Maternal excessive gestational weight gain as a risk factor for autism spectrum disorder in offspring: a systematic review

PRISMA protocol guidelines (2015) were used for this systematic review. A literature search was conducted in the electronic databases of PubMed, SCOPUS, Embase, Cochrane Library and Google Scholar, in addition to Google itself, for all dates up to June 2020 (Table 1). Original articles published in English addressing the relationship between maternal GWG and risk of ASD in offspring were included. Studies on intellectual disability without concurrent ASD diagnosis, neurologic (epilepsy), or other psychiatric diseases (depression, psychosis and anxiety) were excluded. As shown in Table 1, keywords employed for the search were as follows: “gestational weight gain” OR “pregnancy weight gain” OR “weight gain in pregnancy” OR “weight gain during pregnancy” in title-abstract-keywords AND “autism” OR “ASD” OR “developmental delay” OR “developmental disorder” in title-abstract-keywords. The research question was: What is the relationship between maternal GWG and risk of ASD in the child?

Studies retrieved from the search were transferred to an Endnote file and duplicate articles were removed. Titles and abstracts of the remaining articles were assessed by two independent reviewers to screen articles within the scope of this review. Full texts of screened studies were then critically reviewed for eligibility and data extraction. Reference lists of the articles was manually searched to identify additional studies; and when full text of an article was not accessible, it was requested from authors by email. Review articles, animal studies, conference papers, and studies that assessed the relationship of GWG with other health issues such as diabetes or hypertension were excluded. Discrepancies between the two reviewers were resolved through discussion.

Selected studies were assessed for methodological quality by two independent reviewers. As shown in supplementary Tables 3 and 4, the Newcastle-Ottawa quality and risk of bias assessment tool for observational cohort and case-control studies [31] was used to evaluate the quality and risk of bias of the included studies based on three domains: selection of exposed and non-exposed groups and ascertainment of exposure; comparability of groups on the basis of the design or analysis controlled for confounders; and outcome regarding assessment and follow-up time. A star system was applied to classify articles as good, fair, or poor quality. Studies with a total score of 6 or higher were considered high quality.

As shown in Fig. 1, 92 studies were found using the electronic search strategy and four studies were identified through hand searching for a total of 96 that were assessed. Duplicates were deleted, leaving 57 studies. Of those, 13 publications met the topic and scope of the study during screening phase. During the critical review phase, six studies were excluded due to unavailability of full text (n = 1), lack of relevance to the study topic (n = 4), and a summary of Xiang et al. study [30] (n = 1). Hence, a total of seven articles were included in the final review (Fig. 1). The study of Bilder et al. [23] consisted of two different population- and research-based genetic studies, which were considered to be separate studies for the purpose of this review. Therefore, the total count of studies was eight for the final analysis.

Table 2 shows the characteristics of the included studies. All articles were published between 2010 and 2019 and were observational studies (six cohort and three case control). No randomized controlled trial studies were found. All reviewed articles except Dodds et al. study [25] investigated singleton births. Dodds et al. included both singleton and multiple births; however, they adjusted for the effect of multiple births on the relationship between GWG and risk of ASD [25]. All included studies considered the effects of principal confounder factors, including child’s gender, age, birth year, genetic susceptibility, intellectual disability, and mother’s pre-pregnancy BMI, weight, age, education, race/ethnicity, parity, smoking, family income, history of comorbidity, gestational age, as well as father’s age in analyzing the association between maternal GWG and offspring ASD risk. To identify ASD, four of the included studies used ICD (International Statistical Classification of Diseases and Related Health Problems), three studies used DSM-IV-TR (Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision), two studies used ADOS-G (Autism Diagnostic Observation Schedule Generic) and two studies used ADI-R (Autism Diagnostic Inventory-Revised) criteria. The definition of excessive or poor GWG varied across the studies; three studies used Institute of Medicine (IOM) guidelines to identify abnormal GWG. Chinese GWG guideline, GWG ≥18 kg as excess and <  7 kg as poor GWG, weight gained < 0.5 kg/week as poor GWG, and continuous model were other utilized methods to define abnormal GWG.

All included studies were rated as good quality (Tables 3 and 4, Supplementary). The quality scores of all six cohort studies were 9, where 9 represents the lowest degree of bias (Table 3, Supplementary). The quality scores of two case-control studies ranged from 8 to 9 (Table 4, Supplementary).

