Obesity is influenced by life-style, genetic, and environmental risk factors (1) and remains a critical public health issue nationwide (2,3). The burden of the obesity epidemic falls disproportionately on minorities; specifically, Hispanic and non-Hispanic black youth have a higher odds of obesity compared with non-Hispanic white and Asian youth (3). Children diagnosed as obese, based on body mass index (BMI) at or above the 95th percentile, have an increased risk of developing cardiovascular disease and diabetes later in life and becoming obese as adults (4,5).
Increasing evidence suggests that early-life measures of the home environment, including family stability, may influence risk of obesity (6–8). Aspects of the home environment have also been associated with the production of biologically active markers of obesity secreted by adipose tissue, including the adipokines: leptin and adiponectin (9–11). The protein hormone adiponectin, which is primarily derived from adipose tissue, has antidiabetic, antiatherogenic, anti-inflammatory, and insulin-sensitizing properties that mediate the pathogenesis of metabolic and cardiovascular disease (12,13). Studies in adults have shown inverse relationships between self-reported early-life measures of a poor home environment, including abuse, neglect, and household dysfunction, with adiponectin levels assessed in adulthood (9,11,14). Additional research has reported lower adiponectin levels in adult study participants with depressive symptoms in comparison with controls (15–17).
Leptin, another adipokine, is a hormone predominantly secreted by adipose tissue and has also been found in other tissues such as ovaries, placenta, and the brain (18,19). Leptin can moderate energy homeostasis, which can be altered during depression, through interaction with central circuits in the hypothalamus (16,20). Similar to adiponectin, a number of studies in adults have reported associations between leptin levels and early-life family or home dysfunction (9,10,14) and depressive symptoms in adulthood (21). In contrast to adiponectin, the amount of leptin secreted into circulation is higher in adults who have experienced poor home environments or mental illness. However, the only prospective study to date in children (22), using data from 170 12-year-old children from Great Britain, reported suggestive yet insignificant lower levels of leptin in children exposed to early-life maltreatment.
In the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) cohort, we previously reported that 7-year-old children whose mothers consistently demonstrated depressive symptoms had 2.4 times the adjusted odds of overweight/obesity compared with those whose mothers never experienced symptoms of depression (23). Adipokine trajectories from birth to age 9 years have been characterized in a subset of CHAMACOS participants and have shown three distinct risk groups for each adipokine with complex age dynamics through childhood (24). Thus far, studies on aspects of the home and family environment and adipokines have focused primarily on adult populations, are cross-sectional in nature, are predominantly conducted in African American and white populations, and primarily are focused on symptoms of mental health and family dysfunction, rather than protective home environment factors.
In the current study, we examine the relationship of maternal depressive symptoms, as well as the positive influences of social support and a healthy home environment, with BMI and adipokines in children from the CHAMACOS cohort, a Mexican American birth cohort with high levels of childhood obesity (25). We explored these specific psychosocial factors based on previous research that has provided preliminary evidence of how early-life quality of the home environment (26–28), maternal depression (23,29), and offspring social support (30) can impact obesity risk in children, making use of available data measured at multiple time points in early childhood in the CHAMACOS cohort. We hypothesize that children, who were raised in a better home environment with less stress and adversity and whose mothers had better social support and exhibited fewer symptoms of depression, would have increased adiponectin levels and lower BMI and leptin levels through decreased activation of the hypothalamic-pituitary-adrenal (HPA) axis, which can alter levels of hormones that impact obesity risk. We anticipate that recent and previous exposures to home environment factors will influence adipokines and obesity status, given the increasing awareness of how early-life exposures can alter future disease risk. Furthermore, we expand our previous analysis of adipokines to a larger number of participants (n = 326) and extend follow-up from the ages 9 to 14 years.
The CHAMACOS is a longitudinal cohort established to assess the health effects of environmental exposures, such as pesticides, on child growth and development (31). A total of 601 pregnant women were enrolled in the study between October 1999 and October 2000. Mothers enrolled in the CHAMACOS study were older than 18 years, less than 20 weeks of gestation at enrollment, eligible for low-income health insurance (Medi-Cal eligible), Spanish or English speaking, and intended to deliver at the county hospital. Of the 536 live births, the number of children with adipokine measurements at different time points was as follows: delivery (adiponectin, leptin; n = 217, n = 211), age 5 years (n = 227, n = 201), age 9 years (n = 250, n = 244), and age 14 years (n = 233, n = 228). The total number of children included in this study was 326, and most had adipokine data for more than three time points (Figure 1).
All study procedures were approved by the Committee for the Protection of Human Subjects at the University of California, Berkeley. Informed consent was obtained from all mothers; oral assent was obtained from children from ages 7 to 11 years, and written assent was obtained starting at 12 years.
Women were interviewed at approximately 13 and 26 weeks of gestation, after delivery, and during the developmental assessments of the children at ages 6 months and 1, 2, 3.5, 5, 7, 9, 10.5, and 14 years. To accommodate the study participants, questionnaires and assessments were administered in either English or Spanish by bilingual, bicultural interviewers. Information gathered at each interview included sociodemographic factors, health status, diet, and an assessment of exposure risk factors.
