Distinct inflammatory response patterns are evident among men and women with higher depressive symptoms

Abstract

Extensive research links depression and inflammation, with emerging evidence suggesting some differences between males and females in these associations. However, relatively few studies have examined stimulated inflammatory responses (ex vivo) in depression. The present research investigated the associations between depressive symptoms, basal inflammation, and LPS-stimulated production of pro- (IL-1β, IL-6, IL-8, TNF-α) and an anti-inflammatory cytokine (IL-10), with a focus on the extent to which gender moderates these relationships. As part of a larger study, 162 socio-economically and racially diverse subjects (ages 25-65, 67% women) completed extensive self-report measures, including depressive symptoms. Whole blood was quantified for basal inflammation, or incubated with 1μg/mL lipopolysaccharide (LPS) for 2h (at 37°C, 5% CO2) to quantify inflammatory responses to bacterial challenge. We examined the associations between depression and inflammatory markers in regression analyses, controlling for age, BMI, race/ethnicity, income, education, and use of medications. No main effects were observed between depressive symptoms and basal or stimulated levels of inflammation. Moderation analyses revealed a significant interaction between depressive symptoms and gender for stimulated TNF-α, stimulated IL-6 (p < 0.05), and a marginally significant interaction for stimulated IL-10 (p = 0.07). For men, higher depressive symptoms were associated with significantly higher production of TNF-α (p < 0.05) and marginally higher IL-6 (p = 0.07), but not with the anti-inflammatory cytokine IL-10. For women, higher depressive symptoms were associated with significantly lower production of TNF-α and IL-10 (ps <0.05), and marginally lower IL-6 (p = 0.06). These findings provide evidence for gender differences in the association of depressive symptoms with inflammatory response patterns, and highlight the utility of assessing ex vivo immune responses in blood. Implications for health are discussed.

1. Introduction

A large body of evidence indicates that inflammatory processes can play a key role in the pathophysiology of depression [1,2,3]. Both clinical and animal studies show that immunological challenge (e.g., endotoxin administration, interferon therapy) can induce depressive-like symptoms through up-regulation of inflammation [4,5,6,7]. Importantly, this relationship is bidirectional, with evidence demonstrating that depression can lead to inflammation; in several longitudinal studies, systemic inflammation has been found to increase subsequent to depression [8,9]. As depression is a risk factor for the development of medical conditions such as coronary heart disease, type 2 diabetes and hypertension, and given that altered immune responses are a potential mechanism linking depression to such conditions (reviewed by Kiecolt-Glaser and Glaser [10]), the present study sought to examine how inflammatory responses vary in the presence of depressive symptomatology.

For reasons not fully understood, women are almost twice as likely as men to develop depression [11]. Genetic, hormonal, psychological, social, and environmental factors, as well as structural differences in the brain, have been identified as factors that lead to somewhat distinct differential emotional reactions in males and females (e.g., [12]). Emerging evidence further suggests that the relationship between depression and inflammation may also differ across gender. Under conditions of inflammation, women appear more susceptible to developing depression than men (reviewed by Derry et al. [13]). Increased interleukin (IL)-6 levels following in vivo immune challenge have been associated with depressed mood in women, but not in men [14]. Contrary to this, evidence suggests that depression is associated with elevated inflammation more strongly in men compared to women. For instance, large observational studies have shown that clinical depression is associated with elevated CRP and IL-6 levels in men, but not in women [15,16,17]. Clarifying these discrepant findings regarding the role of gender in the association between depression and inflammation is thus an important next step in this research domain.

The majority of studies relating depression to inflammation and health have focused on systemic inflammation (i.e., basal cytokine levels in blood) and have generally shown positive correlations, even at subclinical levels of depression (reviewed by Howren et al. [3]). An alternate approach is to examine inflammatory responses ex vivo. Blood samples in such studies are stimulated with a mitogen, such as lipopolysaccharide (LPS) and/or phytohemagglutinin (PHA), to examine the capacity of immune cells to produce cytokines. Under normal physiological conditions, cytokines often occur in such low levels that they are not detectable; hence the examination of stimulated cytokines ex vivo yields higher detectability and better allows for the determination of abnormal patterns of cytokine production [18].

Relatively few studies have examined the association between depression and ex vivo inflammatory responses, and these studies have yielded contradictory results. Although some ex vivo studies have shown higher production of proinflammatory cytokines in depression [19-25], others have shown reduced proinflammatory cytokine release [26-28], or no reliable relationships [29-33]. Further, very few ex vivo studies have examined anti-inflammatory cytokine production in depression [20,27,30]. Although proinflammatory cytokines, such as IL-1β, IL-6, and tumor necrosis factor (TNF)-α play a critical role in mounting acute inflammatory responses against microbes, the production of anti-inflammatory cytokines, such as IL-10, is important to limit the synthesis of proinflammatory cytokines and resolve inflammation [34]. The balance between pro- and anti- inflammatory cytokines is an important factor in determining the magnitude and duration of inflammatory responses an individual mounts in response to challenge [35]. The current study aimed to characterize the relationship between both pro- and anti- inflammatory cytokine production and depressive symptoms in middle-aged adults.

