher levels of CSF factor 2 (r = 0.414, P = 0.0058), but not the other CSF factors. Nadir CD4 was not related to any of the factors. In a multivariable model containing current CD4, overall support, and their interaction, both overall support (P = 0.0170) and CD4 (P = 0.0330) were significant; the interaction was not. Only 1 participant had a detectable CSF viral load.

As noted above, a subtype of major depressive disorder has been associated with inflammation. Thus, we also modeled the potential contribution of depressive mood to inflammation. In multivariable models predicting the CSF factors from overall social support BDI-II score and their interaction, neither BDI-II nor its interaction with social support was a significant predictor of any of the CSF inflammation factors and did not alter the statistical significance of overall social support index.


Our finding that poorer social support correlated with higher plasma inflammatory biomarkers is consistent with the previous literature.12,23,34,35 However, no previous reports of CSF inflammation markers were available for comparison. Neuroinflammation as indexed by CSF biomarkers was more strongly related to social support than systemic inflammation as indexed by plasma biomarkers. The relationships we found were robust to consideration of potential confounds including demographics, viral suppression, and depressed mood. The associations of social support with systemic and CNS inflammation may be particularly important in the context of HIV, where chronic inflammation occurs despite viral suppression on ART. Both poorer social support and inflammation in HIV are linked to adverse outcomes including cardiovascular disease, neurocognitive impairment, and death.13–16,36,37

As outlined in the Methods section, our selection of biomarkers to assay was informed by previous studies linking them to poorer social support. In addition, the biomarkers selected have been linked previously to HIV disease progression, viral load, and mortality. For example, CRP, IL-6, and D-dimer (plasma factor 1) have been shown to be independent predictors of mortality38–40 and HIV disease progression.40,41 sTNFRII levels (factor 2) were reported to be significantly higher in patients with more advanced HIV disease as indexed by CD4 count,42 and higher levels were associated with increased mortality.43 Increased IP-10 levels can promote HIV disease progression, manifested as loss of CD4+ T cells and increased viral loads.44,45 Serum sUPAR levels predicted early death in 1 study.46 Plasma levels of MCP-1 (factor 3) are increased in patients with higher virus load.47 A significant negative correlation of plasma levels of IL-8 with CD4 counts (cells/μL) was observed in HIV-1 infected ART-naive subjects and IL-8 levels positively correlated with viral load in ART-treated children.48 CD4 counts and viral loads were also significantly associated with serum IL-8 concentrations.49 sCD14, a marker of monocyte response to LPS, is an independent predictor of mortality in HIV infection,50 and high sCD14 levels during primary HIV-1 infection predicted more rapid disease progression.51 In addition, increased levels of MCP-1 in the CNS and CSF (CSF factor 2 in this study) occurred in HIV encephalitis,52–54 which is in turn associated with high mortality and advanced HIV disease.

We used factor analysis to summarize the various biomarkers of inflammation. Factor analysis is a statistical method used to describe variability among observed, correlated variables in terms of a potentially lower number of unobserved variables called factors. Thus, factor analysis is a method for dimensionality reduction and can help control false discovery. It is important to check the identified factors against known physiological relationships. After on this, we found that the biomarker factor patterns emerging from our analyses were consistent with previous literature. For example, the proinflammatory cytokine IL-6 stimulates the production of CRP (plasma factor 1) in the liver.55 In 1 study of PWH, both IL-6, CRP, and D-dimer (see plasma factor 1 and CSF factor 3) were intercorrelated, and each was associated with an increased risk of cardiovascular disease (CVD) independent of other CVD risk factors.56 IL-6, CRP, and ICAM-1 are frequently coupregulated.57 MCP-1 and IL-8 (plasma factor 3) together participate in the adhesion of monocytes to vascular endothelium.58 suPAR levels have been associated with IP-10 and sTNFR-II (plasma factor 2).59 In another study, after carotid artery occlusion, secretion of the cytokines IL-6, and TNF-α led to an upregulation of ICAM-1 on the vascular endothelium, which promotes leukocyte migration across the BBB.60 In a study of subarachnoid hemorrhage, there were strong correlations between CSF levels of IL-6 and MCP-1 (CSF factor 2).61 In HIV infection, sCD14 and sTNFR levels (CSF factor 1) were highly correlated.62 Together, these observations suggest that the factors we identified represent consistent patterns of inflammation.

