Emotional Support Moderates Early Life Adversity’s Impact on High Molecular Weight Adiponectin in Adolescent Females

Abstract

Early life adversity (ELA) has been linked to elevated inflammatory biomarkers and low circulating anti-inflammatory adiponectin levels among adults. Since inflammation is a key contributor to cardiometabolic diseases, lower adiponectin levels may be associated with an increased risk for these conditions. Women have a higher risk of cardiometabolic disease, so identifying mitigating factors specific to this population is crucial. Social support can mitigate the stress effects on cardiometabolic health, and emotional support is a particularly important functional component of social support for women. This study examined whether emotional support moderates the relationship between ELA and adiponectin levels among adolescent females. ELA and emotional support were assessed in 217 female participants (ages 13-17 years, M = 15.3) using standardized instruments, and adiponectin levels were measured from serum samples. Adjusting for age, race/ethnicity, annual income, and adiposity, emotional support moderated the relationship between ELA and adiponectin levels. Participants with more severe ELA who reported higher levels of emotional support exhibited higher adiponectin levels compared to average or lower levels of emotional support. These findings suggest that high emotional support during adolescence is not only a social protective factor but could also provide resilience against the biochemical impacts of ELA, including cardiometabolic diseases, highlighting the benefit of incorporating emotional support strategies in interventions for such high-risk youth.

Early life adversity (ELA) is defined as exposure to an environment during childhood that requires significant psychological, behavioral, or neurological adaptation from the expected environment by an average child (Duffy et al., 2018). Many experiences qualify as ELA, such as physical, sexual, or emotional trauma, parental separation, witnessing domestic violence, etc. In a recent report by the Centers for Disease Control and Prevention (CDC), two-thirds of US adults reported at least one adverse childhood experience, and one in six adults reported four or more experiences (Swedo et al., 2023). Exposure to ELA is associated with numerous negative health outcomes in adulthood, possibly due to the disruption of biological stress regulatory systems during early development, with sustained downstream effects, and engagement in health risk behaviors as individuals gain autonomy (Duffy et al., 2018; Taylor et al., 2011). One of the most significant negative health outcomes is the association between ELA and the development of cardiometabolic conditions (Elsenburg et al., 2023; Suglia et al., 2018; Tan et al., 2024; Zou et al., 2024). ELA programs the immune system toward a pro-inflammatory state, increasing risks for chronic inflammation and diseases in adulthood (Deighton et al., 2018; Suglia et al., 2018; Tan et al., 2024; Taylor et al., 2011). Among adults, ELA is associated with elevated levels of inflammatory biomarkers and low circulating adiponectin levels, which have anti-inflammatory properties (Deighton et al., 2018; Joung et al., 2014). Since inflammation is a key contributor to cardiometabolic diseases, decreased adiponectin levels may contribute to the elevated risk for cardiometabolic conditions (Alfaddagh et al., 2020; Lempesis & Georgakopoulou, 2023; Nguyen, 2020; Tsalamandris et al., 2019).

Adiponectin is a protein hormone in the soluble collagen family secreted by white adipose tissue and exhibits insulin-sensitizing, anti-atherogenic, and anti-inflammatory properties (Choi et al., 2020). High molecular weight (HMW) adiponectin is the most biologically active form (Zhu et al., 2010). Unlike traditional inflammatory markers like C-reactive protein (CRP), which are reactive indicators of current systemic stress, adiponectin acts proactively as an anti-inflammatory mediator, reflecting a physiological buffering response to chronic systemic stress. Adiponectin exerts anti-inflammatory and pro-inflammatory effects through activating AMPK (AMP-activated protein kinase) and NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) signaling pathways, respectively. Under normal conditions, the anti-inflammatory effect is stronger, resulting in enhanced lipid metabolism and decreased TNF-α (tumor necrosis factor - alpha) activation (Choi et al., 2020). Consequently, high levels of adiponectin reduce the risk of developing type 2 diabetes and cardiometabolic disease (Choi et al., 2020; Zhu et al., 2010). Adiponectin is inversely correlated with adiposity, particularly visceral fat, further supporting the impact of adiponectin on cardiometabolic health (Bacha et al., 2004; Guenther et al., 2014). One cross-sectional analysis found that the association between ELA and lower adiponectin levels was stronger among females than males, indicating that further investigation is needed in this population (Pitts et al., 2024).

Rates of cardiometabolic disease have increased worldwide and are linked to a rise in obesity and obesity-associated risk factors (Gerdts & Regitz-Zagrosek, 2019). In the United States, women have a higher prevalence of obesity than men (40.5% vs. 35.2%). Additionally, women are more likely to have undertreated hypertension, hyperlipidemia, and obesity than men, which results in disparities in the prevention and treatment of cardiometabolic diseases (Jones et al., 2023; Wakabayashi, 2017). Given the numerous risk factors influencing cardiometabolic diseases in women, exploring protective and risk factors during early developmental stages is crucial (Wakabayashi, 2017). Racial/ethnic differences in cardiometabolic health are particularly prominent among women, with non-Hispanic Black and Hispanic women experiencing higher rates of obesity compared to non-Hispanic White women (Hales, 2020). These disparities underscore the importance of including a racially and ethnically diverse sample when studying cardiometabolic health outcomes for women.