Most ex vivo studies examining cytokine production in depression have not reported on gender-specific associations, mostly due to the inclusion of only one gender or limited sample size. To date, fifteen studies of which we are aware have focused on the association between stimulated cytokines and either clinical depression or subclinical depressive symptoms. Of these studies, four included only one gender [24,25,26,32], four did not report on gender [19,27,30,31], and seven examined gender effects [20-23,28,29,33]. Among these seven latter studies, only one by Kim et al. [20] found gender-related differences (stimulated levels of IL-6 and TNF-α were lower in female depressed patients compared to male depressed patients). No gender-related differences were observed in the other six studies. However, these studies were generally quite small (typically less than 50) to accurately detect gender effects. Larger studies that examine both genders are needed to clarify gender-specific associations in this line of research.

The present study aimed to investigate the associations between depressive symptoms, basal inflammation, and LPS-induced production of pro- (IL-1β, IL-6, IL-8, TNF-α) and anti-inflammatory cytokines (IL-10), with a focus on whether, and the extent to which, gender moderates these relationships. Given that heightened inflammation has been widely documented in depression, it was hypothesized that greater depressive symptomatology would be associated with higher levels of both basal inflammation and stimulated inflammatory responses. Due to inconsistencies in the literature, a priori hypotheses were not made regarding how gender would affect these associations.

2. Methods

2.1 Participants

This study is part of the larger research study “The Effects of Stress on Cognitive Aging, Physiology and Emotion” (ESCAPE) project. ESCAPE is a prospective longitudinal measurement-burst design study, the protocol for which has been described elsewhere [36]. ESCAPE used systematic probability sampling to recruit a socioeconomically and ethnically/racially diverse sample. Participants were recruited using New York City Registered Voter Lists. Letters of recruitment were sent to identified individuals explaining the study goals, followed by a telephone call to enroll eligible subjects in the study (for more details see [39]). Inclusion criteria included men and women residing in Co-Op City, the Bronx, New York, between the ages of 25 to 65, and being ambulatory, fluent in English and without visual impairment. The ongoing ESCAPE protocol consists of multiple waves of data collection approximately 9 months apart; only data from the first wave of collection are employed here. Of the eligible participants who were recruited, 220 had blood samples from the first wave that were available for assay. Participants were excluded from the present study if they had a history of selected inflammatory-related medical illness (e.g., autoimmune disorders, diabetes, cancer, HIV, chronic infections, cardiovascular disease, kidney and liver diseases), a lifetime history of psychiatric disorders other than depression, or were taking immunosuppressive drugs including oral corticosteroids. This resulted in a subsample of 162 participants (109 women, 53 men). Inhaled corticosteroids, statins, non-steroidal anti-inflammatory drugs (NSAIDs), oral contraceptives (OCPs), and antidepressants were treated as control variables.

2.2 Procedure

After providing informed consent, participants completed a paper-based survey at baseline, which was completed at home. As part of the baseline survey packet, respondents answered questions regarding demographics, physical and mental health conditions, and medication use. Blood samples were collected approximately two weeks after the initial baseline in-clinic visit.

2.3 Demographics

Subjects completed a self-report questionnaire assessing socio-economic and demographic variables including age, gender, race/ethnicity, household income, work, education and marital status. Body mass index (BMI) was calculated based on height and weight measured by a research assistant. Female participants provided self-report of menopausal status. Demographic characteristics are shown in Table 1.

Table 1.

Subject characteristics for total sample and by gender.

