P

PwB

FAC

a

Rb

Ja

R

J

(G

h0

sychoneuroendocrinology (2015) 51, 92—100

Available online at www.sciencedirect.com

ScienceDirect

jou rn al h om epage: www.elsev ier .com/ locate /psyneuen

lasma adiponectin is inversely associatedith antenatal anxiety: Results from arazilian cohort

ernanda Rebeloa,b, Thatiana de Jesus Pereira Pintob,na Beatriz Franco-Senab, Jaqueline Lepschb,amila Benaimb, Claudio José Struchinera, Gilberto Kacb,∗

National School of Public Health, Oswaldo Cruz Foundation, Rua Leopoldo Bulhões, 1480 Manguinhos,io de Janeiro, RJ, BrazilNutritional Epidemiology Observatory, Department of Social and Applied Nutrition, Institute of Nutritionosué de Castro. Rio de Janeiro Federal University, Avenida Carlos Chagas Filho, 367/CCS—Bloco J—2◦

ndar, sala 29. Cidade Universitária—Ilha do Fundão, Rio de Janeiro, RJ, Brazil

eceived 30 June 2014; received in revised form 14 August 2014; accepted 16 September 2014

KEYWORDSAdiponectin;Anxiety;Mental health;Pregnancy;Cohort

Summary Antenatal anxiety may increase the risk of undesirable birth outcomes. Studieshave demonstrated an association between adiponectin and anxiety, but this issue has notbeen investigated during pregnancy. This study aimed to evaluate the association betweenplasma adiponectin, measured throughout gestation, and the occurrence of anxiety at latepregnancy (30—36th weeks). A prospective cohort was investigated in Rio de Janeiro, Brazil.Healthy pregnant women, aged 20—40 years, were evaluated between gestational weeks 5—13,22—26 and 30—36. State anxiety was measured using a validated version of the State-TraitAnxiety Inventory, and women were categorized as high (score ≥ 50, n = 30) or low anxiety(score < 50, n = 129). Plasma samples for all trimesters were analyzed using commercial ELISAkits to determine adiponectin concentrations (U/mL). Statistical analysis involved student’s

t-tests, chi-square, Pearson correlation, multiple logistic regression and linear mixed effects (LME) regression to model longitudinal trends of adiponectin, stratified for anxiety categories.Women with higher anxiety scores had lower mean concentrations of 3rd trimester adiponectincompared with those with lower scores (7.9; 95% CI: 7.0—8.9 vs. 9.9; 95% CI: 9.1—10.7).Women with 3rd trimester adiponectin values within the third tertile (10.47—26.57 U/mL) wereless likely to have high antenatal anxiety (adjusted OR = 0.30; 95% CI: 0.09—0.98) compared

∗ Corresponding author at: Avenida Carlos Chagas Filho, 367, CCS, Bloco J/2, andar, sala 29, Cidade Universitária, Ilha do Fundão, Rio deaneiro, RJ 21941-590, Brazil. Tel.: +55 21 39386595; fax: +55 21 22808343.

E-mail addresses: frebelos@gmail.com (F. Rebelo), thatiana.pinto@gmail.com (T. de Jesus Pereira Pinto), nutrianabeatriz@gmail.comA.B. Franco-Sena), jaquelepsch@gmail.com (J. Lepsch), camilabenaimnutri@gmail.com (C. Benaim), stru@fiocruz.br (C.J. Struchiner),ilberto.kac@gmail.com (G. Kac).

ttp://dx.doi.org/10.1016/j.psyneuen.2014.09.015306-4530/© 2014 Elsevier Ltd. All rights reserved.

Plasma adiponectin and anxiety during pregnancy 93

with those within the first tertile (2.25—7.08 U/mL). Unlike women with low levels of anxiety,those with high levels had a significant decrease of plasma adiponectin throughout pregnancy( = −0.07; 95% CI: −0.13 − [−0.01] vs. ˇ = −0.01; 95% CI: −0.05 to 0.03). Multiple LME modelindicated higher adiponectin throughout pregnancy for women with low anxiety (ˇ = −1.57; 95%CI: −2.78 − [−0.37]). In conclusion, plasma adiponectin throughout pregnancy was inverselyassociated with antenatal anxiety.© 2014 Elsevier Ltd. All rights reserved.

2

2

TwhBai5bd

waawpt(rtaoawdivfi

2ilt2(bAw

2.2. Anxiety symptoms

1. Introduction

Studies conducted in pregnant women have shown thatthe prevalence of antenatal anxiety behaviors ranges from6.6% to 19.8% (Andersson et al., 2003; Berle et al., 2005;Almeida et al., 2012; Rubertsson et al., 2014). Clinical anxi-ety, at this moment of life course, has been associated withadverse outcomes for the mother, such as increased riskof preeclampsia, greater frequency of nausea and vomitingduring pregnancy, and occurrence of postpartum depression(Araújo et al., 2008; Qiu et al., 2009; Rubertsson et al.,2014). Anxiety symptoms may also have a negative impacton the development of the fetus and may increase the riskof low birth weight and preterm delivery (Van den Berghand Marcoen, 2004; Li et al., 2008). Although anxiety duringpregnancy is an important public health problem, this dis-order has received little attention from health professionalsand researchers.