Adolescence is a sensitive period for development defined by heightened biological and psychosocial plasticity. Biologically, the onset of puberty triggers an increase in activity within the hypothalamic-pituitary-adrenal (HPA) axis and the adipose tissue environment, making metabolic biomarkers like adiponectin particularly susceptible to environmental influence (Alberga et al., 2012). Psychosocially, adolescence is defined by the maturation of the social-emotional brain, where adolescents transition from primarily parental support to broader social networks. Emotional support during this period is particularly influential for cardiometabolic outcomes because the psychosocial environment in adolescence has a role in combating the development of adulthood disease processes related to obesity (Narla & Rehkopf, 2019). Social support, a subjective indicator of an individual’s perception of how much support they receive from others, can buffer negative effects of ELA, possibly via diminished cardiometabolic responses to stress, improved immune function, a sense of control and/or health-related lifestyle changes (Christenfeld & Gerin, 2000). Social support has been linked to positive biological profiles, resulting in lower morbidity and mortality rates for various diseases, including cardiometabolic conditions (Barth et al., 2010; Freak-Poli et al., 2023; Strom & Egede, 2012; Uchino, 2006). In a prospective study of long-term cardiometabolic disease risk, participants were followed from age 10 to 29 years and ELA, assessed from age 0 to 10 years, was linked to cardiometabolic risk at age 29 (Lei et al., 2020). Additionally, parental emotional support in adolescence, but not adulthood, buffered ELA effects on cardiometabolic risk, suggesting that social support (emotional support in particular) during adolescence can mitigate the long-term negative effects of ELA on cardiometabolic health (Lei et al., 2020).

Studies examining the association between social support and cardiometabolic health focused primarily on global support, which appears to have less robust effects on ecological measures (Abreu et al., 2024; Bowen et al., 2013; Uchino et al., 2022). It has been argued that some components of social support are beneficial across situations and, therefore, may be more closely linked to cardiometabolic health (Bowen et al., 2013; Cohen & Wills, 1985). Informative, emotional, and instrumental forms of support have been identified as important functional components of social support (Cohen, 1988; Lakey & Cohen, 2000). Specifically, emotional support, which entails empathy, comfort, and validation of self-worth, appears to be the strongest protective factor against cardiometabolic risk and outcomes, especially among women (Bowen et al., 2013; Greenwood et al., 1996; Hosseini et al., 2021; Nordin et al., 2025; Wilson & Ampey-Thornhill, 2001). Compared to men, women tend to prefer, seek out, and respond more positively to emotional support, probably due to biological differences, social roles, and/or emphasis on socialization (Burleson, 2003; Matud et al., 2003; Shumaker & Hill, 1991).

Understanding how the long-term implications of ELA can be mitigated is crucial for improving health outcomes in high-risk populations. Prior research has established a link between ELA and reactive inflammatory markers (e.g., CRP, IL-6), however these often represent end-stage responses to chronic stress or acute reactions to isolated stressful events (Deighton et al., 2018; Joung et al., 2014). In contrast, HMW adiponectin acts as a primary, proactive endocrine buffer, providing insight into how to preserve cardiometabolic health before systemic inflammation becomes clinically evident. Current literature has focused on the impact of ELA on pro-inflammatory markers, but there remains a gap in knowledge regarding how ELA and social support interact to influence the preservation of protective metabolic factors, specifically adiponectin, during adolescence. In this study, we focus on HMW adiponectin as an early indicator of protection against cardiometabolic vulnerability during adolescence. We chose to highlight the emotional support component of social support because of the known association with protection against cardiometabolic risk factors and outcomes for young women (Bowen et al., 2013; Greenwood et al., 1996; Hosseini et al., 2021; Nordin et al., 2025; Wilson & Ampey-Thornhill, 2001). This study examined whether emotional support moderates the effect of ELA on HMW adiponectin levels in adolescent females, predicting that higher levels of emotional support have a protective effect, resulting in higher HMW adiponectin levels, among adolescents who experienced greater ELA.

Method

Data for this report are from a larger study on biobehavioral processes associated with obesity in adolescent females. This study was approved by the institutional review board (IRB #2017-3441) and registered in the ClinicalTrials.gov portal (NCT03369691).

Participants

Two hundred and seventeen adolescent female participants between the ages of 13 - 17 years were recruited for this study between 2017 - 2025 in Orange County, California. Recruitment was done through community outreach events, a local Children’s Hospital’s primary care network, mailers to affiliated community agencies, social media outreach, and word of mouth. An initial phone screening of the parent/legal guardian and adolescent was done to verify basic eligibility criteria. Participants were included in the study if they were African American, Hispanic/Latina, or Non-Hispanic White females between the ages of 13 - 17 years. Stage of puberty was assessed using the Tanner Scale, a standardized five-stage system used to categorize the physical development of secondary sex characteristics. A cutoff of Tanner Stage ≥III was utilized to ensure that all participants had reached mid-puberty, which is specifically characterized by breast development, the start of menstruation, and coarser pubic hair. Participants were excluded if they had a body mass index (BMI) below normal according to the CDC criteria, were trying to lose weight, or were on medications that affect appetite or the HPA axis (CDC, 2024). Participants with food allergies, binge/restrained eating, psychiatric disorders, or chronic medical problems were excluded. Additionally, pregnant females, or those suspected of being pregnant, were excluded. All participants who completed assessments for ELA, social support, and adiponectin, as well as potential covariates including age, race/ethnicity, annual income, and BMI were included for this report (Davis et al., 2016; Gardener et al., 2013; Li et al., 2016; Perng et al., 2019).

