Accelerated head growth during the first years of life is one of the most replicated findings in the biological study of autism spectrum disorder (ASD) (Aylward et al. 2002; Redcay and Courchesne 2005; Sparks et al. 2002). Meta-analyses have identified an early and rapid period of head size growth in ASD, so that brain volume exceeds the population average in an estimated 90% of toddlers subsequently diagnosed with ASD (Courchesne et al. 2001). This accelerated growth ceases by 2–4 years of age (Redcay and Courchesne 2005), with macrocephaly present in around 20% of children with ASD (Fombonne 1999; Lainhart et al. 2006; Miles et al. 2000).

The impact on childhood ASD of fetal brain growth is less well understood. Preliminary evidence suggests that atypical brain development in ASD may commence prenatally. First, while the majority of studies have reported normal head circumference at birth among neonates later diagnosed with ASD (Redcay and Courchesne 2005), the pooled prevalence of macrocephaly at birth among children in these studies is twice that of the typical population (Hobbs et al. 2007). Second, circulated neurotrophins, which are known to be involved in fetal cerebral growth, are elevated in neonates later diagnosed with ASD (Nelson et al. 2001). Third, there are reports that children and adults with ASD have an increase in the number of cortical minicolumns (Casanova et al. 2006) and a decrease in cerebellar Purkinje cells (Bauman and Kemper 1994), which are indicative of abnormalities in brain development during the first trimester of pregnancy.

ASD cannot be reliably diagnosed until after birth. The only published data retrospectively examined fetal head circumference in the second trimester of pregnancy using ultrasonography (M gestational age = 19.83 weeks; SD = 1.94 weeks) in 45 children with ASD compared with 222 controls and found no differences in head circumference between these groups (Hobbs et al. 2007). However, the authors acknowledge that their study was considerably limited by lack of matching between cases and controls for key variables such as gender, race, and exposure to teratogens during pregnancy, all of which can influence fetal growth. Furthermore, scans were performed at different clinical sites using range of ultrasound machines, increasing the chances of inter-observer measurement error. Finally, no postnatal developmental data were available for the control sample, raising the possibility that at least some of these children may have had a condition that affects growth.

In the current study, we examined the prospective relationship between fetal brain growth (as measured by head circumference) and ASD in childhood in a large unselected birth cohort using a case–control design.

Methods

Participants were part of the Western Australian Pregnancy Cohort (Raine) Study, which is a longitudinal investigation of women recruited prior to 18 weeks gestation from the public antenatal clinic at King Edward Memorial Hospital (KEMH) or surrounding private clinics, between May 1989 and November 1991 (Newnham et al. 1993). Approximately 100 unselected antenatal patients per month were enrolled during this period from August 1989 to April 1992, with a final sample of 2,979 women, who delivered 2,868 live births.

Fetal and Infant Biometry

All women enrolled in the Raine study received a fetal ultrasound imaging study at or close to 18 weeks gestation. Gestational age was calculated from the date of the last menstrual period and confirmed via ultrasound biometrics at 18 weeks (Hadlock et al. 1982, 1984).

Ultrasound examinations were completed by a qualified sonographer using one of two General Electric 3600 machines (Milwaukee, USA) with 3.5 MHz linear array and 5 MHz sector transducers. Fetal brain size was estimated by measuring the maximal biparietal head circumference, while overall fetal size was indexed by femur length (both measurements to the nearest mm). A specially trained research midwife measured maximal occiptofrontal head circumference (to the nearest mm) and crown-heel length (to the nearest 5 mm) shortly after birth (within 3 days) with a precise paper tape measure using standard techniques (Cameron 1984).

Diagnosis of ASD

At the 5-, 8-, 10-, 13- and 16-year follow-ups of the Raine cohort, parents were asked whether their child had ever received a diagnosis of ASD by a health professional. Diagnosis of these conditions in Western Australia mandates consensus by a team comprising a Pediatrician, Psychologist and Speech-Language Pathologist under DSM-IV guidelines (American Psychiatric Association 1994). Parent report indicated that 16 children in the Raine cohort had received a diagnosis of ASD, and ultrasound data are available for 14 of these cases: 10 children with a diagnosis of Autism (male = 8, female = 2), one with Pervasive Developmental Disorder—Not Otherwise Specified (male) and three with Asperger’s Syndrome (all male). Each ASD case was from a singleton pregnancy.

