Low Circulating Adropin Concentrations with Obesity and Aging Correlate with Risk Factors for Metabolic Disease and Increase after Gastric Bypass Surgery in Humans

Abstract

Context:

Mouse studies suggest that adropin, a peptide hormone, is required for metabolic homeostasis and prevention of obesity-associated insulin resistance. Whether obesity and insulin resistance are associated with low plasma adropin levels in humans is not known.

Objectives:

Our objective was to investigate the hypothesis that obesity and indicators of insulin resistance are associated with low adropin levels and determine whether weight loss regulates adropin levels.

Design and Participants:

Plasma was obtained from 85 female [age 21–67 yr, body mass index (BMI) 19.4–71.5 kg/m²] and 45 male (age 18–70 yr, BMI 19.1–62.6 kg/m²) volunteers for other clinical studies. The impact of Roux-en-Y gastric bypass was investigated in 19 obese females (BMI 37–65 kg/m²) using samples collected at baseline and 1–12 months after surgery.

Results:

Adropin levels correlate negatively with BMI (r = −0.335, P < 0.001) and age (r = −0.263, P=0.003). Age-adjusted adropin levels are higher in males [4.1 ng/ml; 95% confidence interval (CI) = 3.6–4.6 ng/ml] than females (3.0 ng/ml; 95% CI = 2.6–3.4 ng/ml) (P = 0.001). In all subjects, lower age-adjusted adropin levels were observed in overweight (3.3 ng/ml; 95% CI = 2.8–3.8 ng/ml, P = 0.033) and obese (2.7 ng/ml; 95% CI = 2.1–3.3 ng/ml, P = 0.001) compared with healthy-weight subjects (4.1 ng/ml; 95% CI = 3.6–4.5 ng/ml). This effect was gender specific (weight category × gender, P < 0.001) and was observed in males only. Aging and diagnosis with two or more metabolic syndrome risk factors was associated with low adropin levels, irrespective of sex. Adropin concentrations increased after Roux-en-Y gastric bypass, peaking 3 months after surgery (P < 0.01).

Conclusions:

Although males exhibit higher adropin levels that are reduced by obesity, aging and markers of insulin resistance are associated with low plasma adropin irrespective of sex.

Adropin, first described in 2008, is a peptide hormone encoded by the energy homeostasis-associated (ENHO) gene (1). The adropin open reading frame is unique and produces a small 76-residue secreted peptide that is very highly conserved (97–100%) in the genomes of eutherian (placental) mammals. The adropin transcript is abundant in liver (1), and in mice hepatic expression and circulating adropin concentrations both exhibit rapid regulation by fasting (inhibition) and refeeding (stimulation), suggesting regulation by signals of metabolic status (1, 2). Studies in mouse models of obesity and cell-based systems strongly suggest that adropin is involved in metabolic homeostasis and cardiovascular function (1–4). Importantly, in mice, diet-induced and genetically induced obesity is associated with decreased expression of the adropin transcript in liver and in circulating adropin concentrations (1, 2). A decline in adropin action with obesity may therefore contribute to the development of insulin resistance and dyslipidemia. Supporting this hypothesis, adropin-knockout mice exhibit increased adiposity and fasting triglycerides (TG), hepatic steatosis, insulin resistance, and increased propensity for impaired glucose tolerance with diet-induced obesity (2).

Based on these observations, we hypothesize that low adropin levels may be a risk factor for the development of insulin resistance and other features of the metabolic syndrome such as dyslipidemia that are associated with obesity. To date, only one study has examined circulating adropin concentrations in humans, reporting increased serum adropin levels in aged subjects (60–92 yr) with end-stage heart failure (3). Whether low adropin levels correlate with obesity and insulin resistance in humans has not been investigated. The aim of the current study was therefore to investigate the impact of obesity, metabolic status, and weight loss after bariatric surgery on plasma adropin concentrations in humans. However, as adropin exhibits regulation by food intake in mice (1, 2), we also examined whether plasma adropin levels are regulated by meal intake in humans. However, as adropin levels did not exhibit regulation by feeding, we used fasting samples to compare adropin levels in lean and obese subjects. Collectively, our data suggest that adropin levels are low in obesity and that metabolic risk factors such as dyslipidemia may also be associated with low adropin levels.