Procedures

Comprehensive information on procedures used in the larger study from which these data were collected is described elsewhere (Zurita et al., 2021); therefore, only procedures related to the collection of the data for this report are described. At the beginning of the first visit, parents or legal guardians provided informed consent, and adolescents provided assent for the study and all related procedures. Both parties completed demographic information and standardized interviews assessing adolescents’ exposure to adversity before age 10 years. Adolescents then completed a series of surveys, including social support. After completion of the surveys, adolescents’ height and weight were measured to calculate BMI. Participants completed the assessments and biological data collection over three separate visits spanning on average 3 weeks. The follow-up visit for blood collection was scheduled following the initial interview and at-home assessments, typically occurring within one to three weeks after the baseline visit. The exact timing of the second visit was prioritized based on ensuring an overnight fast and the participant’s schedule. Venous blood draw was used to measure adiponectin levels among other obesity-related biomarkers.

Measures

Early life adversity.

ELA was determined by the validated and reliable Childhood Adversity Interview V2, a semi-structured interview that obtains information on seven domains of adversity commonly experienced in childhood: separation from, or loss of, caregivers; prolonged illness or injury in the adolescent or family members; physical neglect; emotional abuse; physical abuse; sexual abuse/assault; and witnessing violence (Dienes et al., 2006). Each domain was scored by a trained rater on a scale of one to five, where one represents no adversity and five represents extreme adversity. Adolescents and one parent/legal guardian were interviewed separately regarding experiences of the adolescent participant in each domain before age 10. Scores from both interviews were used to make a consensus rating as to the degree of adversity in each domain. Raters were extensively trained and required to achieve a minimum inter-rater reliability of r = 0.85 on a training sample of interviews before interviewing and scoring study participants. Scores from each domain were summed so that possible total scores range from 7 to 35. The ELA measure was treated as an index of exposure across distinct domains; therefore, internal consistency metrics (e.g., Cronbach’s alpha) were not calculated, as items were not expected to be intercorrelated (Streiner, 2003).

Social support.

A modified version of the UCLA Social Support Inventory (SSI) was used to assess adolescent social support (Dunkel-Schetter, 1986). The SSI is designed to assess the various dimensions of support participants receive from their social connections within the past three months. The instrument focuses on three primary sources of support: a significant parental figure (or guardian), a close friend, and, if applicable, a romantic partner. Participants select an individual for each category. They then assess the support they receive from these providers across different areas of support: informational, instrumental, and emotional. Each item is rated on a frequency scale of one to five representing: never, rarely, sometimes, often, and very often. A composite score of each support area is calculated. This report relies on the emotional support subscale for which the score ranges from 16 to 80.

Adiponectin.

We specifically measured high molecular weight adiponectin; henceforth, referred to interchangeably with adiponectin. Serum and plasma were isolated from the fasting blood samples and stored at −80°C until assay. Adiponectin was measured from serum using the Mediagnost ELISA kit for Adiponectin E09, which employs a sandwich-assay technique involving two highly affine antibodies. The sensitivity of the kit was < 0.27 ng/mL (Range 0.094 to 0.59 ng/mL). The intra-assay variability was < 5%. The inter-assay coefficient of variation was 7.5% (SD = 1.6). Adiponectin assay was performed in duplicate using standard techniques (Immuno-Biological Libraries, 2023).

Anthropometry.

Height was measured using a stadiometer, with participants’ posture aligned with the Frankfort horizontal plane. Weight was recorded on a digital scale set to kilograms, ensuring participants wore only light clothing and no shoes. Measurements for height and weight were taken three times each by a trained research assistant, and the average of these readings was calculated to establish the final measurement. BMI was calculated for each participant using the CDC’s BMI Calculator for Children and Teens, which accounts for age, sex, height, and weight (CDC, 2024). Additionally, for a subset of participants (n = 200), visceral fat measurements via dual energy x-ray absorptiometry (DXA) were collected for a more accurate measure of adiposity.

Annual Family Income.

Annual family income was determined by demographics and background information questionnaire completed by the parent/legal guardian. Participants were asked to report the family's estimated gross income in the past year from all sources before taxes. There were nineteen categories with income ranges with less than $2,000 as the minimum and $100,000 and over as the maximum. For simplicity, we grouped our participants into four income categories: < $25,000; $25,000 - $49,999; $50,000 - $99,999; ≥ $100,000.

Race and Ethnicity.

Participants’ race and ethnicity were self-reported on the demographics and background information questionnaire completed by the parent/legal guardian. Racial categories included: Caucasian or White, Black or African American or Haitian, American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and Multiracial. Ethnic categories included Hispanic or Latino, and Not Hispanic or Latino.

