The Interplay Between Loneliness, Cortisol, and NK Cell Function: The Role of Cortisol in NK Cell Dysfunction

Abstract

Chronic psychosocial stressors contribute to cardiovascular disease (CVD) and related risk factors, particularly in under-resourced communities. Psychosocial stressors activate the hypothalamic-pituitary-adrenal (HPA) axis, leading to sustained cortisol action on immune cells. Natural killer (NK) cells are altered in distribution and function in cardio-metabolic diseases; however, pathways linking cortisol as a marker of chronic stress to NK cell dysfunction remain understudied. We recruited 186 African American women from the Step It Up Community-Engaged, Digital Health Physical Activity Intervention study and collected cortisol, NK cell measures, and psychosocial stress measures using validated questionnaires, including chronic stress and loneliness. Plasma cortisol levels were negatively associated with NK cell proportions, and self-reported levels of loneliness modified this relationship (p=0.011); among participants with higher loneliness, higher plasma cortisol levels were associated with lower proliferative NK cell proportions (β=−0.30, p=0.04). In a subset of study participants (n=24), plasma cortisol levels were directly associated with a loss of NK cell degranulation (β=−0.38, p=0.03). Using an RNA sequencing dataset, we found that higher plasma cortisol levels were associated with suppressed gene expression related to NK cell cytotoxicity. Finally, in vitro experiments on freshly isolated primary NK cells revealed that cortisol reduces NK cell degranulation (p=0.006) via FABP-4-induced upregulation of PD-1 expression (p=0.01). In summary, our study provides evidence that in individuals with higher loneliness, cortisol may alter NK cell distribution and function. Furthermore, cortisol-mediated NK cell dysfunction may be facilitated in a FABP-4/PD-1 dependent manner, leading to impaired innate immune function and, potentially, contributing to worsening CVD risk.

Graphical Abstract

1. Introduction:

Cardiovascular disease (CVD) remains the leading cause of death worldwide, with atherosclerotic cardiovascular disease (ASCVD) driving total CVD mortality (Martin et al., 2024). Additionally, CVD risk factors like diabetes, obesity, and hypertension are also increasing in prevalence, further exacerbated by health disparities due to adverse socioeconomic, environmental, and psychosocial factors that constitute the Social Determinants of Health (SDoH) (Powell-Wiley et al., 2022). Improving psychological health and combating adverse SDoH are critical for reducing CVD and its risk factors. Among the various psychosocial stressors, people with depression have an increased risk for CVD events (Senoo et al., 2024), decreased CV health (Kwapong et al., 2023), and a 60% increased risk of developing diabetes (Berge and Riise, 2015). Among other psychosocial stressors, social isolation and loneliness are also associated with a heightened risk of mortality for individuals with pre-existing CHD or a history of stroke (Cené et al., 2022). Moreover, anxiety, depression, dysphoria, and societal withdrawal are often associated with loneliness (Kirkbride et al., 2024). A 2015 meta-analysis by Holt-Lunstad et al. showed a 26%, 32%, and 29% heightened risk of death by living alone, social isolation, and loneliness, respectively (Holt-Lunstad et al., 2015). The 2023 narrative review by Teshale et al. highlights the bidirectional nature between incident CVD, risk factors, and loneliness (Teshale et al., 2023). However, although the mechanistic pathway linking loneliness to CVD is still not fully understood, it is hypothesized to involve loneliness-associated changes to the immune system.

Prolonged periods of chronic stress may profoundly impact the body’s neuroendocrine system. Xia and Li proposed a mechanism by which increased cortisol related to psychosocial stress may promote CVD, stating that the sympathetic nervous system (SNS) and the hypothalamic–pituitary–adrenal (HPA) axis are activated by social isolation and loneliness as well as by alterations in behavior (i.e., reduced physical activity, disruption in sleep, and smoking), ultimately increasing cortisol secretion (Xia and Li, 2018). Amongst the psychosocial stressors, depression (Jia et al., 2019), loneliness (Jopling et al., 2021), or chronic stress due to living in disadvantaged neighborhoods (Rudolph et al., 2014) have been associated with increased cortisol presence. Since most cell types express cortisol receptors, cortisol can have a wide range of effects on behavior, metabolism, and immunity (Ortiz et al., 2022).

Natural Killer (NK) cells are part of the innate immune system, mainly characterized by a lack of CD3 expression and varying levels of CD56 and CD16 expression, allowing for NK cell subclassification into CD56^dim/CD16⁺ and CD56⁺/CD16⁻ NK cell subsets (Poli et al., 2009). The CD56^dim/CD16⁺ subset executes cytotoxicity, primarily via perforin and granzyme release, whereas the CD56⁺/CD16⁻ NK cell subset is often characterized by proliferative characteristics, as evidenced by cytokine release (e.g., IFNγ, TNFα). Indeed, both NK cell subsets contribute to NK cell cytotoxicity and cytokine release, ultimately coordinating the response of surrounding immune cells, such as T cells or monocytes/macrophages (Chen et al., 2024). Recent preclinical and clinical studies have highlighted NK cells in cardiometabolic health, demonstrating their role in diabetes (Kim et al., 2019), obesity (Baumer et al., 2024), and CVD (Ong et al., 2017). For example, Ong et al. demonstrated that NK cells limit the infiltration of eosinophils into myocardial tissue, thereby reducing cardiac inflammation and fibrosis in myocarditis (Ong et al., 2015).

Additionally, NK cells have also been shown to be affected by chronic stress. For example, increasing levels of loneliness and perceived stress have been shown to be negatively associated with NK cell counts (Steptoe et al., 2004), while depression was negatively associated with NK cell activity (Maes et al., 1994). Specifically, recent studies have suggested the potential modifying role of psychosocial factors on associations between inflammation and biological outcomes, hinting that psychosocial factors can alter immune function (Albertini et al., 2025). Loneliness has also been identified as a moderator of the relationship between affective symptoms and behavioral outcomes (Mason, 2024). Taken together, this evidence suggests that loneliness may have a modifying relationship with associations between cortisol and NK cell functioning. Additionally, in vitro experiments have shown that NK cell function and NK cell-activating receptor expression decrease after exposure to cortisol, resulting in a dysfunctional NK cell phenotype (Mavoungou et al., 2005). However, to our knowledge, a direct link between cortisol and NK cell distribution or function in the setting of chronic psychosocial stress has not been reported. Given that the existing literature reports that chronic loneliness impairs the HPA axis, leading to higher cortisol levels despite the body’s attempts to reduce these levels (Cole et al., 2015), and cortisol being cited as an immunosuppressant, more work is needed to examine the modifying effect of loneliness on the impact of cortisol on the immune system (Mavoungou et al., 2005).

