Background
Autism spectrum disorder (ASD) is the collective term for developmental disabilities characterized by impairments in social interaction and verbal and nonverbal communication and by repetitive patterns of behavior. While the biological pathways contributing to ASD remain unclear, current consensus is for a multifactorial etiology, incorporating a constellation of genetic risk variants that may interact with environmental factors [
1]. The in utero hormone environment is critical to development of the structure and function of the central nervous system, and there is increasing evidence that prenatal endocrine factors may also play a key role in the etiology of some cases of ASD [
2]. A particular focus of previous research has been on fetal exposure to elevated levels of androgens, measured via sampling of amniotic fluid during the second trimester [
3] and umbilical cord blood at birth [
4]. Exposure to elevated concentrations of prenatal androgens has been found to be related to a diagnosis of clinical ASD in offspring [
3], as well as autistic-like traits in the general population [
5,
6], particularly communication difficulties [
7,
8]. While it is important to highlight that these findings have not always been replicated [
4,
9], further support for the potential role of the prenatal hormone environment in the etiological pathways contributing to ASD has been observed through associations with factors known to influence maternal hormone status, such as maternal age at menarche [
10], pre-pregnancy BMI [
11], and infertility and its treatments [
10,
12]. This evidence strongly supports the further examination of how prenatal endocrine factors may contribute to ASD.
Nausea and vomiting during pregnancy (NVP) is experienced by approximately 70 to 80% of pregnant women [
13]. The timing, frequency, and duration of NVP varies between women and, in its most extreme form (hyperemesis gravidarum), is characterized by severe and persistent nausea and vomiting that results in a significant loss in body weight [
14]. While NVP is known to involve a heritable component [
15,
16], the biological mechanisms underpinning the maternal experience of severe NVP are thought to involve the dysregulation of maternal human chorionic gonadotropin (hCG), thyroid hormones (particularly thyroxine and thyroid-stimulating hormone), and estrogens [
14,
17]. There are a number of reasons to hypothesize a link between NVP and the ASD phenotype among offspring. First, women who experience hyperemesis gravidarum have a greater risk of having a child with behavioral and neurodevelopmental conditions [
18,
19]. While no study has been sufficiently powered to investigate offspring risk for ASD, an increased risk has been observed for common neurodevelopmental comorbidities for ASD, such as intellectual disability, developmental language disorder, and attentional deficit hyperactivity disorder [
20]. Second, differential levels of hCG during pregnancy have been associated with later ASD in offspring. A recent study of Californian state records examined the association between hCG levels during pregnancy (collected for routine screening between 15 and 20 weeks’ pregnancy) and the prevalence of ASD among offspring [
21]. An increased risk of offspring ASD was identified for women with the highest and lowest decile of hCG concentration distribution, suggesting a link between atypical hCG regulation and offspring ASD. Third, a number of large studies have identified an association between maternal autoimmune thyroid disease during pregnancy and risk for ASD [
22,
23], with a recent meta-analysis reporting a pooled odds ratio of 1.29 (95% confidence interval 1.14–1.45) [
24].
Considered together, the accumulating body of research presents the hypothesis that an atypical profile of pregnancy-related hormones may be involved in the etiological pathway for some cases of ASD, and that this may be indexed by NVP. The current study provides the first investigation of whether maternal NVP relates to symptom severity in children with ASD. Given evidence linking the experience of any nausea or vomiting symptoms during pregnancy to hormone dysregulation [
25], the study examined ASD symptomatology in relation to the full severity of NVP symptoms. Based on previous findings linking NVP to neurodevelopmental delay and disorder [
20], and the association between hormones implicated in NVP and risk for offspring ASD [
21], we hypothesized that increased frequency and severity of maternal NVP would relate to more severe ASD symptoms among children.
Discussion
This is the first study to report the relationship between maternal experience of NVP and ASD severity in offspring. Increasing frequency and severity of NVP was positively related to the severity of social difficulties (SRS total score) and communication impairments (CCC-2 GCC scale) in offspring. The difference in the average SRS total and CCC-2 GCC scores between the offspring of mothers who experienced no NVP and those whose mothers had daily nausea and vomiting was approximately 0.8 of a standard deviation for each measure. While no study has quantified a clinically significant difference using these measures, a difference of this magnitude is highly likely to reflect clinically relevant differences in the severity of each phenotype. The cause of NVP is multifactorial and most likely involves pregnancy-related changes in the maternal hormone milieu. While the current study did not directly measure maternal hormones during pregnancy, there is considerable evidence linking the maternal experience of NVP to dysregulation of hCG, thyroid hormones, and estrogens. Several studies have reported an increased risk for offspring ASD among women experiencing dysregulation of hCG [
21] and thyroid hormones [
22‐
24] during pregnancy. The current study is the first to identify a link between NVP and the severity of ASD phenotype and provides further evidence for the possible involvement of prenatal hormone exposure in the causal pathway(s) for the ASD behavioral phenotype [
3].