Adiponectin is a hormone produced by adipocytes thatsensitizes the body to insulin and that also has anti-inflammatory effects (Weyer et al., 2001; Fallo et al.,2004). Gustafson (2010) indicated the involvement ofadiponectin in the inhibition of important pro-inflammatorycytokines, such as tumor necrosis factor-� and interleukin-6,that have been positively associated with mental dis-orders (Gill et al., 2008). Previous studies have testedassociations between adiponectin concentrations and men-tal health disorders such as anxiety and depression, butevidence of these associations remains inconsistent (Leoet al., 2006; Nagata and Yamada, 2006; Cizza et al.,2010; Diniz et al., 2012; Jeong et al., 2012; Wilhelmet al., 2013). Wilhelm et al. (2013) reported a posi-tive correlation between adiponectin concentrations andanxiety in adults. However, Nagata and Yamada (2006)demonstrated that these variables were inversely correlatedin women with eating disorders. Although the biologicalmechanisms involved in this association are not totallyelucidated, some studies have suggested a strong contri-bution of the immune system and of metabolic factors(Cheng et al., 2012; Wang et al., 2014; Wilhelm et al.,2013).

This study aimed to evaluate the association betweenplasma adiponectin, measured longitudinally during ges-tation, and the occurrence of anxiety in late preg-nancy. We sought to clarify the relationship between

adiponectin concentrations and the occurrence of anx-iety symptoms in pregnancy, during which time thisbehavioral disorder is associated with several adverseoutcomes.

Tat3

. Methods

.1. Study design and eligibility criteria

he present study comprised a prospective cohort designith pregnant women followed at Heitor Beltrão publicealth care center, located in the city of Rio de Janeiro,razil. The women were recruited between November 2009nd October 2011, and those who met the following eligibil-ty criteria were invited to participate in the study: (a) in theth-13th week of gestation at the time of recruitment; (b)etween 20 and 40 years of age; and (c) free of any chronicisease (other than obesity).

The cohort had three follow-up waves at gestationaleeks 5—13, 20—26 and 30—36. Women were interviewednd blood samples were collected for subsequent laboratorynalysis. The initial baseline sample comprised 299 pregnantomen. However, we excluded women who presented twinregnancies (n = 4), who were diagnosed with an infec-ious or non-communicable disease after entering the studyn = 16) or who had a miscarriage or stillbirth (n = 27). Of theemaining 252 women, we had 36 losses at follow-up dueo transference of prenatal unit care (n = 16), change to anddress out of the catchment area of the health unit (n = 5)r absence of the 3rd trimester interview (n = 15). Statisticalnalyses were performed to compare women lost with thoseho remained in the sample, with respect to socioeconomic,emographic, anthropometric and reproductive character-stics. There were no significant differences in any of theseariables between the two groups of women (Supplementalle: Table S1).

After these exclusions and losses, the sample comprised16 women. The analysis of adiponectin was only performedn a subsample of 171 women due to an operational prob-em experienced with the analysis procedure that causedhe loss of two sets of data, one with 23 and the other with2 samples. Finally, we decided to exclude outlying values>30.0 U/mL; n = 12; range of excluded values 33.7—98.9),ased on a normal adiponectin distribution proposed byhima (2006). Our final sample comprised 159 pregnantomen.

he Spielberger State-Trait Anxiety Inventory (STAI) (Biaggiond Natalício, 1979) was used to measure anxiety symp-oms during mid and late pregnancy (weeks 22—26 and0—36, respectively). Although we have evaluated anxiety

9

straS

strsol

2

B7afo−t(0

2

Id(lifpm2wma

(ftapwFawtf(

lpm5

wS0

2

MtsSsPad

cbstwis

wadcswaac

aaeitW

ct

2

TmFaMtocotR

3

AO

4

ymptoms scores in the 2nd trimester, our main objec-ive in this paper was to describe the 3rd trimesteresults. Nevertheless, the results for the 2nd trimesterre available (Supplemental file: Table S2 and Table3).

We assessed state anxiety, which evaluates one’s tran-itory emotional state in relation to a specific situation orime period. This inventory consists of 20 items with fouresponses scored from 1 to 4 according to the intensity ofymptoms (total scores range from 20 to 80). The intensityf anxiety symptoms was categorized as high (score ≥ 50) orow (score < 50).

.3. Adiponectin concentration

lood samples were always collected between 6:50 and:50 am, for all participants, by a nurse technician atll gestational trimesters (weeks 5—13, 20—26 and 30—36)ollowing a 12-h fast. Plasma was separated after 5 minf centrifugation (5000 rpm) and subsequently stored at80 ◦C for later analysis. Plasma adiponectin concentra-

ion (U/mL) was determined using commercial ELISA kitsMillipore, St. Charles, Missouri, USA) with sensitivity of.78 ng/mL.