Taken together, studies are needed to examine underlying signaling pathways linking cortisol to NK cell dysfunction in the setting of chronic psychosocial stress and CVD risk. Therefore, our study’s primary hypothesis was to further investigate the potential impact of cortisol on NK cell distribution and function, and to identify novel signaling pathways underlying a cortisol-induced alteration in NK cell function for African American women from under-resourced neighborhoods who are at risk for CVD. In a secondary aim, we sought to determine potential associations between psychosocial factors and NK cells. As an exploratory aim, we examined the potential moderating effect of psychosocial factors on the cortisol-to-NK cell association.

2. Materials & Methods

2.1. Study Population:

186 African American women from under-resourced neighborhoods of Washington, D.C., and Prince George’s County in Maryland were recruited in the Step It Up Community-Engaged, Digital Health Physical Activity Intervention study (Tamura et al., 2020; Vijayakumar et al., 2022). In accordance with the principles of the Declaration of Helsinki, approval for this study was obtained from the IRB at the National Institutes of Health, USA (NCT 03288207), and the study was carried out according to the International Council on Harmonisation (ICH). Written informed consent was obtained from all participants. At the time of visit to the NIH Clinical Center or the NHLBI Hope Center (a community-based clinical research hub located in Washington, D.C.), enrolled participants completed standardized surveys to assess their sociodemographic, behavioral, and health history, along with psychosocial and environmental factors. Study participants underwent a clinical examination to evaluate their overall health, including weight and height, blood pressure, heart rate, and other relevant parameters. Additionally, clinical and research blood was drawn. Freshly drawn research blood was used for measuring immune cell distribution by flow cytometry. The study participant characteristics of this study population are summarized in Supplementary Table 1.

Prior to the beginning of the main study, a cohort comprised of 24 participants was recruited to the Step It Up Community-Engaged, Digital Health Physical Activity Intervention Pilot study between October 2019 and March 2020 under NCT 03288207 (Vijayakumar et al., 2022) to test the feasibility of the main study protocol, with the study participant characteristics summarized in Supplementary Table 2. Study participants in this Pilot study were all African American women with an average age of 57 years, who presented with overweight or at least class I obesity (average BMI 34.77 ± 6.25 kg/m²) at intermediate Atherosclerotic CVD risk (ASCVD 10-year risk score 8.6 ± 5.58) (Supplementary Table 2). Research blood from these study participants was used to detect immune cell distribution from freshly drawn blood, and to isolate peripheral blood mononuclear cells (PBMCs), which were cryopreserved for future use. These cryopreserved PBMCs were utilized for the ex vivo measurement of NK cell function.

Participants completed validated questionnaires to measure psychosocial stressors during study visits. Depression was determined via the 20-item Center for Epidemiologic Studies Depression-Revised Scale (Eaton et al., 2004), while chronic stress was assessed using the Cohen Perceived Stress scale (Cohen et al., 1983). Social Isolation was measured with the PhenX Toolkit Protocol-Chronic Stress (Turner et al., 1995), and loneliness was assessed using the revised UCLA loneliness scale (Russell et al., 1980). For all psychosocial measures, increased scoring indicates worsening psychosocial stress.

The study participants’ physical activity levels were recorded as daily step counts using Fitbit^™ Charge 2 or Inspire 3 devices. The Healthy Eating Index (HEI), a 24-hour automated dietary assessment tool, was used to evaluate diet quality, with higher scores indicating a better-quality diet (Shams-White et al., 2023).

Covariates for this study included age, socioeconomic status (SES), BMI, self-reported presence of diabetes, self-reported presence of hypertension, self-reported use of lipid-lowering therapy, smoking status, systolic blood pressure, High-Density Lipoprotein concentration (HDL-c), Low-Density Lipoprotein concentration (LDL-c), HEI diet score, daily step count, and presence of inflammatory comorbidities. The Atherosclerotic Cardiovascular Disease (ASCVD) 10-year risk score was calculated using age, sex, race, systolic and diastolic blood pressure, total cholesterol, HDL cholesterol, and LDL cholesterol, as well as history of diabetes, smoking status, and hypertension treatment, lipid-lowering treatment, or aspirin therapy. The presence of inflammatory comorbidities was recorded as self-reported rheumatoid arthritis, HIV, lupus, psoriasis, multiple sclerosis, sarcoidosis, vasculitis, and related inflammatory conditions.

The procedures for characterizing peripheral immune cells, cryopreservation, and characterization of PBMCs are described in detail in the Supplementary Material and Methods.

2.2. In vitro characterization of NK cell function after cortisol exposure

PBMCs from buffy coats of healthy donors (NCT00001846) were isolated as described previously (Baumer et al., 2024) and detailed further in the Supplementary Material and Methods section. Briefly, as NK cells undergo receptor-mediated activation with subsequent target cell killing, NK cell degranulation and cytotoxicity can be determined by exposure of NK cells to a K562 target cell line, with subsequent measurement of degranulation (CD107a surface expression), and detection of intracellular cytokine expression (IFNγ, TNFα, granzyme B, perforin, and GM-CSF).

2.3. NK cell degranulation assay from cryopreserved PBMCs

To evaluate the NK cell function from cryopreserved PBMCs of study participants, the degranulation assay was performed. This assay allows for the determination of receptor-mediated degranulation and cytokine production of NK cells when exposed to a target cell. The degranulation assay was initially standardized using varying effector (PBMC) to target (K562 cells, CCL-243, ATCC) ratios. For this purpose, PBMCs were seeded with K562 cells in a 96-well round-bottom plate at a ratio of 10:1, 5:1, and 1:1 [effector:target (E:T)] with a total amount of 5 × 10^⁵ cells. The resulting cocultures of PBMCs and K562 cells were incubated at 37°C/5% CO₂ for 1 hour. Then Golgi transport inhibitor (Catno: 554724, BD Biosciences, USA) was added to all the wells, and the coculture was further incubated for an additional 4 hours. At the end of incubation, the cells were stained as further described in the Supplementary Material and Methods section.

