1. Introduction
Many studies have examined the associations between maternal obesity and a wide range of mental health problems of their children, including emotional and behavioral problems [
1,
2,
3,
4]. In general, these studies support an association between maternal BMI and poor cognitive performance and increased risk of developing depression and anxiety, however, the relationship between maternal obesity and the risk of developing autism spectrum disorder (ASD) and attention deficit and hyperactivity disorder (ADHD) in their children is currently less clear [
1]. We previously reported that maternal pre-pregnancy BMI ≥ 30 kg/m
2 was a risk factor for increased child behavioral problems [
5]. However, a mechanism connecting observed child behavioral problems and maternal obesity is largely unclear at this time.
One of the possible mechanisms to explain this can be inflammatory status. Obesity is considered a state of chronic inflammation and secretion of inflammatory biomarkers, including interleuikin-6 (IL-6) and tumor necrosis factor-α (TNF-α) [
6]. It has been suggested that elevated levels of inflammatory markers were associated with ASD, ADHD and other neurodevelopmental disorders [
3]. Thus, inflammation is an important potential mechanism. Metabolic hormone-induced programming can be another possible mechanism. It is known that there is maternal fetal transmission of leptin, a metabolism-related hormone [
7]. Several studies have suggested that leptin levels are associated with psychopathology [
8]. Lappas et al. found that leptin activated proinflammatory cytokine release and phospholipid metabolism in human placenta and suggested that leptin may indirectly influence brain development [
9].
There have been studies that investigated levels of metabolic hormones, including adiponectin and leptin, in association with mental and neurobehavioral problems such as ASD [
10,
11] and ADHD [
12,
13]. These studies have suggested that changes in lipid metabolism are a key to understanding the etiology of neurobehavioral problems, yet the mechanism is mostly unknown. Cross-talk between adipokines, including inflammatory cytokines and metabolic hormones secreted from adipose tissue and the central nervous system needs to be further investigated. Fetal adipokines may potentially contribute to child neurobehavioral development. Thus, the aim of this study was to investigate a possible association between fetal adipokine levels and child behavioral problems at preschool age in a prospective study.
3. Results
A comparison of the characteristics of participants in the normal and borderline/clinical groups is shown in
Table 1. All the characteristics shown in the
Table 1, except maternal age and child sex, were not significantly different between the normal and borderline/clinical groups. The mean maternal age was younger in the borderline/clinical group (30.7 ± 4.6) compared to the normal group (32.1 ± 4.3). The borderline/clinical group included higher percentage of boys (63.9%) compared to the percentage in the normal group (49.7%).
The prevalence of obese (BMI ≥ 25 kg/m
2) was 8% overall, and no difference was found between the normal and borderline/clinical groups. A comparison of characteristics between follow-up study population and the present study was shown in
Table S1 (
Supplementary Materials). The distribution of adipokine levels in cord blood is shown in
Table 2.
The median levels of total and HMW adiponectin and leptin were significantly higher in girls compared to these of boys (
Table S2). The median level of IL-6 was significantly higher in boys (
Table S2). The number of children in borderline/clinical groups of total and each subscale was shown in
Table 3. The percentage of children in the borderline/clinical range of TDS was 16.9%. Boys showed a higher prevalence of having problems in TDS, hyperactivity/inattention, and prosocial behavior problems compared to girls. The association between cord blood adipokine levels and child behavioral problems are shown in
Table 4. After adjustments, association between decreased hyperactivity/inattention and increased leptin was still significant (OR = 0.23, 95% CI: 0.06–0.89). Increased leptin was marginally associated with prosocial behavioral problems (OR = 0.38, 95% CI: 0.12–1.17) without statistical significance. Cord blood TNF-α and IL-6 levels were not associated with any of the child behavioral problems. The results of stratification by child sex are shown in
Table S3. Increased leptin was significantly associated with decreased total problems and hyperactivity/inattention in boys. No significant association was found in girls.
4. Discussion
We found that increased cord blood leptin level was associated with decreased hyperactivity/inattention at preschool age. To our knowledge, this is the first prospective epidemiological study investigated associations between fetal adipokine levels and child behavioral problems. Child behavioral problems in association with obesity, overweight and high BMI have been reported through cross-sectional studies [
24,
25,
26]. Similarly, several previous studies investigated child behavioral problems and cognitive and motor development in association with leptin levels in cross-sectional studies [
27,
28]. However, these cross-sectional studies could only provide relationships between obesity or biomarker levels and child mental and behavioral problems at one point, but they cannot find causal relationships.
The mean cord blood adiponectin levels of this study (17.1 µg/mL) was in a similar range as previous studies (21.3 µg/mL for a Taiwanese study, and 18.23 µg/mL for a Canadian study, respectively) [
29,
30]. Contrary, the mean leptin level of this study (4.9 ng/mL) was comparable to levels in a Taiwanese study (4.6 ng/mL) [
30], however, much lower compared to the study in Canada (19.8 ng/mL) [
29]. Adiponectin and leptin levels vary among ethnicities according to the literature [
31], this could explain relatively the lower leptin among the Asian population. Additionally, it is possible that difference in the prevalence of overweight women in this study and others may explain varied leptin levels among studies. The median level of cord blood TNF-α in this study (2.45 pg/mL was lower compared to the previous report from Hong Kong (7.44 pg/mL) [
32] and from Taiwan (5.47 pg/mL) [
33]. The median level of cord blood IL-6 in this study (1.06 pg/mL) was slightly higher but comparable to the report from Hong Kong (0.65 pg/mL) [
32], however, lower than levels in Taiwanese (3.70 pg/mL) [
33].