Subjects and Methods

Adropin levels were measured in plasma samples obtained from volunteers for five studies performed at the Pennington Biomedical Research Center (PBRC) (Baton Rouge, LA), the New York Obesity Nutrition Research Center (NYONRC) (St. Luke's-Roosevelt Hospital Center, New York, NY) and the University of California (UC) Davis Clinical and Translational Science Center Clinical Research Center (CCRC). The studies were reviewed and approved by the Institutional Review Boards of the PBRC, St. Luke's-Roosevelt Hospital Center, and UC Davis. All participants provided written informed consent to participate. Age, body mass index (BMI), and blood chemistry data (when available) for these subjects grouped by study are shown in Table 1. The use of medications for dyslipidemia or hypertension was an exclusion criterion for these studies.

Table 1.

Gender, age, BMI, lipids, insulin, and glucose data for the study subjects

Study (center)	Sex	n	Age (yr)	BMI (kg/m2)	Blood chemistry
					TG (mg/dl)	FFA (mEq/liter)	Total cholesterol (mg/dl)	HDL (mg/dl)	LDL (mg/dl)	ApoA1 (mg/dl)	ApoB (mg/dl)	Insulin (ng/ml)	Glucose (mg/dl)
A (NYONRC)	Female	12	33.8 ± 1.3	22.7 ± 0.3
B (PBRC)	Female	12	41.8 ± 2.5	37.7 ± 5.2
	Male	4	49.5 ± 6.1	38.8 ± 8.9
C (UCD-CCRC)	Female	21	29.2 ± 1.4	26.8 ± 1.0	102 ± 10	0.38 ± 0.03	165 ± 5	52 ± 3	87 ± 5	131 ± 5	71 ± 4	19 ± 3	88 ± 2
	Male	25	26.8 ± 1.5	24.6 ± 0.6	115 ± 8	0.32 ± 0.03	157 ± 7	43 ± 2	91 ± 6	117 ± 3	73 ± 5	16 ± 1	89 ± 1
D (UCD-CCRC)	Female	16	54.4 ± 1.5	29.9 ± 0.8	152 ± 19	0.44 ± 0.04	195 ± 8	41 ± 2	124 ± 9	131 ± 7	82 ± 7	15 ± 2	89 ± 1
	Male	15	53.1 ± 2.5	28.6 ± 0.6	138 ± 17	0.33 ± 0.02	177 ± 7	37 ± 3	115 ± 5	122 ± 8	80 ± 5	14 ± 2	87 ± 1
E (UCD-CCRC)	Female	24	42.2 ± 1.8	45.8 ± 1.3	123 ± 28	0.28 ± 0.03	178 ± 8	42 ± 2	104 ± 8	155 ± 6	111 ± 7	29 ± 3	99 ± 5

Data are mean ± sem.

Study A (NYONRC)

Plasma adropin was measured in samples obtained from 12 lean female volunteers (four Caucasians, four Hispanic, two African-Americans, one Asian, and one mixed race) for an inpatient study investigating the effects of sleep on metabolic homeostasis (5). The samples used for the current study had been collected at 0800, 0815, 0830, 0900, and 0930 h; every 2 h from 1000–22-00 h; and at 0730 h the next day. Meals supplying 30% of resting energy requirement (30% fat, 15% protein, 55% carbohydrates) were served at 0800, 1200, and 1900 h; a snack supplying 10% was given at 1600 h.