As study participants’ PBMCs are limited in quantity, detection of the optimal effector-to-target cell ratio (Supplementary Figure 1A,B) was critical. This was done by co-culturing PBMCs and K562 (target cancer cell line) initially at three ratios: 1:1, 5:1, and 10:1. The 5:1 and 10:1 ratios displayed greater degranulation (CD107a expression, column 4 in Supplementary Figure 1B), and cytokine expression (GranzymeB and IFNγ, column 5 and 6 in Supplementary Figure 1B), when compared to the 1:1 ratio. Additionally, participants’ body mass index (BMI) was significantly associated with degranulation (CD107a expression) at a 5:1 ratio (Supplementary Figure 1C), confirming previous findings from our lab (Baumer et al., 2024). Therefore, ex vivo analysis of the study participants’ PBMC samples was performed at the 5:1 E:T ratio.

2.4. Cortisol measurement from human plasma

Cortisol was measured from frozen EDTA plasma aliquots obtained from fasting participants between 7 and 8 am utilizing the MyBioSource human cortisol ELISA Kit per the manufacturer’s recommendations (Catno: MBS700592, MyBioSource, USA). The sensitivity of this ELISA is < 1.56 ng/ml, and the detection range is 3.12 – 200 ng/ml per the manufacturers’ information.

2.5. Statistical analysis:

Demographic data for both study populations were calculated as means and standard deviations for continuous variables. For categorical variables, frequencies of affirmative responses were used to calculate percentages of the study population with various health conditions and behaviors. Using the Step It Up Community-Engaged, Digital Health Physical Activity Intervention cohort data, associations between cortisol and NK cell distribution were examined using multivariable linear regression. Associations between psychosocial stressors and NK cells, as well as between psychosocial stressors and cortisol, were also determined using multivariable regression. Unadjusted and stepwise adjusted multivariable linear regression modeling were performed, and adjustments for age, SES, BMI, presence of diabetes, presence of hypertension, use of lipid-lowering therapy, smoking status, systolic blood pressure, HDL-c, LDL-c, HEI diet score, daily step count, and presence of inflammatory comorbidities were included. Additionally, psychosocial factors, specifically chronic stress and loneliness, were examined as moderators of associations between cortisol and NK cells. Variables with significant p-interaction effects were categorized into tertiles based on psychosocial factor scores. Stratified, multivariable linear regression analyses were conducted to investigate the relationships between cortisol and NK cells in relation to various psychosocial factors. Using the Step It Up Community-Engaged, Digital Health Physical Activity Intervention Pilot cohort data, associations examining the relationship between cortisol levels and NK cell function, as determined by an ex vivo experimental setup, were analyzed using multivariable regression in both unadjusted and adjusted models (BMI and ASCVD 10-year risk score). For all studies involving associations, P values of <0.05 were considered statistically significant. All multivariate and moderation analyses were conducted using STATA Release 12 (StataCorp., College Station, TX, USA).

For in vitro experiments, statistical analysis was performed using PRISM 10.0 (GraphPad) software. Data are represented as mean ± the standard error of the mean. Every in vitro dataset was first tested for normality in distribution to determine subsequent appropriate statistical testing (RM one-way ANOVA, Friedman or Wilcoxon test). Additionally, all statistical testing of in vitro experiments was performed with multiple comparison adjustment using Dunn’s rank-sum or Sidak step-down procedures. Statistical approach for each dataset is given in each Figure Legend. Statistical significance was established at a p-value of p < 0.05 (and shown as *p < 0.05, **p < 0.001, ***p < 0.0001). Bioinformatics analysis is described in detail in the Supplementary Material and Methods section.

3. Results

3.1. Study Participant Characteristics

The study participants (n = 186, Supplementary Table 1) were all African American women with a mean age of 57.4 ± 12.4 years. They had class II obesity with a mean BMI of 36.3 ± 6.8 kg/m² and were at intermediate risk for atherosclerotic cardiovascular disease (ASCVD 10-year risk score 9.8 ± 9.2%). The intermediate CVD risk was further substantiated by a hypertension prevalence of 58.6%, diabetes prevalence of 22%, hyperlipidemia prevalence of 37.6% with only 26.3% of the study participants receiving lipid-lowering treatment, and 17.7% of the study participants reporting a smoking history. The average annual household income, as a measure of SES, was $70,200 ± 30,380.

3.2. Cortisol, Chronic Stress, and Loneliness are associated with NK cell Proportions

In a first step, we aimed to determine potential associations between fasting morning plasma cortisol levels and NK cell distribution in the peripheral blood of our study participants. Multivariable regression analysis revealed that plasma cortisol levels were not associated with overall CD3-/CD56+ NK cell or CD56low/CD16+ cytotoxic NK cell proportions in either the unadjusted or fully adjusted model (Figure 1A–C). However, plasma cortisol levels associated negatively with CD56⁺/CD16⁻ proliferative NK cell proportions in the unadjusted and fully adjusted model (β=−0.20, p=0.008 and β=−0.21, p=0.006, respectively). As sleep duration and quality can significantly affect morning cortisol levels (Backhaus et al., 2004), we also included an adjustment for hours of sleep as a sensitivity analysis. Adjustment for sleep did not change the observed associations (Figure 1A–C, indicated in orange).

Figure 1. The association between cortisol and proliferative NK cells is moderated by loneliness.

(A-C): Regression analysis graphical display of the association between cortisol and CD3⁻/CD56⁺ NK cells (A) and subsets: CD56^low/CD16⁺ cytotoxic (B) & CD56⁺/CD16⁻ proliferative (C) NK cells. The graphs display this unadjusted model 1, which is also indicated in black font. In contrast, blue font indicates the standardized beta and p-value for the fully adjusted model 2 (age + SES + presence of diabetes + systolic blood pressure + hypertension treatment + lipid-lowering treatment + HDL-c + LDL-c + former and current smoking status + BMI + HEI + Daily Step Count + existing inflammatory comorbidities). Model 3 is indicated in orange font and displays the results of model 2 with additional adjustment for ‘hours of sleep’. (D) Display of the graphical summary of the moderation analysis tested in this Figure. (E/F) Graphical display of linear prediction curves of the impact of chronic stress (E) or loneliness (F) on the association between cortisol levels and CD56⁺/CD16⁻ proliferative NK cell presence. (G-I) Graphical display of cortisol associations to NK cell measures after splitting all participants into two groups (low loneliness and medium/high loneliness). (D) Graphical display of the results for the overall CD3⁻/CD56⁺ NK cell population. In (E), the graphs for CD56^low/CD16⁺ cytotoxic NK cells are shown, while (F) displays the associations for CD56⁺/CD16⁻ proliferative NK cells. The standardized betas followed by the p-value are listed above each graph for the unadjusted model 1 in black and the fully adjusted model 2 in blue. (Significance is indicated by the asterisk (*) and was assumed when a p-value reached below 0.05)

Chronic stress was negatively associated with overall CD3⁻/CD56⁺ NK cell distribution in the unadjusted model (β=−0.16, p=0.028), which remained after adjustment for age, SES, diabetes prevalence, systolic blood pressure, hypertension treatment, lipid-lowering treatment, HDL-c, LDL-c, smoking history, BMI, HEI, daily step count, and existing inflammatory conditions (β=−0.17, p=0.023) (Table 1). Loneliness was negatively associated with overall CD3⁻/CD56⁺ NK cell distribution in unadjusted and adjusted models (β=−0.14, p=0.049 and β=−0.16, p=0.041, respectively). Neither depression nor social isolation was associated with overall CD3⁻/CD56⁺ NK cell distribution. None of the psychosocial stress measures were associated directly with plasma cortisol levels (Supplementary Table 3).