.4. Assessment of covariates

nterviews with structured questionnaires were conducteduring pregnancy to obtain the following information: ageyears), education (years), marital status (lives/does notive with a partner), skin color (white/brown/black), par-ty (0/1/≥2 deliveries), duration of sleep (hours/night),amily history of depression (yes/no), and unplannedregnancy (yes/no). Gestational age (weeks) was deter-ined based on the first ultrasound performed prior to

6 weeks of pregnancy (n = 156). When this informationas not available (n = 3), the reported date of the lastenstrual period was used to calculate the gestational

ge.Early pregnancy body mass index [BMI = weight

kg)/height (m)2] was calculated based on weight measuredrom 5th to 13th weeks of gestation and on the average ofwo measurements of height. Both measures were obtainedt the first follow-up visit. Height was measured using aortable stadiometer (Seca Ltd., Hamburg, Germany) andeight was obtained using a digital scale (Filizzola PL 150,ilizzola Ltda, Brazil), according to standardized proceduresnd recorded by trained interviewers. Early pregnancy BMIas classified according to cutoff points proposed by

he Institute of Medicine (Institute of Medicine, 2009) asollows: normal weight (18.5—24.9 kg/m2), overweight25.0—29.9 kg/m2) or obese (≥30 kg/m2).

Weight gain (kg) was calculated as the difference of theast weight measured before delivery (obtained at the lastrenatal visit at 34—42 gestational weeks) and the weighteasured at the first follow-up visit (gestational weeks

—13).

Plasma insulin (U/mL) and leptin (ng/dL) concentrations

ere also determined using commercial ELISA kits (Millipore,t. Charles, Missouri, USA) with sensitivities of 2 �U/mL and.5 ng/dL, respectively.

iCwd

F. Rebelo et al.

.5. Statistical analyses

eans with 95% confidence intervals (CIs) and propor-ions were used to describe characteristics of the sample,tratifying for late pregnancy anxiety symptom categories.tudent’s t-test and chi-square were used to compare theample distribution between high and low anxiety groups.earson correlations were estimated between 3rd trimesterdiponectin levels and anxiety scores and a scatter plotepicting this relationship was constructed.

Logistic regressions, both unadjusted and adjusted foronfounders, were performed to evaluate the associationetween plasma adiponectin (categorized in tertiles of theample distribution; reference category = 1st tertile) in eachrimester with anxiety symptoms. The following variablesere considered potential confounders and were included

n the adjusted analyses: 3rd trimester duration of sleep,kin color, early pregnancy BMI and marital status.

A longitudinal linear mixed-effect (LME) regression modelas employed to investigate whether changes in plasmadiponectin throughout pregnancy (continuous variable)iffered between outcome groups, with adjustments foronfounding variables (selected according to biological plau-ibility). LME models capture changes both between andithin individuals, allow the inclusion of time-dependentnd time-independent covariates and take into consider-tion that repeated measures in the same subject areorrelated.

Gestational age was included in the LME regression models a time variable and was modeled both as a random ands a fixed effect. All of the covariates (age, education,arly pregnancy BMI, family history of depression and plasmansulin [change over time]) and the dependent variable (3rdrimester anxiety score) were considered as fixed effects.e used the unstructured covariance matrix to fit the model.

An LME graph was constructed to evaluate longitudinalhanges in plasma adiponectin during pregnancy based onhe occurrence of antenatal anxiety symptoms.

.6. Ethical approval

he study protocol was approved by the research ethics com-ittee of the Institute of Psychiatry of the Rio de Janeiro

ederal University (Protocol number: 0012.0.249.000-09)nd the Municipal Secretary of Health of Rio de Janeirounicipality (Protocol number: 0139.0.314.000-09). All par-

icipants signed a two-way term of consent, which wasbtained freely and spontaneously after all necessary clarifi-ations had been provided. Participants did not get any typef compensation. All ethical procedures of this study relatedo research involving human beings followed the Brazilianesolution 466/2012.

. Results

total of 159 pregnant women were included in the study.f these women, 19% (n = 30) reported high anxiety scores

n the 3rd trimester. The mean anxiety score was 37.7 (95%I: 36.7—38.7) and 53.5 (95% CI: 52.4—54.7) for the groupsith high and low anxiety scores, respectively. All socio-emographic, reproductive, anthropometric, behavioral and

Plasma

adiponectin and

anxiety during

pregnancy

95

Table 1 Characteristics of the sample according to anxiety score categories at the third trimester of pregnancy. Rio de Janeiro, 2009—2012.