	Total sample	Women	Men
N (%)	162	109 (67)	53 (33)
Age [years]: Mean (SD)	44.4 (11.1)	44.3 (11.2)	44.7 (11.0)
BMI [kg/m2]: Mean (SD)	31.3 (8.0)	32.1 (8.4)	29.9 (6.9)
Ethnicity/Race: N (%)
Non-Hispanic white	13 (8)	7 (6)	6 (11)
Black/African-American	104 (64)	71 (65)	33 (62)
Hispanic	34 (21)	23 (21)	11 (21)
Other	11 (7)	8 (8)	3 (6)
Household income: N (%)
≥ $40,000	85 (53)	58 (53)	27 (51)
< $40,000	70 (43)	45 (41)	25 (47)
Missing	7 (4)	6 (6)	1 (2)
Marital status: N (%)
Married or living with partner	64 (40)	39 (36)	25 (47)
Divorced/Separated	25 (15)	19 (17)	6 (11)
Never married	56 (35)	39 (36)	17 (32)
Widowed	2 (1)	0	2 (4)
Other	6 (4)	5 (5)	1 (2)
Missing	9 (5)	7 (6)	2 (4)
Education: N (%)
Completed college or higher	76 (47)	53 (49)	23 (43)
Some college or less	86 (53)	56 (51)	30 (57)
Depressive scores: Mean (SD)
Raw score	15.5 (7.0)	15.3 (7.2)	16.0 (6.7)
T-score	52.5 (9.2)	52.0 (9.3)	53.3 (8.8)
PROMIS depression cutoff scores: N (%)
Non or minimal depression: [less than 55.0]	91 (56)	65 (60)	26 (49)
Mild depression: [55.0 – 59.9]	40 (25)	24 (22)	16 (30)
Moderate depression: [60.0 – 69.9]	29 (18)	18 (16)	11 (21)
Severe depression: [70 and over]	2 (1)	2 (2)
Medication use: N
Antidepressants	8	6	2
Statins	14	10	4
OCPs*	5	5	0
Inhaled corticosteroids	7	6	1
NSAIDs*	5	2	3
Smoking status: N (%)
Smokers	35 (22)	20 (18)	15 (28)
Non-smokers	125 (77)	88 (81)	37 (70)
Missing	2 (1)	1 (1)	1 (2)
High blood pressure: N (%)	37 (23)	26 (24)	11 (21)

^*OCPs= Oral contraceptives,

^*NSAIDs=Non-steroidal anti-inflammatory drugs

2.4 Depression

Depressive symptoms were assessed using the Patient Reported Outcome Measurement Information System – Depression (PROMIS Depression) short-form scale. The PROMIS depression scale has been shown to validly assess a broad range of depressive symptoms with high precision in measurement [37]. The depression questionnaire was given to subjects as part of the baseline survey. Respondents answered questions regarding their depressive symptoms during the past 7 days. The PROMIS depression short-form consists of eight items that focus on negative mood (4-items) and negative views of the self (4-items). Each item is rated on a 5-point scale, “1 (never)”, “2 (rarely), “3 (sometimes)”, “4 (often)’, “5 (always)”. The total summed raw score (ranging from 8 to 40) was converted to a standardized T-score (ranging from 38.2 to 81.3), which was used as the final depression score for each participant [38]. This T-score is a well-established metric which is centered based on the U.S. general population, with a mean of 50 and a standard deviation (SD) of 10 [39]. Hence, a T-score of 60 indicates one standard deviation higher than the U.S. average. For descriptive purposes, scores 0.5 – 1.0 SD higher than the mean were considered mild symptoms (55.0-59.9), scores 1.0 – 2.0 SD higher than the mean were considered moderate symptoms (60.0 – 69.9), and scores > 2.0 SD higher than the mean were considered severe symptoms (70 and over) [40].

2.5 Cytokine assays

Blood samples were drawn between 7 AM and 11 AM by a certified phlebotomist at the Albert Einstein College of Medicine. Following 12-h of fasting, 5 ml of blood was collected into sodium heparin tubes to assess both basal and stimulated cytokine levels. To determine a measure of basal inflammation, whole blood was centrifuged at 3000g for 15 min at room temperature. The supernatant was aliquoted and stored at −80°C. To determine stimulated cytokine levels (e.g., stimulated IL-6), 1 ml of whole blood was incubated with 1 μg/mL of bacterial LPS (E. Coli 055:B5, Sigma-100mg) on a rotational shaker at 37°C in 5% CO2 for 2 hours. All cytokines are expected to show increases from basal (unstimulated) values in response to LPS. After incubation, samples were centrifuged at 3000g for 15 min at room temperature. The supernatant was aliquoted and stored at −80°C for future analysis. Basal (IL-1β, IL-6, IL-8, IL-10, TNF-α) and LPS-stimulated (IL-1β, IL-6, IL-8, IL-10, TNF-α) cytokines were quantified using multiplex magnetic bead arrays (Life Technologies, Grand Island NY). The minimum detection limit for these assays is less than 0.5 pg/ml for each analyte and inter-assay CVs are 4.4 %–8.6 %. All assays were performed in duplicate.

2.6 Statistical analyses

Data analyses were conducted using IBM SPSS Statistics, Version 22.0 (Armonk, NY). A value of p <0.05 was considered statistically significant. Independent-samples t-tests were conducted to compare men and women in terms of depression scores, and Chi square tests were used to compare men and women in terms of income and education levels. Multiple linear regression analyses were conducted to determine the relationship between reported depression scores and each basal and stimulated cytokine level. Because the distribution of cytokine data was skewed, logarithmic (log) transformation was conducted prior to analyses. For basal cytokines, a log (x+1) formula was applied because many basal values were below the value of one. In the main-effects models, the PROMIS depression score was entered as a predictor variable and each log-transformed cytokine level was entered as separate outcome variable. In the moderation models, gender was entered as a moderator (0=female, 1=male) to examine whether gender influenced the relationships between depression scores and log-transformed cytokine levels. In order to conduct the moderation analyses, depression scores were mean centered, and product terms were computed between the mean centered depression scores and gender. When the interaction of depression scores by gender on cytokine levels was significant, follow-up linear simple effects analyses were conducted (i.e., separately in males and females).