Maternal depressive symptoms and protective factors, including social support and HOME scores, were measured throughout childhood and early adolescence in CHAMACOS participants. Maternal depressive symptoms were assessed using the Center for Epidemiological Studies Depression Scale (CES-D) (32) when the children were 1, 3.5, 7, and 9 years old (Figure 1). The recommended cutoff score of 16 (33) was used to identify mothers exhibiting symptoms of depression and the continuous CES-D scores (max range of 0–60, with a score of 60 representing the worst possible score) were used in regression analyses. Social support was characterized in mothers when children were 1 and 5 years using the Duke-University of North Carolina Functional Social Support Questionnaire, with scores representing an average of responses on a 1 to 5 scale to a validated eight-item instrument designed to measure functional social support. Higher scores indicate increased social support (34). Characteristics of the home environment conducive to child development were determined at ages 6 months and 1, 2, 3.5, 7, 9, and 10.5 years using the Home Observation for the Measurement of the Environment (HOME) (35). For earlier visits (ages 6 months–2 years), we used the Infant-Toddler HOME inventory that combines observations of the home environment and mother child interactions by trained bilingual interviewers with questions about toys, books, and the overall home learning environment that were asked of the mothers. The total HOME scores are a combination of the responsivity, avoidance, learning, and involvement subscales with a range of 16 to 45 points. At later time points (3.5–10.5 years), we used a short form of the HOME inventory that was based only on maternal questionnaires that was composed of emotional and cognitive subscales with a range of 5 to 24 points. Higher social support and HOME scores indicate more protective environments, whereas increased depression scale scores refer to more severe maternal depressive symptoms. Cronbach's α, a measure of internal consistency, was 0.77, 0.57, and 0.79 for maternal depressive symptoms, social support, and HOME scores, respectively, measured throughout childhood in the CHAMACOS cohort.
Children were weighed and measured by trained research staff at ages 5, 9, and 14 years. We measured the children's barefoot standing height to the nearest 0.1 cm using a stadiometer. Standing weight was measured to the nearest 0.1 kg using an electronic scale at age 5 years (Tanita 1582) and a foot-to-foot bioimpedence scale at ages 9 and 14 years (Tanita TBF-300A Body Composition Analyzer). BMI was calculated as kilogram per square meter. The Centers for Disease Control and Prevention age- and sex-specific percentiles were used to categorize children as underweight, normal weight, overweight, or obese and BMI z-scores were used in analyses (36).
Adiponectin and leptin were assessed in blood specimens stored at −80°C collected from CHAMACOS participants at birth (cord blood) and repeatedly at ages 5, 9, and 14 years. We selected samples for analysis based on individuals with plasma available at a given CHAMACOS assessment. The subset of CHAMACOS participants used in this analysis was not significantly different from the main cohort in most demographic characteristics (child sex, baseline poverty, education, gestational age, parity, child overweight status). However, mothers in this sample tended to be slightly older at delivery and had higher average BMI than those not included in the analyses. Plasma adipokines were measured using enzyme-linked immunoassay (ELISA) RayBiotech Human Adiponectin and Human Leptin kits (RayBiotech Inc, Norcross, GA). The protocol developed by the manufacturer was used, with some previously described modifications (37) to increase precision for assessment in children: (a) the standard curve for adiponectin was narrowed to obtain better resolution at smaller readings, (b) the standard curve for leptin was broadened, and (c) plasma samples in the leptin analysis were more diluted. Final dilutions for the RayBio Human Leptin ELISA and the Human Acrp30 ELISA were 1:70 and 1:30,000, respectively. Absorbance values were obtained at a wavelength of 450 nm, with an upper absorbance cutoff of 4.0 optical density units. The minimum detectable concentrations for adiponectin and leptin were 10 and 6 pg/ml, respectively. Internal laboratory controls were included on each plate to ensure reproducibility between the experimental runs, with average intra- and interplate coefficients of variations of less than 5% and 14%, respectively. All plasma samples were measured in duplicate.
We log10-transformed leptin concentrations in CHAMACOS study participants for use in regression and correlation analyses because they were right skewed. Adiponectin concentrations did not require this adjustment. The interrelationships between adiponectin and log-transformed leptin levels measured at birth and ages 5, 9, and 14 years were assessed by Pearson correlation coefficients. Differences in mean adipokine levels by sex were determined by t tests, and longitudinal trends of adipokines across childhood were characterized by generalized estimating equations (GEEs) with an exchangeable correlation structure (38). Because leptin demonstrated a difference by sex, we adjusted for sex in the GEE models with leptin as the outcome.
Linear regression models with robust estimates of the SEs were used to determine the association of continuous measures of maternal scores on the CES-D, HOME scores, and total social support scores with concurrent or future child adipokine levels and BMI z-scores, which were assessed at ages 5, 9, or 14 years. CES-D maternal depressive symptom, Duke-University of North Carolina social support, and raw HOME scores were assessed as continuous variables in regression models. We included additional covariates in the models that were identified in previous literature in CHAMACOS (39) and in research assessing the relationship between adversity and obesity trends (6).
Specifically, two regression models were performed to assess the relationship between each independent home environment variable (i.e., continuous measures of maternal depressive symptoms, raw HOME scores, and total social support) and outcome (adiponectin, log-transformed leptin, and BMI z-score): Model 1, the unadjusted model, and Model 2, adjusting for child sex and maternal pregnancy variables. Maternal pregnancy variables in Model 2 included continuous (i.e., maternal pre-pregnancy BMI, years in the United States, pregnancy sugar-sweetened beverage consumption per week), categorical (maternal education level, poverty status), and binary (smoking and alcohol consumption) variables. The maternal education variable was coded as shown in Table 1. Maternal baseline poverty was coded as at or below poverty, greater than poverty to 200% of the poverty level, and greater than 200% of the poverty level. As a sensitivity analysis, we also ran Model 2 with an adjustment for poverty assessed concurrently with the outcome of interest (adipokines or BMI z-score). The results were similar to those of Model 2; thus, they were not presented in the tables.