Variable n (%) Low anxiety scores (<50)a High anxiety scores (≥50)a p-Valueb

n (%) Mean 95% CI n (%) Mean 95% CI

State anxiety score 159 129 37.7 36.7—38.7 30 53.5 52.4—54.7 <0.001Age (years) 159 129 26.5 25.5—27.4 30 27.0 24.8—29.2 0.330Skin color 159 129 30

White 42 (26.4) 37 (28.7) 5 (16.7)Brown 76 (47.8) 62 (48.1) 14 (46.7) 0.219Black 41 (25.8) 30 (23.3) 11 (36.7)

Education (years) 159 129 8.6 8.1—9.1 30 9.0 8.0—10.0 0.261Early pregnancy BMI (kg/m2) 159 129 25.5 24.8—26.3 30 24.7 23.1—26.2 0.166

Normal weight [<25] (Ref) 88 69 (53.5) 19 (63.3) 0.599Overweight [25—30] 49 41 (31.8) 8 (26.7)Obese [≥30] 22 19 (14.7) 3 (10.0)

Marital status 159 129 30Lives with a partner 137 (86.2) 114 (88.4) 23 (76.7) 0.094Does not live with a partner 22 (13.8) 15 (11.6) 7 (23.3)

Parity (number of deliveries) 159 129 1.1 0.9—1.3 30 1.3 0.8—1.8 0.1920 56 (35.2) 47 (36.4) 9 (30.0) 0.6111 61 (38.4) 50 (38.8) 11 (36.7)≥2 42 (26.4) 32 (24.8) 10 (33.3)

Familiar history of depression 156 129 27Yes 40 (25.6) 32 (24.8) 8 (29.6) 0.602No 116 (74.4) 97 (75.2) 19 (70.4)

Unplanned pregnancy 159 129 30Yes 92 74 (57.3) 18 (60.0) 0.792No 67 55 (42.6) 12 (40.0)

Duration of sleep (hours) 159 129 7.7 7.4—8.1 30 6.8 6.0—7.7 0.021Weight gain (kg) 159 129 11.33 10.6—12.1 30 11.33 9.72—12.9 0.4133rd trimester gestational age (weeks) 159 129 32.3 31.9—32.7 30 32.6 31.7—33.5 0.2413rd trimester insulin (U/mL) 159 129 7.9 6.6—9.2 30 7.8 3.9—11.7 0.5113rd trimester leptin (ng/mL) 159 129 34.7 30.4—39.0 30 31.4 25.0—37.8 0.2421st trimester adiponectin (U/mL) 158 128 10.1 9.3—10.9 30 9.4 8.0—10.7 0.2082nd trimester adiponectin (U/mL) 158 128 11.3 9.1—13.4 30 8.2 7.0—9.3 0.0863rd trimester adiponectin (U/mL) 159 129 9.9 9.1—10.7 30 7.9 7.0—8.9 0.014

Note: BMI = body mass index; CI = confidence interval.a Refers to the results of the State Anxiety Inventory, where low is considered a score < 50 and high represents a score ≥ 50.b p-Value refers to t-tests for continuous variables and chi-square tests for categorical variables.

9

baastwpswh

tfees(taalafi

wdp[

ancaC−−[

4

Tcstiticaof

qTitoolls

acectmrbtrbea

sutttoa

tesie(Y

dchee2(endtp(

oacmmib(

acp

6

iochemical co-variables were comparable between highnd low anxiety score groups, except for 3rd trimesterdiponectin and sleep duration. Women with high anxietycores had lower mean concentrations of adiponectin inhe 3rd trimester (7.9 [95% CI: 7.0—8.9] U/mL) comparedith those with low anxiety (9.9 [95% CI: 9.1—10.7] U/mL,

= 0.014). Women with high anxiety scores reported shorterleep duration (6.8 [95% CI: 6.0—7.7] hours/night) comparedith women with low anxiety scores (7.7 [95% CI: 7.4—8.1]ours/night) (Table 1).

Adiponectin concentrations in the highest tertile duringhe 3rd trimester of pregnancy represented a protectiveactor against the occurrence of high antenatal anxi-ty (OR = 0.32 [0.11—0.98]; p = 0.045). The adiponectinffect persisted even after adjusting for the duration ofleep, early pregnancy BMI, marital status and skin coloradjusted OR = 0.30 [0.09—0.97]; reference category = 1stertile, p = 0.044) (Table 2). Plasma adiponectin in the firstnd second trimesters was not significantly associated withntenatal anxiety (Supplemental file: Table S4). The corre-ation between third trimester adiponectin levels and thenxiety score was −0.155 (p-value = 0.051) (Supplementalle: Figure S1).

As opposed to women with low anxiety levels, womenith high anxiety during the 3rd trimester, had a significantecrease in plasma adiponectin concentrations throughoutregnancy (ˇ = −0.07 [95% CI: −0.13 to −0.01] vs. ˇ = −0.0195% CI: −0.05 to 0.03]) (Fig. 1).

The multiple LME model indicated higher values ofdiponectin throughout pregnancy for women with low ante-atal anxiety (ˇ = −1.57 [95% CI: −2.78 to −0.37]). Otherovariates that were significantly associated with plasmadiponectin during pregnancy included age (ˇ = −0.16 [95%I: −0.25 to −0.07]), early pregnancy BMI (ˇ = −0.29 [95% CI:0.39 to −0.19]), insulin concentrations (ˇ = −0.05 [95% CI:0.10 to −0.01]) and family history of depression (ˇ = 1.31

95% CI: 0.25—2.36]) (Table 3).

