Associations of gestational age and birth anthropometric indicators with brain white matter maturation in full‐term neonates

Abstract

Newborn assessments, including gestational age (GA) and anthropometric measurements (birth weight, crown‐heel length, head circumference) are routinely performed in pediatric settings, being used as important indicators in assessing neonatal development. Close associations of these birth indicators with later cognitive abilities were also reported. However, specific associations of these indicators with white matter (WM) development during the neonatal period remain unclear, as well as the extent to which they influence WM maturation. To address this issue, 51 full‐term neonates (GA range, 37–42 weeks) with no abnormalities on MRI were retrospectively recruited. Specific correlations between birth indicators and WM maturation, quantified by diffusion tensor imaging (DTI)‐metrics (fractional anisotropy, mean diffusivity, axial diffusivity, radial diffusivity), were identified by using DTI tract‐based spatial statistics and automated fiber‐tract quantification. Our findings suggest that (a) higher GA, birth weight, and crown‐heel length may indicate greater WM maturation in full‐term neonates, while head circumference presented weak correlation with WM maturation during early newborn period; (b) among the four indicators examined, GA was the one most associated with WM maturation. We believe that this study advances our knowledge of specific correlations between birth indicators and neonatal brain development and provides a valuable reference for future neonatal studies.

Abbreviations

AD

axial diffusivity

AR

auditory radiation

CST

corticospinal tract

DTI

diffusion tensor imaging

FA

fractional anisotropy

GA

gestational age

GLM

general linear model

MD

mean diffusivity

OR

optic radiation

RD

radial diffusivity

TBSS

tract‐based spatial statistics

thal‐PSC

thalamus‐primary somatosensory cortex

WM

white matter

1. INTRODUCTION

Newborn assessments, including gestational age (GA) and anthropometric measurements (e.g., birth weight, crown‐heel length, and head circumference) are routinely performed in pediatric settings, being used as important indicators in assessing neonatal development. Among these, GA at birth has been demonstrated to be closely linked to neonatal brain maturation (Dubois et al., 2014). Newborns with low GA, especially those born preterm (GA < 37 weeks) may have abnormalities in brain white matter (WM) maturation (Berman et al., 2005; Kelly et al., 2015; Larroque et al., 2003). Such abnormalities in WM development would continue throughout infancy (Deoni et al., 2011) and persist into adulthood (Nosarti et al., 2014). In addition, close associations between longer gestation and greater brain WM maturation have also been found (Broekman et al., 2014; Ou et al., 2017). Besides GA, anthropometric indicators at birth reflecting the fetal growth, provide important information regarding the newborn development (Fok, 2003). Birth weight was shown to be closely linked to brain maturation (Walhovd et al., 2012) and neurodevelopmental outcome (e.g., cognitive gains) in childhood and adulthood (Matte, Bresnahan, Begg, & Susser, 2001; Wade, Browne, Madigan, Plamondon, & Jenkins, 2014). In comparison with infants born with low birth weight (<2,500 g), higher birth weight is indicative of more advanced brain maturation (Eikenes, Løhaugen, Brubakk, Skranes, & Håberg, 2011; Skranes et al., 2007; Taylor et al., 2011). Moreover, similar associations between crown‐heel length or head circumference and intelligence quotient (IQ) have also been reported (Gale, O'Callaghan, Bredow, & Martyn, 2006; Lira et al., 2010; Ranke, Krägeloh‐Mann, & Vollmer, 2014). These findings support the notion that GA and clinical anthropometric indicators at birth indeed show certain associations with brain WM maturation in the early and/or later periods.