In addition, regression was used to assess the relationship between longitudinal trends of measures of the home environment and the outcomes (adipokines and BMI). We used GEE-derived SEs based on exchangeable working correlation models to provide robust inference in the presence of repeated measures. For both the maternal depressive symptom and HOME scores, we generated a new variable by using the values for each that were measured at the same time or before the assessment of adipokines or BMI. Specifically, maternal depressive symptom measurements at child age 3 years corresponded to obesity measurements at age 5 years, and depressive symptoms at age 9 years corresponded to adipokine and BMI measurements at ages 9 and 14 years. We used the same time points for the HOME score variable, with the exception of the correspondence of the 10.5-year HOME score with the 14-year outcome assessment. We used social support values from the 5-year child assessment because it was the closest value for all time points of the outcomes. Henceforth, we will designate these variables as time-varying measures of the home environment to distinguish them from the original variables. We used GEE models to assess the relationship between the time-varying variables for maternal depressive symptoms, social support, or home environment and repeated measurements of adipokines and BMI from ages 5, 9, and 14 years. In the models, we adjusted for age and child sex, because they have been previously shown to influence adipokine and BMI trends and would be the most likely confounders in the current analysis. The following model was used:
where Yij is either the BMI z-score or adipokine level for the i-th subject at the j-th anthropometric or adipokine assessment in childhood (j = 5, 9, 14). X 1ij is the time-varying home environment value (i.e., HOME, social support, or maternal depressive symptom score) for the i-th subject and the j-th assessment, where j is time point of the home environment measurement that is either the same time as the adipokine or BMI assessment or the measurement prior. X 2, X 3 and b 2, b 3 represent the covariates and their respective slopes for the age and child sex variables. eij is the residual error term.
As a sensitivity analysis, we ran additional models, including one with all the time-varying exposure variables (maternal depressive symptoms, HOME score, and social support) and another model with poverty at outcome assessment. We also examined the potential of social support to moderate the relationship between home environment factors and obesity. We generated interaction terms by multiplying the maternal depressive or HOME time-varying scores by the social support variable and included the new variable in analyses. We also ran mixed-effect regression models to confirm GEE findings. Given the sample size, more complicated longitudinal models did not seem to have sufficient power to detect changes in measures of obesity related to previous or concurrent home environment parameters.
We controlled for the false discovery rate (FDR) in the analysis of individual assessments of measures of the home environment with child adipokines and BMI. All statistical analysis was performed in STATA (version 12.1; STATA Corp, College Station, TX). Statistical significance was set at p < .05.
Study Sample Characteristics
Demographic and anthropometric data for CHAMACOS mothers and children (n = 326) with adipokines assessed at least once during childhood (delivery or ages 5, 9, or 14 years) are shown in Table 1. At pregnancy, most mothers were young, had lived in the United States for 10 or fewer years, did not obtain a high school diploma, and had overweight or obesity. In our study, 48% of participants were boys and 52% were girls. More than half of the children at each time point had overweight or obesity: 53% of 5 year olds, 55% of 9 year olds, and 54% of 14 year olds.
Age and Sex Adipokine Trends
Adiponectin levels were highest at birth (M (SD) = 101.2 (32.9) μg/ml), lower as children got older (age 5 years: M (SD) = 36.4 (16.2) μg/ml; age 9 years: M (SD) = 40.7 (18.1) μg/ml), and were lowest in adolescence (age 14 years: M (SD) = 23.7 (11.0) μg/ml) (Table 2, Figure 2). Leptin levels at birth were high (M (SD) = 18.6 (18.7) ng/ml), which could indicate an influence of maternal levels of leptin. Leptin levels were much lower at age 5 years (M (SD) = 3.6 (3.3) ng/ml) and were higher as the children aged (age 9 years: M (SD) = 15.0 (16.7) ng/ml; age 14 years: M (SD) = 22.2 (17.2) ng/ml). Adiponectin levels did not vary by sex, with the exception of a borderline significant (p = .049) difference at age 14 years with higher levels in girls compared with boys (Table 2). Log-leptin levels from delivery through age 14 years were consistently higher in girls. Adiponectin levels were moderately and positively correlated across different time points, becoming stronger as children aged (Table 3). Adiponectin levels at a given time point were most correlated with measurements at adjacent time points (Table 3). Similar trends were observed in the leptin measurements. At each assessment, adiponectin levels were significantly and inversely associated with leptin levels, with the exception of delivery levels, when the negative trend was not significant (p = .70) (Figure 3). Age was a strong predictor (p < .001) of leptin and adiponectin levels in the GEE model. For adiponectin, B value was close to −5.0 (GEE B = −4.975, robust SE = 0.153), and for leptin, B value was 0.013 (0.003), respectively.
Maternal Depressive Symptoms and Home Environment Protective Factors
Among the mothers, 50%, 43%, 26%, and 26% were at risk of clinical depression with CES-D scores of 16 or higher when children were assessed at 12 months, 3.5 years, 7 years, and 9 years, respectively (Table 4). Scores reported at 12 months could also reflect symptoms of postpartum depression. Median maternal social support scores were high (1 year: 4.0; 5 years: 4.3). Median HOME scores varied across the collections, reflective of the use of the Infant-Toddler HOME inventory for earlier visits (ages 6 months–2 years) and the short form of the HOME inventory for the later time points (3.5–10.5 years). Specifically, median scores ranged from 26 to 36 and 14 to 18 for the earlier and later visits, respectively.