The elevated inflammation factors associated with social support differed partially between plasma (MCP-1, IL-8, and VEGF) and CSF (MCP-1 and IL-6). MCP-1 was common between the 2. In 1 study, lonelier diabetics without HIV had raised plasma MCP-1 concentrations after a stressful stimulus.63 In another, plasma MCP-1 levels were higher with stress in patients with a recent myocardial infarction.64 Stimulated production of interleukin-8 was higher in participants with lower levels of social support,31 but CSF was not studied in this report. Similarly, women with ovarian carcinoma who reported lower levels of social well-being had higher levels of VEGF.33

As is true for any cross-sectional study, our data cannot determine causality, nor can it determine the direction of causality if it does exist—in other words, whether lack of social support causes inflammation or inflammation leads—possibly indirectly, through effects on cognition or other aspects of behavior—to social isolation. Indeed, it is conceivable that both social support and inflammation may be linked to a third, unobserved variable. We did not see evidence that the links were attributable to depressed mood. Causality may be informed by future longitudinal studies that can evaluate whether changes in social support may relate to changes in systemic inflammation. Several observations deserve consideration here. Inflammation, such as pain, produces physical withdrawal; similarly, inflammation might lead to social withdrawal. In 1 study, an inflammatory challenge (endotoxin) led to higher levels of perceived social disconnection in healthy volunteers.65 Alternatively, lack of social support might trigger brain pathways involved in the processing of physical pain. Thus, activation of brain regions associated with the salience of pain and pain-related distress—the dorsal anterior cingulate cortex, insula, and ventral prefrontal cortex—show similar patterns of activation in paradigms designed to elicit social pain through social exclusion.66,67 In addition, cutaneous heat pain sensitivity is associated with distress reported after a social exclusion paradigm.48

Because of the importance of social cooperation to individual success during human evolution, it is believed that social connectedness co-opted pre-existing brain systems responsible for processing physical pain.68 By triggering pain systems in the brain, social isolation became an unpleasant experience to be avoided. If it could not be avoided, similar to what happens in chronic pain, inflammation resulted.

This study is limited by the relatively small cohort size, particularly for CSF, and by the cross-sectional design, the imbalance of males and females between PWH and the HIV− group, by inability to control for comorbid medical illness, and by the limited number of biomarkers assessed. The study was underpowered for testing interactions.

These results provide a springboard for future research on the mechanisms linking social support and inflammation. For example, a pilot study might be performed to assess whether social support interventions might reduce inflammation, possibly ameliorating its adverse health consequences, especially in PWH. An important next step might be a randomized clinical trial to determine whether interventions leading to improved social functioning and social satisfaction in PWH would result in reduced inflammation. On the other hand, if inflammation drives social isolation, then knowing which specific inflammatory mediators are critical might lead to targeted anti-inflammatory therapies to treat this condition.