Although these studies revealed the associations between birth indicators and WM maturation, most studies have focused on abnormal populations, for example, infants born preterm or with low birth weight (Kelly et al., 2015; Skranes et al., 2007). It is worth noting that although positive correlations between GA and WM maturation have been detailed in healthy full‐term neonates (Broekman et al., 2014; Ou et al., 2017), some inconsistency still existed. Broekman et al. (2014) observed GA‐related WM maturation in some regions, such as corpus callosum, right anterior corona radiata, anterior limb of the internal capsule, and external capsule; while Ou et al. (2017) observed the associations in more extensive WM regions. In this regard, there remains to be insufficient quantitative characterization of brain WM maturation in typical developing full‐term neonates. In particular, the specific associations of WM maturation with GA and anthropometric indicators at birth remain unclear, as well as the extent to which they relate to WM maturation. Diffusion tensor imaging (DTI) is a typical neuroimaging technique that allows quantitative characterization of water diffusion patterns for in‐vivo investigation of brain microstructure development (Dubois, Hertz‐Pannier, Dehaene‐Lambertz, Cointepas, & Le Bihan, 2006; Qiu, Mori, & Miller, 2015). Particularly, DTI‐derived metrics (i.e., fractional anisotropy, FA; axial diffusivity, AD; mean diffusivity, MD; and radial diffusivity, RD) are reportedly good markers for quantifying WM microstructural changes (Dubois et al., 2007). Using DTI‐metrics, we therefore aimed to investigate the associations between birth indicators and WM maturation in full‐term neonates; and their weight in relation to WM maturation was further assessed.

2. MATERIALS AND METHODS

This study was approved by the internal review board of the First Affiliated Hospital of Xi'an Jiaotong University. Written informed consent was obtained from the parents of all participants.

2.1. Participants

From January 2011 to June 2013, 211 full‐term neonates born in Department of Neonatology of the First Affiliated Hospital of Xi'an Jiaotong University who underwent magnetic resonance (MR) imaging examinations were retrospectively recruited. Of all recruited subjects, 106 fulfilled the following inclusion criteria: (a) had term gestation at birth (GA range, 37–42 weeks), (b) underwent completed MR imaging examinations, (c) had a 10‐min APGAR of ≥7; (d) were born with size (birth weight, crown‐heel length, and head circumference) appropriate for GA (Villar et al., 2014).

Neonates with abnormalities on MR images and/or neonatal diseases, such as cerebral infection, congenital malformation, metabolic disorders, punctate WM lesions, T2 hyperintensity, periventricular leukomalacia, cortical infarction, hydrocephalus, or intracranial hemorrhage were excluded (as for infarction, hemorrhage, or vascular malformation, diffusion weighted imaging and/or susceptibility weighted imaging were implemented to identify these diseases if necessary). Neonates with MR image artifacts were also excluded.

2.2. MRI data acquisition

All MR imaging data were acquired using a 3.0‐T scanner (Signa HDxt, GE Healthcare, Milwaukee, WI) with an eight‐channel head coil. Neonates underwent the brain MR imaging during natural sleep (i.e., adopting sleep deprivation and/or feeding protocols), while for neonates who could not remain still, sedation by oral administration of 10% chloral hydrate (dose: 25–50 mg/kg) was used to reduce head motion during the examination. Given the potential risks of chloral hydrate, patient selection, monitoring, and management were strictly performed following the established guidelines (Cote & Wilson, 2016); adverse drug reactions within 24 hr following sedation were also monitored. Earplugs and sponge mats were used for hearing protection. Heart rate, transcutaneous oxygen saturation, and respiration rate were synchronously recorded. MR imaging protocol included three‐dimensional T1‐weighted imaging, T2‐weighted imaging, and DTI (Table 1).

Table 1.

Magnetic resonance imaging protocols and scanning parametersa

Scanning parameters	3D‐T1WI	T2WI	DTI
Repetition time (ms)	10	4,200	5,500
Echo time (ms)	4.6	120	95
Field of view (cm)	24	18	18
Matrix acquisition	240 × 240	256 × 256	128 × 128
Slice thickness (mm)	1	4	4
Imaging resolution (mm3)	1 × 1 × 1	0.7 × 0.7 × 4	1.4 × 1.4 × 4
Number of excitations	1	1.5	1
Direction number	‐	‐	35
b value (s/mm2)	‐	‐	1,000

Abbreviations: DTI, diffusion tensor imaging; T2WI, transverse fast spin‐echo T2‐weighted imaging.

^a3D‐T1WI: three‐dimensional fast spoiled gradient echo T1‐weighted imaging.

The MR images were independently reviewed by two experienced radiologists with 5 years of experience in pediatric MR images. Disagreements regarding image findings were resolved by discussion and mutual agreement.