Analytic Plan

Statistical analyses were conducted with the Statistical Package for the Social Sciences (SPSS) 29 software package and the PROCESS macro version 5.0 add-on. The study variables included continuous variables: ELA, emotional support, HMW adiponectin, BMI, visceral fat, age, and categorical variables: annual family income and race/ethnicity (Figure 1). Inter-item correlations were calculated using Pearson’s correlation coefficient to determine relationships among the continuous variables, with a significance set at p ≤ .05. A regression analysis was performed to investigate the moderating effect of emotional support on the relationship between ELA (independent variable) and HMW adiponectin (dependent variable). Age, annual income, BMI, visceral fat, and race/ethnicity served as covariates to adjust for the potential confounding influence of these sociodemographic factors on adiponectin levels and cardiometabolic health (Davis et al., 2016; Gardener et al., 2013; Li et al., 2016; Perng et al., 2019). Traditionally, BMI is used as a measure of adiposity, and given the relationship between adiposity and adiponectin, we chose to utilize it as a covariate in this analysis. Given that visceral fat is also a measure of adiposity, in the subset of participants with this measurement, in a separate regression model, we substituted BMI for visceral fat as a covariate to compare the effects of both covariates on adiponectin. The Johnson-Neyman technique was used to identify threshold emotional support scores that discriminated statistically significant from non-significant effects of ELA on adiponectin levels.

Figure 1. Conceptual Moderation Model.

This model demonstrates the hypothesized moderation effect of emotional support on the relationship between Early Life Adversity (ELA) and high molecular weight (HMW) adiponectin. Solid arrows represent the primary hypothesized pathways, and dashed arrows indicate the group of sociodemographic covariates (age, race, BMI, visceral fat, and income) used as statistical controls. Visceral fat is used as a covariate, substituting for BMI as a measure of adiposity, for a subset of the sample (n=200).

Prior to primary analyses, all continuous variables were screened for normality using Shapiro-Wilk tests and visual inspection of histograms and Q-Q plots. All continuous variables were assessed for skewness and kurtosis following the established acceptable parameters for linear regression (e.g., Skewness < ∣1.0∣; Kurtosis < ∣3.0∣) (George & Mallery, 2010). Some variables fell within the acceptable parameters: HMW Adiponectin (Skew: 0.500, Kurtosis: −0.310); Emotional Support (Skew: 0.616, Kurtosis: 0.908); and Age (Skew: −0.210, Kurtosis: −1.360). However, the primary independent variable, ELA (Skew: 1.733, Kurtosis: 3.065), and covariates [BMI (Skew: 1.731, Kurtosis: 4.209); and Visceral Fat (Skew: 2.336, Kurtosis: 9.626)] exhibited higher skewness. All variables were analyzed using their original raw scales to preserve a clinical interpretation of the results. To protect against this distributional skewness, bootstrapping with 5,000 bias-corrected resamples was used across all multivariate estimations. The predictors ELA and ES were automatically mean-centered via the PROCESS macro add-on before interaction model construction. Step-by-step analytic pipeline of primary and subgroup models is available in Supplementary Materials.

Results

Sociodemographic Characteristics

Participants included 217 adolescent females (ages 13-17 years, M = 15.27, SD = 1.45), with diverse racial/ethnic backgrounds, BMI, and annual family income. A significant proportion, 60.4%, experienced ELA in the mild range, possibly because participants with a history of psychiatric illness or medical problems were excluded. Additionally, over one-third of the sample reported more than one type of ELA. Common types of adversity included separation from or loss of a caregiver, physical abuse, illness, injury, or loss of a non-caregiver, and emotional abuse (Table 1). The estimated prevalence of ELA among children in California varies depending on age and the informant, with available data focusing on childhood abuse and/or neglect. In California, parent-reported exposure to ELA (primarily child abuse) was 36% in 2022 (CDPH, 2016). The categories of ELA reported in our study that were comparable to California statistics were emotional abuse, sexual abuse, physical abuse, and physical neglect, with a prevalence of 37.8%.

Table 1.

Sociodemographic Characteristics of Participants

Variables	Proportion within Sample (N = 217)
	n	%
BMI
Normal (18.5 – 24.9)	124	57.1
Overweight (25.0 – 29.9)	56	25.8
Obese (≥ 30)	37	17.1
Annual Family Income
< $25,000	26	12.0
$25 – 49,999	41	18.9
$50 - 99,999	55	25.3
≥ $100,000	95	43.8
Race/Ethnicity
Non-Hispanic White	63	29.0
African American	65	30.0
Hispanic/Latina	89	41.0
Reported Adversity
No Adversity	86	39.6
Any Type of Adversity	131	60.4
Two or More Types of Adversity	73	33.6
Types of Adversity
Separation/Loss from Caretaker	64	29.5
Physical Abuse	49	22.6
Illness/Injury/Non- Caretaker Loss	50	23.0
Emotional Abuse/Assault	46	21.2
Witnessing Violence	23	10.6
Physical Neglect	15	6.9
Sexual Abuse/Assault	14	6.5

Note. Reported adversity types are not mutually exclusive, so the combined total exceeds 100%