. Discussion

he present study identified lower plasma adiponectinoncentrations during pregnancy in women with higherymptoms of anxiety at late pregnancy, compared withhose with lower scores of anxiety. Women in the high anx-ety group also had a significant decrease of this adipokinehroughout pregnancy, a change that was not observedn women with low levels of anxiety. Moreover, higheroncentrations of adiponectin in the third trimester weressociated with a 70% lower likelihood of the occurrencef antenatal anxiety, even when adjusting for possible con-ounders.

Some limitations need to be highlighted for more ade-uate interpretation and generalizability of these results.he study had losses at follow-up, a common occurrence

n cohort studies, which amounted to 14.3% (36/252) ofhe initial sample. Additionally, we had a limited number

f blood samples that could be analyzed for measurementf plasma adiponectin (79.2%; 171/216). However, as thisoss was random, this can be considered as a not selectiveoss that did not compromise the internal validity of thetudy.

oaece

F. Rebelo et al.

Some authors have already described the potentialssociation between the circadian clock and adiponectinoncentrations (Gómez-Abellán et al., 2010; Hashinagat al., 2013). It is important to highlight that the bloodollection was performed always at the same time early inhe morning, which may be considered a strength that caninimize this confound. However, it is possible that the dis-

uption of the circadian clock would drive the associationetween adiponectin level and anxiety. We were unable toest this hypothesis, as the present study lacks a more accu-ate measure of sleep deprivation especially in the day thelood was drawn. Due to this limitation, we are not able tovaluate differences in sleep deprivation, adiponectin andnxiety.

The longitudinal measurement of adiponectin is atrength of this study. Although sectional analyses werenable to detect differences in plasma adiponectin duringhe 1st and 2nd trimester between low and high antena-al anxiety groups, a repeated measures analysis was ableo capture this difference and also provided evidence thatnly the high anxiety group showed a significant decrease indiponectin concentrations.

The novelty of this study lies in the fact that it is the firsto evaluate the relationship between adiponectin and anxi-ty during pregnancy. Even among other population groups,tudies related to this topic are rare, with limited evidencen adults (both sexes, mean age: 47.9 ± 13.4 years) (Wilhelmt al., 2013), and in non-pregnant women with diabetesBrennan et al., 2009) and with bulimia nervosa (Nagata andamada, 2006). These studies have shown conflicting results.

The inverse association between adiponectin and mentalisorders has been reported more frequently for other out-omes than anxiety. Lower concentrations of adiponectinave previously been associated with major depression (Leot al., 2006; Diniz et al., 2012), alexithymia (Honkalampit al., 2014), obsessive compulsive disorder (Ari et al.,012), anorexia (Amitani et al., 2013) and bulimia nervosaNagata and Yamada, 2006). In an experimental study, Liut al. (2012) observed an antidepressant effect of exoge-ous administration of adiponectin in normal weight andiet-induced obese diabetic mice. These authors suggestedhat this effect occurs because adiponectin stimulates hip-ocampal neurogenesis, based on evidence from Zhang et al.2011).

Zhang et al. (2011) conducted an in vitro experiment andbserved that adiponectin treatment increased the prolifer-tion of cultured adult hippocampal neural stem/progenitorells in a dose- and time-dependent manner. This regulationay be the pathway that protects against the develop-ent of anxiety symptoms observed in our study, as the

nvolvement of hippocampal neurogenesis in anxiety-relatedehaviors has previously been reported by other authorsBergami et al., 2009; Revest et al., 2009).

It is important to notice that plasma adiponectin wasssociated only with 3rd trimester anxiety. This finding indi-ates that the pathway for anxiety development duringregnancy differs and depends on gestational age. While webserved a greater influence of social factors, such as parity

nd marital status in the mid-pregnancy, women with anxi-ty in the late pregnancy seems to be more influenced by theontinuous decrease of adiponectin over pregnancy, whichnded up with significant lower levels of this hormone.

Plasma

adiponectin and

anxiety during

pregnancy

97

Table 2 Association of plasma adiponectin in the third trimester of pregnancy with antenatal anxiety, unadjusted and adjusted. Rio de Janeiro, 2009—2012.

Variable n Anxiety score (n)a Prevalence of highanxiety (%)

OR 95% CI p-Valueb AORc 95% CI p-Valueb

Low (<50) High (≥50)

3rd trimester adiponectin (U/mL)1st tertile [2.25—7.08] 53 40 13 24.53 1.00 — — 1.00 — —2nd tertile [7.10—10.36] 53 41 12 22.64 0.90 0.37—2.21 0.819 0.61 0.23—1.67 0.3393rd tertile [10.47—26.57] 53 48 5 9.43 0.32 0.11—0.98 0.045 0.30 0.09—0.97 0.044

3rd trimester duration of sleep (h)≤6 49 33 16 32.65 1.00 1.00 — —>6 110 96 14 12.73 0.30 0.13—0.68 0.004 0.25 0.10—0.62 0.002