2.7 Covariates

Prior research has shown that older age and higher BMI are associated with higher basal levels of proinflammatory cytokines [41,42], and lower stimulated levels of proinflammatory cytokines [26,43]. Also, basal inflammation has been linked to household income, education levels and race/ethnicity [44]. Thus, age, BMI, income, education, race/ethnicity (with Black/African-American race used as a reference group) included as covariates in our model. Further, we statistically controlled for medications that may affect inflammation including inhaled corticosteroids, statins, NSAIDs, OCPs, and antidepressants. We confirmed in later analyses that removing a subsample of participants who were taking antidepressants (N=8) did not change the results. In addition, when menopausal status was added to the analysis in women it did not change the results (39 out of 109 were postmenopausal). Finally, additional adjustment for smoking status (n = 35 smokers) or high blood pressure (n = 37) as covariates did not change the results; thus, these factors were not included in the final analyses.

3. Results

Subject characteristics are shown in Table 1. The study sample consisted of 162 socio-economically and ethnically/racially diverse subjects with a mean age of 44.4 ± 11.1 years (67% women). The racial/ethnic composition of the sample was 64% Black/African-American, 21% Hispanic, 8% non-Hispanic White, and 7% other. The mean depression score was 52.5, which is close to the US general population mean (i.e., 50), with scores ranging from 38.2 to 81.3. No differences were observed between men and women in demographic characteristics or in depression scores (52.0 vs. 53.3, respectively; all p > 0.3). No significant correlations were found between basal and stimulated levels of cytokines with the exception of IL-10. Significant correlations were observed between basal and stimulated levels of IL-10 in the entire sample (r = 0.40, p < 0.001), and in men (r = 0.44, p < 0.001) and women (r = 0.40, p < 0.001) separately.

3.1 Basal inflammation

The mean values of non-transformed basal cytokine levels are presented in Table 2A.

Table 2.

A. Non-log transformed levels of basal cytokines, Mean (SEM) (pg/ml).
	Total sample	Women	Men
IL-1β	0.009 (0.002)	0.006 (0.001)	0.014 (0.005)
IL-6	0.111 (0.032)	0.067 (0.012)	0.200 (0.093)
IL-8	0.453 (0.047)	0.371 (0.032)	0.623 (0.127)
IL-10*	0.295 (0.058)	0.226 (0.047)	0.436 (0.147)
TNF-α	0.132 (0.045)	0.076 (0.019)	0.246 (0.130)

B. Non-log transformed levels of LPS-stimulated cytokines, Mean (SEM) (pg/ml).
	Total sample	Women	Men
IL-1β	436 (40)	438 (45)	431 (45)
IL-6	1283 (146)	1278 (206)	1294 (139)
IL-8	6745 (521)	7178 (714)	5854 (606)
IL-10*	74 (17)	79 (24)	63 (16)
TNF-α	1613 (106)	1482 (139)	1881 (151)

^*Anti-inflammatory cytokine

3.1.1 Main-effects analyses

Regression analysis revealed no significant main effects of depressive symptoms on any basal cytokine levels (p > 0.3), in both unadjusted and fully adjusted (i.e., controlling for age and BMI, household income, education level, race/ethnicity, and medication use) models. Results are presented in Table 3.

Table 3.

Multiple regression analyses of depressive symptoms on log-transformed basal cytokine levels^***.

		IL-1β		IL-6		IL-8		IL-10		TNF-α
		β	p	β	p	β	p	β	p	β	p
Unadjusted Model	Depressive symptoms	−0.04	0.63	0.08	0.31	−0.04	0.60	−0.08	0.32	−0.01	0.87
Full adjusted Model	Age	−0.14	0.12	−0.08	0.40	0.12	0.20	0.03	0.77	−0.80	0.37
	BMI	0.06	0.50	0.10	0.30	−0.05	0.60	−0.03	0.67	−0.003	0.96
	Income	0.01	0.90	−0.03	0.70	−0.05	0.60	0.009	0.91	−0.03	0.74
	Education	0.02	0.80	−0.02	0.80	−0.12	0.17	−0.07	0.45	−0.06	0.53
	Hispanic	0.01	0.90	0.10	0.23	0.06	0.43	0.15	0.06†	0.02	0.85
	White	−0.01	0.90	−0.02	0.85	−0.07	0.43	−0.17	0.40	0.01	0.87
	Antidepressants	0.11	0.20	0.12	0.14	0.14	0.09†	0.25	0.002**	0.21	0.01*
	Statins	0.15	0.09†	0.12	0.20	−0.10	0.23	0.18	0.04	0.08	0.35
	OCPs	−0.05	0.60	−0.06	0.50	0.06	0.48	−0.05	0.53	−0.06	0.47
	Inhaled corticosteroids	−0.06	0.50	−0.04	0.65	0.006	0.94	−0.03	0.70	−0.08	0.35
	NSAIDs	0.02	0.80	0.04	0.70	−0.05	0.55	−0.08	0.34	−0.03	0.75
	Depressive symptoms	−0.05	0.60	0.08	0.32	−0.04	0.62	−0.06	0.44	−0.04	0.64