1. Bonett S, Meanley S, Stevens R, et al. The Role of networks in racial disparities in HIV incidence among men who have sex with men in the United States. AIDS Behav. 2020;24:2781–2796.
2. Chandran A, Benning L, Musci RJ, et al. The longitudinal association between social support on HIV medication adherence and healthcare utilization in the women's interagency HIV study. AIDS Behav. 2019;23:2014–2024.
3. Chapman Lambert C, Westfall A, Modi R, et al. HIV-related stigma, depression, and social support are associated with health-related quality of life among patients newly entering HIV care. AIDS Care. 2019:1–8.
4. Enriquez M, Mercier DA, Cheng AL, et al. Perceived social support among adults struggling with adherence to HIV care and treatment. J Assoc Nurses AIDS Care. 2019;30:362–371.
5. Emlet CA. An examination of the social networks and social isolation in older and younger adults living with HIV/AIDS. Health Soc Work. 2006;31:299–308.
6. Leserman J, Petitto JM, Golden RN, et al. Impact of stressful life events, depression, social support, coping, and cortisol on progression to AIDS. Am J Psychiatry. 2000;157:1221–1228.
7. Richardson TM, Friedman B, Podgorski C, et al. Depression and its correlates among older adults accessing aging services. Am J Geriatr Psychiatry. 2012;20:346–354.
8. Ma L, Sun F, Tang Z. Social frailty is associated with physical functioning, cognition, and depression, and predicts mortality. J Nutr Health Aging. 2018;22:989–995.
9. Greene M, Hessol NA, Perissinotto C, et al. Loneliness in older adults living with HIV. AIDS Behav. 2018;22:1475–1484.
10. Zanoni BC, Sibaya T, Cairns C, et al. Barriers to retention in care are overcome by adolescent-friendly services for adolescents living with HIV in South Africa: a qualitative analysis. AIDS Behav. 2019;23:957–965.
11. Mazonson P, Berko J, Loo T, et al. Loneliness among older adults living with HIV: the older old may be less lonely than the younger old. AIDS Care. 2020:1–8.
12. Smith KJ, Gavey S, NE RI, et al. The association between loneliness, social isolation and inflammation: a systematic review and meta-analysis. Neurosci Biobehav Rev. 2020;112:519–541.
13. Grund B, Baker JV, Deeks SG, et al. Relevance of interleukin-6 and D-dimer for serious non-AIDS morbidity and death among HIV-positive adults on suppressive antiretroviral therapy. PLoS One. 2016;11:e0155100.
14. Duprez DA, Neuhaus J, Kuller LH, et al. Inflammation, coagulation and cardiovascular disease in HIV-infected individuals. PLoS One. 2012;7:e44454.
15. Ross AC, Rizk N, O'Riordan MA, et al. Relationship between inflammatory markers, endothelial activation markers, and carotid intima-media thickness in HIV-infected patients receiving antiretroviral therapy. Clin Infect Dis. 2009;49:1119–1127.
16. Ford ES, Greenwald JH, Richterman AG, et al. Traditional risk factors and D-dimer predict incident cardiovascular disease events in chronic HIV infection. AIDS. 2010;24:1509–1517.
17. Kuller LH, Tracy R, Belloso W, et al. Inflammatory and coagulation biomarkers and mortality in patients with HIV infection. PLoS Med. 2008;5:e203.
18. Tavazzi E, Morrison D, Sullivan P, et al. Brain inflammation is a common feature of HIV-infected patients without HIV encephalitis or productive brain infection. Curr HIV Res. 2014;12:97–110.
19. Garden GA. Microglia in human immunodeficiency virus-associated neurodegeneration. Glia. 2002;40:240–251.
20. Chaillon A, Nakazawa M, Anderson C, et al. Effect of cannabis use on human immunodeficiency virus DNA during suppressive antiretroviral therapy. Clin Infect Dis. 2020;70:140–143.
21. Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med. 1991;32:705–714.
22. Heo S, Moser DK, Pressler SJ, et al. Depressive symptoms and the relationship of inflammation to physical signs and symptoms in heart failure patients. Am J Crit Care. 2014;23:404–413.
23. Slavich GM, Way BM, Eisenberger NI, et al. Neural sensitivity to social rejection is associated with inflammatory responses to social stress. Proc Natl Acad Sci U S A. 2010;107:14817–14822.
24. Miller GE, Rohleder N, Cole SW. Chronic interpersonal stress predicts activation of pro- and anti-inflammatory signaling pathways 6 months later. Psychosom Med. 2009;71:57–62.