Bivariate Correlations

ELA did not correlate with emotional support or age. However, ELA was significantly correlated with BMI (r = .205, p = .002) and visceral fat (r = .238, p < .001). Emotional support was not correlated with ELA, age, or BMI. There was a modest, but statistically significant, negative correlation between emotional support and visceral fat (r = −.200, p = .004). Notably, HMW adiponectin did not exhibit a significant correlation with ELA (r = .025, p = .717) or emotional support (r = .051, p = .453). This lack of bivariate association provided the statistical rationale for exploring more complex multivariate relationships, including potential moderation and suppression effects. Individual adiponectin scores varied within our study sample (range: 0.01 to 6.85 μg/mL), but the majority were within the reference range (1.54 to 3.80 μg/dL). There are no established normal levels for HMW adiponectin, especially among adolescents, and the values vary depending on age, sex, BMI, and stage of puberty (Andersen et al., 2007; Cunha et al., 2023; Nakatani et al., 2008). BMI and visceral fat correlated negatively with adiponectin (r = −.417, p ≤.001; r = −.364, p ≤.001, respectively). BMI was significantly positively correlated with visceral fat, as expected, (r = .851, p ≤.001) and age (r = .209, p = .002) (Table 2).

Table 2.

Descriptive Statistics and Intercorrelations for Continuous Study Variables

		ELA	ES	HMWA	BMI	Age	VF†
ES	r	.093	--
	p	.174
HMWA	r	.025	.051	--
	p	.717	.453
BMI	r	.205*	−.120	−.417**	--
	p	.002	.078	<.001
Age	r	.094	.108	−.092	.209*	--
	p	.168	.113	.177	.002
VF†	r	.238**	−.200*	−.364**	.851**	.121	--
	p	<.001	.004	<.001	<.001	.088
BMI Model	Mean	8.32	48.71	2.85	25.45	15.27	--
	SD	1.77	10.21	1.53	6.18	1.45	--
	Variance	3.13	104.20	2.33	38.18	2.11	--
	Range	7 - 16	19 - 77	0.01 - 6.85	19 - 77	13 - 17	--
VF Subset Model †	Mean	8.340	48.870	2.880	--	15.230	247.150
	SD	1.770	9.950	1.540	--	1.450	162.990
	Variance	3.120	98.94	2.370	--	2.100	26566.080
	Range	7 - 16	25 - 77	0.010 - 6.850	--	13 - 17	64.510 – 1336.100

Note. N = 217

^†N = 200

^*p ≤ .05

^**p ≤ .001. ELA = Early Life Adversity, ES = Emotional Support, HMWA = High Molecular Weight Adiponectin, BMI = Body Mass Index, VF = Visceral Fat, SD = Standard Deviation

Moderation Analysis

BMI as a measure of adiposity.

The hypothesis was supported as the moderation model revealed a significant interaction of emotional support and ELA on adiponectin (b = 0.012, SE = 0.004, p = .005). The inclusion of the interaction term improved model fit (F = 7.946, p = .005), indicating the impact of the interaction is more than can be attributed to random variance alone. Consistent with the bivariate model, there were no significant main effects of ELA (b = 0.064, SE = 0.058, p = .264) and emotional support (b = −0.006, SE = 0.010, p = .565) on adiponectin within the multivariate model. This indicates that ELA and emotional support do not independently predict HMW adiponectin levels on their own; rather, the biological impact of ELA is conditional upon the level of emotional support available in the environment. Notably, while sociodemographic factors were included in the model, neither race/ethnicity (p = .704) nor annual family income (p = .884) emerged as significant predictors of adiponectin levels. This suggests that the observed interaction between ELA and emotional support operates independently of these sociodemographic factors within this sample (Table 3). Further investigation into the moderation effect through simple slopes analysis centering emotional support scores (M = 48.71) revealed that participants who reported higher levels of emotional support (SD = 10.207, score = 58.917) and higher ELA exhibited higher levels of adiponectin (b = 0.189, SE = 0.065, p = .004). In contrast, no significant relationship between ELA and adiponectin was observed at average (score = 48.71; b = 0.064, SE = 0.058, p = .264) or low (SD = −10.207, score = 38.503; b = −0.060, SE = 0.079 p = .447) levels of emotional support (Figure 2).

Table 3.

Linear Regression Table for Primary Model (BMI)

BMI Model		Unstandardized		Standardized	t	p	95% CI
		b	SE	b			Lower	Upper
1	(Constant)	5.928	1.116		5.314	<.001	3.729	8.127
	Race	−.030	.130	−.015	−.234	.815	−.286	.225
	Age	−.011	.067	−.010	−.160	.873	−.143	.122
	Income	−.081	.095	−.056	−.851	.396	−.267	.106
	BMI	−.104	.016	−.421	−6.513	<.001	−.135	−.072
2	(Constant)	6.103	1.101		5.541	<.001	3.931	8.274
	Race	−.048	.127	−.024	−.380	.704	−.300	.203
	Age	−.038	.067	−.036	−.559	.577	−.170	.095
	Income	−.014	.096	−.010	−.146	.884	−.204	.176
	BMI	−.103	.016	−.416	−6.329	<.001	−.135	−.071
	ELA	.066	.057	.077	1.156	.249	−.047	.179
	ES	−.005	.009	−.036	−.571	.568	−.024	.013
	ELA x ES	.012	.004	.183	2.819	.005	.004	.021

Note. SE = Standard Error, CI = Confidence Interval, BMI = Body Mass Index, ELA = Early Life Adversity, ES = Emotional Support

Figure 2. Simple Slopes Analysis of the ELA and Emotional Support Interaction.