2.3. Measurements of birth indicators

GA was calculated by ultrasound findings before 20 weeks gestation or maternal last menstrual data. Birth weight and crown‐heel length were measured by an infant physical examination instrument (WS‐RTG‐1G, Wuhan Computer Software Development Co., Ltd., China). Using an inelastic tape, head circumference was measured as the maximal circumference around the head at the most anterior protuberance above the eyebrows and the most posterior one of occipital bone (see more in Supporting Information).

2.4. DTI processing and statistical analysis

2.4.1. DTI processing and tract‐based spatial statistics (TBSS)

DTI data were processed using the FMRIB software library (http://www.fmrib.ox.ac.uk/fsl). The Brain Extraction Tool was used to extract the brain and FMRIB's Diffusion Toolbox was used to correct the eddy currents and head motion‐induced distortions, to estimate the diffusion tensor, and to obtain scalar DTI maps. Tract‐based spatial statistics (TBSS; Smith et al., 2007) was performed using a pipeline optimized for neonates (Li, Gao, Wang, Wan, & Yang, 2016; see more in Supporting Information). All scalar DTI maps were normalized to Johns Hopkins neonatal template (Oishi et al., 2011). The aligned scalar DTI map of each subject was projected onto the mean FA skeleton (threshold = 0.15).

GLM was used to assess the relationships between DTI‐metrics and birth indicators (GA, birth weight, crown‐heel length, and head circumference). Relationships between DTI‐metrics and background characteristics (maternal age, education level, and family socioeconomic status) were first evaluated to determine whether these characteristics should be considered as covariates in addition to the set covariates (postnatal age and sex). Anthropometric indicators were adjusted for GA because of their confirmed strong relationships (p < .05). The number of permutations was set to 5,000. The results of all tests were considered significant when p < .05 after family‐wise error rate correction with threshold‐free cluster enhancement.

2.4.2. Pearson correlations between birth indicators and DTI‐metrics

3D automated fiber‐tract quantification (AFQ; Yeatman, Dougherty, Myall, Wandell, & Feldman, 2012) was used to detail the Pearson correlations between birth indicators and DTI‐metrics. Given the developmental sequence of neonatal brain WM, four visual, auditory, and sensorimotor WM tracts, that is, corticospinal tract (CST; from pons to the motor cortex), optic radiation (OR; from caudal parts of the thalamus to the occipital lobe), auditory radiation (AR; from thalamus to the transverse temporal gyrus of the cerebral cortex), and thalamus‐primary somatosensory cortex tract (thal‐PSC; from thalamus to the primary somatosensory cortex), were selected as tracts of interest. By using 3D AFQ and Johns Hopkins probabilistic maps of fiber tracts (Akazawa et al., 2016), DTI‐metrics along the bilateral OR, AR, CST, and thal‐PSC were extracted. The selected WM tracts were divided into 100 segments (sections 1–100), and mean DTI‐metrics at the plane perpendicular to each section were calculated. Pearson partial correlations of each birth indicator with mean DTI‐metrics were performed, with other three indicators, postnatal age, and sex controlled.

2.4.3. Multiple linear regression of birth indicators related to DTI‐metrics at selected WM ROIs

To further explore the weights of birth indicators in relation to DTI‐metrics, multiple linear regression analysis was performed. By Johns Hopkins probabilistic maps of fiber tracts (Akazawa et al., 2016), CST, OR, AR, and thal‐PSC were selected as ROIs. Mean FA, AD, MD, and RD were calculated in all ROIs for all individuals (see more in Supporting Information). Multiple linear regression was used to investigate the associations between DTI‐derived metrics and birth indicators (GA, birth weight, birth crown‐heel length, and birth head circumference), with postnatal age and sex controlled. Besides, multicollinearity in multiple regression analysis has been tested (see more in Supporting Information).

All regression analyses were performed using SPSS software (SPSS version 17.0; SPSS, Inc.). The correlation analyses were performed using MATLAB software (MATLAB version R2012b; The Mathworks, Inc.). p < .05 was considered statistically significant.

3. RESULTS

3.1. Participant demographics

Of all 211 neonates, 51 full‐term neonates with postnatal age of 1–14 days were included (Figure 1 and Table 2).

Figure 1.

Flow chart for determining the eligible subjects based on the inclusion and exclusion criteria. Abbreviations: GA, gestational age; HIE, hypoxic ischemic encephalopathy; PWML, punctate white matter lesion

Table 2.