Individual Assessments of Measures of the Home Environment With Child Adipokines and BMI
We assessed the relationship of measures of the home environment across childhood at individual time points with obesity, as measured by BMI z-score, and adipokines. Maternal depressive symptoms at child ages 3.5 and 9 years were positively associated with BMI z-score at ages 9 and 14 years (Table 5 and Supplemental Table 2, http://links.lww.com/PSYMED/A530). A one-unit increase in CES-D score when children were 3.5 and 9 corresponded to a 0.01-point increase in BMI z-score at age 9 years (3.5 years: 95% CI = 0.003 to 0.02; 9 years: 95% CI = 0.002 to 0.02) and age 14 years (3.5 years: 95% CI = 0.003 to 0.02; 9 years: 95% CI = 0.006 to 0.03). Social support and HOME scores were not significantly associated with 9-year BMI z-scores. However, a unit increase in social support at age 5 years corresponded to a significant 0.2-point decrease in BMI z-score at age 14 years (95% CI = −0.3 to −0.05). In addition, a one-unit increase in HOME score at age 9 years was associated with a 0.05-point decrease in BMI z-score at age 14 years (95% CI = −0.09 to −0.005). The associations of maternal depressive symptoms at age 9 years and social support at age 5 years with BMI z-score at age 14 years remained significant after FDR adjustment (all p < .01).
In models with adiponectin as the outcome, we observed a consistent trend across models of higher HOME score, indicating a more enriched home environment, with increased adiponectin at age 9 years. Positive associations with similar magnitudes of effect sizes were with HOME scores at ages 3.5, 7, and 9 years, but only age 3.5 years was significant in crude models, indicating that a one-unit increase in HOME score at age 3.5 years corresponded to a 0.9-point increase in adiponectin at age 9 years (95% CI = 0.03 to 1.8). Similarly, HOME scores were positively associated with adiponectin at age 14 years, with larger-effect sizes observed in crude models with the 7-, 9-, and 10.5-year HOME scores. Specifically, a one-unit increase in the total HOME score at ages 7, 9, and 10.5 years corresponded respectively to a 0.6-point increase (95% CI = 0.1 to 1.2), 0.6-point increase (95% CI = 0.001 to 1.2), and a 0.5-point increase (95% CI = 0.02 to 1.0) in adiponectin at age 14 years.
We identified inverse associations of maternal depressive symptoms and 9-year adiponectin levels; however, results did not reach statistical significance. In the 14-year adiponectin models, a one-unit increase in CES-D score when children were 7 and 9 years corresponded to a 0.2-point decrease (7 years: 95% CI = −0.3 to −0.03; 9 years: 95% CI = −0.3 to −0.07) in adiponectin.
Social support at age 5 years had a positive relationship with adiponectin at age 9 and 14 years. Specifically, a one-unit increase in social support at age 5 years corresponded to a 1.6-point increase in 9-year adiponectin levels (95% CI = −0.9 to 4.0), which was not significant but was similar in magnitude to a borderline significant 1.3-point increase in 14-year adiponectin (95% CI = 0.003 to 2.7). For the adiponectin models, only the relationship between maternal depressive symptoms and adiponectin at age 14 years remained significant after FDR adjustment (p = .003).
We also observed significant associations between symptoms of maternal depression, HOME scores, and social support, predominantly in early childhood, and leptin levels in CHAMACOS children. In crude models, a one-point increase of maternal depressive symptoms at child aged 3.5 years corresponded to a 0.007-point increase in log leptin at age 9 years (95% CI = 0.001 to 0.013), whereas a unit increase in scores at age 7 years significantly related to a 0.007 increase in 14-year log-leptin levels (95% CI = 0.001 to 0.013). A unit increase in HOME score at age 2 years corresponded to a 0.03-point increase (95% CI = 0.003 to 0.055) in log-leptin levels at age 9 years. In the 9- and 14-year leptin models, the direction of association between social support and leptin was consistently negative.
We also examined the relationship between positive and negative aspects of the home environment and adipokines at age 5 years (Supplemental Table 1, http://links.lww.com/PSYMED/A530). Results were similar in magnitude to the 9- and 14-year adipokine and BMI models but did not reach statistical significance, suggesting that obesity status and biomarkers of obesity in later childhood may be more sensitive to the effects of maternal depressive symptoms or protective home environment factors.
Longitudinal Assessment of Measures of the Home Environment With Child Adipokines and BMI
Longitudinal models of adipokines and BMI z-score showed significant relationships with the most recent measures of maternal depressive symptoms and social support (Table 6). Maternal depressive symptoms were significantly associated with increased BMI z-score (B = 0.007; 95% CI = 0.0004 to 0.014) and decreased adiponectin (B = −0.248; 95% CI = −0.376 to −0.119). Social support at age 5 years, which was the closest assessment to all childhood adipokine measurements (age 5, 9, and 14 years) was negatively associated with BMI z-score (B = −0.162; 95% CI = −0.273 to −0.050) and leptin (B = −0.056; 95% CI = −0.096 to −0.016). Cumulative HOME score was not significantly associated with longitudinal levels of adiponectin across childhood (B = −0.285; 95% CI = −0.694 to 0.124; p = .17), contrary to the results observed in the individual models. In the sensitivity analysis, we found no significant differences in the model including all the exposure variables (maternal depressive symptoms, HOME score, and social support) and another model with poverty at outcome assessment. The results of the analyses of a potential social support moderation of the relationships of HOME score and maternal depressive symptoms with obesity indicated a lack of interaction. Mixed-effect models produced similar results to the GEE models.
