Fetal brain growth and infant autistic traits

Approximately four weeks after conception, the formation of the cerebral hemispheres can be observed [1], with spontaneous movements of the head and trunk being seen from as early as 9 weeks’ gestation age (GA) [2]. By 37–42 weeks’ GA, there is evidence of specific cognitive functions, such as the ability to discriminate between sounds [3] and to anticipate directed movements [4]. Thus, during the 40 weeks of gestation it may be that the foundations for later cognition and behavior are being programmed, likely by an interaction between genetic and environmental factors.

Longitudinal research of the relationship between fetal growth and development in relation to later outcomes is still limited. Fetal biometrics are used to predict which newborn infants may be at risk for low birth weight due to intrauterine growth restriction (IUGR) or are small or large for gestation age (SGA, LGA). These biometrics are predictors of adverse fetal and infant outcomes. Numerous studies have explored the relationship between deviation in fetal growth (e.g., SGA, LGA) and birth weight, with mixed findings for later child outcomes (e.g., increased likelihood of developing sensory issues [5], autism [6, 7] and increased incidence of cognitive disabilities [8]). However, most research examining the relationship between the prenatal period and later outcomes has only measured pregnancy outcomes or overall growth (e.g., birthweight, IUGR and LGA), with few studies measuring the longitudinal relationship between fetal growth and development of outcomes in infancy.

Autism is strongly heritable [9] and genetic variants (both rare and common) affect brain development. The genetics of autism overlaps with genetic variance associated with sex differences in growth and anthropometric measures [10]. Early brain overgrowth is a prominent theory in autism aetiology research [11‐14] with numerous studies observing a relationship between early (0–36 months) head growth and later diagnosis of autism. However, some studies find no difference in brain growth in toddlers diagnosed as autistic. While Constantino et al. [11] reported a slight acceleration in head growth during the first 2 years of life, this difference is not big enough to be considered a reliable marker of increased autism likelihood. Many studies using HC as a proxy for brain growth have observed no difference in longitudinal head growth from birth to infancy [15, 16], concluding that enlarged HC size is not reliably associated with autism [12] in postnatal development. However, this debate remains unresolved. The largest single-site lifespan study showing early-age ASD brain overgrowth is Courchesne et al. [17] with N = 586 subjects; the largest multi-site study showing brain overgrowth in ASD is Bedford et al. [14] with N = 1327 subjects; and the largest multi-site meta-analysis showing lifespan ASD brain overgrowth is Sacco et al. [18] with N = 3085 MRI subjects and N = 5225 HC subjects.

The association between fetal growth, pregnancy outcomes and later autism likelihood is still unclear [9]. Studies exploring the relationship between prenatal brain growth and later infant outcomes have produced mixed findings. Growth velocity in HC (both pre- and postnatally) has also been studied in relation to later developmental outcomes including autism [19]. Abel et al. [20] and Bonnet-Brilhault et al. [21] both observed atypical prenatal head growth trajectories in children later diagnosed as autistic, finding overgrowth in prenatal HC during the second and third trimesters. Conversely, no significant difference in HC has been found between autistic children, or siblings with an increased genetic likelihood for autism, compared to non-autistic controls, in studies spanning the perinatal period [22‐24]. Unwin et al. [23] suggested that, rather than using HC as a proxy measure of brain growth, examining the growth of subregions of the fetal brain may reveal differences between fetuses who later receive an autism diagnosis, compared to those who do not. While HC has been commonly used by researchers as a proxy for brain growth, studies have yet to investigate subregions of the developing fetal brain in relation to a later outcome of autism.

The cerebellum is one of the earliest structures to develop, emerging from the roof of the rhombencephalon between 4 and 6 weeks post-conception [25], which places it at increased vulnerability to a range of developmental effects during prenatal development [25, 26]. Transcerebellar diameter (TCD) in fetuses measuring below the 5th centile is associated with anomalies such as a high rate of fetal malformations, chromosomal anomalies, severe IUGR and genetic disorders [27]. Reduced fetal cerebellar diameter has been associated with atypical movement in infants at 1 and 3 months old [28] and atypical motor and cognitive functions [29]. Similarly, studies have shown that prenatal isolated ventriculomegaly defined as enlargement of the brain ventricles due to build up of cerebrospinal fluid, is associated with later developmental delay (e.g., in fine motor and expressive language skills [30]) and psychiatric diagnoses (autism, ADHD and schizophrenia) [31‐34].

To date only one other study has examined sub-regional brain size using prenatal ultrasound [19], finding a higher rate of fetal ultrasound anomalies in HC but not in the cerebellum or ventricular atrium (VA) size in infants later diagnosed as autistic. Here we take this one step further, testing longitudinal growth of head and brain regions measured by 2D ultrasound sonography during pregnancy. The regions of interest were: the cerebrum (using HC as a proxy) [35], the cerebellum [36] and the ventricular atrium [37, 38] of fetuses. Subsequently, infants were followed up to test if there is an association between these regions of the developing fetal brain and the emergence of autistic traits (Q-CHAT scores) at 18–20 months of age.