Table 1: Associations between psychosocial stress variables and NK cells in the Step It Up Community-Engaged, Digital Health Physical Activity Intervention.

Univariate and multivariable regression analysis were performed to examine the relationship between individual psychosocial stress variables (depression, chronic stress, social isolation, or loneliness as independent/exposure variables) and NK cells, as well as NK cell phenotypes (NK cell proportions were used as the dependent/outcome variables). Associations listed are shown with the following models: Model 1 (unadjusted) and Model 2 (age + BMI + presence of diabetes + systolic blood pressure + hypertension treatment + lipid-lowering treatment + HDL-c + LDL-c + former and current smoking status + HEI + Daily Step Count + existing inflammatory comorbidities + SES). Model 3 displays the results of the association in Model 2 with additional adjustment for ‘hours of sleep’. The data are displayed as standardized betas (β) followed by the p-value in parentheses for 186 study participants.

	NK Cells			CD56dim/CD16+ NK cells (Cytotoxic NK cells)			CD56hi/CD16− NK cells (Proliferative NK cells)
	Model 1	Model 2#	Model 3##	Model 1	Model 2#	Model 3##	Model 1	Model 2#	Model 3##
Depressive Symptoms	−0.059 (0.439)	−0.069 (0.389)	−0.078 (0.348)	0.024 (0.758)	0.017 (0.838)	−0.007 (0.931)	−0.055 (0.474)	−0.043 (0.600)	−0.036 (0.675)
Chronic Stress	−0.16 (0.028)*	−0.17 (0.023)*	−0.170 (0.030)*	0.01 (0.931)	0.01 (0.947)	−0.001 (0.988)	−0.01 (0.916)	0.01 (0.926)	0.006 (0.937)
Social Isolation	−0.11 (0.124)	−0.13 (0.100)	−0.128 (0.106)	−0.095 (0.204)	−0.11 (0.150)	−0.049 (0.543)	0.00035 (0.996)	0.0024 (0.976)	0.033 (0.684)
Loneliness	−0.14 (0.049)*	−0.16 (0.041)*	−0.178 (0.025)*	−0.02 (0.748)	−0.04 (0.622)	−0.061 (0.458)	0.07 (0.324)	0.09 (0.267)	0.104 (0.203)

Significance was assumed when the p-value was below 0.05 and is indicated by the asterisk (*) and bold font.

^#Model 2 covariates: age + SES + presence of diabetes + systolic blood pressure + hypertension treatment + lipid-lowering treatment + HDL-c + LDL-c + former and current smoking status + BMI + HEI + Daily Step Count + existing inflammatory comorbidities

^##Model 3 covariates: Model 2 + hours of sleep

3.3. Loneliness Moderates the Relationship between Cortisol and Proliferative NK cells

We evaluated whether loneliness or chronic stress could moderate the cortisol-NK cell relationship (Figure 1D). While chronic stress was not a significant moderator overall for any NK cell metric (interaction terms for Model 2’s all non-significant), (Table 2, Figure 1E), we found a significant interaction between loneliness and plasma cortisol levels on CD56⁺/CD16⁻ proliferative NK cells in the unadjusted and fully adjusted model (p=0.011, Table 2). The inverse association between plasma cortisol levels and CD56⁺/CD16⁻ proliferative NK cells was observed (purple and green lines, Figure 1F) among study participants who reported moderate to high loneliness as compared to those who reported low loneliness (blue line).

Table 2: Cortisol associates with NK cells via Loneliness in the Step It Up Community-Engaged, Digital Health Physical Activity Intervention.

Multivariable regression analysis was performed between plasma cortisol levels, NK cells, and NK cell phenotypes (NK cell proportions were used as the dependent variables). Interaction effects between stress and plasma cortisol levels, as well as between loneliness and plasma cortisol levels, were also assessed. Associations listed are shown with the following models: Model 1 (included cortisol, psychosocial factor, and their interaction to NK cell measures with no further adjustments), Model 2 (Model 1 with covariate adjustment), and Model 3 (Model 2 with adjustment for ‘hours of sleep’). Log-transformed cortisol levels were used in all regression models. The data are displayed as standardized betas (β) followed by the p-value in parentheses.

	NK Cells			CD56dim/CD16+ NK cells (Cytotoxic NK cells)			CD56hi/CD16− NK cells (Proliferative NK cells)
	Model 1	Model 2#	Model 3##	Model 1	Model 2#	Model 3##	Model 1	Model 2#	Model 3##
Cortisol (log)	−0.013 (0.930)	−0.072 (0.638)	−0.043 (0.776)	−0.084 (0.576)	−0.11 (0.480)	−0.049 (0.753)	−0.16 (0.278)	−0.14 (0.369)	−0.151 (0.329)
Chronic Stress	−0.66 (0.610)	−1.22 (0.358)	−1.139 (0.392)	0.29 (0.823)	0.041 (0.976)	−0.069 (0.960)	0.35 (0.786)	0.79 (0.557)	0.598 (0.659)
Cortisol x Stress	0.50 (0.700)	1.05 (0.430)	0.974 (0.466)	−0.29 (0.827)	−0.034 (0.980)	0.069 (0.960)	−0.36 (0.781)	−0.78 (0.562)	−0.591 (0.664)
Cortisol	−0.30 (0.336)	−0.41 (0.208)	−0.408 (1.149)	−0.23 (0.476)	−0.26 (0.425)	−0.228 (0.488)	0.58 (0.063)	0.58 (0.069)	0.603 (0.057)
Loneliness	−1.57 (0.235)	−2.00 (0.142)	−2.145 (0.117)	−0.52 (0.700)	−0.69 (0.624)	−0.871 (0.540)	3.39 (0.010)*	3.54 (0.010)*	3.682 (0.008) *
Cortisol x Loneliness	1.44 (0.280)	1.87 (0.175)	2.000 (0.149)	0.49 (0.716)	0.65 (0.648)	0.819 (0.570)	−3.38 (0.011)*	−3.52 (0.011)*	−3.643 (0.009)*

Significance was assumed when the p-value was below 0.05, indicated by the asterisk (*) and bold font, while italicized font indicates values approaching significance.