Participant demographics

	Term neonates (n = 51)
Neonatal clinical characteristics
Boys/girls	32/19
Gestational age (weeks)a	39.57 ± 1.22
Postnatal age at MRI (days)a	6.57 ± 2.47
Birth weight (g)a	3,260.80 ± 479.26
Adjusted birth weight (g)a	3,250.60 ± 457.79
Crown‐heel length (cm)a	50.10 ± 2.18
Adjusted crown‐heel length (cm)a	50.20 ± 2.12
Head circumference (cm)a	34.00 ± 1.51
Adjusted head circumference (cm)a	34.00 ± 1.43
5‐min APGARa	8.92 ± 1.30
10‐min APGARa	9.47 ± 0.77
Background characteristics
Maternal age (years)a	28.63 ± 3.80
Mother with no high school educationb	32 (62.75)
Prenatal smoking exposureb	0 (0)
Minority ethnicityb	0 (0)
Family socioeconomic statusc
Professional/managerialb	18 (35.29)
Technical/skilledb	7 (13.73)
Semiskilled/unskilled/unemployedb	26 (50.98)
Clinical diagnosis
Mild asphyxiab , d	29 (56.86)
Metabolic acidosisb , e	10 (19.61)
Physiologic jaundiceb	8 (15.69)
Pneumoniab	15 (29.41)
Scalp hematomab	8 (15.69)

^aData represented as mean ± SD.

^bData represented as number (percentage, %).

^cAssessed by the Elley–Irving socioeconomic index (Elley & Irving, 2003).

^dAssessed by ICD‐10 categories of the diagnosis “perinatal asphyxia” defined by clinical signs and 1‐min APGAR score (World Health Organisation, 1999); 1‐min APGAR score of 4–7 was categorized as mild asphyxia; the mean 1‐min APGAR score of 29 newborns was 5.68 ± 1.09.

^eAssessed by blood‐gas analysis (PH < 7.0 and base excess < −12.0 mmol/L; Graham, Holcroft, Rai, Donohue, & Allen, 2004); mean PH and base excess of 10 newborns were 7.30 ± 0.09 and −11.04 ± 3.96, respectively.

3.2. Relationship between birth indicators and brain WM DTI‐metrics

The background characteristics (i.e., maternal age, education level, and family socioeconomic status) showed no association with DTI‐metrics.

GA positively correlated with FA and negatively correlated with MD, AD, and RD in most WM structures (p < .05, corrected; Figure 2a).

Figure 2.

Relationships between diffusion metrics (i.e., FA, AD, MD, and RD) and (a) gestational age, (b) birth weight, (c) birth crown‐heel length, (d) birth head circumference. Green labeling indicates the fibrous skeleton of brain white matter; the warm‐toned (scale on the top) and cool‐toned labels (scale on the bottom) indicate the positive and negative correlations, respectively. AD, axial diffusivity; FA, fractional anisotropy; MD, mean diffusivity; RD, radial diffusivity [Color figure can be viewed at http://wileyonlinelibrary.com]

Birth weight showed positive correlations with FA in regional WM, including corpus callosum, left external capsule, left posterior thalamic radiation, and left corona radiata (p < .05, corrected). Birth weight was negatively correlated with MD, AD, and RD in regional WM such as the bilateral corona radiata, external capsule, and posterior limb of the internal capsule (p < .05, corrected). Besides, the correlation distributions of MD, AD, and RD in most WM structures were bilateral, whereas that of FA was asymmetrical, particularly in the left corona radiata and posterior thalamic radiation (Figure 2b).

Crown‐heel length positively correlated with FA in regional WMs, for example, genu of the corpus callosum, external capsule, left corona radiata, left splenium of the corpus callosum, and left posterior thalamic radiation (p < .05, corrected); negative correlations with MD and RD were observed in regional WMs, for example, genu of the corpus callosum, external capsule, corona radiata, and posterior thalamic radiation (p < .05, corrected); negative correlations with AD were only found in the posterior thalamic radiation (Figure 2c). Furthermore, correlation distribution of FA in most WM structures was bilateral, while hemispheric asymmetry (i.e., in the left hemisphere) was also observed in the corona radiata and posterior thalamic radiation.