In this longitudinal study, we examined adipokine levels and BMI throughout childhood and assessed their relationship with early-life measures of the home environment in a Mexican American birth cohort. Adiponectin levels in CHAMACOS children were high at birth and, on average, lower as the children aged. Leptin levels were high at birth and during adolescence, but lower during early childhood (age 5 years). Previous mixture modeling analyses that reported trajectory clusters for adiponectin and leptin in a subset of CHAMACOS children up to age 9 years observed similar patterns in adipokine levels, including the large difference in adiponectin levels at birth and early childhood (24). We found consistent positive associations between HOME score in late childhood and adiponectin at ages 9 and 14 years indicating that a protective home environment is associated with beneficial profiles of adiponectin in children. Significant associations were observed between maternal depressive symptoms and HOME score in early childhood and leptin at ages 9 and 14 years, but effect sizes were relatively small. We also report positive associations between maternal depressive symptoms when children were 3 and 9 years with BMI z-scores at ages 9 and 14 years, suggesting that increased maternal depressive symptoms are related to higher BMI z-scores in offspring. These data indicate that aspects of the home environment of CHAMACOS children could lead to altered obesity status and the expression of biomarkers of obesity. Clinical implications of these findings could include the direct impact of psychosocial factors on child health status and indirect effects, such as increased medical costs due to obesity comorbidities.
In addition to previous adipokine trajectories research in the CHAMACOS cohort (24), there have been only a few studies that have assessed trends of repeated measurements of adipokines throughout childhood. Mantzoros et al. (40) examined the relationship between cord blood and 3-year adipokine levels in 588 children in the Project Viva cohort. The authors found that although there was not a significant relationship between adiponectin levels at birth and age 3 years, leptin levels at the assessments were positively associated (crude model p = .0003). Gruszfeld et al. (41) studied trends in adipokines in 459 children from infancy to age 8 years, observing a decrease in adiponectin and leptin levels from infancy, with an increase around school age for adiponectin only. This finding is similar to our observations in the CHAMACOS cohort for a longer period from birth to 14 years. Another study in 519 Japanese children aged 9 to 10 years observed positive associations (p < .001) between adipokine levels at baseline and values measured at the 3-year follow-up (42). In the current study, we also observed positive associations between adipokine assessments throughout childhood, with the strongest correlation observed for consecutive measurements. At each time point, we found that leptin values were higher in girls compared with boys, which is consistent with previous reports (42,43). The sex difference observed for leptin, but not for adiponectin, could be a result of differences in body composition, as well as the potential of serum androgen levels to reduce the amount of leptin in boys (44).
In addition to assessment of trends in adipokine levels across childhood and by child sex, we examined the relationship of BMI, adiponectin, and leptin with maternal depressive symptoms and protective factors of the home environment, including social support and HOME scores. Previous research has identified relationships of early-life aspects of the home environment (26–28), maternal depression (23,29), and offspring social support (30) with obesity risk in children. In the CHAMACOS cohort, we identified consistent and positive associations between maternal depressive symptoms at child aged 3 and 9 years with BMI in adolescence. The positive relationship identified in the current study is similar to a previous study conducted by Audelo et al. (23) in 332 7-year-old CHAMACOS children, where youth whose mothers demonstrated depressive symptoms at the 1-, 3.5-, and 7-year child assessments had 2.4 times the adjusted odds of overweight/obesity relative to those whose mothers never experienced symptoms of depression. A similar study (29), including 1090 children with anthropometric assessments at three time points during early childhood, reported that children with mothers exhibiting depressive symptoms, assessed by CES-D when the child was 1 month, as well as 2 and 3 years old, were 1.7 times more prone to be overweight than their counterparts whose mothers did not report depressive symptoms. Taken together, these studies suggest a possible role of maternal depression in offspring obesity status and a potential avenue for obesity intervention.
Most of the research regarding the relationship between adversity and adipokine measurements has been conducted in cross-sectional assessments of adult populations. Positive associations were found between abuse, family instability, and depressive symptoms, in early childhood or adulthood, and adult leptin levels (9,10,21,45), whereas negative relationships were observed between home and family measures and adiponectin in adulthood (9,11,15–17,46). In our prospective analysis in Mexican American children, we also saw positive associations between maternal depressive symptoms and leptin levels throughout the child's youth and negative associations with adiponectin, particularly in older children. In addition, the positive associations we observed between HOME score and adiponectin at 9 and 14 years concur with trends seen in the literature; in this case, as HOME scores decrease, indicating poor home learning environment, adiponectin levels decline. Because adiponectin has many beneficial immunological and metabolic properties, a profile of lower levels is less desirable. The association between HOME scores and adiponectin in the longitudinal models did not reach statistical significance (p = .17) and thus was unable to completely corroborate the individual time point findings.
Although the current study relies on maternal assessment of adversity as children age into adolescence, it is plausible that maternal depressive symptoms, social support, and home environment could impact mental well-being, and biomarkers of obesity, of their children through several pathways. Numerous studies have demonstrated associations between stress and depressive symptoms in mothers and altered risk of obesity in their children, potentially mediated by parenting behaviors (i.e., through unhealthy dietary practices or sedentary lifestyles) or parent-child interactions (47–50). Maternal stress can induce a subsequent stress response in their children, as a result of augmented levels of cortisol (51). In addition, pervasive stress during childhood has been shown to activate the HPA axis, resulting in the release of hormones such as glucocorticoids that have downstream effects on obesity risk (52).