Study B (PBRC)

These samples were taken from a larger clinical trial comparing the impact of three bariatric surgery procedures with a low-calorie diet on insulin sensitivity and energy expenditure (BARIA, www.ClinicalTrials.gov identifier NCT00936130). Samples from age-, sex-, and ethnicity-matched lean controls were collected specifically for the current study. The samples used for this study were from eight morbidly obese (BMI = 52.2 ± 4.6 kg/m², age = 44.8 ± 3.6 yr) and eight age-, sex- (two males, six females), and ethnicity-matched (four Caucasians, four African-Americans) lean subjects (BMI = 23.7 ± 0.6 kg/m², age = 42.6 ± 3.6 yr). Blood samples were collected at baseline (fasting) [time −15 min) (t⁻¹⁵)] and 60 (t⁺⁶⁰) and 120 (t⁺¹²⁰) minutes after a 400-kcal liquid breakfast (40% fat, 20% protein, 40% carbohydrates).

Study C (UC Davis CCRC)

Baseline samples were obtained from 27 male and 21 female volunteers (32 Caucasian, five Hispanic, four African-American, four Asian, one Native American, and two Asian/Pacific Islander) for a clinical study investigating the effect of dietary sugars on insulin resistance and lipid metabolism over a 2-wk period (6, 7). For this study, we used baseline samples that had been collected at 0800, 0830, and 0900 h after an overnight fast and then pooled. Samples collected at 2200, 2300, and 2400 h, the period of peak TG concentrations, were used to assess adropin levels during the postabsorptive phase.

Study D (UC Davis, CCRC)

Samples were obtained from 15 male and 16 female volunteers (23 Caucasians, five Hispanic, and three Asian) in a study investigating the impact of dietary sugars on metabolic risk factors over a 10-wk period (7). The samples used for this study were baseline samples that had been collected at 0800, 0830, and 0900 h after an overnight fast.

Study E (UC Davis, CCRC)

These samples had previously been collected from a clinical study of the response of 19 morbidly obese Caucasian females after Roux-en-Y gastric bypass (RYGB) surgery. The samples used for this study had been collected after an overnight fast at baseline and 1, 3, 6, and 12 months after surgery. Study design and a description of the anthropometric, metabolic, and endocrine responses in these patients have been reported previously (8–10).

Adropin measurements

Serum or plasma adropin was measured in duplicate using a commercially available ELISA (Peninsula Laboratories, Bachem, San Carlos, CA). The samples used had never been previously thawed. There is 100% conservation of the amino sequence of human and mouse adropin (1), and the assay detects immunoreactivity in both mouse and human plasma. We previously used plasma from wild-type and adropin-deficient knockout mice to test for assay specificity; adropin immunoreactivity was not detected in plasma obtained from adropin-knockout mice (2). We also spiked human plasma with synthetic adropin^34–76 (Phoenix Pharmaceuticals, Inc., Burlingame, CA) using a range of concentrations (2.8–27.0 ng/ml), observing a linear rate of recovery (r² = 0.9957 between observed and predicted concentration of adropin) and a recovery rate of more than 100%. The lowest detection limit was 0.2 ng/ml; the intraassay coefficient of variation (CV) determined using quality-control human plasma samples with adropin values ranging from 1.1–2.4 ng/ml was 8.6%, whereas the interassay CV varied between 19% (PBRC) and 29.6% (UC Davis). We also calculated the CV of the duplicate samples in the low (0.5–2.5 ng/ml, CV = 8.9%), mid (2.5–4.5 ng/ml, CV = 6.7%), and high (4.5–22.0 ng/ml, CV = 5.1%) ranges.

Study A samples were collected in EDTA-coated tubes with a dipeptidyl peptidase-4 inhibitor and aprotinin. Samples for study B were collected with a protease inhibitor cocktail (Roche Diagnostics, Dallas, TX; Millipore Corp, Bedford, MA) that provides a broad spectrum of protease inhibition that is used to avoid possible interference from EDTA. For studies C–E, samples were collected in EDTA tubes.