Skin colorWhite 42 37 5 11.90 1.00 — — 1.00 — —Brown 76 62 14 18.42 1.67 0.56—5.02 0.360 1.76 0.54—5.71 0.345Black 41 30 11 26.83 2.71 0.85—8.67 0.092 3.01 0.86—10.56 0.086

Early pregnancy BMI (kg/m2)Normal weight [<25] 88 69 19 21.59 1.00 — — 1.00 — —Overweight [25—30] 49 41 8 16.33 0.71 0.28—1.76 0.459 0.75 0.28—2.02 0.567Obese [≥30] 22 19 3 13.64 0.57 0.15—2.14 0.409 0.59 0.14—2.49 0.473

Marital statusLives with a partner 137 114 23 16.79 1.00 — — 1.00 — —Does not live with a partner 22 15 7 31.82 2.31 0.85—6.31 0.101 2.67 0.90—7.88 0.076

Note: OR = odds ratio; AOR = adjusted odds ratio; CI = confidence interval; BMI = body mass index.a Refers to the results of State Anxiety Inventory, where low is considered a score < 50 and high represents a score ≥ 50.b p-Values refer to Wald tests.c Estimates obtained were adjusted by all variables included in the final model (plasma adiponectin, sleep duration, skin color, early pregnancy BMI, and marital status).

98 F. Rebelo et al.

F pred xed ev ); (b

betvlm

1wh

igure 1 Longitudinal changes in plasma adiponectin duringe Janeiro, 2009—2012. Notes: values are fitted by a linear miariation for women: (a) with low anxiety score (p value = 0.625

Further studies could improve on the current resultsy analyzing the multiple isoforms of adiponectin. Amitanit al. (2013) found differences in the direction of

he association between adiponectin and anorexia ner-osa when analyzing each isoform separately (trimericow-molecular-weight [LMW] form; hexameric middle-olecular-weight [MMW] form; and dodecameric and

sio

Table 3 Multiple linear mixed effects model for repeated measuoccurrence of antenatal anxiety. Rio de Janeiro, 2009—2012.

Plasma a

ˇa

Fixed effectAnxiety score (low/high)c −1.57Age (years) −0.16

Education (years) 0.13

Early pregnancy BMI (kg/m2) −0.29

Family history of depression (no/yes)d 1.31

Insulin (U/mL) −0.05

Gestational age (weeks) −0.02

Random effectGestational age (weeks) 0.02

Intercept 20.89

Residual 8.26

Note: BMI = body mass index.a

refers to longitudinal linear regression coefficient.b p-Value refers to restricted maximum likelihood estimator.c Refers to the results of the State Anxiety Inventory, where low is co

score ≥ 50.d Reference category = no.

gnancy according to the occurrence of antenatal anxiety. Rioffects model; ˇ1 represents the gestational age coefficient of) with high anxiety score (p value = 0.032).

8-meric high-molecular-weight [HMW] form). The patientsith anorexia nervosa presented lower levels of HMW, butigher concentrations of LMW, compared with controls.

Since there are studies that used non-fastingerum/plasma samples to measure adiponectin, it ismportant to clarify that previous studies have alreadybserved no impact of fasting on adiponectin levels neither

res of plasma adiponectin during pregnancy, according to the

diponectin (U/mL) p-Valueb

95% CI

−2.78—[−0.37] 0.011−0.25—[−0.07] <0.001−0.04—0.30 0.134−0.39—[−0.19] <0.001

0.25—2.36 0.015−0.10—[−0.01] 0.043−0.05—0.02 0.320

0.01—0.0216.87—24.917.03—9.70

nsidered a score < 50 (reference category) and high represents a

A

A

B

B

B

B

C

C

D

F

G

G

G

H

H

I

Plasma adiponectin and anxiety during pregnancy

in rats (Kmiec et al., 2005) nor in humans (Shand et al.,2006).

The main finding raised by our results is the possible roleof adiponectin in the pathophysiology of anxiety during preg-nancy. In the short-term, this result will help contribute toknowledge about the etiology of anxiety. In the long-term,we hope to encourage more research in this area, so thatadiponectin may possibly be used as a marker or even as atherapeutic approach in the control of mood disorders.

Role of funding source

Funding for this study was provided by CNPq (grantnumbers: 471196/2010-0, 304182/2013-3) and FAPERJ(grant numbers: E-26/111.400/2010, E-26/110.681/2012; E-26/112.181/2012; E-26/111.698/2013). These agencies hadno further role in study design, collection, analysis and inter-pretation of data, in the writing of the paper or in thedecision to submit the paper for publication.

Conflicts of interest statement

None declared.

Acknowledgments

We would like to thank the funding agencies that have madepossible the realization of this study. FR received a scholar-ship from the National School of Public Health/Oswaldo CruzFoundation. GK and CJS received scholarships from NationalCouncil for Scientific and Technological Development (CNPqin the Portuguese acronym). TJPP received a scholarshipfrom Carlos Chagas Filho Foundation for Research Sup-port of Rio de Janeiro State (FAPERJ in the Portugueseacronym). ABFS and CB were recipients of scholarships fromthe Brazilian Coordination for the Training of UniversityLevel Personnel (CAPES in the Portuguese acronym).

