Main

Over the last decades, there is emerging evidence of early sexual dimorphism of the brain. Studies in animals have shown that fetal testes secrete testosterone independently of gonadotrophins and the male brain is exposed to high levels of these hormones (1,2). Aromatase, the enzyme converting testosterone to estradiol, plays a crucial role in sexually dimorphic prenatal brain development (3–5), and there are male-female differences in estrogen receptors during fetal development (6–8). Furthermore, genetic studies have reported that the genes associated with sex chromosomes directly affect brain sexual differentiation, suggesting that genetic influence on early brain development precedes or overlaps with the epigenetic action of gonadal hormones (9–12). In humans, MRI can be used to in vivo investigate early morphologic and structural differences between males and females in the developing brain.

Based on the literature, it is evident that hormonal effects and genetic mechanisms influence sex-related brain developmental phenomena. However, the pattern of early sexual differentiation of the human brain has not been explored and is not fully understood. It is not clear whether and how sex-related differences in early brain development are related to growth and maturational phenomena such as somatic growth, cerebral growth, and cortical folding.

This study aimed to investigate the pattern of cerebral growth and cortical folding in infants at a preterm stage of development, comparing the findings between males and females. To remove the influence of age, a significant determinant factor of early brain development (13,14), a case-controlled study was conducted, matching the infants in the two groups for age.

METHODS

Study design and participants.

Written informed consent was given by the parents and ethics approval was received from the University of Western Ontario Health Sciences Research Ethics Board. Preterm infants with weight appropriate for gestational age (AGA) below 1.5 kg were included in the study for research purposes (15). Exclusion criteria were congenital infections or malformations, metabolic diseases or hypoglycemia, CNS infections, brain malformations, seizures, hypoxic-ischemic injury, necrotizing enterocolitis, bowel perforation, and recurrent sepsis (defined as >2 episodes of positive blood cultures).

All infants enrolled in the study had serial cranial ultrasonography studies. Scans were performed by qualified radiology technicians using an ATL, HDI 3500 scanner (Bothell, USA) with a broadband transducer of 8-5 MHz. Each scan received clinical reading by neuroradiologist and neonatologist. Infants with cystic periventricular leukomalacia, parenchymal involvement of hemorrhage, posthemorrhagic hydrocephalus, and ventriculomegaly (defined as >4 mm above 97th centile) (16) were excluded before the MRI, on the basis of ultrasonography findings. Males were matched against females one-to-one, based on the postmenstrual age at MRI scan (±3 d).

MRI.

Infants had the MRI scan using a dedicated neonatal 3 T MRI system (IMRIS Winnipeg, Canada). Sound levels were measured to be less than 85 dB at the center of the coil. Neonates wore molded earplugs (EarClassic; Aearo, Indianapolis, IN) for further sound attenuation. After feeding, they were transferred to the scanner in a cot and a single dose of chloral hydrate (50 mg/kg) was administered orally for sedation, before scanning. Three-dimensional images of the cerebrum were acquired in 6.5 min with a center-out acquisition, magnetization-prepared rapid gradient echo sequence, with timing parameters optimized for neonatal brain imaging (17–19). Imaging parameters included intersegment repeat time 5200 ms, inversion time 2250 ms, repeat time 10 ms, echotime 5 ms, image bandwidth 33.3 kHz, flip angle 10 degrees, matrix size 120 × 120 × 75, and field of view = 160 × 160 × 100 mm giving an isotropic resolution of 1.3 mm.

The image analysis was performed on axial slices using Analyze 4.0 software (Biomedical Imaging Resources, Mayo Foundation, Rochester, MN) by a single blinded analyst. To obtain the cerebral tissue volume (CRB; gray and white matter), the skull, extracerebral cerebrospinal fluid, ventricular cerebrospinal fluid, and the cerebellum were extracted. This segmentation and extraction was based on signal intensity thresholds and was performed automatically. The brainstem was traced below the level of the anteroposterior commissure and was also extracted from the cerebrum. The CRB was calculated from the voxel volume and the number of voxels remaining after the extractions. The borderline between cortical gray matter and white matter was manually traced based on the signal intensity difference between the two tissue types (Image 1). Adding the measurements of all slices, the area between cortical gray matter and white matter was calculated and is been referred to here as the folding area (FA) of the cortex. The intraobserver variability for the FA was tested in 10 repeated measurements of the same scan.

Image 1
figure 4

Trace of FA.

Full size image

Data analyses and statistics.

Statistical analysis was performed with SPSS 15.0 for Windows (SPSS Inc, Chicago, IL). CRBs of males were compared with those of the females with an ANOVA. Potential confounding factors—covariates were tested for significance using regression analyses. They were tested first independently, then using a stepwise regression (20). A p value of 0.2 for entry criteria and of 0.1 for exit was used. The FA, defined here as the surface area of the interface between cortical gray matter and white matter, was used as the relevant index of cortical convolution. Because in any three-dimensional object surface is related to volume (for a sphere s = avb, s stands for surface area, v for volume, a is a constant, b is 2/3) (21), the raw and log-transformed data of FA were evaluated in relation to the raw and log-transformed data of CRB in males and females using linear regression. A test for homogeneity of slopes and intercepts was performed to determine whether the regression lines for males were significantly different from that of the females.