^**p< 0.01,

^*p < 0.05,

^†p < 0.1.

^***Regression analysis revealed no significant main effects of depressive symptoms on basal cytokine levels.

3.1.2 Moderation analyses

Unadjusted analyses revealed no significant interaction of depressive symptoms × gender on basal cytokine levels (p > 0.1). These results remained the same in the fully adjusted model (Table 4).

Table 4.

Multiple regression analyses of depressive symptoms by gender interaction on log-transformed basal cytokine levels^***.

		IL-1β		IL-6		IL-8		IL-10		TNF-α
		β	p	β	p	β	p	β	p	β	p
Unadjusted Model	Depressive symptoms	−0.11	0.22	0.04	0.65	−0.008	0.93	−0.11	0.25	−0.11	0.25
	Gender1	0.15	0.06†	0.15	0.06†	0.16	0.05†	0.11	0.15	0.15	0.05†
	Dep. symptoms × gender	0.12	0.20	0.05	0.62	−0.08	0.39	0.04	0.64	0.15	0.11
Full adjusted Model	Age	−.15	0.09†	−0.07	0.44	0.13	0.16	0.02	0.80	−0.10	0.29
	BMI	.07	0.41	0.10	0.20	−0.003	0.97	−0.02	0.80	0.01	0.90
	Income	0.03	0.70	−0.02	0.78	−0.04	0.60	0.01	0.86	−0.008	0.92
	Education	0.02	0.83	−0.004	0.96	−0.11	0.20	−0.05	0.53	−0.05	0.50
	Hispanic	0.02	0.81	0.09	0.25	0.06	0.47	0.14	0.07†	0.02	0.82
	White	−0.04	0.62	−0.06	0.53	−0.08	0.34	−0.10	0.23	−0.008	0.93
	Antidepressants	0.12	0.20	0.13	0.11	0.16	0.05†	0.26	0.002**	0.21	0.01*
	Statins	0.16	0.08†	0.11	0.20	0.14	0.11	0.18	0.04*	0.09	0.30
	OCRs	−0.02	0.80	−0.02	0.78	0.09	0.30	−0.03	0.70	−0.04	0.65
	Inhaled corticosteroids	−0.03	0.72	−0.03	0.71	−0.01	0.87	−0.04	0.60	−0.06	0.47
	NSAIDs	0.009	0.92	0.03	0.72	−0.06	0.47	−0.08	0.30	−0.05	0.59
	Depressive symptoms	−0.13	0.18	0.04	0.70	0.006	0.95	−0.09	0.34	−0.13	0.18
	Gender	0.17	0.05	0.17	0.04*	0.18	0.03	0.13	0.11	0.16	0.06†
	Dep. symptoms × gender	0.13	0.20	0.03	0.76	−0.11	0.24	0.03	0.75	0.15	0.12

^**p < 0.01,

^*p < 0.05,

^†p < 0.1.

¹Male=1, Female=0

^***Moderation analyses revealed no significant interaction of depressive symptoms × gender on basal cytokine levels.

3.2 Stimulated inflammatory responses

As expected, levels of all cytokines (IL-1β, IL-6, IL-8, IL-10, and TNF-α) were significantly elevated in response to LPS relative to basal levels (p < 0.001 for each). Mean values of non-transformed stimulated cytokine levels are presented in Table 2B.

3.2.1 Main-effects analyses

Analyses revealed a marginally significant main effect of depressive symptoms on stimulated IL-10 in both unadjusted (β = −0.15, t = −1.92, partial r² = 0.02, p = 0.06), and fully adjusted (β = −0.15, t = −1.19, partial r² = 0.02, p = 0.06) models, suggesting that higher depressive symptoms were marginally associated with lower stimulated levels of IL-10. No significant main effects of depressive symptoms on other stimulated cytokines were observed (p > 0.1) in either model.