The biological pathways whereby family instability, stress, and depression in an individual can affect their adipokine profiles are well documented in humans. As an anti-inflammatory adipokine, adiponectin can inhibit the function of macrophages and tumor necrosis factor-α and increase the secretion of anti-inflammatory cytokines, including interleukin-10 and interleukin-1 receptor antagonist (53–55). However, during symptoms of depression, proinflammatory cytokines, such as interleukin 6, can alter HPA activity, releasing glucocorticoids that can prevent adiponectin from exerting its anti-inflammatory effects (56). Conversely, glucocorticoid secretion as a response to stressors enhances the production of leptin in humans (57,58).
The main strengths of the current study include the use of data collected longitudinally to assess the relationship between maternal depressive symptoms, social support, and the home environment and adipokine trends in a well-characterized Mexican American birth cohort. Understanding the relationship between the home environment and biomarkers of obesity in this population is particularly relevant because of the high prevalence of both parental pre-pregnancy and childhood obesity. However, we recognize certain limitations, because we did not examine more data on child behavior, adverse experiences, and emotional well-being, which could mediate the relationship between maternal psychosocial factors and child obesity risk. In addition, other home environment factors that were not examined in the current study, such as single parent household and crowding, could influence adipokine levels in children. Given the projected growth in the Hispanic population in the United States (59) and the high prevalence of obesity in Mexican American children (3), which we have observed in the CHAMACOS cohort, this research can have health implications in the general population and in minority populations at high risk of obesity. This study provides novel support for the relationship between early-life exposure to maternal depressive symptoms and protective factors, such as social support and an enriching home environment, and their association with adipokines. This research can inform potential public health strategies to address early childhood factors that contribute to the obesity epidemic and comorbidities of obesity.
The authors thank the CHAMACOS participants and the laboratory and field staff.
Source of Funding and Conflicts of Interest: This study was supported by grants from the Environmental Protection Agency (R82670901 and RD83451301), the National Institute of Environmental Health Science (NIEHS) (P01 ES009605, R01ES021369, R01ES017054, R24ES028529, F31ES027751), the National Institute on Drug Abuse (NIDA) (5R01DA035300), the National Institutes of Health (NIH) (UG3OD023356), and the JPB Foundation through the JPB Research Network on Toxic Stress, a project of the Center on the Developing Child at Harvard University. The content is solely the responsibility of the authors and does not necessarily represent the official views of the EPA, JPB Foundation, NIEHS, or the National Institutes of Health. The authors report no conflicts of interest.
1. Comuzzie AG, Allison DB. The search for human obesity genes. Science 1998;280:1374–7.
2. Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in obesity among adults in the United States, 2005 to 2014. JAMA 2016;315:2284–91.
3. Ogden CL, Carroll MD, Lawman HG, Fryar CD, Kruszon-Moran D, Kit BK, Flegal KM. Trends in obesity prevalence among children
and adolescents in the United States, 1988-1994 through 2013-2014. JAMA 2016;315:2292–9.
4. NHLBI. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults—The Evidence Report. National Institutes of Health. Obes Res 1998;(6 suppl 2):51S–209S.
5. Serdula MK, Ivery D, Coates RJ, Freedman DS, Williamson DF, Byers T. Do obese children
become obese adults? A review of the literature. Prev Med 1993;22:167–77.
6. Morris TT, Northstone K, Howe LD. Examining the association between early life social adversity and BMI changes in childhood: a life course trajectory analysis. Pediatr Obes 2016;11:306–12.
7. Bzostek SH, Beck AN. Familial instability and young children
's physical health. Soc Sci Med 2011;73:282–92.
8. Schmeer KK. Family structure and obesity in early childhood. Soc Sci Res 2012;41:820–32.
9. Joung KE, Park KH, Zaichenko L, Sahin-Efe A, Thakkar B, Brinkoetter M, Usher N, Warner D, Davis CR, Crowell JA, Mantzoros CS. Early life adversity is associated with elevated levels of circulating leptin
, irisin, and decreased levels of adiponectin
in midlife adults. J Clin Endocrinol Metab 2014;99:E1055–60.
10. Farr OM, Ko BJ, Joung KE, Zaichenko L, Usher N, Tsoukas M, Thakkar B, Davis CR, Crowell JA, Mantzoros CS. Posttraumatic stress disorder, alone or additively with early life adversity, is associated with obesity and cardiometabolic risk. Nutr Metab Cardiovasc Dis 2015;25:479–88.
11. Tietjen GE, Khubchandani J, Herial NA, Shah K. Adverse childhood experiences are associated with migraine and vascular biomarkers. Headache 2012;52:920–9.
12. Goldstein BJ, Scalia R. Adiponectin
: a novel adipokine linking adipocytes and vascular function. J Clin Endocrinol Metab 2004;89:2563–8.
13. Kershaw EE, Flier JS. Adipose tissue as an endocrine organ. J Clin Endocrinol Metab 2004;89:2548–56.
14. Crowell JA, Davis CR, Joung KE, Usher N, McCormick SP, Dearing E, Mantzoros CS. Metabolic pathways link childhood adversity to elevated blood pressure in midlife adults. Obes Res Clin Pract 2016;10:580–8.
15. Leo R, Di Lorenzo G, Tesauro M, Cola C, Fortuna E, Zanasi M, Troisi A, Siracusano A, Lauro R, Romeo F. Decreased plasma adiponectin
concentration in major depression. Neurosci Lett 2006;407:211–3.