RESULTS

Ten male and 10 female infants were included in the study. The median gestational age and range at birth was 28.7 wk (24.6, 31.7) in males and 28 wk (24.6, 29.3) in females. The postnatal clinical course characteristics are presented in Table 1.

Table 1 Neonatal clinical data
Full size table

The CRB was larger (p = 0.012) in males than in females. The infants' weight at scan was a significant covariate for the CRB but the difference between males and females remained significant (p = 0.039) after adjusting for weight differences between males and females (Table 2 and Fig. 1). Other parameters such as gestational age at birth, weight at birth, days on ventilation, bronchopulmonary dysplasia, days to full feeds, germinal matrix hemorrhage, although not different between the groups, were also entered in the regression model and tested for possible covariance and no significance was demonstrated (all p values >0.3). The intraobserver variability (SD-SD) for the FA was <2.5%.

Table 2 Cerebral tissue volume (CRB) data displayed as mean ± SD
Full size table
Figure 1
figure 1

Scatterplot for CRB with weight at MRI scan (R2 = 0.39).

Full size image

The relationship of the cortical FA with the cerebral tissue for males and females is shown in Figure 2. The differences between the slopes were not significant (p = 0.95; F < 0.01; pooled slope = 2.66). However, the intercepts were significantly different (p = 0.019; F = 6.78). The relationship between the log-transformed data of the FA and the CRB is shown in Figure 3. The differences between the slopes were not significant (p = 0.87; F < 0.05; pooled slope = 1.35). However, the intercepts were significantly different (p = 0.018; F = 6.80).

Figure 2
figure 2

Scatterplot and best-fit line for CRB and FA. Males (□): FA = 2.62 × CRB − 209.1, R2 = 0.48; Females (○): FA = 2.68 × CRB −172.7, R2 = 0.95.

Full size image
Figure 3
figure 3

Scatterplot and best-fit line for logCRB and logFA. Males (□): FA = 1.40 × CRB − 0.71, R2 = 0.49; Females (○): FA = 1.33 × CRB − 0.49, R2 = 0.94.

Full size image

DISCUSSION

Previous studies in humans have also shown that the cerebrum is larger in males than in females. This difference is present in adults (22,23), in children (24,25), and in infants (26). In preterm neonates, at an earlier stage of life than this study, it has been observed that males have larger cerebrum than females, and females have a higher sulcation index than males only after adjusting for total brain volume/surface (27). This study investigated these brain developmental aspects taking into account the somatic growth differences in males and females, and also explored the direct relationship of cerebral volume with cortical FA measurements in males and females at near term age. It is important to note that the male and female infants in our study had similar antenatal and postnatal characteristics and no major brain injury. In addition, by matching the male and female subjects according to age, the influence of the very active growth and maturation that occurs at this stage was removed, creating a homogenous cohort. However, the possibility of subtle or diffuse white matter injury cannot be excluded. Brain development involves growth and maturation phenomena, which are strongly linked and intertwined. At a preterm stage, these phenomena are highly active and complicated. It is interesting but not surprising at this age that the weight, which is an indicator of the overall somatic growth, was found to account significantly for variation in cerebral growth. However, the difference in cerebral growth between males and females persisted after adjusting for the influence of this parameter, indicating that sexual dimorphism is an independent determinant of brain development in early human life.

During phylogeny and brain ontogeny, the convoluted external cortical surface area follows a power law that approximately lies between 8/9 and 4/3 (28,29). This study showed that the cortical FA—interface between cortical gray matter and white matter—increases in relation to CRB by a power of 4/3 (if log-transform the data: log s = b log v + log a, then the b value is the slope: 1.35 ≈ 4/3), a value that was not dissimilar between males and females. Strikingly, the regression line plotted significantly higher in females than males meaning that for similar CRBs the corresponding cortical FA was larger in females. These findings of females having smaller CRB than males but having disproportionately larger cortical FA, present a sexually dimorphic brain development with a relatively more “compact” female brain. The “compact” female model is further supported by the stronger correlation between CRB and cortical FA in females than in males.

The more “compact,” compared with males, early female brain development model is an interesting finding on brain macrostructure and refers potentially to underlying sexual differences in brain microstructure. The sexual differentiation of the developing brain in oligodendrocytes, astrocytes, and related processes or mechanisms is evident in the rodent brain and highly possible in humans (30–32). Considering the tension-based theory of brain morphogenesis, these microstructural phenomena and differences between males and females can lead to differences in tension along axons and tissue anisotropy which in turn may well result in different cortical FA and pattern in males and females, as found in this study (33,34). Furthermore, the more “compact” female brain development model can be associated with lower aggregate length of axonal and dendritic wiring and compactness of neural connectivity and circuitry. This may contribute to differences in domains of cognitive function and developmental performance early or later in life, underlining sex as an independent influencing factor (35,36).

The highly complex process of brain development is especially challenging to investigate during this dynamic phase of fetal or early life period. The very active growth of cerebral tissue, combined with essential multiple tissue organization phenomena, continues to raise fundamental questions. This study provides novel macroscopic in vivo evidence of a sexually dimorphic brain developmental model at an early stage of human life. Females have smaller CRBs independently to overall size, but disproportionately larger cortical FA than males following a more “compact” brain developmental model. Longitudinal studies in humans shall provide more insight into the pattern of sexual brain dimorphism and how this evolves.