Associations between plasma adropin and metabolic risk factors

Data available from studies C–E included plasma lipids [TG, free fatty acids (FFA), total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), apolipoproteins (Apo) B and A1], glucose, insulin, and blood pressure (6–13). The metabolic syndrome risk factor (MSRF) score (a numerical score of the number of facets of the metabolic syndrome) was calculated by scoring each subject a value of 1 for fasting TG at least 150 mg/dl, HDL no higher than 40 mg/dl for males or 49 mg/dl for females, waist circumference at least 102 cm for males or 88 cm for females, fasting glucose at least 100 mg/dl, and systolic blood pressure at least 130 mm Hg or diastolic blood pressure at least 85 mm Hg. The criteria used for MSRF score were those defined by the American Heart Association/National Heart, Lung, and Blood Institute (14–16).

Statistical analysis

Data were analyzed using SPSS version 19. Correlations between adropin and anthropomorphic and blood chemistry data were assessed using the Pearson product-moment correlation coefficient (r) and multiple regression. Adropin levels over the day were analyzed using repeated-measures ANOVA. The effects of study and gender on plasma adropin levels were assessed using a multivariate ANOVA using normal or transformed data with Bonferroni correction for post hoc comparisons. In situations where Levene's test of equality of error variances indicated unequal variance between groups, data transformation was employed. In situations where data transformation failed to produce equal variance, nonparametric tests were used (Kruskal-Wallis ANOVA). Unless stated otherwise, data are presented as mean ± sem.

Results

Plasma adropin levels do not exhibit meal- or circadian-related changes

Plasma adropin levels in mice exhibit regulation by fasting and refeeding (1, 2). We therefore measured adropin levels throughout the daytime to determine whether it is regulated by meal- or circadian-related factors. In healthy-weight females, plasma adropin levels did not exhibit meal- or circadian-related changes (study A, Fig. 1). In study B, adropin levels did not change 60 and 120 min after breakfast in healthy-weight subjects (t⁻¹⁵, 2.6 ± 0.4; t⁺⁶⁰, 2.7 ± 0.4; t⁺¹²⁰, 2.7 ± 0.4 ng/ml) or obese subjects (t⁻¹⁵, 2.2 ± 0.3; t⁺⁶⁰, 2.3 ± 0.3; t⁺¹²⁰, 2.0 ± 0.3 ng/ml) (time effect, P = 0.838; body weight effect, P = 0.290; interaction between time and body weight, P = 0.392). Adropin levels also did not differ between fasting and postprandial conditions in study C (4.2 ± 0.3 vs. 4.3 ± 0.2 ng/ml, respectively, P = 0.332).

Fig. 1.

Meal intake and overnight fasting have no significant effect on plasma adropin levels in female volunteers. Plasma adropin levels were measured throughout the day and after an overnight fast. Meal times are indicated by letters (B, breakfast; L, lunch; S, snack; D, dinner).

These data suggest that levels of adropin in plasma are stable over time and do not exhibit marked diurnal or meal-related variations. Studies examining the impact of obesity on plasma adropin levels therefore used samples collected after an overnight fast.

Decreased adropin concentrations with high BMI and aging

In pooled samples from studies A–E (n = 130), a significant negative association between BMI and fasting adropin concentrations was observed (r = −0.335, P < 0.001, Fig. 2A). Significant associations between adropin concentrations and BMI were observed when males (r = −0.318, P = 0.033, n = 45) and females (r = −0.291, P = 0.007, n = 85) were analyzed independently, although for females, the decline with obesity was less severe. Analysis of the impact of ethnicity with age and BMI as covariates indicated no significant effect (P = 0.423).

Fig. 2.