As head circumference increased, MD, AD, and RD significantly increased in regional WMs, for example, the corona and posterior thalamic radiations (p < .05, corrected; Figure 2d).

Postnatal age and sex showed no correlations with DTI‐metrics.

3.3. Pearson correlations between birth indicators and DTI‐metrics along the selected WM tracts

In the left CST, GA showed positive correlations with FA and negative correlations with MD, AD, and RD in the middle and terminal parts of WM tract (sections 40–100). Birth weight was correlated with DTI‐metrics in the initial and middle parts (FA, sections 29–34; MD, AD, sections 14–67; RD, sections 1–37). Crown‐heel length negatively correlated with MD, AD and RD in the terminal parts (MD, AD, sections 56–73; RD, sections 29–84; Figure 3a).

Figure 3.

Pearson partial correlation analysis of birth indicators related to diffusion tensor imaging (DTI)‐metrics at selected left hemispheric white matter tracts. (a) Left corticospinal tract (left CST), (b) left optic radiation (left OR), (c) left auditory radiation (left AR), and (d) left thalamus‐primary somatosensory cortex (left thal‐PSC). The selected fiber tract was divided into 100 segments denoted as sections 1–100. Pearson correlations of DTI‐metrics with gestational age, adjusted birth weight, adjusted head circumference, and adjusted crown‐heel length were separately analyzed. For each birth indicator, other three indicators, postnatal age and sex were controlled in partial correlation analysis. Birth weight, head circumference, and crown‐heel length were first adjusted for gestational age because of their confirmed strong relationships (p < .05). r _{_GA}, r _{_BW}, r _{_CHL}, and r _{_HC} denote the Pearson correlation coefficients of gestational age, birth weight, crown‐heel length, and head circumference, respectively [Color figure can be viewed at http://wileyonlinelibrary.com]

In the left OR, significant correlations of GA with DTI‐metrics were found almost in entire tract (sections 9–100). Birth weight correlated with DTI‐metrics in the initial parts (FA, sections 46–51; MD, AD, RD, sections 12–26). Crown‐heel length correlated with DTI‐metrics in the middle and terminal parts (FA, sections 60–98; MD, AD, RD, sections 28–100; Figure 3b).

In the left AR, GA presented significant correlations with DTI‐metrics in the initial and middle parts (FA, sections 1–69; MD, RD, sections 1–100; AD, sections 19–100). Birth weight was correlated with DTI‐metrics in the initial and middle parts (FA, sections 1–69; MD, RD, sections 1–100; AD, sections 19–100), while crown‐heel length was correlated with FA, MD, and RD in the terminal parts (FA, sections 54–79, 92–100; MD, RD, sections 49–82, 89–100; Figure 3c).

In the left thal‐PSC, GA showed significant correlations with DTI‐metrics almost in entire tract (FA, sections 1–82, MD, AD, RD, sections 1–100). Birth weight was correlated with MD, AD, and RD in the initial parts (sections 1–39), while crown‐heel length was correlated with FA, MD, and RD in the terminal parts (FA, sections 91–100; MD, AD, RD, sections 42–100; Figure 3d).

Similar results were observed in the right CST, OR, AR, and thal‐PSC (Figure 4). GA showed higher positive correlations with FA and lower negative correlations with MD, AD, and RD at a wider tract range than birth weight and crown‐heel length.

Figure 4.

Pearson partial correlation analysis of birth indicators related to diffusion tensor imaging (DTI)‐metrics at selected right hemispheric white matter tracts. (a) Right corticospinal tract (right CST), (b) right optic radiation (right OR), (c) right auditory radiation (right AR), and (d) right thalamus‐primary somatosensory cortex (right thal‐PSC). The selected fiber tract was divided into 100 segments denoted as sections 1–100. Pearson correlations of DTI‐metrics with gestational age, adjusted birth weight, adjusted head circumference, and adjusted crown‐heel length were separately analyzed. For each birth indicator, other three indicators, postnatal age and sex were controlled in partial correlation analysis. Birth weight, head circumference, and crown‐heel length were first adjusted for gestational age because of their confirmed strong relationships (p < .05). r _{_GA}, r _{_BW}, r _{_CHL}, and r _{_HC} denote the Pearson correlation coefficients of gestational age, birth weight, crown‐heel length, and head circumference, respectively [Color figure can be viewed at http://wileyonlinelibrary.com]

3.4. Multiple linear regression of birth indicators related to DTI‐metrics at selected WM ROIs

Multiple linear regression analysis showed consistent results with Pearson correlation analysis. Specifically, GA was positively correlated with FA and negatively correlated with MD, AD, and RD in all four ROIs. Birth weight and crown‐heel length showed similar results in some ROIs, for example, OR, and thal‐PSC. Head circumference showed positive correlations with MD, AD, and RD in OR and AR.