As shown in Table 2, seven studies investigated the association of maternal high GWG with risk of ASD in offspring. All seven studies indicated that high GWG was significantly associated with increased risk of ASD. Dodds et al. [25], in a retrospective longitudinal cohort study over 129,733 births, diagnosed 924 children with ASD after 1 to 17 years follow up from birth, and found that GWG ≥ 18 kg was significantly correlated with an increased risk of ASD. Gardner et al. [26], in a cohort study with 333,057 children ≥4 years, of whom 6420 were identified with ASD, reported that excessive GWG independently correlated with risk of ASD. In a case-control study, Shen et al. [28] found that greater GWG was associated with risk of ASD in the whole sample, also enhanced the risk of ASD in overweight/obese mothers. Windham et al. [29], also in a case-control study (ASD: n = 540, control: n = 776, aged 2 to 5 years), reported that excessive GWG was associated with risk of ASD, and, further, each 5-pound increase in GWG was correlated with 6% increased odds of ASD. Similar findings were also reported by two other studies (Table 2) [23, 30]. Bilder et al. [23] in a population-based cohort study (ASD: n = 128, control: n = 10,920) and in a genetic research-based study (ASD: n = 288, unaffected siblings: n = 493) on singleton births found that each 5 pounds of weight gained was significantly associated with increased risk of ASD. Xiang et al. [30] in a retrospective longitudinal cohort study on 322,323 singleton births (ASD: n = 3388) reported that each4 kg increase in GWG was positively associated with increased risk of ASD.

As shown in Table 2, five studies investigated the association of inadequate GWG with risk of ASD in offspring. Four indicated that insufficient GWG was not associated with increased risk of ASD [24, 25, 28, 29]. Burstyn et al. [24], in a cohort study over 218,890 births, after 4–10 years follow up, reported that poor weight gain (26–36 weeks, < 0.5 kg/week) was not associated with increased risk of ASD. Dodds et al. [25], in a cohort study (total children: n = 129,733; ASD: n = 924; ages 1–17 years) found that inadequate GWG (< 7 kg) was not associated with increased risk of ASD. Shen et al. [28], in a case-control study (control: n = 2236; ASD: n = 705; ages 2–9 years) found that inadequate GWG (based on a Chinese recommendation) was not associated with increased risk of ASD. Windham et al. [29], in a case-control study (control: n = 776; ASD: n = 540; ages 2–5 years) reported that insufficient GWG (based on IOM recommendation) was not associated with increased risk of ASD.

However, Gardner et al. [26], in a cohort study (total children: 333,057; ASD: 6420; ages ≥4 years) reported that insufficient GWG (based on IOM recommendation) was independently associated with increased risk of ASD.

Two of the included studies considered the confounding effect of the child’s genetic susceptibility on the GWG-ASD relationship. One study reported that the significant relationship between GWG and ASD was independent of genetic susceptibility [25], while the other reported that the association may be influenced by the confounder factor [26].

Three reviewed studies addressed the role of child’s intellectual disability in the GWG-ASD association [23, 26, 29]. All showed that excessive GWG increased the odds of ASD in ASD children with or without intellectual disability.

Of the included studies, seven studies examined the association of maternal pre-pregnancy BMI (n = 5) or weight (n = 2) with risk of ASD in offspring (Table 2). Five studies found that maternal pre-pregnancy BMI and weight ≥ 90 kg were not associated with increased risk of ASD in offspring [23 (population-based study and research-based study),24,28,29]. Bilder et al. [23] reported that maternal pre-pregnancy continuous BMI was not associated with ASD risk and concluded that GWG-ASD association was independent of pre-pregnancy BMI. Burstyn et al. [24] found that maternal pre-pregnancy weight ≥ 90 kg was not associated with increased risk of ASD in offspring. Shen et al. [28] showed that maternal pre-pregnancy low (< 18.5 kg/m2) and high (> 24 kg/m2) BMI were not significantly associated with risk of ASD. The author concluded that the maternal pre-pregnancy BMI might not be independently associated with ASD risk. Windham et al. [29] reported that maternal pre-pregnancy BMI (in all categories; < 18.5, 25.0–29.9 and ≥ 30 kg/m2) was not associated with risk of ASD.

However, two studies reported contrary findings [25, 30]. Dodds et al. [25] showed that maternal pre-pregnancy high weight (≥ 90 kg) was associated with increased risk of ASD. Xiang et al. [30] found that maternal pre-pregnancy continuous BMI and BMI ≥30were positively associated with ASD risk.

Nevertheless, in all studies that assessed the GWG-ASD relationship, the effect of maternal pre-pregnancy BMI or weight were considered and the observed link between GWG and ASD was independent.