25. Chiang JJ, Eisenberger NI, Seeman TE, et al. Negative and competitive social interactions are related to heightened proinflammatory cytokine activity. Proc Natl Acad Sci U S A. 2012;109:1878–1882.
26. Fuligni AJ, Telzer EH, Bower J, et al. A preliminary study of daily interpersonal stress and C-reactive protein levels among adolescents from Latin American and European backgrounds. Psychosom Med. 2009;71:329–333.
27. Ford ES, Loucks EB, Berkman LF. Social integration and concentrations of C-reactive protein among US adults. Ann Epidemiol. 2006;16:78–84.
28. Loucks EB, Sullivan LM, D'Agostino RB Sr, et al. Social networks and inflammatory markers in the framingham heart study. J Biosoc Sci. 2006;38:835–842.
29. Grosgebauer K, Salinas J, Sharkey M, et al. Psychosocial correlates of monocyte activation and HIV persistence in methamphetamine users. J Neuroimmune Pharmacol. 2019;14:16–22.
30. Barrett B, Hayney MS, Muller D, et al. Meditation or exercise for preventing acute respiratory infection (MEPARI-2): a randomized controlled trial. PLoS One. 2018;13:e0197778.
31. Marsland AL, Sathanoori R, Muldoon MF, et al. Stimulated production of interleukin-8 covaries with psychosocial risk factors for inflammatory disease among middle-aged community volunteers. Brain Behav Immun. 2007;21:218–228.
32. Wu X, Liu BJ, Ji S, et al. Social defeat stress promotes tumor growth and angiogenesis by upregulating vascular endothelial growth factor/extracellular signal-regulated kinase/matrix metalloproteinase signaling in a mouse model of lung carcinoma. Mol Med Rep. 2015;12:1405–1412.
33. Lutgendorf SK, Johnsen EL, Cooper B, et al. Vascular endothelial growth factor and social support in patients with ovarian carcinoma. Cancer. 2002;95:808–815.
34. Grant N, Hamer M, Steptoe A. Social isolation and stress-related cardiovascular, lipid, and cortisol responses. Ann Behav Med. 2009;37:29–37.
35. Nersesian PV, Han HR, Yenokyan G, et al. Loneliness in middle age and biomarkers of systemic inflammation: findings from Midlife in the United States. Soc Sci Med. 2018;209:174–181.
36. Hernandez R, Carnethon M, Giachello AL, et al. Structural social support and cardiovascular disease risk factors in Hispanic/Latino adults with diabetes: results from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Ethn Health. 2018;23:737–751.
37. Kokab F, Greenfield S, Lindenmeyer A, et al. The experience and influence of social support and social dynamics on cardiovascular disease prevention in migrant Pakistani communities: a qualitative synthesis. Patient Educ Couns. 2018;101:619–630.
38. Feldman JG, Goldwasser P, Holman S, et al. C-reactive protein is an independent predictor of mortality in women with HIV-1 infection. J Acquir Immune Defic Syndr. 2003;32:210–214.
39. Wada NI, Bream JH, Martinez-Maza O, et al. Inflammatory biomarkers and mortality risk among HIV-suppressed men: a multisite prospective cohort study. Clin Infect Dis. 2016;63:984–990.
40. Borges AH, O'Connor JL, Phillips AN, et al. Interleukin 6 is a stronger predictor of clinical events than high-sensitivity C-reactive protein or D-dimer during HIV infection. J Infect Dis. 2016;214:408–416.
41. Lau B, Sharrett AR, Kingsley LA, et al. C-reactive protein is a marker for human immunodeficiency virus disease progression. Arch Intern Med. 2006;166:64–70.
42. Hober D, Benyoucef S, Delannoy AS, et al. High plasma level of soluble tumor necrosis factor receptor type II (sTNFRII) in asymptomatic HIV-1-infected patients. Infection. 1996;24:213–217.
43. Ajello F, La Licata R, Lodato M, et al. Soluble tumor necrosis factor alpha receptors (sTNF-Rs) in HIV-1-infected intravenous drug users: change in circulating sTNF-R type II level and survival for AIDS patients. Eur J Epidemiol. 2000;16:209–216.
44. Lei J, Yin X, Shang H, et al. IP-10 is highly involved in HIV infection. Cytokine. 2019;115:97–103.
45. Mhandire K, Mlambo T, Zijenah LS, et al. Plasma IP-10 concentrations correlate positively with viraemia and inversely with CD4 counts in untreated HIV infection. Open AIDS J. 2017;11:24–31.