The figure illustrates that the relationship between ELA and adiponectin is significantly moderated by the level of emotional support. The intersection point represents the “activating” point where the protective effect of emotional support becomes more pronounced as ELA increases. The ELA x Emotional Support Interaction is displayed across values one standard deviation below the average and one standard deviation above the average score for ELA and emotional support. Multivariable regression analysis controlled for age, income, BMI, and race.

Further analysis of the conditional effects of emotional support using the Johnson-Neyman technique identified two distinct thresholds for statistical significance across our sample. First, there is an upper emotional support score threshold of 3.854 points above the mean (score = 52.56), where the conditional effect of ELA on adiponectin becomes significantly positive (b = 0.112, SE = 0.057, p = .05). In this sample, 28.57% of participants scored above the upper threshold, indicating that the protective upregulation of adiponectin is active among individuals with high levels of emotional support. Second, a lower threshold emerged at 29.535 points below the mean (score = 19.18), below which the conditional effect turned significantly negative (b = −0.297, SE = 0.151, p = .05). Only a negligible fraction of the sample (0.46%) fell below this lower threshold, indicating that a severe lack of emotional support begins to unmask a negative biological impact from ELA on adiponectin levels. Ultimately, these dual thresholds further highlight the powerful role of emotional support as a moderator of the association between ELA and adiponectin levels (Table 4).

Table 4.

Conditional Effects of ELA at Levels of Emotional Support

Centered ES Score	b	SE	t	p	95% CI
Model A: Adjusted for BMI (N = 217; Mean ES Score = 48.71)
Sample Minimum (−29.714)	−0.299	0.151	−1.977	.049	[−0.597, −0.001]
Lower Threshold (−29.535)	−0.297	0.150	−1.971	.050	[−0.593, 0.000]
−1 SD Lower Support (−10.208)	−0.060	0.079	−.762	.447	[−0.217, 0.096]
Sample Mean (0.000)	0.064	0.058	1.120	.264	[−0.049, 0.178]
Upper Threshold (3.854)	0.112	0.057	1.971	.050	[0.000, 0.223]
+1 SD Higher Support (10.208)	0.189	0.065	2.903	.004	[0.061, 0.318]
Sample Maximum (28.286)	0.410	0.125	3.291	.001	[0.164, 0.656]
Model B: Adjusted for Visceral Fat (N = 200; Mean ES Score = 48.87)
Sample Minimum (−23.875)	−0.200	0.142	−1.413	.159	[−0.479, 0.079]
−1 SD Lower Support (−9.947)	−0.043	0.087	−.499	.618	[−0.215, 0.128]
Sample Mean (0.000)	0.069	0.063	1.092	.276	[−0.055, 0.193]
Upper Threshold (4.720)	0.122	0.062	1.972	.050	[0.000, 0.243]
+1 SD Higher Support (9.947)	0.180	0.069	2.600	.010	[0.043, 0.317]
Sample Maximum (28.125)	0.385	0.134	2.876	.005	[0.121, 0.649]

Note. ES = Emotional Support, SE = Standard Error, CI = Confidence Interval, BMI = Body Mass Index, SD = Standard Deviation

Visceral Fat as a measure of Adiposity.

Visceral fat was collected for a subset of participants in this study (N = 200). In this model, the hypothesis was also supported as the moderation model revealed a statistically significant interaction of emotional support and ELA on adiponectin (b = 0.011, SE = 0.005, p = .019). The inclusion of the interaction term improved model fit (F = 5.624, p = .018) for this subgroup analysis as well. Similarly to the BMI group, there was no main effect of emotional support (b = −0.006, SE = 0.011, p = .553) and ELA (b = 0.069, SE = 0.063, p = .276) on adiponectin levels in multivariate analysis (Table 5). Through the simple slopes analysis, a similar pattern to the primary BMI model emerged: for participants with high levels of emotional support (SD = 9.947 score = 58.82), higher ELA was significantly associated with higher adiponectin levels (b = 0.181, SE = 0.069, p = .01). No significant relationship between adversity and adiponectin was observed at average (score = 48.87; b = 0.069, SE = 0.063, p = .276) or low (SD = −9.947; score = 38.92; b = −.043, SE = 0.087, p = .618) levels of emotional support. Analysis of the conditional effects using the Johnson-Neyman technique within this subset only identified an upper emotional support threshold score 4.72 points above the mean (score = 53.59) as the threshold for statistical significance (b = 0.122, SE = 0.062, p = .05). In this sample, 26% of participants scored above this threshold, further demonstrating that the protective upregulation effect of emotional support on adiponectin is conditional on the level of ELA (Table 4).

Table 5.