It is important to highlight that non-hormonal factors have also been linked to the onset of NVP. For example, twin studies have reported high heritability for NVP both within the normal range [
32] and at the clinical extreme (
hyperemesis gravidarum) [
33]. A potentially fruitful line of research would be to examine how NVP may co-occur with heritable autistic traits—the so called broader autism phenotype [
34]—in mothers of children with ASD. The co-occurrence of these two phenotypes at above chance levels may afford further insights into the pattern of findings identified in the current study. Furthermore, several studies have found that between 90 and 95% of women with hyperemesis gravidarum are seropositive for
Helicobacter pylori (
H. pylori) compared to between 45 and 50% of women who do not experience NVP [
35,
36]. While these prevalence figures have not always been replicated [
37], a meta-analysis of 25 case-control studies reported a pooled odds ratio of 3.32 (95% confidence interval 2.25–4.90) for
H. pylori infection in women with hyperemesis gravidarum [
38]. Unlike hormonal theories for NVP, the biological pathways linking maternal
H. pylori infection and disrupted fetal neurodevelopment are less clear. Recent evidence from both preclinical [
39] and human studies [
40] suggests that maternal infection with
H. pylori is associated with a higher risk of fetal growth restriction, which may implicate poor placental function [
41]. Furthermore, numerous epidemiological studies have found that maternal
H. pylori infection increases risk for pre-eclampsia [
42,
43], potentially through negative effects on early placental development and function [
44]. However, it is important to emphasize that
H. pylori was not measured in this study and the rate of gestational hypertension and preeclampsia in the study sample was low, and thus, it was difficult to ascertain its relationship with maternal NVP. An alternative potential mechanism is the possible effect that a maternal inflammatory reaction (in response to
H. pylori infection) may have on the developing fetus [
45]. There is a strong literature describing the effects of maternal immune activation during pregnancy on fetal neurodevelopment [
46], with a particular focus on the increased risk of offspring ASD [
47‐
49]. However, no study has examined maternal
H. pylori infection in relation to offspring ASD, and it remains unclear how bacteria that infects the stomach of half of the human population [
50] may cause an inflammatory response that disrupts neurodevelopment in only a relatively small proportion of children.
Another area of growing research interest is how the sex of the fetus may influence vulnerability to differences in the prenatal environment. While the majority of research in this area has highlighted the vulnerability of male fetuses to prenatal perturbations, the effects of NVP and associated hormones on fetal outcomes has predominantly identified effects for female offspring. Numerous epidemiological studies have found that NVP is more common when gestating a female fetus [
51‐
53], and this finding was replicated in the current study. Several studies have also reported that maternal experience of NVP [
53] and increased hCG concentrations during pregnancy [
54] influence growth rates in female but not male fetuses. The mechanisms by which biological sex may impart vulnerability or protection on the fetal nervous system remain largely unknown [
55]. Fetal growth rate is reflective of intrauterine conditions, and fetal growth restriction is known to be associated with poor placental function [
56]. There is increasing evidence that the placenta functions in a sex-specific manner and has different interactions with steroid pathways, growth factor pathways, and the fetal-placental immune system depending on the sex of the fetus [
57]. While the current study did not have adequate statistical power to conduct analyses separately for male and female offspring (see Additional file
1), this background evidence suggests that an investigation of whether offspring sex moderates the relationship between NVP severity and ASD symptomatology is an important priority for future research.
A strength of the study design was the availability of standardized and widely used assessments of symptomatology on a large cohort of children with ASD that were assessed in a single geographic location (Perth, Western Australia), thus limiting bias in data collection methods. However, we note two important limitations of the study design. First, the scale used to measure NVP was developed for the purpose of this study and has not undergone broader validation. The most widely used prospective scale—the Pregnancy-Unique Quantification of Emesis and nausea (PUQE) [
58]—asks women about NVP symptoms in the immediately preceding 12-h period. Given the retrospective nature of the current study, the PUQE was not considered a reliable measure of NVP, and the scale used in the current study incorporated broader NVP categories asking about daily NVP experiences (compared to hourly experiences in the PUQE). However, a major limitation of this scale is the lack of information on the timing and duration of NVP experienced by the mothers. While it is estimated that 91% of women experience NVP prior to 20 weeks’ gestation [
59], there is considerable variability in the onset (and therefore duration) of symptoms. This information was not recorded in the current study, which limits the conclusions that can be drawn from the dataset. Furthermore, while the current study identified significant linear trends across NVP categories of increasing frequency and different symptomatology (nausea vs vomiting), it remains unclear whether these represent a true gradient of NVP severity. However, we note that treatment studies for NVP often differentiate between nausea and vomiting in defining NVP symptoms of increasing severity [
60], and epidemiological studies have identified differential relationships between these categories and a number of pregnancy [
61] and offspring [
62,
63] outcomes.
A second limitation is that the obstetric information, including information on NVP, was collected via retrospective maternal report, which may be subject to a recall bias [
64]. While the maternal experience of hyperemesis gravidarum is routinely noted in medical records, the full spectrum of NVP symptomatology is not prospectively recorded, which makes it challenging to determine the accuracy of maternal recall in this study. An average of 8 years had passed since the experience of NVP in the current maternal sample. While there is evidence that mothers have moderate levels of accuracy in recalling severe obstetric events (e.g., hyperemesis gravidarium) across this time period [
65], there is less evidence of the validity of recall of the broader spectrum of NVP symptoms. Previous studies have also reported that a child’s health status can bias maternal recall [
66], such that mothers of children with greater health or developmental difficulties may be less accurate in recalling obstetric events. While there is no evidence for this phenomenon in the ASD population, the possibility that mothers of children with more severe ASD symptoms inaccurately recalled more severe NVP symptoms than they actually experienced remains an important consideration in the interpretation of the current findings.