46. Sidenius N, Sier CF, Ullum H, et al. Serum level of soluble urokinase-type plasminogen activator receptor is a strong and independent predictor of survival in human immunodeficiency virus infection. Blood. 2000;96:4091–4095.
47. Weiss L, Si-Mohamed A, Giral P, et al. Plasma levels of monocyte chemoattractant protein-1 but not those of macrophage inhibitory protein-1alpha and RANTES correlate with virus load in human immunodeficiency virus infection. J Infect Dis. 1997;176:1621–1624.
48. Eisenberger NI, Jarcho JM, Lieberman MD, et al. An experimental study of shared sensitivity to physical pain and social rejection. Pain. 2006;126:132–138.
49. Pananghat AN, Aggarwal H, Prakash SS, et al. IL-8 alterations in HIV-1 infected children with disease progression. Medicine (Baltimore). 2016;95:e3734.
50. Sandler NG, Wand H, Roque A, et al. Plasma levels of soluble CD14 independently predict mortality in HIV infection. J Infect Dis. 2011;203:780–790.
51. Krastinova E, Lecuroux C, Leroy C, et al. High soluble CD14 levels at primary HIV-1 infection predict more rapid disease progression. J Infect Dis. 2015;212:909–913.
52. Cinque P, Vago L, Mengozzi M, et al. Elevated cerebrospinal fluid levels of monocyte chemotactic protein-1 correlate with HIV-1 encephalitis and local viral replication. AIDS. 1998;12:1327–1332.
53. Conant K, Garzino-Demo A, Nath A, et al. Induction of monocyte chemoattractant protein-1 in HIV-1 Tat-stimulated astrocytes and elevation in AIDS dementia. Proc Natl Acad Sci U S A. 1998;95:3117–3121.
54. Kelder W, McArthur JC, Nance-Sproson T, et al. Beta-chemokines MCP-1 and RANTES are selectively increased in cerebrospinal fluid of patients with human immunodeficiency virus-associated dementia. Ann Neurol. 1998;44:831–835.
55. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med. 1999;340:448–454.
56. Reinfuss M. [Kaposi's sarcoma. Clinical picture, treatment and prognosis. I. Clinical forms of Kaposi's sarcoma]. Nowotwory. 1987;37:195–206.
57. Yang YL, Cheng X, Li WH, et al. Kaempferol attenuates LPS-induced striatum injury in mice involving anti-neuroinflammation, maintaining BBB integrity, and down-regulating the HMGB1/TLR4 pathway. Int J Mol Sci. 2019;20:491.
58. Gerszten RE, Garcia-Zepeda EA, Lim YC, et al. MCP-1 and IL-8 trigger firm adhesion of monocytes to vascular endothelium under flow conditions. Nature. 1999;398:718–723.
59. Hoenigl M, Moser CB, Funderburg N, et al. Soluble urokinase plasminogen activator receptor is predictive of non-AIDS events during antiretroviral therapy-mediated viral suppression. Clin Infect Dis. 2019;69:676–686.
60. Zareie M, Johnson-Henry K, Jury J, et al. Probiotics prevent bacterial translocation and improve intestinal barrier function in rats following chronic psychological stress. Gut. 2006;55:1553–1560.
61. Niwa A, Osuka K, Nakura T, et al. Interleukin-6, MCP-1, IP-10, and MIG are sequentially expressed in cerebrospinal fluid after subarachnoid hemorrhage. J Neuroinflammation. 2016;13:217.
62. Lien E, Aukrust P, Sundan A, et al. Elevated levels of serum-soluble CD14 in human immunodeficiency virus type 1 (HIV-1) infection: correlation to disease progression and clinical events. Blood. 1998;92:2084–2092.
63. Hackett RA, Poole L, Hunt E, et al. Loneliness and biological responses to acute stress in people with Type 2 diabetes. Psychophysiology. 2019;56:e13341.
64. Lima BB, Hammadah M, Wilmot K, et al. Posttraumatic stress disorder is associated with enhanced interleukin-6 response to mental stress in subjects with a recent myocardial infarction. Brain Behav Immun. 2019;75:26–33.
65. Eisenberger NI, Inagaki TK, Mashal NM, et al. Inflammation and social experience: an inflammatory challenge induces feelings of social disconnection in addition to depressed mood. Brain Behav Immun. 2010;24:558–563.
66. Eisenberger NI, Lieberman MD, Williams KD. Does rejection hurt? An FMRI study of social exclusion. Science 2003;302:290–292.
67. Craig AD. Significance of the insula for the evolution of human awareness of feelings from the body. Ann N Y Acad Sci. 2011;1225:72–82.
68. Eisenberger NI. The neural bases of social pain: evidence for shared representations with physical pain. Psychosom Med. 2012;74:126–135.

HIV; social isolation; inflammation

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.