Linear Regression Table for Subgroup Model (Visceral Fat)

VF Model		Unstandardized		Standardized	t	p	95% CI
		b	SE	b			Lower	Upper
1	(Constant)	5.192	1.172		4.431	<.001	2.881	7.504
	Race	−.132	.136	−.067	−.973	.332	−.401	.136
	Age	−.044	.071	−.041	−.616	.539	−.184	.097
	Income	−.202	.104	−.137	−.936	.054	−.407	.004
	VF	−.004	.001	−.386	−5.667	<.001	−.005	−.002
2	(Constant)	5.360	1.167		4.593	<.001	3.058	7.663
	Race	−.150	.135	−.076	−1.107	.270	−.416	.117
	Age	−.069	.072	−.065	−.961	.338	−.211	.073
	Income	−.129	.106	−.088	−1.219	.224	−.338	.080
	VF	−.004	.001	−.389	−5.483	<.001	−.005	−.002
	ELA	.069	.063	.079	1.091	.277	−.055	.193
	ES	−.007	.011	−.043	−.616	.539	−.028	.015
	ELA x ES	.011	.005	.170	2.371	.019	.002	.021

Note. SE = Standard Error, CI = Confidence Interval, VF = Visceral Fat, ELA = Early Life Adversity, ES = Emotional Support

Discussion

This study examined the moderating influence of emotional support on the relationship between ELA and HMW adiponectin in female adolescents. Adjusting for age, annual family income, adiposity, and race/ethnicity, multivariate analysis revealed that the relationship between ELA and HMW adiponectin is conditional upon perceived emotional support. Utilizing both measures of adiposity, namely BMI and visceral fat, we obtained the same pattern of results, further strengthening our findings. In depth analysis using the Johnson Neyman technique confirmed distinct regions of significance across the levels of emotional support demonstrating the protective upregulation effect of high emotional support and the consequences when emotional support is severely lacking. Higher levels of ELA were positively associated anti-inflammatory adiponectin concentrations when emotional support was high, demonstrating the robust protective upregulation of psychosocial environment. On the other hand, when emotional support was extremely low, a trend of down-regulation of adiponectin with higher levels of ELA emerged.

Conditional Effect of ELA on Adiponectin

The lack of significant standalone main effect of ELA on adiponectin within the multivariable model initially appears to contradict existing literature, which established ELA as a risk factor for poor health outcomes and the development of cardiometabolic conditions in adulthood, such as obesity, hypertension, type 2 diabetes, and cardiometabolic disease (Suglia et al., 2018). Clinical models have frequently proposed that suppressed anti-inflammatory adipokine levels serve as a possible mechanism for the link between ELA and cardiometabolic disease (Deighton et al., 2018; Joung et al., 2014). Adiponectin exerts anti-inflammatory and insulin-sensitivity effects on the body, resulting in a lower baseline risk for cardiometabolic diseases (Alfaddagh et al., 2020; Lempesis & Georgakopoulou, 2023; Nguyen, 2020; Tsalamandris et al., 2019). Our study findings reveal that the relationship between ELA and anti-inflammatory adiponectin levels is more nuanced than a uniform, linear, downregulation relationship within the adolescent population. The conditional impact of emotional support on the relationship between ELA and adiponectin in adolescence extends the evidence on the association between ELA and physical health, which is focused primarily on adult samples. Capturing a snapshot of this mechanism during adolescence, a sensitive period when several risk and protective factors impact the development of adulthood disease or resilience, our findings offer insight into the importance of psychosocial environment (Alberga et al., 2012; Narla & Rehkopf, 2019). A deeper understanding of the unique factors that have an influential role during this developmental period offers a perspective on lowering risk for future adverse health outcomes.

Moderation Effect of Emotional Support

The moderation effect of emotional support on the relationship between ELA and adiponectin is consistent with the literature in adults demonstrating the role of emotional support as a protective factor, resulting in lower morbidity and mortality rates associated with cardiometabolic conditions (Greenwood et al., 1996; Nordin et al., 2025). Current evidence has established a link between ELA and susceptibility to cardiometabolic disease in adulthood (Elsenburg et al., 2023; Suglia et al., 2018; Tan et al., 2024; Zou et al., 2024). However, this study contributes a unique cross-sectional snapshot of the physiological mechanisms influencing cardiometabolic health during early development. We demonstrated that the moderating influence of emotional support is detectable at the biomarker level before the onset of clinical cardiometabolic disease. Our findings provide insight on how emotional support may not just improve psychosocial outcomes but actively preserves anti-inflammatory mechanisms during a critical period of adipose tissue maturation, which can influence reproductive health and cardiometabolic status into adulthood. The protective role of emotional support in our sample of adolescent females further supports current literature highlighting that women tend to seek out and respond to emotional support more positively than their male counterparts (Burleson, 2003; Matud et al., 2003; Shumaker & Hill, 1991). Additionally, emotional support tends to have a higher impact among females than males in reducing the morbidity and mortality related to cardiometabolic diseases (Bowen et al., 2013; Hosseini et al., 2021; Wilson & Ampey-Thornhill, 2001). Overall, social support is associated with higher levels of physical activity and overall well-being among adolescents (Gill et al., 2018; King et al., 2008), which potentially can reduce the risk for obesity and cardiometabolic diseases in adulthood.