In the regression models, GA showed markedly higher regression coefficients (β) than the other three indicators (Table 3).

Table 3.

Multiple linear regression of birth indicators related to DTI‐metrics at selected white matter regions of interest

DTI‐metrics		Model variables								R 2
		GA		Birth weight		Crown‐heel length		Head circumference
		β	P value	β	P value	β	P value	β	P value
CST	FA	0.488	<.001	0.327	.082	0.270	.106	0.270	.106	0.359
	MD	−0.539	<.001	−0.313	.072	−0.286	.065	0.175	.366	0.455
	AD	−0.519	<.001	−0.300	.093	−0.248	.117	0.136	.494	0.421
	RD	−0.539	<.001	−0.313	.072	−0.296	.056	0.189	.331	0.454
OR	FA	0.520	<.001	0.185	.294	0.380	.018	−0.169	.397	0.422
	MD	−0.589	<.001	−0.183	.286	−0.411	.009	0.402	.042	0.452
	AD	−0.561	<.001	−0.163	.362	−0.380	.020	0.442	.033	0.403
	RD	−0.596	<.001	−0.191	.260	−0.421	.007	0.379	.052	0.469
AR	FA	0.435	<.001	0.268	.164	0.321	.064	−0.206	.342	0.317
	MD	−0.537	<.001	−0.406	.028	−0.265	.103	0.430	.040	0.393
	AD	−0.496	<.001	−0.415	.032	−0.173	.306	0.444	.043	0.330
	RD	−0.540	<.001	−0.391	.032	−0.298	.065	0.412	.046	0.406
Thal‐PSC	FA	0.494	<.001	0.196	.289	0.281	.092	−0.065	.755	0.366
	MD	−0.574	<.001	−0.337	.048	−0.323	.034	0.269	.158	0.480
	AD	−0.541	<.001	−0.368	.038	−0.292	.062	0.316	.111	0.441
	RD	−0.577	<.001	−0.318	.061	−0.331	.029	0.244	.198	0.483

Abbreviations: AD, axial diffusivity; AR, auditory radiation; CST, corticospinal tract; DTI, diffusion tensor imaging; FA, fractional anisotropy; GA, gestational age; MD, mean diffusivity; OR, optic radiation; RD, radial diffusivity; thal‐PSC, thalamus‐primary somatosensory cortex.

β denotes the standardized coefficient in the regression model.

4. DISCUSSION

In this study, multivariable regression analysis was used to explore the associations of birth indicators with early brain WM maturation, with potential confounding factors such as postnatal age and sex controlled. Our results indicated that in the newborn stage, aspects of WM maturation in full‐term neonates may be predicted by GA, birth weight, and crown‐heel length. Specifically, higher GA, birth weight, and crown‐heel length may indicate greater WM maturation. Moreover, among these four indicators, GA showed a predominant correlation with WM maturation characterized by DTI‐metrics.

It is known that from birth, brain WM undergoes an extremely rapid development during the neonatal period. In particular, as summarized in one systematic review (Li et al., 2019), neonatal brain MR images present reduced tissue contrast and dynamic changes greatly different from adult brain MR images. All these bring the technical challenges associated with image processing and analysis in neonates, for example, target identification and normalization in TBSS. To address such issue, an optimized neonatal TBSS pipeline developed by our previous work (Li et al., 2016) was employed. With the aid of optimized group‐wise target choice, this TBSS pipeline showed more competitive efficiency and reliability in neonatal brain DTI analysis and also ensured the accuracy of the results.