3.2.2 Moderation analyses

Moderation analyses revealed a significant interaction of Depressive Symptoms × gender for stimulated TNF-α (β = 0.24, t = 2.71, partial r² = 0.04, p = 0.007) and stimulated IL-6 (β = 0.24, t = 2.58, partial r² = 0.04, p = 0.01). A marginally significant interaction was also evident for stimulated IL-10 (β = 0.17, t = 1.79, partial r² = 0.02, p = 0.07). After controlling for all covariates, this association remained significant for stimulated TNF-α (β = 0.24, t = 2.46, partial r² = 0.04, p = 0.015) and stimulated IL-6 (β = 0.21, t = 2.19, partial r² = 0.03, p = 0.03), but became non-significant for stimulated IL-10 (β=0.14, t = 1.55, partial r² = 0.01, p = 0.12). No other significant interactions were observed.

3.2.3 Gender-stratified regression analyses

Based on the finding of significant gender moderation, separate regression analyses were conducted to examine the relationships between depressive symptoms and stimulated TNF-α, IL-6 and IL-10 levels by gender. Results of unadjusted and fully adjusted analyses are presented in Table 5.

Table 5.

Multiple regression analyses of depressive symptoms predicting log-transformed stimulated cytokine levels in women and men.

		Women						Men
		Stimulated TNF-α		Stimulated IL-10		Stimulated IL-6		Stimulated TNF-α		Stimulated IL-6
		β	SE	β	SE	β	SE	β	SE	β	SE
Unadjusted Model	Depressive symptoms	−0.24*	0.005	−0.25**	0.006	−0.20*	0.005	0.23†	0.006	0.26†	0.006
Full adjusted Model
	Age	0.04	0.005	0.12	0.006	0.12	0.006	0.12	0.006	0.13	0.006
	BMI	0.09	0.006	−0.05	0.007	0.15	0.007	−0.18	0.009	−0.13	0.009
	Income	0.02	0.05	0.21*	0.06	0.12	0.6	0.05	0.06	−0.06	0.07
	Education	−0.08	0.1	0.02	0.11	0.05	0.11	0.09	0.14	0.06	0.14
	Hispanic	−0.009	0.12	0.09	0.13	0.07	0.13	−0.20	0.15	0.01	0.15
	White	0.11	0.21	−0.01	0.24	0.07	0.24	−0.16	0.22	0.04	0.23
	Antidepressants	0.05	0.21	0.16	0.24	0.09	0.24	−0.16	0.32	−0.13	0.34
	Statins	0.05	0.18	0.05	0.21	−0.02	0.21	−0.14	0.24	−0.11	0.25
	OCPs	−0.05	0.24	−0.002	0.28	−0.005	0.28
	Inhaled corticosteroids	0.06	0.21	−0.05	0.24	0.04	0.24	−0.06	0.43	0.009	0.45
	NSAIDs	0.07	0.36	0.07	0.41	0.04	0.41	−0.07	0.29	0.04	0.31
	Depressive symptoms	−0.26*	0.005	−0.22*	0.006	−0.19†	0.006	0.34*	0.007	0.30†	0.007

^**p < 0.01,

^*p < 0.05,

^†p < 0.1.

Men

As shown in Figure 1A, fully adjusted model analysis revealed a significant positive association between depressive symptoms and stimulated TNF-α (β = 0.34, t = 2.09, partial r² = 0.09, p = 0.04) and a marginally significant positive association with stimulated IL-6 (β = 0.30, t = 1.83, partial r² = 0.07, p = 0.07). There was no association between depressive symptoms and stimulated IL-10 (β = −0.04, t = 0.29, p = 0.7) in men. The inclusion in the model of smoking status or high blood pressure as covariates did not change these results.

Figure 1.

Associations between depressive symptoms and log-transformed LPS-stimulated levels of TNF-α, IL-6 and IL-10 in men (A) and women (B). (A) In men, higher depressive symptoms were associated with significantly higher levels of TNF-α and marginally higher IL-6 (p = 0.07), with no association observed for IL-10 (p = 0.7). (B) In women, higher depressive symptoms were associated with significantly lower levels of TNF-α, IL-10 and marginally lower IL-6 (p = 0.06).

Women

As shown in Figure 1B, analysis of the fully adjusted model revealed a significant (negative) association between depressive symptoms and stimulated TNF-α (β = −0.26, t = −2.59, partial r² = 0.06, p = 0.01), stimulated IL-10 (β = −0.22, t = −2.26, partial r² = 0.05, p = 0.02), and a marginally significant effect for stimulated IL-6 (β = −0.19, t = −1.91, partial r² = 0.04, p = 0.06). Adding menopausal status, smoking status, or high blood pressure to the model did not change these results.

4. Discussion

The present study examined whether depressive symptoms were associated with basal (systemic) inflammation and ex vivo inflammatory responses to LPS, and whether gender influenced these associations. Interestingly, LPS challenge induced distinct patterns of inflammatory responses within men and women in relation to reported depressive symptoms. In the present research, men showed a higher production of proinflammatory cytokines as depressive symptoms increased; however, production of the anti-inflammatory cytokine IL-10 showed no such association. In contrast, women exhibited a reduced production of proinflammatory cytokines and IL-10 as depressive symptoms increased. No associations were found between depressive symptoms and basal levels of inflammation.