^#Model 2 covariates: age + SES + presence of diabetes + systolic blood pressure + hypertension treatment + lipid-lowering treatment + HDL-c + LDL-c + former and current smoking status + BMI + HEI + Daily Step Count + existing inflammatory comorbidities

^##Model 3 covariates: Model 2 + hours of sleep

To further explore loneliness as a potential moderator of the cortisol-to-NK cell relationship (Figure 1G–I, Supplementary Table 5), we divided the study cohort into tertiles based on loneliness. We merged the medium and high loneliness groups to investigate the potential impact of elevated loneliness levels in a broader spectrum. There were no significant associations between plasma cortisol and any of the NK cell measures in the low loneliness group (Figure 1G–I, top row). However, among individuals reporting medium to high loneliness levels (Figure 1G–I, bottom row), there was a negative association between plasma cortisol and CD56⁺/CD16⁻ proliferative NK cells in both the unadjusted and fully adjusted models. A trend towards a negative association with CD56^low/CD16⁺ cytotoxic NK cells is observed in the unadjusted model (p=0.077). Sensitivity analyses with further adjustment for sleep did not alter any of these associations (Supplementary Table 4). As a sensitivity analysis, we conducted a simple slope analysis to test the hypothesis that loneliness may serve as a graded or continuous moderator of the association between cortisol and CD56⁺/CD16⁻ proliferative NK cells (Supplementary Table 5). These data further indicate that the relationship between cortisol and CD56⁺/CD16⁻ proliferative NK cells differs as individuals report greater loneliness. In particular, the slope in the fully adjusted Model 3 is negative for those with loneliness at or above the mean.

3.4. Plasma cortisol levels Associate with Less NK cell Degranulation: An Ex Vivo Approach

To explore the hypothesis that cortisol levels may also associate with NK cell function, we examined the relationship between cortisol and NK cell degranulation in a small pilot study cohort (Figure 2A–D). The Pilot Study participants display characteristics comparable to those of the main study participants, as summarized in Supplementary Table 2. Multivariable linear regression analysis of plasma cortisol levels and CD107a expression on NK cells after exposure to K562 target cells, an indicator of degranulation, displayed a significantly negative association in the unadjusted as well as the BMI and ASCVD 10-year risk score adjusted models (β=−0.45, p=0.026 and β=−0.38, p=0.046, respectively; Figure 2B). Additionally, intracellular expression of IFNγ, an important modulator of NK cell killing capacity and immune cell activation, also displayed a negative association in the unadjusted and adjusted models (β=−0.52, p=0.011 and β=−0.521, p=0.026, respectively; Figure 2C). However, intracellular TNFα expression, which synergistically regulates surrounding immune cell function with IFNγ, did not show any significant association (Figure 2D).

Figure 2: Determining the potential impact of cortisol on NK cell function utilizing two different approaches.

(A) Graphical display of the aims in this Figure. (B-D) NK cell degranulation and cytokine presence were measured utilizing cryopreserved PBMCs. Multivariable regression analysis was performed between plasma cortisol levels and (B) NK cell degranulation as measured by surface CD107a expression, as well as NK cell intracellular expression of IFNγ (C) and TNFα (D). Associations are graphed as unadjusted models with the tabular display of unadjusted and BMI and ASCVD 10-year risk score adjusted standardized beta followed by the p-value. (Significance was assumed when the p-value was below 0.05 and is indicated by the asterisk (*)). (E) Volcano plot of the genes profiled in the high versus low plasma cortisol level NK cell RNA sequencing dataset after adjustment for BMI. Genes indicated by a red dot fall within the threshold of p-value ≤ 0.05 and absolute FC ≥ 1.5. (F) The top 10 significantly enriched KEGG pathways (adjusted p-value ≤ 0.1; absolute normalized enrichment score ≥ 1.5) are displayed. Each dot represents a pathway labeled on the y-axis, with the size of each dot scaled to the number of concordant DE genes overlapping with the pathway. The color of each dot is scaled to the adjusted p-value. (Abbreviations: DE – differentially expressed, FC – fold change, IFN – interferon, TNF – tumor necrosis factor, KEGG – Kyoto Encyclopedia of Genes and Genomes)

3.5. Cortisol Decreases NK cell Degranulation and IFNγ and TNFα Expression in Primary NK Cells in vitro

To explore in an unbiased hypothesis-generating way whether plasma cortisol levels may influence dysregulation of NK cells, we performed principal component analysis (PCA) using the study participants’ morning plasma cortisol values as the outcome variable with input variable from a previously published NK cell RNA sequencing dataset (Baumer et al., 2024) (Figure 2A). The low cortisol group (n=4) was characterized by mean cortisol levels of 24.8 ± 5.4 ng/ml, a mean BMI of 30.0 ± 7.1 kg/m², and a mean age of 64.0 ± 6.1 years, while the high cortisol group (n=3) was characterized by mean cortisol levels of 183.0 ± 21.1 ng/ml, a mean BMI of 31.5 ± 2.9 kg/m², and a mean age of 63.7 ± 5.4 years. PCA of the normalized, BMI-adjusted expression profiles revealed clear separation between the high- and low-cortisol participants (Supplementary Figure 1D). After adjusting for BMI, differential gene expression analysis identified 466 differentially expressed genes (DEGs) between the high- and low-cortisol participant study groups (Supplementary Table 6). Of these genes, 200 were down-regulated in the high cortisol group, while 266 were found to be up-regulated in the high cortisol group (Figure 2E). Pathway enrichment analysis showed significant enrichment of lysosome-related genes and NK cell cytotoxicity-related genes in the high cortisol group (Figure 2F, Supplementary Table 7). Given the small sample size and the hypothesis-generating nature of the study, both pathways were suppressed, supporting the potential role of cortisol in NK cell function loss.