Appendix A. Supplementary data

Supplementary material related to this article can befound, in the online version, at http://dx.doi.org/10.1016/j.psyneuen.2014.09.015.

References

Ahima, R.S., 2006. Metabolic actions of adipocyte hormones: focuson adiponectin. Obesity 14 (Suppl. 1), 9S—15S.

Almeida, M.S., de Nunes, M.A., Camey, S., Pinheiro, A.P., Schmidt,M.I., 2012. Mental disorders in a sample of pregnant womenreceiving primary health care in Southern Brazil. Cad. SaudePublica 28, 385—394.

Amitani, H., Asakawa, A., Ogiso, K., Nakahara, T., Ushikai, M.,Haruta, I., Koyama, K., Amitani, M., Cheng, K., Inui, A., 2013.The role of adiponectin multimers in anorexia nervosa. Nutrition29, 203—206.

Andersson, L., Sundström-Poromaa, I., Bixo, M., Wulff, M.,Bondestam, K., Åström, M., 2003. Point prevalence of psychi-atric disorders during the second trimester of pregnancy: apopulation-based study. Am. J. Obstet. Gynecol. 189, 148—154.

J

99

raújo, D.M.R., Pacheco, A.H.deR.N., Pimenta, A.M., Kac, G., 2008.Prevalence of anxiety symptoms and associated factors in acohort of pregnant women attending a health care center in themunicipality of Rio de Janeiro. Rev. Bras. Saúde Materno Infant.8, 333—340.

ri, M., Ozturk, O.H., Bez, Y., Arica, S., Can, Y., Erduran, D., 2012.Serum adiponectin and resistin levels in patients with obsessivecompulsive disorder. J. Affect. Disord. 136, 979—982.

ergami, M., Berninger, B., Canossa, M., 2009. Conditional deletionof TrkB alters adult hippocampal neurogenesis and anxiety-related behavior. Commun. Integr. Biol. 2, 14—16.

erle, J.Ø., Mykletun, A., Daltveit, A.K., Rasmussen, S., Hol-sten, F., Dahl, A.A., 2005. Neonatal outcomes in offspring ofwomen with anxiety and depression during pregnancy: a linkagestudy from The Nord-Trøndelag Health Study (HUNT) and Med-ical Birth Registry of Norway. Arch. Women’s Mental Health 8,181—189.

iaggio, A.M.B., Natalício, L., 1979. Manual para o inventário deansiedade Traco-Estado (IDATE). Rio Jan. Cepa, 15.

rennan, A.M., Fargnoli, J.L., Williams, C.J., Li, T., Willett, W.,Kawachi, I., Qi, L., Hu, F.B., Mantzoros, C.S., 2009. Phobic anxi-ety is associated with higher serum concentrations of adipokinesand cytokines in women with diabetes. Diabetes Care 32,926—931.

heng, X., Folco, E.J., Shimizu, K., Libby, P., 2012. Adiponectininduces pro-inflammatory programs in human macrophages andCD4+ T cells. J. Biol. Chem. 287, 36896—36904.

izza, G., Nguyen, V.T., Eskandari, F., Duan, Z., Wright, E.C.,Reynolds, J.C., Ahima, R.S., Blackman, M.R., POWER StudyGroup, 2010. Low 24-hour adiponectin and high nocturnalleptin concentrations in a case—control study of community-dwelling premenopausal women with major depressive disorder:The Premenopausal, Osteopenia/Osteoporosis, Women, Alen-dronate. Depression (POWER) Study. J. Clin. Psychiatry 71,1079—1087.

iniz, B.S., Teixeira, A.L., Campos, A.C., Miranda, A.S., Rocha, N.P.,Talib, L.L., Gattaz, W.F., Forlenza, O.V., 2012. Reduced serumlevels of adiponectin in elderly patients with major depression.J. Psychiatr. Res. 46, 1081—1085.

allo, F., Scarda, A., Sonino, N., Paoletta, A., Boscaro, M., Pagano,C., Federspil, G., Vettor, R., 2004. Effect of glucocorticoids onadiponectin: a study in healthy subjects and in Cushing’s syn-drome. Eur. J. Endocrinol. 150, 339—344.

ill, J., Vythilingam, M., Page, G.G., 2008. Low cortisol, high DHEA,and high levels of stimulated TNF-alpha, and IL-6 in women withPTSD. J. Trauma. Stress 21, 530—539.