Our findings among male participants are broadly consistent with existing literature. Suarez et al. [24] reported higher levels of LPS-induced TNF-α and IL-6 in men with higher depressive symptoms compared to men with lower depressive symptoms (no women were included in this study). In another study, clinically depressed men showed increased TNF-α and IL-6 production and decreased anti-inflammatory IL-4 production after stimulation with a combination of PHA and LPS, compared to non-depressed men [20]. Although anti-inflammatory cytokines are critical for resolving inflammatory responses, evidence regarding ex vivo anti-inflammatory responses (e.g., IL-10) in depression is scarce. In the present study, although men with higher depressive symptoms exhibited higher inflammation, IL-10 levels did not vary across depression scores in males. Given the anti-inflammatory nature of IL-10, the lack of a concordant increase in IL-10 in the presence of higher inflammation in males raises the possibility of an uncoupling of pro- and anti- inflammatory responses with increasing depressive symptoms in men.

The present results in men show an association between higher depressive symptomatology and greater inflammatory responses, and mirror what has been reported in the literature in men with respect to systemic inflammation. It should be noted that such associations have been reported less consistently in women. For instance, epidemiologic data from the Third National Health and Nutrition Examination Survey reported higher C-reactive protein (CRP) levels in US young adult men (age 18-30 years) with a history of major depression relative to men without a history of depression [16]; no such associations were observed in women. Other studies with large sample sizes have shown similar results in men, but again not in women [15,17,45]. Importantly, in a large longitudinal study with an average follow-up of 7.7 years, men with both depression and elevated CRP levels (3-20mg/L) had an increased risk of cardiovascular mortality and all-cause mortality, but no such associations were observed in women [46]. Considered together, these data suggest that elevated inflammation may help explain the link between depression and negative health outcomes, such as coronary artery disease or mortality, more commonly in men than in women. Future work is needed to test this possibility.

In the present study, women exhibited lower stimulated pro- and anti-inflammatory responses as depressive symptoms increased. Consistent with these findings, Cyranowski et al. [26] showed that women with higher depressive symptoms produced less PHA-stimulated IL-6, and less LPS-stimulated IL-1β and TNF-α, than women with lower depressive symptoms. As cytokine production is one of the earliest responses of the innate immune system to infection, reduced inflammation to mitogenic challenge may be indicative of a weakened or delayed inflammatory response to immune challenge in general.

Although the clinical significance of reduced ex vivo production of cytokines is not yet fully understood, there is some evidence indicating an association to poorer health. Reduced IL-1β production to LPS has been associated with delayed wound healing in Alzheimer’s caregivers who experienced chronic psychological stress relative to age-matched non-caregiver controls [47]. A follow-up study by Kiecolt-Glaser et al. [48] showed that dementia caregivers also mounted a weaker antibody response to influenza vaccination in vivo, and a weaker inflammatory response to LPS ex vivo (IL-1β, IL-2). These studies, both conducted in only women, suggest that a reduction in inflammatory responses may herald negative health effects such as slower healing and susceptibility to infection (at least in the context of chronic stress). Interestingly, both clinical depression and subclinical depressive symptoms have been related to negative health outcomes that include slower wound healing, infectious-disease susceptibility, and delayed recovery from infection [49-52]. Altered immunity is a potential pathway through which depression can lead to health problems [10]; hence in the present study, the reduced inflammatory responses observed in women with higher depressive symptoms may denote a higher risk for slower healing or delayed recovery from infection.

Some prior findings are not consistent with the present results in women. Suarez et al. [25] found an up-regulation of LPS-stimulated TNF-α and IL-8 in women with higher versus lower depressive symptoms. Another study [20] reported higher levels of IL-6 after ex vivo stimulation with a combination of PHA and LPS in depressed female patients compared to controls. Conversely, Miller et al. [32] reported no differences in stimulated inflammatory responses between depressed and non-depressed women. Overall, the current literature relating ex vivo inflammatory responses to depression is less consistent in women than in men.

Some possible reasons for these inconsistencies in past studies, that have reported on stimulated inflammatory responses (ex vivo) and depression, include: (1) heterogeneity of the sample (i.e., clinical depressed inpatients or outpatients, different types of clinical depression, non-clinically depressed individuals), (2) different methodologies for cytokine stimulation (e.g., type of mitogen used [LPS, PHA, both], incubation period of blood with mitogen(s), use of isolated peripheral blood mononuclear cells or whole blood), (3) cytokines measured (i.e., pro- and/or anti- inflammatory cytokines), (4) some studies included a single gender, some controlled for gender, some did not report gender differences, (5) small sample sizes that limited the ability to examine gender differences. It is important to note that the present study would have reported a null finding if gender was not specifically examined, which highlights the utility and importance of assessing the depression-inflammation relationship separately in men and women.