To investigate if the observed associations in our study participant cohorts and RNA sequencing analysis may be the consequence of cortisol impacting NK cells directly, we conducted in vitro work utilizing freshly isolated primary NK cells (Figure 3). After exposure of primary NK cells to cortisol for 24 hours, we subjected control and cortisol-treated NK cells to K562 target cells to determine NK cell degranulation and cytotoxicity. Primary NK cells exposed to cortisol displayed a decreased expression of CD107a (a marker of degranulation) at baseline conditions (Figure 3A, 8.4 ± 1.4% vs 4.9 ± 1.1% CD107a⁺ NK cells). This decrease in degranulation also was also detected when a cell activator (which pushes NK cells to their maximum potential) was present (Figure 3A, 42.3 ± 4.3% vs 37.6 ± 3.4% CD107a⁺ NK cells). Intracellular expression of IFNγ was not decreased under baseline conditions after cortisol exposure but was decreased by 50.2% in the presence of a cell activator (Figure 3B, 35.9 ± 7.5% vs. 17.9 ± 5.0% IFNγ⁺ NK cells). Intracellular expression of TNFα was decreased by 62% under baseline (p=0.08) and 65% under cell activator (p=0.004) conditions (Figure 3C). No significant differences were seen for intracellular granzyme B or perforin expression after cortisol treatment (Figure 3D/E). These data indicate that cortisol exposure of NK cells in vitro may lead to decreased degranulation and cytokine production, resulting in decreased NK cell activity, NK cell-mediated killing, and immune cell activation.

Figure 3: Cortisol impacts NK cell function in vitro.

Primary freshly isolated NK cells from healthy blood bank donors were incubated ± cortisol for 24 hours and subsequently subjected to the degranulation assay in the presence of K562 target cells. Each marker was determined under baseline conditions (no activator) or under activating conditions (+ activator) to determine the maximum response rate of control or cortisol-incubated NK cells. Subpanels accompany an example zebra flow plot and the summarized graph results. (A) Evaluation of extracellular CD107a expression as a marker of degranulation (n=9, RM one-way ANOVA with Sidak step-down correction). (B-E) Intracellular expression of various cytokines as indicated (n=9; IFNγ/TNFα/GRZB: RM one-way ANOVA with Sidak step-down correction; Perforin: Friedman test with Dunn’s correction). (Significance is indicated by the asterisk (*) and was assumed when a p-value was below 0.05; Abbreviations: Act – cell activator, FSC-A – forward scatter area, GRZB – granzyme B, RM ANOVA – Repeated Measures Analysis of Variance)

3.6. Cortisol-induced NK cell function loss is in part mediated via PD-1 and FABP4

The relationship between the exploratory RNA sequencing dataset and cortisol in Figure 2F suggested involvement of T cell receptor signaling, and the implicated T cell receptors are also expressed on NK cells (Quatrini et al., 2021). We compared our analysis with the genes for immune checkpoint inhibitors expressed in T cells and identified as significantly upregulated in NK cells from atherosclerotic lesions, as published by Barcia Durán et al. in 2024 (Barcia Durán et al., 2024) (Figure 4A). We identified six genes that were commonly regulated across both datasets (Figure 4B). However, four of these six genes were regulated in different directions (indicated in red). Two genes of immune checkpoint inhibitors were upregulated in both datasets, namely, PDCD1 (PD-1) and CD274 (PD-L1) (green bars). In the next step, we investigated whether cortisol-treated primary NK cells would display an upregulation of PD-1 or PD-L1 surface expression by flow cytometry after cortisol exposure (Figure 4C). Incubation of primary NK cells with cortisol significantly increased PD-1 surface expression by 4.2-fold, while PD-L1 surface expression was significantly decreased by more than 50% when compared to vehicle control-treated NK cells. Blockade of PD-1 in cortisol-incubated primary NK cells significantly partially rescued CD107a expression as compared to cortisol treatment alone (Figure 4E), while PD-L1 blockade did not impact cortisol-mediated NK cell degranulation. Neither PD-1 nor PD-L1 blockade rescued either cortisol-induced decrease in intracellular IFNγ or TNFα expression (Figure 4F/G).

Figure 4: FABP4 may regulate cortisol-induced PD-1 expression and NK cell degranulation.

(A) To identify pathways potentially leading to cortisol-induced NK cell dysfunction, the high vs low cortisol dataset was subjected to a publicly available dataset characterizing immune cells within the atherosclerotic lesion of humans. This public dataset encompasses sequencing data for NK cells within the human atherosclerotic plaque utilized in this analysis approach. (B) Graphical display of genes common in both datasets, with green bars displaying significantly upregulated genes in both datasets, while red bars display genes that are downregulated in the cortisol dataset, opposing the public dataset’s directionality. (C) Primary freshly isolated NK cells were incubated with cortisol or vehicle control and the surface expression of PD-1 and PD-L1 was determined by flow cytometry (n=10, Wilcoxon test for each receptor pair). (D) Graphical display of the hypothesis driving the results shown in E-K. (E-K) Primary freshly isolated NK cells were treated with or without cortisol for 23 hours. One hour before starting the degranulation assay in the presence of K562 cells either a PD-1 or PD-L1 inhibitor was added. Degranulation results are shown in (E) (n=9, RM one-way ANOVA with Sidak step-down correction), while intracellular cytokine expression is shown in (F/G) (n=9, IFNγ: Friedman test with Dunn correction; TNFα: RM one-way ANOVA with Sidak step-down correction). (H-K) Determining the impact of FABP4 inhibitor on cortisol-induced NK cell function loss utilizing primary freshly isolated NK cells. (H) Measurement of PD-1 surface expression by flow cytometry after indicated treatments (n=9, Friedman test with Dunn correction). (I) Display of degranulation potential of NK cells with indicated treatments (n=5, RM one way ANOVA with Sidak step-down correction), while J/K summarize the results of intracellular cytokine expression after indicated treatments (n=6, IFNγ: Friedman test with Dunn correction; TNFα: RM one way ANOVA with Sidak step-down correction). (Significance is indicated by the asterisk (*) and was assumed when a p-value was below 0.05)

FABP4 was within the top 20 highest upregulated genes when comparing high versus low cortisol RNA sequencing dataset (Supplementary Table 6) with a 14.06-fold change upregulation. FABP4 is overexpressed in PD-1-resistant tumors (Freitas-Cortez et al., 2025) and has recently been found to be upregulated in the plasma of individuals with loneliness (Shen et al., 2025). The addition of an FABP4 inhibitor fully rescued the cortisol-induced upregulation of PD-1 on primary NK cells to control levels (p=0.014 cortisol vs cortisol+FABP4-i; p= 0.999 control vs cortisol+FABP4-I; Figure 4H). Additionally, in the presence of the FABP-4 inhibitor, cortisol-induced reduction in degranulation was fully restored (Figure 4I). However, the FABP4 inhibitor did not rescue the cortisol-induced reduction in the intracellular presence of IFNγ or TNFα expression (Figure 4J/K). Nevertheless, these data suggest that the cortisol-induced reduction in NK cell degranulation may, in part, be dysregulated by FABP4-mediated increased expression of PD-1 on primary NK cells.