ómez-Abellán, P., Gómez-Santos, C., Madrid, J.A., Milagro, F.I.,Campion, J., Martínez, J.A., Ordovás, J.M., Garaulet, M., 2010.Circadian expression of adiponectin and its receptors in humanadipose tissue. Endocrinology 151, 115—122.

ustafson, B., 2010. Adipose tissue, inflammation and atheroscle-rosis. J. Atheroscler. Thromb. 17, 332—341.

ashinaga, T., Wada, N., Otabe, S., Yuan, X., Kurita, Y., Kakino,S., Tanaka, K., Sato, T., Kojima, M., Ohki, T., Nakayama,H., Egashira, T., Tajiri, Y., Yamada, K., 2013. Modulation byadiponectin of circadian clock rhythmicity in model mice formetabolic syndrome. Endocr. J. 60, 483—492.

onkalampi, K., Viinamäki, H., Niskanen, L., Koivumaa-Honkanen, H., Valkonen-Korhonen, M., Elomaa, A.-P.,Harvima, I., Herzig, K.-H., Lehto, S.M., 2014. Reduced serumadiponectin levels in alexithymia. Neuroimmunomodulation 21,234—239.

nstitute of Medicine, 2009. Weight Gain During Pregnancy: Reex-amining the Guidelines. National Academies Press, Washington,

DC.

eong, H.-G., Min, B.J., Lim, S., Kim, T.H., Lee, J.J., Park, J.H.,Lee, S.B., Han, J.W., Choi, S.H., Park, Y.J., Jang, H.C., Kim,K.W., 2012. Plasma adiponectin elevation in elderly individuals

1

K

L

L

L

N

Q

R

R

S

V

W

W

W

Z

00

with subsyndromal depression. Psychoneuroendocrinology 37,948—955.

miec, Z., Pokrywka, L., Kotlarz, G., Kubasik, J., Szutowicz, A.,Mysliwski, A., 2005. Effects of fasting and refeeding on serumleptin, adiponectin and free fatty acid concentrations in youngand old male rats. Gerontology 51, 357—362.

eo, R., Di Lorenzo, G., Tesauro, M., Cola, C., Fortuna, E., Zanasi,M., Troisi, A., Siracusano, A., Lauro, R., Romeo, F., 2006.Decreased plasma adiponectin concentration in major depres-sion. Neurosci. Lett. 407, 211—213.

i, D., Liu, L., Odouli, R., 2008. Presence of depressive symp-toms during early pregnancy and the risk of preterm delivery:a prospective cohort study. Hum. Reprod. 24, 146—153.

iu, J., Guo, M., Zhang, D., Cheng, S.-Y., Liu, M., Ding, J., Scherer,P.E., Liu, F., Lu, X.-Y., 2012. Adiponectin is critical in determiningsusceptibility to depressive behaviors and has antidepressant-like activity. Proc. Natl. Acad. Sci. 109, 12248—12253.

agata, T., Yamada, H., 2006. Psycho-neuro-immunological aspectsof eating disorders. Int. Congr. Ser. 1287, 279—284.

iu, C., Williams, M.A., Calderon-Margalit, R., Cripe, S.M.,Sorensen, T.K., 2009. Preeclampsia risk in relation to maternalmood and anxiety disorders diagnosed before or during earlypregnancy. Am. J. Hypertens. 22, 397—402.

evest, J.M., Dupret, D., Koehl, M., Funk-Reiter, C., Grosjean, N.,Piazza, P.V., Abrous, D.N., 2009. Adult hippocampal neurogene-sis is involved in anxiety-related behaviors. Mol. Psychiatry 14,959—967.

F. Rebelo et al.

ubertsson, C., Hellström, J., Cross, M., Sydsjö, G., 2014. Anxietyin early pregnancy: prevalence and contributing factors. Arch.Women’s Mental Health 17, 221—228.

hand, B., Elder, P., Scott, R., Frampton, C., Willis, J., 2006. Bio-variability of plasma adiponectin. Clin. Chem. Lab. Med. 44,1264—1268.

an den Bergh, B.R., Marcoen, A., 2004. High antenatal maternalanxiety is related to ADHD symptoms, externalizing problems,and anxiety in 8-and 9-year-olds. Child Dev. 75, 1085—1097.

ang, Y., Wang, X., Lau, W.B., Yuan, Y., Booth, D., Li, J.-J., Scalia,R., Preston, K., Gao, E., Koch, W., Ma, X.-L., 2014. Adiponectininhibits tumor necrosis factor-�-induced vascular inflammatoryresponse via caveolin-mediated ceramidase recruitment andactivation. Circ. Res. 114, 792—805.

eyer, C., Funahashi, T., Tanaka, S., Hotta, K., Matsuzawa, Y., Prat-ley, R.E., Tataranni, P.A., 2001. Hypoadiponectinemia in obesityand type 2 diabetes: close association with insulin resistance andhyperinsulinemia. J. Clin. Endocrinol. Metab. 86, 1930—1935.

ilhelm, C.J., Choi, D., Huckans, M., Manthe, L., Loftis, J.M., 2013.Adipocytokine signaling is altered in flinders sensitive line rats,and adiponectin correlates in humans with some symptoms ofdepression. Pharmacol. Biochem. Behav. 103, 643—651.

hang, D., Guo, M., Zhang, W., Lu, X.-Y., 2011. Adiponectin stimu-

lates proliferation of adult hippocampal neural stem/progenitorcells through activation of p38 mitogen-activated proteinkinase (p38MAPK)/glycogen synthase kinase 3 (GSK-3)/-cateninsignaling cascade. J. Biol. Chem. 286, 44913—44920.