In each case, the diagnosis of ASD was strengthened by prospectively collected phenotypic data from the Infant Monitoring Questionnaire (Bricker and Squires 1989) and Child Behavior Checklist (Achenbach et al. 1987). Age 2 years was the first follow-up at which a near complete set of data were obtained from the current participants. A greater proportion of children in the ASD group did not consistently display appropriate eye contact (ASD: 4/13 children; Control: 8/51) or engage in pretend play (ASD: 8/13; Control: 14/51) at age 2 years, and produce noun–verb phrases (ASD: 4/13 children; Control: 7/56) or successfully carry out two-step instructions (5/13 children with ASD; Control: 1/55) at age 3 years. Fisher’s exact test found these group differences to be significant at the level of p < 0.05. Further analyses found that significantly (p < 0.05) more ASD participants than control participants were reported to display at least one of these four behaviors inconsistently (ASD: 12/14, Typical: 27/56) or never (ASD: 7/14, Typical: 5/56). Individual data are included in “Appendix A”.

Case Matching

Each ASD case was matched with four control children from the Raine cohort according to maternal age, gender, gestational age, race (all Caucasian), maternal age at conception (within 2 years), and maternal smoking (no smoking; 1–10 cigarettes per day; or 11+ cigarettes per day) and alcohol consumption (no alcohol consumed; drink alcohol once/per week or less; drink alcohol several times per week or daily) during pregnancy (see “Appendix B”). Each control child was from a singleton pregnancy and had not been diagnosed with a developmental or learning disorder at their last contact with Raine investigators (age 13: n = 17; age 16: n = 39). Controls were not limited to only those children who participated in all of the follow-up visits. “Appendix B” also provides all available language and IQ data on these participants at the 2-, 5-, 8- and 10-year follow-up.

Biostatistical Methods

Independent samples t-tests found no significant group difference in the gestational age at which the second trimester (p = 0.56) and birth (p = 0.8) measurements were obtained. However, because there was minor variability between subjects in gestational age at both time points, the raw measurements were converted to standardized z-scores using conventional methods: (participant’s measurement—reference data mean at participant’s gestational age)/reference data standard deviation at participant’s gestational age. Z-scores for fetal head circumference (zFHC) and femur length (zFFL) were calculated using the most widely used reference data for fetal growth (Hadlock et al. 1982, 1984). Although small gender differences in body size emerge during the final trimester of pregnancy, there are currently no Australian norms that provide head circumference reference data separately for males and females. Accordingly, raw measurements of birth head circumference were converted to z-scores (zBHC) using gender-stratified reference data from the public antenatal clinic at the John Radcliffe Hospital, Oxford (UK) (Yudkin 1987), which, like KEMH, predominantly served a Caucasian population at the time these norms were generated. Z-scores for birth length (zBL) were calculated using the gender-stratified Australian norms of Beeby et al. (1996). Consistent with literature on fetal and infant growth (Fombonne 1999), the ‘normal’ range for each measurement was considered to be a z-score between −2 and +2, and macrocephaly was defined as a zFHC or zBHC equal to or greater than +2.

Results

Z-scores for fetal and birth measurements are presented in Table 1. Data on head circumference and length at birth were missing for one ASD participant. One participant with ASD had macrocephaly at the second trimester ultrasound (ASD11: zFHC = 2.6), but had a relatively small head circumference when measured again at birth (zBHC = −1.5). All control participants had a head circumference within the normal range at the second trimester ultrasound, while macrocephaly was observed for two control participants at birth (zBHC = 2.6 and 2.98). The femur length of all ASD and control participants at the second trimester ultrasound were within the normal range, while one control participants had a zBL of +2. Independent samples t-tests found no difference between cases and controls in zFHC, t(15.47) = 1.46, p = 0.164, and zFFL, t(68) = 1.31, p = 0.153, at the second trimester, and zBHC, t(67) = 0.67, p = 0.508, and zBL, t(67) = 0.09, p = 0.925, at birth.