4. Discussion

This study provides insight into mechanisms by which cortisol, as a potential marker of psychosocial stress, associates with changes in NK cell distribution in peripheral blood among African American women with CVD risk and recruited from under-resourced neighborhoods in the Washington D.C. area and Maryland. We found an association between plasma cortisol levels and the proliferative NK cells, and loneliness as a psychosocial stressor modified this relationship. Study participants’ plasma cortisol levels were also associated with NK cell function, and RNA sequencing analysis of NK cells from participants with high versus low plasma cortisol levels revealed that cortisol may mediate decreased NK cell function. In vitro experiments on freshly isolated primary NK cells showed that cortisol decreased NK cell degranulation and cytotoxicity. Our data suggest that cortisol at least partially mediates the loss in NK cell degranulation via increasing PD-1 in a fully FABP4-dependent manner. In summary, our data suggest that among individuals with increasing self-reported loneliness, higher cortisol is associated with lower NK cell numbers, as well as decreased NK cell function, worsening their risk for CVD (Ong et al., 2017) and cancer (Bald et al., 2020).

The literature on the impact of psychosocial stressors (such as adverse SDoH) on NK cell distribution and function is scarce and can appear contradictory (Katz et al., 2025). Variability in the duration of stress may explain these discrepancies (Katz et al., 2025). Acute stress has been reported to increase NK cell numbers and/or proportions and potentially even NK cell function. For example, study participants given a public-speaking task showed an increased number of cytotoxic CD56dim natural killer (NK) cells without altering regulatory CD56bright NK cell counts (Bosch et al., 2005). Another study employing acute stress demonstrated a similar increase in NK cells; however, the increase in NK cell response was blunted in individuals reporting medium or high levels of loneliness (Steptoe et al., 2004). These findings suggest that the changes observed in acute stress situations may also depend on underlying lifelong experiences of chronic stressors. In contrast, chronic stress often is associated with a decline (Nakamura et al., 1999) or unchanged (Arranz et al., 2009) NK cell numbers and/or proportions and a reduction in NK cell function [reviewed in (Katz et al., 2025)].

Our findings suggest that loneliness as a chronic stress modifies the relationship between cortisol and NK cells. This finding is supported by a study showing that a decline in NK cell activity among medical students during their final exams was exacerbated in students who reported higher levels of loneliness (Kiecolt-Glaser et al., 1984). Similarly, NK cell activity of women who underwent coronary bypass grafting was lower in those with depression (Doering et al., 2008). Among this population of women with CVD, decreased NK cell activity mediated the relationship between depression and non-wound infections, highlighting the real-life pathological consequences of chronic stress-induced NK cell dysfunction. Interestingly, our data identified loneliness as an effect modifier, whereas chronic stress did not moderate the cortisol-to-NK cell associations. This may stem from fundamental differences between these two measures of chronic psychosocial stress. While both measures are highly interconnected psychosocial constructs, both assess a subjective, negative, and enduring state with significant health impacts. The UCLA Loneliness Scale captures a concept of social isolation and emotional lack of connection, highlighting being alone, misunderstood, and a lack of companionship as a trait-like, stable, chronic state. In contrast, Cohen’s Perceived Stress Scale provides a subjective evaluation of stress (not the stressor itself) by measuring the degree of life’s unpredictability, overload, and uncontrollability, particularly over the past month. However, chronically lonely individuals have been shown to display hypervigilance for other stressors, potentially further enhancing states of chronic stress and subsequent consequences on immune and overall health (Hawkley and Cacioppo, 2010).

In our study, we did not observe a direct relationship between psychosocial stressors (especially loneliness) and fasting morning plasma cortisol levels. As our study focuses on women only, these results are less surprising, as other studies have reported that the association of elevated awaking cortisol levels may be sex dependent. For example, in an all-White cohort of older adults, an elevated awaking cortisol response was not observed in women (Díaz-Mardomingo et al., 2023). A recent study in African American adults measured hair cortisol levels and found that sex significantly modified the association between reported loneliness and hair cortisol levels. Using a simple slope analysis, the authors demonstrated that a link between increased loneliness and hair cortisol levels is particularly strong among male study participants (Knauft et al., 2026). Sex-based specificity in the stress-induced cortisol response may not be limited to loneliness, as other studies examining the relationships between chronic or acute stressors and cortisol displayed lower response rates in women (Henze et al., 2021; Reschke-Hernández et al., 2017). Therefore, one explanation for the lack of an association between psychosocial stressors and cortisol in the Step It Up intervention cohort may be that the cohort was all-female and cross-sectional.

While human data connecting cortisol to NK cell distribution and NK cell activity in the context of chronic stress are scarce, a study performed in rats with cancer suggested that glucocorticoid-induced suppression of NK cell function occurred under prolonged periods of stress (Rosenne et al., 2014). A study performed in women exposed to varying levels of airplane noise found that with increasing noise-related stress, cortisol levels increased, which was associated with a) increasing numbers of circulating NK cells and b) decreasing NK cell activity (Hartono, 2010), which supports our ex vivo and in vitro findings that indicate cortisol-induced NK cell dysfunction. Finally, among young adults who experienced early childhood adversity, NK cells displayed decreased CD107a and IFNγ expression when exposed to a target cell line (Fernandes et al., 2021). In the Step It Up cohort, the association between cortisol and circulating NK cells was observed only in the medium/high loneliness group, which may be due to glucocorticoid receptor resistance (GCR), a phenomenon known to occur with prolonged, constant exposure to chronic stress. It is possible that in individuals with medium and high loneliness levels, cortisol is persistently secreted due to a sustained activation of the HPA axis (Cohen et al., 2012) as chronic loneliness (trait loneliness) has been demonstrated to flatten the diurnal cortisol rhythm in children and adults alike (Knauft et al., 2026; Zilioli and Jiang, 2021; Zilioli et al., 2017). While GCR is generally connected to pro-inflammatory responses, this seems to differ for NK cells, which may display dampened reactivity in the setting of steroid resistance in chronic inflammatory lung diseases (Hodge and Hodge, 2019). However, very little is known about how GCR specifically impacts NK cells, highlighting the need for future studies to better understand the consequences of chronic-stress-associated GCR and NK cells.