Analysis of the impact of obesity and aging on plasma adropin levels in males and females. A, Plasma adropin levels as a function of BMI; B, plasma adropin levels as a function of age; C, age-adjusted plasma adropin levels in subjects categorized as being of healthy weight (BMI = 18.5–24.9 kg/m²), overweight (BMI = 25–29.9 kg/m²), or obese (BMI > 30 kg/m²). Multivariate analysis indicated that the effect of obesity is sex specific. There was a significant effect of gender (P < 0.05), with higher levels in males (plasma adropin levels = 3.7 ± 0.2 ng/ml) relative to females (3.0 ± 0.2 ng/ml). Significance (*, P < 0.05; **, P < 0.01) was determined using Bonferroni post hoc analysis. D, BMI-adjusted plasma adropin levels by age group. The difference between those under 30, 40–50, and 50 yr and older was significant in all subjects (*, P < 0.05 vs. <30 yr by post hoc analysis using Bonferroni's analysis). There was a trend (P = 0.058) for an effect of gender on plasma adropin levels.

Subjects were then categorized as healthy body weight (BMI = 18.5–24.9 kg/m², n = 41); overweight (BMI = 25–29.9 kg/m², n = 35), or obese (BMI > 30 kg/m², n = 54). Because age significantly affected adropin levels (Fig. 2B), age-adjusted plasma adropin levels were used for the analysis. Adropin levels were lower in the overweight (3.4 ng/ml; 95% CI = 2.8–4.0 ng/ml) and obese groups (3.1 ng/ml; 95% CI = 2.5–3.7 ng/ml) compared with healthy-weight subjects (4.1 ng/ml; 95% CI = 3.6–4.7 ng/ml) (Fig. 2C). When analyzed using gender and weight category as fixed factors and age as a covariate, the effect of weight did not achieve statistical significance (P = 0.060). However, there was a significant effect of gender (P = 0.001) and a significant interaction between gender and weight category (P = 0.005). Adropin levels were higher in males (4.1 ng/ml; 95% CI = 3.6–4.6 ng/ml) than females (3.0 ng/ml; 95% CI = 2.6–3.4 ng/ml). When analyzed separately, there was a significant effect of weight category in males (P = 0.024) but not females (P = 0.127). Adropin levels were lower (P = 0.021) in overweight (3.3 ng/ml; 95% CI = 2.5–4.2 ng/ml, n = 24) compared with healthy-weight males (5.6 ng/ml; 95% CI = 4.3–6.9 ng/ml, n = 13). There was no significant difference between healthy-weight and obese males (3.5 ng/ml; 95% CI = 2.0–5.0 ng/ml, n = 8, P = 0.169).

Two outliers were observed in the overweight and obese males. A subject with a BMI of 29.1 kg/m² had an adropin concentration of 9.3 ng/ml (+6.5 sd from the average). Another subject with BMI of 31.7 kg/m² had an adropin value of 10.0 ng/ml (+8.4 sd from the average). Excluding these subjects led to the weight-category effect achieving significance in all subjects (P = 0.020); gender (P = 0.031) and the interaction between gender and weight category (P = 0.003) were still significant. When males were analyzed separately, age-adjusted adropin levels were significantly lower in overweight (3.1 ng/ml; 95% CI = 2.5–3.7 ng/ml, n = 23, P = 0.001) and obese (2.8 ng/ml; 95% CI = 1.6–3.9 ng/ml, n = 7, P = 0.006) males compared with healthy-weight males (5.4 ng/ml, 95% CI = 4.5–6.3 ng/ml, n = 13) (Fig. 2C).

Adropin levels decreased with age (Fig. 2B), with a significant negative association between plasma adropin levels and age (r = −0.251, P = 0.004) that was still significant when controlled for BMI (r = −0.195, P = 0.027). Plasma adropin concentrations were then analyzed in groups categorized as being 30 yr or younger (n = 34), 30–40 yr (n = 39), 40–50 yr (n = 30) or over 50 yr (n = 27). Individuals aged 30 yr or younger were grouped together due to a low number of participants aged under 2 0 yr (five males). Plasma adropin levels are highest in individuals aged 30 yr or younger (Fig. 2D). Multivariate analysis with BMI as a covariate indicated significant effects of age (P = 0.050) and gender (P = 0.011), with no interaction between gender and age (P = 0.995). When the outliers mentioned previously were excluded from the analysis, age effects remained significant (P = 0.007); however, gender was no longer significant (P = 0.058).