16. Cizza G, Nguyen VT, Eskandari F, Duan Z, Wright EC, Reynolds JC, Ahima RS, Blackman MR. POWER Study Group. Low 24-hour adiponectin
and high nocturnal leptin
concentrations in a case-control study of community-dwelling premenopausal women with major depressive disorder: the Premenopausal, Osteopenia/Osteoporosis, Women, Alendronate, Depression (POWER) study. J Clin Psychiatry 2010;71:1079–87.
17. Diniz BS, Teixeira AL, Campos AC, Miranda AS, Rocha NP, Talib LL, Gattaz WF, Forlenza OV. Reduced serum levels of adiponectin
in elderly patients with major depression. J Psychiatr Res 2012;46:1081–5.
18. Margetic S, Gazzola C, Pegg GG, Hill RA. Leptin
: a review of its peripheral actions and interactions. Int J Obes Relat Metab Disord 2002;26:1407–33.
19. Esler M, Vaz M, Collier G, Nestel P, Jennings G, Kaye D, Seals D, Lambert G. Leptin
in human plasma is derived in part from the brain, and cleared by the kidneys. Lancet 1998;351:879.
20. Koerner A, Kratzsch J, Kiess W. Adipocytokines: leptin
—the classical, resistin—the controversical, adiponectin
—the promising, and more to come. Best Pract Res Clin Endocrinol Metab 2005;19:525–46.
21. Morris AA, Ahmed Y, Stoyanova N, Hooper WC, De Staerke C, Gibbons G, Quyyumi A, Vaccarino V. The association between depression and leptin
is mediated by adiposity. Psychosom Med 2012;74:483–8.
22. Danese A, Dove R, Belsky DW, Henchy J, Williams B, Ambler A, Arseneault L. Leptin
deficiency in maltreated children
. Transl Psychiatry 2014;4:e446.
23. Audelo J, Kogut K, Harley KG, Rosas LG, Stein L, Eskenazi B. Maternal depression and childhood overweight in the CHAMACOS Study of Mexican-American Children
. Matern Child Health J 2016;20:1405–14.
24. Volberg V, Heggeseth B, Harley K, Huen K, Yousefi P, Dave V, Tyler K, Vedar M, Eskenazi B, Holland N. Adiponectin
trajectories in Mexican-American children
from birth to 9 years of age. PLoS One 2013;8:e77964.
25. Harley KG, Berger K, Rauch S, Kogut K, Claus Henn B, Calafat AM, Huen K, Eskenazi B, Holland N. Association of prenatal urinary phthalate metabolite concentrations and childhood BMI and obesity. Pediatr Res 2017;82:405–15.
26. O'Brien M, Nader PR, Houts RM, Bradley R, Friedman SL, Belsky J, Susman E. The ecology of childhood overweight: a 12-year longitudinal analysis. Int J Obes (Lond) 2007;31:1469–78.
27. Holdsworth EA, Schell LM. Maternal-infant interaction as an influence on infant adiposity. Am J Hum Biol 2017;29:e23023.
28. Strauss RS, Knight J. Influence of the home environment
on the development of obesity in children
. Pediatrics 1999;103:e85.
29. Wang L, Anderson JL, Dalton Iii WT, Wu T, Liu X, Zheng S, Liu X. Maternal depressive symptoms and the risk of overweight in their children
. Matern Child Health J 2013;17:940–8.
30. Serlachius A, Elovainio M, Juonala M, Shea S, Sabin M, Lehtimaki T, Raitakari O, Keltikangas-Jarvinen L, Pulkki-Raback L. High perceived social support protects against the intergenerational transmission of obesity: The Cardiovascular Risk in Young Finns Study. Prev Med 2016;90:79–85.
31. Eskenazi B, Bradman A, Gladstone EA, Jaramillo S, Birch K, Holland N. CHAMACOS, a longitudinal birth cohort study: lessons from the fields. Environ Health Perspect 2003;1:3–27.
32. Radloff L. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Measur 1977;1:385–401.
33. Weissman MM, Sholomskas D, Pottenger M, Prusoff BA, Locke BZ. Assessing depressive symptoms in five psychiatric populations: a validation study. Am J Epidemiol 1977;106:203–14.
34. Broadhead WE, Gehlbach SH, de Gruy FV, Kaplan BH. The Duke-UNC Functional Social Support Questionnaire. Measurement of social support in family medicine patients. Med Care 1988;26:709–23.
35. Caldwell B, Bradley R. Home Observation for Measurement of the Environment. Little Rock, AR: University of Arkansas; 1984.
36. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, Mei Z, Wei R, Curtin LR, Roche AF, Johnson CL. 2000 CDC growth charts for the United States: Methods and development. National Center for Health Statistics. Vital Health Stat 2002;11:1–190.
37. Volberg V, Harley KG, Aguilar RS, Rosas LG, Huen K, Yousefi P, Dave V, Phan N, Lustig RH, Eskenazi B, Holland N. Associations between perinatal factors and adiponectin
in 9-year-old Mexican-American children
. Pediatr Obes 2013;8:454–63.
38. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13–22.
39. Volberg V, Harley K, Calafat AM, Dave V, McFadden J, Eskenazi B, Holland N. Maternal bisphenol a exposure during pregnancy and its association with adipokines in Mexican-American children
. Environ Mol Mutagen 2013;54:621–8.
40. Mantzoros CS, Rifas-Shiman SL, Williams CJ, Fargnoli JL, Kelesidis T, Gillman MW. Cord blood leptin
as predictors of adiposity in children
at 3 years of age: a prospective cohort study. Pediatrics 2009;123:682–9.