The divergent patterns of inflammatory responses in men and women with higher depressive symptoms were surprising; specifically, we did not expect to see an attenuated inflammatory response in women with higher depressive symptoms. Although the mechanisms underlying such differences were not investigated in this study, possible explanations exist based on the literature. An increased permeability of the gut to bacteria has been shown among depressed patients relative to non-depressed controls, indicating that an increased translocation of endotoxin into the bloodstream occurs in depression [53]. We speculate that this pre-exposure of blood to endotoxin might be responsible, at least in part, for the lower immune responses to LPS that were observed in women with higher depressive symptoms in this study. LPS tolerance is characterized by a reduced production of proinflammatory cytokines to endotoxin challenge. Importantly, females have been shown to develop tolerance to repeated exposures of LPS much more quickly than males [54]. Lower levels of cell surface expression of Toll-like receptor 4 (TLR 4) on macrophages in response to endotoxin has been demonstrated in females compared to males, and has been suggested to be a contributing factor to the higher development of LPS tolerance in females [55]. Thus, in the present study, women with higher depressive symptoms may have developed some tolerance to LPS due to a pre-exposure of blood to endotoxin via this “leaky” gut mechanism. Microbiome studies that examine such gender differences are needed to help explicate this issue.

Although a meta-analysis has shown a dose-dependent relationship between depression and basal inflammation (CRP, IL-6) in community based samples [3], no such associations were observed in the present study. Consistent with our results, Cyranowski et al. [26] and Steptoe et al. [56] found no associations between depressive symptoms and basal inflammation in middle-aged community samples. It has been suggested that a large sample size (larger than the present study) may be needed to reliably detect differences in basal (systemic) inflammation across depressive symptomatology when the sample is not clinically depressed [56]. Another potential reason for the lack of associations between depressive symptoms and basal cytokines in our study might be that the PROMIS depression 8-item scale does not include measures of somatic symptoms, and there is evidence that somatic symptoms of depression are particularly associated with inflammation [57]. This might be particularly relevant to our sample given that the majority of participants were Black/African-American (64%); prior studies have shown ethnic/racial differences in depressive symptoms, with depressed individuals who identify as Black/African-American reporting more somatic symptoms than White individuals [58-60]. Further exploration of the relationship between the symptom dimensions of depression and inflammation, as well as examining the moderating role of race/ethnicity in such associations, are needed in future studies.

There were several limitations in the current study other than those noted. First, blood was incubated with LPS for 2 hours; from this single time point, we cannot differentiate whether a reduction in the magnitude, and/or a delay in the release, of cytokines occurred in relation to higher depression symptomatology in women (i.e., TNF-α, IL-6 and IL-10) and men (i.e., IL-10). Future research with multiple time points of blood incubation with LPS would clarify this point. Second, because we did not have data on the menstrual phase of premenopausal female participants, this variable could not be included as a control variable in the analyses. We did include menopause status (pre, post) as a covariate and this did not change any of the results. Third, the sample consisted of individuals with varying depressive symptomatology that have not been confirmed to be clinically depressed; therefore, these results may not be generalizable to clinically depressed patients. Finally, non-biological explanations for gender differences in inflammatory responses may exist, such as psychosocial and behavioral factors that can vary by gender due to differences in socialization and experiences that occur from an early age [61]; these were beyond the scope of the current research, but will be important considerations for future research.

In summary, greater inflammatory responses without a concordant increase in anti-inflammatory cytokines were observed in men reporting more depressive symptoms, suggesting heightened inflammation in the face of immune challenge in men with higher depression. Such elevated inflammatory responses may be responsible, at least in part, for the reported negative health outcomes (e.g., cardiovascular disease, mortality) associated with depression in men. Conversely, a blunted inflammatory response was observed in women reporting more depressive symptoms. Although the clinical significance of this is unclear, lower inflammatory responses in women with higher depressive symptoms may be indicative of a reduced ability of immune cells to mount an appropriate response to challenge. Overall, these findings provide evidence of divergent patterns of inflammatory responses across depressive symptomatology in men and women. This study highlights the importance of examining both between-gender and within-gender effects when studying depression and inflammation. Finally, examining inflammation in a dynamic fashion (e.g., via antigen-stimulated immune responses) as it relates to depression may yield meaningful data beyond what can be learned from more static measures such as basal (systemic) inflammation.

Study highlights.

• Distinct inflammatory patterns occur across depressive symptoms in men and women.

• In men, higher depressive symptoms predict greater inflammatory responses.

• In women, higher depressive symptoms predict reduced inflammatory responses.