This is the first study to examine sub-regional brain growth across several time-points in utero and test for associations with early neurodevelopmental outcomes in infants. We found that TCD in the 2nd trimester is significantly associated with autistic traits measured using the Q-CHAT at 18 to 20 months of age (Fig. 3). Longitudinal ultrasound measures of fetal brain parameters showed varying rates of growth between trimesters and correlations between them (Additional file 1: Fig. S1). HC and TCD were significantly correlated, particularly in the 3rd trimester. HC increased at a stable rate from the 1st to the 3rd trimester, which was mirrored by increases in TCD between the 2nd and 3rd trimester (Additional file 1: Fig. S2). Sex differences were largely confined to the late 2nd trimester, with males measuring significantly larger at both. This could be attributed to potential trophic effects from the sex steroid surge which occurs in male fetuses and peaks around the 16th week of gestation [41]. A reduction of sex differences in the 3rd trimester suggests a potential “catching-up” for female fetuses before the end of pregnancy, suggesting that studies of postnatal brain parameters may overlook prenatal sex differences.

Observations of fetal ventricular width during gestation have shown size increases until approximately 20 weeks GA and then remains stable, or steadily declines afterwards [42]. Therefore, we would expect to see either stability or decrease in VA size between the 2nd and 3rd trimester scans. As expected, we found no significant changes in VA size between the 2nd and 3rd trimester. This is in line with findings from Regev et al. [19] who also found no association between prenatal VA at 20–24 weeks gestation and a diagnosis of autism. However, unlike that study [19] we found TCD was correlated with Q-CHAT scores in the second trimester, after controlling for multiple covariates. We also found a significant effect of HC in the third trimester but not in the second (Table 2). Between the second and third trimester the cerebellum has been observed to have the greatest relative maturation rate (12.8% per week), when compared to other regions (e.g., brain stem) and total brain growth [43], which could explain why our finding of an association with TCD predates the association with HC. The observed pattern may also indicate a faster rate of brain growth in pregnancies of children with high autistic traits, which is first evident in the cerebellum but then may plateau, compared to the other pregnancies that ‘catch-up’ by 28 weeks. Longitudinal modeling that utilizes multiple time-points, as well as standardized, sex-specific and population-specific growth curves, may be better suited to capture this phenomenon.

The cerebellum has an important role in neurodevelopment, as it regulates the formation of networks in the developing cortex [44]. It has been suggested that early cerebellar dysfunction may result in autism, by preventing the integration of sensory stimuli and the maturation of the social association network in later life [45]. This is consistent with analyses showing that genes associated with autism are more active during prenatal cerebellar development, when these cortical projections are established [46]. Fetal MRI could potentially reveal more microscopic increases in neuronal density [47] at this developmental time-point and whether these are associated with later autistic traits. The difference in findings between the above-mentioned studies could be attributed to a potential ‘sensitive’ period of influence during the prenatal period where observable changes in fetal brain structure are apparent.

Over and above genetics, specific factors in the prenatal environment could be driving both TCD growth as well as an increase in autistic traits. For example, estradiol increases the density of neuronal fibres and spines in the developing cerebellum [48], as well as the likelihood of an autism diagnosis [49]. In addition, given the cerebellum’s rapid growth in late pregnancy, increases in TCD could be an adaptive response to prenatal adverse conditions which can also affect neurodevelopment [50, 51].

The brain undergoes rapid growth during the first nine months of development, unsurprisingly this time of rapid growth is a time of shaping and potential but also increased vulnerability. Existing research has separated potential ‘sensitive’ periods to alterations to development. For example first trimester has been implicated in greater risks of major birth defects [52, 53], as it is during this period major structures of the body is formed (e.g., spine, limbs and heart). It is during the second and third trimesters to that research has observed more growth problems, minor birth defects and influences on later developmental outcomes [54‐56] (e.g., preterm delivery, small-for-gestational age and functional deficits such as cognitive delay). Our study suggests a need to study variability in prenatal growth across pregnancy to gain a greater understanding of the time window of differences, sensitive periods and its longitudinal effect on later neurological and behavioral development.

This cohort of pregnant participants was relatively homogeneous in terms of clinical history and pregnancy parameters. Maternal factors such as polycystic ovary syndrome (PCOS), high BMI and age in the mother were not significantly associated with Q-CHAT score in the child. This does not necessarily contradict previously reported associations with autism [57, 58], as the current study did not include clinically diagnosed cases and was a limited, community sample of 179 mother–child pairs with a Q-CHAT score. The Q-CHAT is not specific to or diagnostic of autism, but rather captures a wider spectrum of early neurodevelopmental traits. Similarly, there was no observed sex difference in autistic traits at 18–20 months of age, which could be attributed to a smaller sample size compared with previous studies or that effect sizes related to sex differences have been small [59]. Due to the nature and location of the study at the Autism Research Centre, ascertainment bias could also have contributed to the recruitment of volunteer mothers with a higher than typical range of autistic traits and/or with a significant interest in the neurodevelopment of their children.

In conclusion, this study demonstrated that prenatal brain growth is associated with neurodevelopmental outcomes in infants. Specifically, we found that a larger TCD as early as week 20 is associated with autistic traits in both males and females at 18–20 months. This finding further supports the potential inclusion of prenatal TCD as an early marker of later neurodevelopmental outcomes. The observed positive relationship between prenatal TCD size at 20 weeks’ gestation and Q-CHAT score at 18–20 months could be introduced into early postnatal screening as a potential early marker for autism. In the future this could aid with early diagnosis and therefore could improve access to interventions and/or support early in development for those that want these services. However further longitudinal exploration is needed to validate this potential early marker, as well as its specificity to autistic traits. Finally, this finding may improve our understanding of how autism and related traits are associated with prenatal brain development.