In this respect, only two studies considered this point with two contrasting findings. Dodds et al. [25] reported that a significant relationship between GWG and ASD was independent of the child’s genetic susceptibility, since the association remained significant when the analyses were restricted to children with high genetic susceptibility (having an autistic sibling or a mother with a psychiatric or neurologic disease). The result suggests that the observed association is unlikely to be attributable to familial confounding. Gardner et al., in a large study in Sweden [26], performed matched sibling analyses to compare affected children with their unaffected pairs to assess whether observed correlations between maternal GWG and offspring ASD might be the consequence of confounding by shared familial factors. The analysis showed that the association did not remain significant and this may indicate that the result could be influenced by residual confounding, although the authors did not reach this conclusion and suggested an elevated risk of ASD with excessive GWG in matched sibling analyses. Taken together, child’s genetic susceptibility connection to the GWG-ASD association remains debatable and this area requires further investigation.

Three of the included studies considered the role of child’s intellectual disability in the association between maternal excessive GWG and offspring ASD. Bilder et al. [23] found that a significant GWG-ASD relationship was independent of the existence of comorbid intellectual disability among individuals with ASD, because the significant association persisted when analyses were limited to ASD children with a normal IQ. Windham et al. [29] found that elevated GWG similarly increased the odds of ASD in ASD children with or without intellectual disability. Similarly, Gardner et al. [26] showed that increased GWG similarly affected the odds of ASD in ASD children with or without intellectual disability across the entire studied population, while the odds of ASD in children without intellectual disability significantly increased in mothers with normal baseline BMI. Taken all together, it appears that the association between maternal excessive GWG and offspring ASD is independent of child’s intellectual disability.

Although most studies considered the confounding role of maternal pre-pregnancy BMI in GWG-ASD relationship and indicated that excessive GWG might increase the risk of ASD independent of BMI, however, several studies suggested an interaction between maternal BMI and GWG leading to increased risk of ASD in offspring. Shen et al. [28] found that the risk of ASD significantly rose in mothers with high pre-pregnancy BMI along with greater GWG. The authors concluded that the interaction between maternal BMI and GWG was significantly related to offspring ASD risk. Windham et al. [29] reported that the GWG-ASD association did not change by BMI in a continuous model. However, mothers with high BMI were more likely to have excessive weight gain (~ 55%) than mothers of normal weight (~ 35%) and the relation of ASD to greater GWG was more obvious, although non-significant, in overweight/obese mothers. Nevertheless, Gardner et al. [26] found when the analysis of GWG-ASD association was limited to mothers with a normal pre-pregnancy BMI for differentiating potential impacts of GWG from baseline BMI, the risk pattern did not change compared to the entire sample. The authors suggested that excessive GWG was independent of BMI correlated with ASD. Collectively, it is speculated that the association of GWG-ASD independent of maternal BMI is prevailing; however, further investigations are warranted.

Most of the studies used the WHO categorization of BMI [32]; however, two studies used continuous BMI for analysis and two used pre-pregnancy weight instead of BMI, which limits comparability of the results with respect to the BMI-GWG-ASD association.

ASD is a sex-specific neurological condition which affects more males than females. This may indicate that underlying mechanisms of sex differentiation may contribute to the neurobiology of ASD. Sex-specific differences in the occurrence of ASD may be attributable to sex-hormone specific effects. High perinatal testosterone levels have been suggested as a potential risk factor for neurodevelopmental disorders, including ASD [33‐37]. In this regard, male fetuses would be more affected since testosterone concentration in the amniotic fluid of male fetuses is significantly higher than those of female fetuses at all stages of gestation [38]. The results of this review showed that high GWG may be a risk factor for ASD. According to some evidence, high GWG is accompanied by high testosterone levels [39]. This may indicate that child neurodevelopment may be associated with GWG in a sex-specific manner. It is suggested that fetal testosterone level to be taken into account in the analyzing of the relationship between GWG and risk of ASD.

None of the articles reviewed mechanistically studied the role of GWG on ASD risk and the underlying mechanisms connecting maternal excessive GWG to offspring ASD remain unclear. However, the dysregulation of steroid hormones, leptin, and pro-inflammatory cytokines during gestation have been projected as possible mechanisms involved in psychopathology. Endogenous steroid hormones (for example, testosterone, estrogen, progesterone, or cortisol) released by mother, placenta, and fetal gonads and adrenal glands generate a fetal steroid environment [40]. These steroid hormones, as environmental factors, may influence fetal gene transcription and expression through DNA binding during susceptible stages of embryonic development [40]. Any impairment in the fetal steroid hormone environment can lead to dysfunction in the body or even death. Inappropriate levels of fetal steroid hormones such as increased fetal testosterone levels have been reported to be associated with autistic traits [33‐35]. As well, a positive relationship has been reported between increased levels of amniotic steroid hormones and ASD [41]. Lof et al. [42], in a longitudinal study of Swedish women, found that plasma levels of progesterone during pregnancy were positively associated with GWG.