It has been demonstrated that DTI‐metrics, that is, FA, MD, AD, and RD could respectively reflect brain WM microstructural integrity, water diffusivities, axonal growth and myelination (Dubois et al., 2014). For the full‐term neonates, as GA increased, FA significantly increased while MD, AD, and RD decreased in the whole brain WM. Such changes in DTI metrics may suggest pre‐ and on‐going myelinations, seen as increments of axon density, cell membrane, and myelination, and decrements of brain water content and extracellular spaces (Dubois et al., 2014). A previous study with term neonates just observed GA‐related FA changes in some regional WMs, for example, corpus callosum, right anterior corona radiata, anterior limb of the internal capsule, and external capsule (Broekman et al., 2014). The minor inconsistency with our results of whole‐brain WM may be attributable to the older GA range (39.6 ± 1.2 weeks vs. 38.8 ± 0.9 weeks) that indicates more advanced WM maturation.

Previous studies have reported significant associations of birth weight with FA and MD in preterm infants at term‐equivalent ages (Lepomäki et al., 2012). In this study, with respect to full‐term neonates, both birth weight and crown‐heel length positively correlated with FA and negatively correlated with MD, AD, and RD, indicating more advanced WM maturation. These findings further suggest an association of birth weight and crown‐heel length with early brain maturation. Specific nutrients important for fetal growth and brain development could be a possible mechanism (Georgieff, 2007; Prado & Dewey, 2014). In details, as anthropometric indexes of nutritional background, birth weight and crown‐heel length showed certain correlations with nutrient intake (Doyle, Crawford, Wynn, & Wynn, 1990). In addition, animal studies have demonstrated the importance of adequate nutrition for the neurodevelopmental processes during pregnancy and infancy, such as neuron proliferation and myelination (Prado & Dewey, 2014). In this regard, the observed associations between brain WM maturation and anthropometric indexes may reflect the influences of nutrients on early brain WM development. Notably, besides the above maturation differences, hemispheric asymmetry in the FA correlations was also observed. Specifically, higher birth weight and crown‐heel length showed more advanced WM maturation in left‐hemispheric regions, such as corona and posterior thalamic radiations. This structural asymmetry may underpin the lateralization of the somatosensory response that is detected at birth (Erberich et al., 2006). Although the specific relationships between brain structural asymmetry and functional lateralization remain unclear (Deoni et al., 2011; Dubois et al., 2014; Paus et al., 1999), the observed asymmetry in the corona and posterior thalamic radiations may presumably support the increased development of sensorimotor ability (Dubois et al., 2008). Beyond, cortical gray matter asymmetries around the peri‐Sylvian region, superior temporal sulcus and parieto‐occipital sulcus have also been found (Li et al., 2014). These structural asymmetries may be likely attributable to genetic programs during prenatal stage, as asymmetry of gene expression in the human embryonic cortex has been found at gestational of 12 weeks (Erberich et al., 2006). It was worth noting that the observed WM asymmetries at corona and posterior thalamic radiations appeared to be partially inconsistent with cortical findings (e.g., rightward asymmetry at parieto‐occipital sulcus). Given the complexity in brain development, such inconsistency should be further clarified by exploring the synergetic development in both brain WM and gray matter, and their relationships associated with brain structural asymmetries.

Differently, as head circumference increased, FA remained unchanged, while MD, AD, and RD significantly increased in localized WM, for example, posterior thalamic radiation. This result may be rooted in the WM developmental status during early postnatal age of 1–14 days. During this period, glial cells progressively increase when majority of the observed WMs, for example, OR and external capsule undergo premyelination (Dubois et al., 2014). In this regard, it may be relatively higher water content in brain volume with higher head circumference that led to higher MD, AD, and RD. Although previous studies suggested head circumference as a marker of in‐utero brain development (Cheong et al., 2008) and found that head circumference correlated with neurodevelopmental outcome (Broekman et al., 2009; Dekhtyar et al., 2015), its associations with early WM maturation have rarely been investigated. In full‐term neonates aged ≤2 weeks, we observed no significant changes in brain WM maturation relating to head circumference. However, this preliminary finding may be limited by the insufficient data; more neonate data are required to further explore and clarify this issue.