The results from this study suggested that fetal leptin levels may possibly be a predictor of child hyperactivity/inattention problems at preschool age even after controlling for maternal pre-pregnancy BMI. In this study, increased maternal BMI was correlated with increased cord blood leptin levels (
Table S2) and increased leptin levels were associated with reduced risk of child hyperactivity/inattention problems. Two of the previous cross-sectional studies found relationships between higher leptin levels and more problems or decrease in cognitive development in children, which was different from our finding. One possible reason for the difference was the difference in the study design. Other possible reason was the difference in the prevalence of overweight. According to the Global Health Observatory data of 2015 from the World Health Organization (WHO), both countries in previous studies (Brazil and Germany) have much higher prevalence of overweight among females (74% and 75%, respectively), whereas prevalence of overweight was only 20% in Japan. Moreover, the trend of prevalence of overweight differed between these two countries and Japan. The prevalence increased in these two countries, contrary, it decreased in Japan. This difference in prevalence of overweight could contribute to different findings among studies. Leptin is known to play a critical role in gestation. The placenta synthesizes leptin, as indicated by the presence of high amount of leptin messenger RNA [
34,
35]. Other possible sources of leptin for the fetus include fetal membranes and the umbilical cord that co-express leptin and leptin receptor genes during pregnancy [
36] and amnion cells that secret leptin into the circulating amniotic fluid [
37]. Newborn infant leptin levels are much higher than that of children and adults [
38] and the levels decreases significantly after birth [
39]. Elevated cord blood leptin levels are associated with increased adiposity and hyperleptinemia of offspring [
40].
Leptin levels were not only associated with physiological changes, but may also affect the brain development of the fetus [
8]. Although studies have suggested the importance of fetal leptin levels regarding fetal development, little is known in human studies and thus, animal studies are necessary. Leptin has been well studied its role of regulating neuroendocrine system and brain development [
41]. In human studies, leptin treatment increased grey matter concentration in areas such as the anterior cingulate gyrus, inferior parietal lobule, and the cerebellum which have roles in emotion, attention and motivation [
42] and improved cognitive development [
43]. The amygdala is well known for its role in anxiety and stress response and the presence of leptin receptors, and their projections to this brain region indicate that leptin may play a role in the mediation of emotions and behaviors [
8]. Leptin regulates several neurotransmitter systems and brain regions critical in behavioral regulation in both rodents and human [
44,
45]. The emerging picture of leptin interaction with lateral hypothalamic area (LHA) suggests that the LHA is not merely regulating feeding, but is a crucial integrator of energy balance and motivated behavior [
46]. It is possible that leptin has effects on fetal development and psychopathology via inflammation according to some evidences [
9,
47]. However, in this study, cord blood inflammatory biomarker levels including TNF-α and IL-6 were found to be not associated with child behavioral problems. This indicated that cord blood leptin level was independently associated with child behavioral problems.
In animal models [
48], neonatal leptin deficiency increased cerebral cortex leptin receptor expression and reduces frontal cortex volumes in association with increased adult locomotor activity. Similarly, falling levels of leptin resulted in increased activity levels in rat model [
49]. Furthermore, in growth restricted male mice, physiologic leptin replacement improves adult neuro outcomes [
50]. Our results agree with these findings from animal studies.
In an adult study, higher plasma adiponectin levels were associated with neurodegeneration and cognitive decline [
51]. Contrary, a recent review article mentioned that adiponectin had several protective functions in the peripheral tissues including insulin sensitizing, anti-inflammatory and anti-oxidative effects that may benefit neurodegenerative diseases [
52]. The role of adiponectin is still controversial. Moreover, no prospective study has shown the association between adiponectin and child neurobehavioral problems to our knowledge.
The previous study showed that elevated levels of IL-6, but not TNF-α were significantly associated with a decrease in motor score in the first year of life [
51]. Their finding suggested that markers of inflammation could serve as prognostic indicators. However, in this study, we did not find any association between either IL-6 or TNF-α and child behavioral problems. Even though both cross-sectional and prospective studies have been conducted, the findings were inconclusive. Thus, more longitudinal studies are needed to confirm whether these biomarkers can be a predictor of child neurobehavioral problems or not.
The strength of this study was a prospective study design and it was able to report longitudinal associations. Our study also has limitations. Although this study was population based, the findings may not be fully representative for the total population with regard to sociodemographic characteristics. The percentages of parental education ≥13 years were higher in this study population compared to the follow-up study population (n = 3896), however, not very different from the returned SDQ population (n = 2079). The percentage of annual family income during pregnancy ≥5 million Japanese yen was also higher in this study population compared to the follow-up population but not that much from the returned SDQ population. This implied that null bias existed between the population with (the present study, n = 361) and without (returned SDQ, n = 2079) adipokine measurements. Contrary, the percentage of maternal smoking during pregnancy was much lower in this study population compared to the follow-up study population and also slightly lower compared to the returned SDQ population (
Table S1). Therefore, the present study population may represent a higher socioeconomic status population. Since parental education and annual family income were inversely correlated with SDQ scores and maternal smoking during pregnancy was positively associated with SDQ scores (
Table S4), we may have underestimated the effects. One other issue was that we did not assess maternal mental health including anxiety or depression in this study. Maternal mood and mental health may contribute to child behavioral problems, however, we were unable to control for the statistical analysis. In addition, maternal history of hyperactivity/inattention was not obtained and not controlled in the statistical analysis. Thus, our findings may possibly be hindered. Finally, we should address that the prevalence of obese pregnant women in this study was not high (8%), and thus, this study population may not be appropriate to examine maternal obesity induced inflammation programing of offspring neurodevelopmental disorders.