Practical and Clinical Implications

In the current study, participants with higher levels of ELA who also had higher levels of emotional support exhibited higher adiponectin levels. Thus, emotional support, in the context of high ELA, could reduce the risk of type 2 diabetes and cardiometabolic disease in this population (Choi et al., 2020; Zhu et al., 2010). Extant literature on ELA and associated poor health conditions does not account for gradations in ELA severity, and our finding suggests that the severity of ELA matters when considering emotional support’s moderating effects on the adiponectin biomarker. The protective effect of emotional support on adiponectin being activated at higher levels of ELA, has potential implications for targeted interventions for adolescent females with a history of significant ELA. Our findings suggest that low-intensity social interventions may be insufficient to counteract the biological consequences of ELA, and programs should aim to bolster high emotional support networks for adolescents. Overall, social support interventions (i.e., trauma-informed family therapy or peer-mentorship programs) should be seen not only for their impact on mental health but also for their potential to preserve the body’s anti-inflammatory mechanisms and impact future cardiometabolic health. Clinically, these results suggest that health care providers should view emotional support in adolescents as more than a psychological benefit, but also as a shield for metabolic health. Identifying adolescents with a history of adversity who also lack adequate emotional support may help providers prioritize those at the highest risk for future cardiometabolic disease, allowing for early-stage physiological monitoring. An unexpected finding of this study was the negative main effect of emotional support on adiponectin levels. However, on further examination (see Johnson Neyman data output cut off points, and Figure 2), only participants with a low level of emotional support in the context of high ELA exhibited lower adiponectin levels (negative beta value), whereas participants with a high level of emotional support exhibited higher adiponectin levels.

Statistical Considerations

Our study has several statistical nuances that may warrant clarification regarding the observed predictive value and relationships among variables. While the interaction coefficient (b = 0.012) may appear numerically small, it must be interpreted within the context of the differing scales of our study variables. This small interaction coefficient is a mathematical consequence of mapping psychosocial survey scores (i.e., 58-point range for emotional support scale) onto the narrow clinical range of HMW adiponectin (0.01 to 6.85 μg/mL). The impact of the interaction is better reflected by the F-score, which indicates that the ES x ELA interaction is more than can be attributed to random variance, refuting the idea that the effect size is too small to be meaningful. Furthermore, utilizing the Johnson-Neyman technique, multiple specific thresholds of emotional support were identified where robust support has a protective upregulation effect and a lack of support reveals the detrimental downregulation effect, highlighting emotional support as a powerful moderator.

Limitations

This study recruited adolescent females who had no major medical or psychiatric diagnoses and were not taking any medications that could impact adiponectin or other biomarker levels, suggesting that even in a young, healthy sample, the negative association between ELA and adiponectin levels can be observed. Despite our novel findings, several limitations must be acknowledged. Given the cross-sectional study design, we cannot definitively establish a causal pathway between early life adversity, emotional support, and adiponectin levels. Consequently, our findings should be interpreted as evidence of a significant association rather than a definitive causal pathway.

Secondly, our sample was characterized by a relatively high socioeconomic status, with 43.8% reporting an annual family income exceeding $100,000, which may limit the generalizability of our findings to lower-income populations who often face compounded systemic stressors. Of note, this study was conducted in Southern Orange County, California, an area known for having one of the highest costs of living in the country. Additionally, a large proportion of our participants came from families with multiple dependents within the home. Notably, for a family of four in Orange County, CA, an annual income of $135,350 and below is considered low income (HCD, 2025). Therefore, considering these additional factors, the annual income of $100,000 for this sample may still be considered relatively low income. Additionally, annual family income was controlled for in the regression analysis and did not emerge as a significant predictor of adiponectin levels, suggesting that the moderating role of emotional support functions regardless of financial resources. Furthermore, the observation of significant biological protective upregulation in a relatively low-risk, healthy sample suggests that these effects may be even more pronounced in higher-risk populations where the physiological weathering effect of sociodemographic stressors is more pronounced.

Thirdly, this study’s reliance on ELA retrospective reports introduces the possibility of a recall bias, especially as adolescents and parents/guardians were asked to recall events prior to the age of 10. While both informants provided information independently in a structured interview in which, for the most part, they corroborated each other’s information, prospective studies would be ideal for validating these associations. Additionally, the range of scores for ELA was relatively low (probably because we excluded participants with a history of medical or psychiatric conditions) and limited to this specific sex and age group. Therefore, these findings may not generalize to males or to those with more severe clinical histories of trauma. Regardless, the protective upregulation effect of emotional support on adiponectin levels within the limited range of ELA, adjusting for potential confounding sociodemographic variables, offers some confidence in the finding and supports further examination of emotional support’s moderating influence on a wider range of ELA. Finally, our measure for emotional support was assessed over the past three months rather than over a longer period, which provides only a snapshot of the participants' support systems and not a comprehensive view of the support received over the lifespan.

Conclusion

The findings from the current study, in conjunction with the extant literature on cardiometabolic health, offer insights into how social support intervention strategies might serve as buffers against cardiometabolic disease among adolescent females exposed to ELA. This preliminary finding in a relatively healthy, non-referred sample is promising, but further research is needed to assess whether these findings will apply to a clinical population. Future research should include a broader demographic range, investigating the effects of emotional support on adiponectin in male adolescents or adults with a history of ELA. Different sources of emotional support and other forms of social support should also be considered. For example, it is essential to understand which source of support (from family, friends, or partners) is most effective in mitigating the adverse effects of ELA on cardiometabolic health. Additionally, it is important to further explore which types of ELA are most predictive of lower adiponectin levels. Expanding the evaluation period for emotional support to include both the recent timeframe and over the lifespan would provide a better understanding of emotional support’s short- and long-term effects. Prospective studies examining the influence of emotional support on adiponectin levels, in the context of ELA, and its impacts on vulnerability to, or protection against, cardiometabolic conditions are likely to have a greater clinical impact.