Our experiments also show that the FABP4-related increase in PD-1 may at least partially account for cortisol’s impact on NK cell dysfunction. These results are supported by a previous study showing an immunosuppressive role of glucocorticoid signaling and cortisol on NK cell function. A study using expanded NK cells demonstrated an increase in PD-1⁺ NK cells after dexamethasone exposure in the presence of activating cytokines (Quatrini et al., 2021). This study also demonstrated that under these experimental conditions, degranulation and IFNγ were reduced in a PD-1-dependent manner. Similar to our data, blocking PD-1 rescued degranulation capacity. In contrast to in vitro experiments in Quatrini et al., we did not observe the cortisol-induced reduction of IFNγ to be PD-1 dependent. The differences may be that our study utilized freshly isolated, non-expanded NK cells versus cytokine-expanded NK cells, and cortisol versus dexamethasone for in vitro experiments. As confirmation of our results, this study also demonstrated that PD-1 is expressed on unstimulated NK cells, albeit at low levels (Quatrini et al., 2021). Another study linked cortisol-mediated reduction of NK cell degranulation to a decrease in NK cell-activating receptors, specifically NKp46 and NKp30 (Mavoungou et al., 2005). Future studies should extensively characterize NK cell surface marker expression to further investigate the underlying signaling pathways linking cortisol and NK cell dysfunction in individuals experiencing lifelong chronic stress.

Our finding that an FABP4 inhibitor fully rescued the cortisol-induced upregulation of PD-1 on primary NK cells and reduction in degranulation, coupled with the recent findings from UK Biobank data, which demonstrated overexpression of FABP4 protein in individuals with loneliness (Shen et al., 2025), potentially suggests that FABP4 may be the proximate mediator of cortisol effects on NK cell function. If larger studies corroborate these results, FABP4 may be a potential target to improve NK cell function in patients with CVD (van der Ark-Vonk et al., 2024).

The role of health behavior interventions should also be considered in combating chronic stress-induced NK cell dysfunction and associated clinical pathologies. NK cell parameters improve from various health behavior interventions. For example, high-intensity exercise increases overall NK cell proportions and cell subset distribution, as indicated by increases in the cytotoxic subset and decreases in the proliferative subset (Quintana-Mendias et al., 2023). Similarly, mindfulness-based stress reduction in women with breast cancer is associated with increased NK cell activity (Witek Janusek et al., 2019). Several of these behavioral interventions, such as physical activity, are associated with lower cortisol levels (De Nys et al., 2022). Lifestyle changes are also associated with a decrease in FABP4 plasma levels, particularly in patients with CVD risk (Lázaro et al., 2012). Taken together, these data suggest that future studies should examine the longitudinal impact of health behavior interventions on NK cell distribution, function, and signaling pathways in diverse cohorts experiencing chronic stress.

Our study has both clear strengths and some limitations. The cross-sectional nature of the clinical trial data does not establish directionality of effects. Also, as this study focused on African American women, the generalizability of our findings may be limited. Despite this, our results provide mechanistic insights into this understudied population at risk of CVD. Additionally, although our sample size is sufficiently powered for the current analyses and to control for various traditional CVD risk factors, overall, this is a relatively small observational study. Our findings warrant expansion to larger, diverse populations studied longitudinally. Furthermore, we acknowledge that measuring morning plasma cortisol has limitations compared to measuring saliva-based cortisol several times a day, hair cortisol, or 24-hour urinary cortisol, as plasma cortisol fails to capture diurnal rhythms and dynamic changes of cortisol secretion, providing only a snapshot of daily cortisol levels. Additionally, several factors can significantly impact circulating plasma cortisol levels. To account for the impact of sleep, the associations of our cohort data were further adjusted for hours of sleep. Similarly, because some medications, especially corticosteroids or HIV medications, can lower cortisol levels, we adjusted for anti-inflammatory conditions in all cohort-based analyses. Our ex vivo results, while promising, should be interpreted as preliminary results due to the relatively small sample size of the Pilot study cohort.

We believe that the strengths of our study approach (e.g., inclusion of study participants at the highest CVD risk and traditionally underrepresented in research; community-engaged research design; translational component) mitigate these limitations and improve our understanding of how chronic stress may influence innate immune cells relevant to CVD development.

5. Conclusion

The present study sheds light on the impact of cortisol on the innate immunity of the African American women with CVD risk and residing in under-resourced neighborhoods. Increased cortisol levels are associated with a significant shift in the proportions and distribution of NK cells and their subsets, particularly when reporting higher levels of loneliness. Our mechanistic analysis reveals that the stress hormone cortisol can significantly suppress NK cell function. Targeting PD-1 and FABP4 restored the NK cytolytic activity, highlighting potential therapeutic avenues. We hypothesize that these results are broadly relevant to all individuals enduring disproportionate levels of chronic stress. Thus, future research should define the mechanisms linking stress, NK cell dysfunction, and immunometabolism through the action of stress hormones.

Highlights.

• Cortisol, as a marker of psychosocial stress, associates with altered NK cell distribution.

• The interplay between cortisol and NK cells is moderated by loneliness.

• Cortisol impairs NK cell degranulation and cytokine production.

• Cortisol may mediate an increase in PD-1 surface expression on NK cells.

• Cortisol-mediated PD-1 increase and NK function loss are FABP4 dependent.

Abbreviations

ASCVD

Atherosclerotic Cardiovascular Disease

CD

Cluster of Differentiation

CVD

Cardiovascular Disease

FABP

Fatty Acid Binding Protein

HDL

High-Density Lipoprotein

HPA

Hypothalamic–Pituitary–Adrenal

IFN

Interferon

IL

Interleukin

LDL

Low-Density Lipoprotein

NK

Natural Killer

PBMC

Peripheral Blood Mononuclear Cells

PD

Programmed Cell Death Protein

PDL

Programmed Cell Death Protein Ligand

RNA

Ribonucleic Acid

SES

Socioeconomic Status

SNS

Sympathetic Nervous System

TNF

Tumor Necrosis Factor

Data Availability

Data presented in this study are available from the corresponding author upon reasonable request and will be uploaded to FigShare. RNA sequencing data were uploaded to NCBI Gene Expression Omnibus (Geo, GSE278321).