To investigate whether age and BMI interact to affect plasma adropin levels, we performed a multiple regression analysis using BMI, age, and gender as independent variables. Inclusion of all three variables produced the strongest association (r = 0.454, P < 0.001), with the formula adropin levels = 6.008 − (0.036 × BMI) − (0.026 × age) − (0.959 × gender), where gender = 0 for males or 1 for females.

Data from an anthropometric analysis (body fat, percent body fat, height, and weight) were also available for participants of studies C and D (n = 45). Participants in study D were nearly twice as old as those in study C (Table 1). Subjects in study D were also more obese because BMI below 25 kg/m² was an exclusion criterion for the study. Partial correlation analysis controlling for age, gender, and study indicated significant associations between plasma adropin and BMI (r = −0.235, P = 0.044) and adropin and body weight (r = −0.252, P = 0.030) for all subjects. However, no significant correlations were evident between adropin and body fat content when age and gender were controlled for.

Associations between plasma adropin and risk factors for metabolic disease

We next investigated whether plasma adropin levels are reduced in subjects with a MSRF score of 1 or more (Fig. 3A). Analysis of plasma adropin using MSRF score and gender as fixed variables and age as a covariate indicated a significant effect of MSRF score (P < 0.001) but no effect of gender (P = 0.893) and no interaction between gender and MSRF score (P = 0.144). Men with a MSRF score of 2 or more had plasma adropin concentrations that were 50% lower compared with subjects with no MSRF score (P < 0.01). In women, plasma adropin levels were similar in subjects with MSRF score of 0 or 1; however, adropin was lower in subjects with a MSRF score of 3 or more.

Fig. 3.

Low plasma adropin levels in individuals with metabolic risk factors. A, Plasma adropin levels (age-adjusted) in subjects grouped by MSRF score (0, 1, 2, or more than 3). Significantly different from MSRF = 0 or as indicated: *, P < 0.05; **, P < 0.01 (post hoc analysis using Bonferroni's correction for multiple comparisons). B, Negative correlation between fasting plasma adropin and fasting TG in males (shaded circles) and females (open circles). The data used shown in this analysis are natural log transformed.

We then examined the relationship between plasma adropin and individual metabolic risk factors. Plasma lipids (fasting TG, FFA, total cholesterol, HDL, LDL, ApoB, and ApoA1), measures of glucose homeostasis [fasting insulin, fasting glucose, homeostatic model assessment of insulin resistance (HOMA-IR) and HOMA-B], and measurements of diastolic and systolic blood pressure were available for 99 subjects (43 males, 56 females). Partial correlation analysis controlling for gender found significant negative associations between plasma adropin levels and TG, ApoB, LDL, glucose, and blood pressure; positive associations were also observed between plasma adropin and HDL and FFA (Table 2).

Table 2.

Correlation matrix showing associations (Pearsons correlation coefficient, r) between plasma adropin levels and markers of metabolic risk (dyslipidemia, glucose homeostasis, blood pressure, and obesity)

	r	P
Fatty acid metabolism
TG	−0.245	<0.001
Ln(TG)	−0.341	<0.001
FFA	−0.232	0.029
Ln(FFA)	−0.253	0.025
Cholesterol metabolism
Cholesterol	−0.179	NS
Ln(cholesterol)	−0.218	0.033
HDL	0.258	0.011
Ln(HDL)	0.334	0.001
ApoA1	−0.008	NS
Ln(ApoA1)	0.057	NS
ApoB	−0.277	0.006
Ln(ApoB)	−0.341	0.001
LDL	−0.209	0.041
Ln(LDL)	−0.165	NS
Glucose homeostasis
Insulin	−0.081	NS
Ln(insulin)	−0.185	NS
Glucose	−0.210	0.040
Ln(glucose)	−0.269	0.008
HOMA-IR	−0.123	NS
Ln(HOMA-IR)	−0.221	0.031
HOMA-B	0.056	NS
Ln(HOMA-B)	−0.018.	NS
Blood pressure
dBP	−0.276	0.006
Ln(dBP)	−0.324	0.001
sBP	−0.251	0.014
Ln(sBP)	−0.274	0.007
Morphometry
BMI	−0.371	<0.001
Ln(BMI)	−0.477	<0.001
BW	−0.383	<0.001
Ln(BW)	−0.466	<0.001
Age	−0.252	0.013
Ln(age)	−0.319	0.002