Disturbance of leptin is another suggested mechanism contributing to psychopathology. Leptin has a vital role in fetal growth and development during pregnancy [43]. Any dysregulation in fetal leptin levels leads to mental health impairments and neurodevelopmental disorders, including ASD [44]. High plasma leptin in early childhood has been proposed as a potential biomarker for ASD [45] and increased circulating leptin is consistently observed in individuals with ASD [44, 46, 47]. According to previous investigations, fetal leptin levels are affected by maternal leptin levels [48, 49]. Walsh et al. [49] demonstrated an association between maternal and fetal leptin levels at each time point of pregnancy period. Elevated levels of leptin have also been reported in the placental vascular endothelial cells which are associated with maternal obesity [50]. Variation in the interchange of leptin among mother, placenta, and fetus may influence the development of the fetus and increase the risk of disease in later life [51]. Higher maternal leptin levels have also been shown to be associated with greater GWG [52]. The results of two birth cohort studies found that excessive GWG correlated with greater leptin levels in either cord blood or post-delivery maternal serum [53]. Patro-Małysza et al. [54] also reported greater umbilical cord leptin levels in neonates born to mothers with excessive GWG. Further, Vargas-Aguirre at al [55]. found that umbilical cord blood leptin levels were higher in neonates born to mothers with high GWG. Collectively, it is thought that excessive GWG may likely associate with in-utero leptin disturbance, which in turn may lead to increased risk of ASD. However, further studies are needed to support the association of fetal leptin levels and risk of ASD.

Leptin is able to cross the blood-brain barrier and enter the brain. As shown in Fig. 2, the brain leptin-glutamate or leptin-serotonin interactions have been proposed as possible mechanisms involved in the development of neurodevelopmental disorders, including ASD. Serotonin and glutamate are well-known neurotransmitters that have critical roles in neural activities and social behaviors. Clinically, abnormal levels of these neurotransmitters are linked with many neurodevelopmental disorders, including ASD [56, 57]. Serotonin regulates neural development and alterations in its concentration during development can have life-long effects. Exposure to abnormal levels of serotonin during the prenatal period may contribute to behavioral impairments in adulthood [58]. Excessive levels of glutamate have also been reported to have neurotoxic properties [59]. Calapai et al. [60] demonstrated that leptin injections caused a dose-dependent increase of serotonin in mice. Yadav et al. [61] showed that leptin decreased serotonin synthesis and prevented neuronal activity of serotonergic neurons in the ventromedial hypothalamus. Haque and Haleem [62] indicated that serum leptin was inversely associated with circulating serotonin in working men. Fuente-Martín et al. [63] reported that leptin regulated astrocyte-specific glutamate. Further, Yu and Cai [64] showed that excess central leptin increased glutamatergic pathway activation.

Interaction of steroid hormones with the neurotransmitters is another possible pathway for the development of neurodevelopmental disorders (Fig. 2). Zlotnik et al. [65], in a study on men and women, showed that blood glutamate levels were negatively associated with plasma levels of estrogen and progesterone. Raz et al. [66] found that serotonin was regulated by sex steroid hormones and could influence mood. Ladisich [67] found that progesterone injections dose-dependently increased serotonin turnover. According to the evidence, steroid hormones are linked to the regulation of leptin and the leptin receptor [68‐70]. Thus, steroid hormones might indirectly contribute to the pathogenesis of ASD through the leptin interface. However, more research is required to confirm the interaction of leptin or steroid hormones with the neurotransmitters involved in occurrence of ASD, in the prenatal period. The role of neuroinflammation in the projected pathways should also be a focus for future study.

This study highlights that pregnancy excessive weight gain, which exhibits the metabolic and nutritional condition of the mother throughout pregnancy, may have long-term neurodevelopmental effects on children and could be employed as a serious predictive biomarker of ASD. Focusing attention on helping women adhere to GWG guidelines is imperative to achieving a healthy weight gain during pregnancy.

Using large sample sizes and consideration of the role of potential confounding factors in the GWG-ASD association in the included studies were strengths of the study.

Various criteria were used to identify ASD and GWG across the studies, which may influence the comparability of the results. None of the studies investigated the role of biochemical markers such as leptin, steroid hormones, neurotransmitters, or neuropeptides involved in the pathogenesis of ASD in the prenatal period to determine underlying mechanisms connecting maternal excessive GWG with offspring ASD.