Table 1 Details of gestational age (GA), as well as measurements of head circumference and body size (femur length and birthweight) at the second trimester ultrasound and at birth
Full size table

To examine head circumference relative to body size, a ‘difference score’ was calculated by subtracting zFFL from zFHC at the second trimester (Fig. 1) and zBL from zBHC at birth (Fig. 2). A positive score indicated greater standardized head circumference relative to body size (standardized for gestational age), while a negative score signified the reverse. Rank ordering of these indices revealed that participants with ASD had the five highest difference scores at the second trimester and the two highest at birth. However, independent t-tests found no significant group difference in these indices at the second trimester ultrasound, t(15.06) = 1.25, p = 0.232, or at birth, t(67) = 0.53, p = 0.599.

Fig. 1
figure 1

Distribution of ‘difference scores’ at the second trimester ultrasound, calculated by subtracting an individual’s femur length z-score from their head circumference z-score

Full size image
Fig. 2
figure 2

Distribution of ‘difference scores’ at birth, calculated by subtracting an individual’s birth length z-score from their head circumference z-score

Full size image

A final analysis examined final trimester growth in head circumference and body size by subtracting zFHC from zBHC and zFFL from zBL. A positive index would indicate accelerated growth during the final trimester, while a negative score would suggest a slower rate of growth. An independent samples t-test found no between groups difference in the growth index for head circumference (ASD: M = −0.04, SD = 1.44; M = −0.11, SD = 1.14), t(67) = 0.39, p = 0.699, or body size (ASD: M = −0.42, SD = 0.86; M = −0.37, SD = 0.87), t(67) = 0.18, p = 0.85.

Discussion

This is the first study to examine fetal head size in children later diagnosed with ASD in comparison to tightly-matched typically developing controls. While one participant with ASD had macrocephaly at the second trimester ultrasound, all children in this group had a head circumference within the normal range at birth, and there was no significant difference between-groups in the mean head circumference or body size at either time-point. Examination of ‘difference scores’, which quantified head circumference relative to body size, revealed that a small number of children with ASD had a disproportionately large head size at the second trimester (n = 5) and birth (n = 2) assessments. However, there was no statistically significant group difference on this index at either time-point in this small sample. A final analysis found no systematic difference between-groups in head circumference or overall body growth during the final trimester.

The main strengths of the current study are the prospective design, the consistent methods of measuring fetal growth, and the close matching of ASD cases with a large number of control participants on a range of parameters known to influence fetal development. However, we also acknowledge limitations. A clinical diagnosis of ASD had been provided in each case by a pediatrician, psychologist and speech pathologist. While we were able to demonstrate increased levels of developmental difficulties in the ASD group (see Appendices), the historical nature of the current study prevented us from confirming the expert diagnoses via ‘gold-standard’ observational methods, such as the Autism Diagnostic Observation Scale—Generic (Lord et al. 2000). We also acknowledge the relatively small size of the ASD sample. The clinical participants were drawn from a large, unselected birth cohort who had received an ultrasound scan in the second trimester of gestation; the proportion of individuals with ASD among the Raine cohort (16/2868, 0.6%) was similar to prevalence estimates of ASD in the general population (~1%) (Baird et al. 2006). Future prospective studies that monitor fetuses at increased genetic risk for ASD, such as pregnancies to parents with a family history of ASD, may generate a larger clinical sample. Finally, while ultrasonography is the most-widely used and reliable technique for providing a measure of overall fetal brain growth, this methodology does not enable detailed analysis of cortical development. Methodological advances in fetal magnetic resonance imaging (Lan et al. 2000), as well as the development of new imaging techniques, such as Diffusion Tensor Imaging and tractography (Kasprian et al. 2008), which provide greater spatial resolution than ultrasonography, may provide further insights into the prenatal neurodevelopment of children with ASD.

In this study, we found no statistical differences between a small cohort of children with ASD and a tightly-matched control group in fetal head circumference size at approximately 18 weeks gestation or at birth. However, qualitative inspection (Figs. 1, 2) found that a small number of children with ASD had disproportionately large head circumference relative to body size at both time points. The lack of a group-difference on these indices may be due to low statistical power when examining the small but phenotypically heterogeneous ASD group. Statistical differences between ASD and control groups may emerge when samples are pooled across studies, as has been the case with investigations of birth head circumference in ASD (Hobbs et al. 2007). The findings of this study suggest that prenatal growth in ASD warrants further investigation in other birth cohorts as well as children at increased genetic risk for ASD, in order to determine whether enlarged fetal head circumference is an early risk factor for ASD.