Correlations are shown for raw and natural log (Ln) transformed data [e.g. Ln(adropin) compared with Ln(TG)]. This analysis controlled for sex; italicized values are not significant. The results shown are from a partial correlation that controlled for gender. BW, Body weight; dBP, diastolic blood pressure; NS, not significant; sBP, systolic blood pressure.

Age and obesity are risk factors for insulin resistance and other facets of the metabolic syndrome, and both exhibited a negative correlation with plasma adropin levels. We therefore repeated the analysis using a partial correlation controlling for gender, age, and BMI. The results from this analysis suggest that the interaction between plasma adropin levels and most indicators of metabolic risk cannot be dissociated from the effects of aging and increased BMI. However, the exception was between plasma adropin and fasting TG (Fig. 3B), which persisted when and age and BMI were controlled for (r = −0.305, P = 0.003). For comparison, an analysis was performed comparing fasting insulin and TG levels, because insulin resistance is known to be a risk factor for dyslipidemia (17, 18). As predicted, a positive correlation was observed between TG and insulin (r = 0.399, P < 0.001), glucose (r = 0.497, P < 0.001), and HOMA-IR (r = 0.513, P < 0.001).

Assay data assessing liver function was available for study C. Data from mice suggests that the liver may be a significant source of circulating adropin (1, 2). We therefore analyzed whether there was any correlation between the levels of adropin and transminases in the circulation. No correlations were observed between plasma adropin levels and either aspartate aminotransferase (r = −0.178, P = 0.241), alanine aminotransferase (r = −0.123, P = 0.420) or γ-glutamyl transferase (r = −0.015, P = 0.923) when age, BMI and sex were controlled for.

Increase in plasma adropin levels after RYGB

RYGB resulted in marked weight loss (Fig. 4A) and improvements in glucose homeostasis and blood lipids (8–10). Plasma adropin concentrations increased after RYGB, peaking 3 months after surgery (Fig 4B). Despite continued weight loss, adropin levels had returned to baseline at 12 months. Baseline adropin levels predicted the decrease in plasma TG after surgery (r = −0.631, P = 0.004), likely related to the negative correlation between adropin levels and TG observed when baseline level of adropin and TG were analyzed for study E participants (r = −0.462, P = 0.047).

Fig. 4.

Gastric bypass surgery is associated with increased plasma adropin levels in obese females (n = 19). A, Impact of gastric bypass on obesity. All subjects lost weight in the 12 months after surgery. B, Impact of RYGB on plasma adropin levels. The effect of surgery on adropin levels was significant (P < 0.01), with plasma adropin levels significantly higher 3 months after surgery relative to levels at baseline and 1 month (*, P < 0.05 vs. baseline; **, P < 0.01 vs. baseline and 1 month).

Discussion

These data are the first to indicate an association between low plasma adropin levels with obesity and increased metabolic risk factors in humans. The decline in adropin production in mouse models of obesity is a secondary consequence of weight gain (1, 2). Whether low adropin levels in obese humans are also secondary to weight gain was not determined in these studies. Plasma adropin levels ranged from less than 1 to 10 ng/ml in normal-weight subjects. The lower average plasma adropin level in obese subjects was due to fewer individuals with values in the high end of the range. Additional studies are required to determine whether low adropin levels observed in the healthy-weight volunteers predict future weight gain or risk of metabolic disorders. However, results from studies using mice do not support a major role for adropin in regulating body weight. Deletion of the Enho gene encoding adropin in mice has a modest impact on fat mass but is not associated with hyperphagia or altered response to diet-induced obesity (2).