Both Pearson correlations and multiple linear regression analysis indicated that GA showed higher associations with DTI‐metrics than the three anthropometric indicators. This may suggest that among the four birth indicators examined, GA may be most strongly associated with early WM maturation in full‐term newborns. As GA increases during the term period, brain WM continues to mature in the order of posterior to anterior and central to peripheral (Kinney, Ann brody, Kloman, & Gilles, 1988). Along the CST, higher correlation coefficients at the posterior limb of the internal capsule may be rooted in its earlier myelination initiation (Kinney et al., 1988). In contrast to OR and thal‐PSC, significant correlations of FA with GA were mainly found at the initial and middle parts of AR. Such findings are in line with the evolution sequence of WM myelination, that is, from central/proximal to peripheral/distal areas (Kinney et al., 1988). Beyond, a previous study found that AR has a much longer duration of WM myelination than OR (Dubois et al., 2014). This may be the reason why we found no significant correlations of FA with GA at the distal part of AR during the first 2 weeks of postnatal life.

Our study had some limitations. First, the sample size was small. A larger sample size will be needed to examine the associations of birth indicators with WM maturation, for example, whole or regional changes and lateralization in brain maturation. Second, some neonates with history of diseases such as mild asphyxia, metabolic acidosis, and physiologic jaundice, were included due to difficulty in recruiting entirely healthy neonates. We have performed the follow‐up of part of subjects' neurodevelopment outcome by a questionnaire for assessing developmental milestones items (social, language, cognitive, and movement; Center for Disease Control and Prevention, 2018). Although our follow‐up and previous findings (Graham et al., 2004; Low, Lindsay, & Derrick, 1997; Ullah, Rahman, & Hedayati, 2016; van Handel, Swaab, de Vries, & Jongmans, 2007) indicated that these diseases rarely correlated with abnormal neurodevelopment, more data of entirely healthy neonates are required to further verify our findings. Third, a DTI protocol with thick slices and nonisotropic voxels was used in consideration of neonatal poor tolerance to long‐time and noisy MRI scan. As the advances of fast MRI acquisition techniques, a high‐resolution DTI protocol recommended by Developing Human Connectome Project (dHCP) and UNC/UMN Baby Connectome Project (BCP; Howell et al., 2019), should be performed to obtain more accurate results. Fourth, given brain development's complexity, perinatal factors (e.g., maternal antenatal steroids and preeclampsia) potentially relating to WM maturation should be considered in future analyses. Last, this was a single‐center study. Multi‐center longitudinal studies would be of great value in revealing the associations of birth indicators with early brain development.

In conclusion, this study presented some interesting findings regarding neonatal brain development: (a) full‐term neonates with higher birth indicators such as GA, birth weight, and crown‐heel length may show greater brain microstructure maturation; (b) among the four birth indicators examined, GA was the one most strongly associated with WM maturation; (c) higher birth weight and crown‐heel length with leftward superiority at the corona and posterior thalamic radiations may presumably support earlier motor function. Taken together, such findings to a certain extent reveal the associations of birth indicators with brain WM maturation, thereby providing insights into the understanding of brain developmental mechanisms.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

SFigure 1. Flowchart of the optimized tract‐based spatial statistics (TBSS) pipelines for neonates. FA = fractional anisotropy.

SFigure 2. Scatterplots and linear fitting of DTI‐metrics with gestational age (GA). CST = Corticospinal tract, OR = optic radiation, AR = auditory radiation, thal‐PSC = thalamus‐primary somatosensory cortex; FA = fractional anisotropy, MD = mean diffusivity, AD = axial diffusivity, RD = radial diffusivity.

SFigure 3. Scatterplots of DTI‐metrics with postnatal age (days). CST = Corticospinal tract, OR = optic radiation, AR = auditory radiation, thal‐PSC = thalamus‐primary somatosensory cortex; FA = fractional anisotropy, MD = mean diffusivity, AD = axial diffusivity, RD = radial diffusivity.

STable 1. Multiple linear regression of birth indicators related to fractional anisotropy at optical radiation.

STable 2. Multiple linear regression of birth indicators related to DTI‐metrics at selected white matter regions of interest.

Jin C, Li Y, Li X, et al. Associations of gestational age and birth anthropometric indicators with brain white matter maturation in full‐term neonates. Hum Brain Mapp. 2019;40:3620–3630. 10.1002/hbm.24620

DATA AVAILABILITY

Data sharing is not applicable to this article as no new data were created or analyzed in this study.