41. Gruszfeld D, Kułaga Z, Wierzbicka A, Rzehak P, Grote V, Martin F, Poncelet P, Closa-Monasterolo R, Escribano J, Verduci E, Riva E, Koletzko B. Leptin
serum levels from infancy to school age: factors influencing tracking. Child Obes 2016;12:179–87.
42. Nishimura R, Sano H, Matsudaira T, Morimoto A, Miyashita Y, Shirasawa T, Kokaze A, Tajima N. Changes in body mass index, leptin
in Japanese children
during a three-year follow-up period: a population-based cohort study. Cardiovasc Diabetol 2009;8: 30-2840-8-30.
43. Blum WF, Englaro P, Hanitsch S, Juul A, Hertel NT, Muller J, Skakkebaek NE, Heiman ML, Birkett M, Attanasio AM, Kiess W, Rascher W. Plasma leptin
levels in healthy children
and adolescents: dependence on body mass index, body fat mass, gender, pubertal stage, and testosterone. J Clin Endocrinol Metab 1997;82:2904–10.
44. Wabitsch M, Blum WF, Muche R, Braun M, Hube F, Rascher W, Heinze E, Teller W, Hauner H. Contribution of androgens to the gender difference in leptin
production in obese children
and adolescents. J Clin Invest 1997;100:808–13.
45. Pasco JA, Jacka FN, Williams LJ, Henry MJ, Nicholson GC, Kotowicz MA, Berk M. Leptin
in depressed women: cross-sectional and longitudinal data from an epidemiologic study. J Affect Disord 2008;107:221–5.
46. Lehto SM, Huotari A, Niskanen L, Tolmunen T, Koivumaa-Honkanen H, Honkalampi K, Ruotsalainen H, Herzig KH, Viinamäki H, Hintikka J. Serum adiponectin
and resistin levels in major depressive disorder. Acta Psychiatr Scand 2010;121:209–15.
47. Tate EB, Wood W, Liao Y, Dunton GF. Do stressed mothers have heavier children
? A meta-analysis on the relationship between maternal stress and child body mass index. Obes Rev 2015;16:351–61.
48. McConley RL, Mrug S, Gilliland MJ, Lowry R, Elliott MN, Schuster MA, Bogart LM, Franzini L, Escobar-Chaves SL, Franklin FA. Mediators of maternal depression and family structure on child BMI: parenting quality and risk factors for child overweight. Obesity (Silver Spring) 2011;19:345–52.
49. Topham GL, Page MC, Hubbs-Tait L, Rutledge JM, Kennedy TS, Shriver L, Harrist AW. Maternal depression and socio-economic status moderate the parenting style/child obesity association. Public Health Nutr 2010;13:1237–44.
50. O’Connor SG, Maher JP, Belcher BR, Leventhal AM, Margolin G, Shonkoff ET, Dunton GF. Associations of maternal stress with children
's weight-related behaviours: a systematic literature review. Obes Rev 2017;18:514–25.
51. Essex MJ, Klein MH, Cho E, Kalin NH. Maternal stress beginning in infancy may sensitize children
to later stress exposure: effects on cortisol and behavior. Biol Psychiatry 2002;52:776–84.
52. Charmandari E, Kino T, Souvatzoglou E, Chrousos GP. Pediatric stress: hormonal mediators and human development. Horm Res 2003;59:161–79.
53. Ouchi N, Kihara S, Funahashi T, Matsuzawa Y, Walsh K. Obesity, adiponectin
and vascular inflammatory disease. Curr Opin Lipidol 2003;14:561–6.
54. Yokota T, Oritani K, Takahashi I, Ishikawa J, Matsuyama A, Ouchi N, Kihara S, Funahashi T, Tenner AJ, Tomiyama Y, Matsuzawa Y. Adiponectin
, a new member of the family of soluble defense collagens, negatively regulates the growth of myelomonocytic progenitors and the functions of macrophages. Blood 2000;96:1723–32.
55. Wolf AM, Wolf D, Rumpold H, Enrich B, Tilg H. Adiponectin
induces the anti-inflammatory cytokines IL-10 and IL-1RA in human leukocytes. Biochem Biophys Res Commun 2004;323:630–5.
56. Fallo F, Scarda A, Sonino N, Paoletta A, Boscaro M, Pagano C, Federspil G, Vettor R. Effect of glucocorticoids on adiponectin
: a study in healthy subjects and in Cushing’s syndrome. Eur J Endocrinol 2004;150:339–44.
57. Newcomer JW, Selke G, Melson AK, Gross J, Vogler GP, Dagogo-Jack S. Dose-dependent cortisol-induced increases in plasma leptin
concentration in healthy humans. Arch Gen Psychiatry 1998;55:995–1000.
58. Askari H, Liu J, Dagogo-Jack S. Hormonal regulation of human leptin
in vivo: effects of hydrocortisone and insulin. Int J Obes Relat Metab Disord 2000;24:1254–9.
59. U.S. Census Bureau. National Population Projections 2014;2017.
adiponectin; children; home environment; leptin; Mexican American; BMI = body mass index; CES-D = Center for Epidemiological Studies Depression Scale; CHAMACOS = Center for the Health Assessment of Mothers and Children of Salinas; FDR = false discovery rate; GEE = generalized estimating equation; HPA = hypothalamic-pituitary-adrenal; HOME = Home Observation for the Measurement of the Environment
Supplemental Digital Content
Copyright © 2019 by American Psychosomatic Society