Adropin-deficient mice exhibit insulin resistance, dyslipidemia, and more pronounced hyperinsulinemia and impaired glucose tolerance with diet-induced obesity (2). Moreover, administration of recombinant adropin reverses insulin resistance and dyslipidemia in mice (1). In the current study, we observed a significant negative correlation between plasma adropin and fasting TG concentrations after controlling for age and BMI. In addition, subjects with lower plasma adropin levels exhibited the largest reductions in plasma TG after RYGB. The association between plasma adropin and TG is comparable to that observed between insulin and TG. Collectively, these results suggest that, as observed for mice, adropin may have functions in humans that impact TG metabolism (synthesis or clearance).

A significant increase in plasma adropin levels was observed after RYGB, suggesting that reversal of the metabolic syndrome associated with obesity is also associated with a reversal of low plasma adropin levels. However, weight loss per se does not appear likely to be the main factor driving the increase, because the response occurred after significant weight loss had been achieved. The time of the increase (3 months after surgery) coincides with a period of dynamic change in metabolic homeostasis. The initial changes in variables such as blood pressure and blood chemistries observed 1 month after surgery begin to reverse at this time. Additional studies are needed to investigate the significance of the change in plasma adropin that is observed with RYGB and to determine whether lifestyle-induced weight loss has a similar effect.

Circulating adropin concentrations were also lower in individuals with a MSRF score of at least 2. This observation suggests that conditions where the risk of developing metabolic diseases is increased are associated with low plasma adropin levels. Additional studies examining whether low plasma adropin levels are a risk factor for developing type 2 diabetes and cardiovascular disease are needed.

In normal-weight individuals, women had lower plasma adropin levels than men. The impact of obesity on plasma adropin levels was also far less severe in women. The physiological significance of this gender difference is unclear at this time. All studies of the role of adropin in metabolic homeostasis using mice published thus far have been conducted in males (1, 2). Future investigations comparing the phenotype of male and female adropin-deficient mice to investigate sex differences may be informative. Another unanticipated outcome was the decline in adropin with aging. As with obesity, this effect appeared to be more pronounced in men (Fig. 1E). If, as data obtained from mouse models suggest, adropin has functions that are essential for maintaining metabolic homeostasis, then a decline with age might have been anticipated. Aging is well known to be associated with deteriorating metabolic homeostasis, including reductions in hormone production and increased risk of insulin resistance and impaired glucose tolerance (19). An age-related decline in adropin production could thus contribute to the deterioration of metabolic homeostasis associated with aging.

We also assessed plasma adropin levels throughout the daytime and after an overnight fast. No circadian-, meal-, or fasting-related changes were observed, suggesting that adropin levels are not regulated by acute signals of nutrient intake in humans. It is possible that more prolonged negative energy balance will alter plasma adropin levels as observed for fibroblast growth factor-21, a liver-secreted hormone that exhibits acute regulation by fasting in mice but that requires longer-term fasting in humans (20).

In summary, obesity and aging are associated with lower circulating adropin concentrations in humans. Analysis of adropin-deficient mice suggests that this peptide hormone is required for metabolic homeostasis, specifically for maintaining insulin sensitivity and preventing dyslipidemia and protecting against impaired glucose tolerance. In the current study, we observed that low adropin levels are associated with high fasting TG concentrations, consistent with a similar role in humans. In addition, adropin levels increase after RYGB, and this is related to changes in TG after surgery. Additional studies investigating the impact of weight changes on adropin regulation and whether plasma adropin levels predict risk for metabolic disorders such as dyslipidemia and type 2 diabetes are clearly warranted.