Reduced corticomotor excitability and motor skills development in children born preterm

Abstract

The mechanisms underlying the altered neurodevelopment commonly experienced by children born preterm, but without brain lesions, remain unknown. While individuals born the earliest are at most risk, late preterm children also experience significant motor, cognitive and behavioural dysfunction from school age, and reduced income and educational attainment in adulthood. We used transcranial magnetic stimulation and functional assessments to examine corticomotor development in 151 children without cerebral palsy, aged 10–13 years and born after gestations of 25–41 completed weeks. We hypothesized that motor cortex and corticospinal development are altered in preterm children, which underpins at least some of their motor dysfunction. We report for the first time that every week of reduced gestation is associated with a reduction in corticomotor excitability that remains evident in late childhood. This reduced excitability was associated with poorer motor skill development, particularly manual dexterity. However, child adiposity, sex and socio-economic factors regarding the child's home environment soon after birth were also powerful influences on development of motor skills. Preterm birth was also associated with reduced left hemisphere lateralization, but without increasing the likelihood of being left handed per se. These corticomotor findings have implications for normal motor development, but also raise questions regarding possible longer term consequences of preterm birth on motor function.

Key points

• Children born preterm commonly experience motor and cognitive difficulties, but the physiology underlying this dysfunction is unknown.

• We used transcranial magnetic stimulation techniques and age-appropriate assessments of motor skills development to investigate neurodevelopment in 151 children born between 25 and 41 weeks of gestation.

• Reduced gestational age at birth was associated with a reduction in corticomotor excitability that persisted in late childhood, poorer development of manual dexterity skills and reduced hemispheric lateralization of hand preference.

• We suggest this reduced corticomotor excitability at least partly reflects reduced white matter integrity and functional connectivity in the brain regions subserving movement control.

• These findings show that preterm birth, which is increasingly common, impacts neuromotor development and related physiology into adolescence. Whether this altered neurophysiology and motor function persists in adulthood is unknown.

Introduction

Impaired motor and cognitive development remain the two major adverse outcomes of preterm birth despite advances in neonatal care (Hack & Fanaroff, 2000). Preterm birth is defined as one following a pregnancy of less than 37 completed weeks (or 259 days) of gestation (GA; WHO, 1992). In Australia in 2008, 79% of preterm infants (or about 7% of all births) were born ‘late preterm’ between 32 and 36 weeks GA, yet long-term follow-up studies of outcomes typically confine their focus to the very/extremely preterm (≤32 weeks GA) or very low birth weight, who accounted for only 21% of preterm babies (or 1.7% of all births). The statistics are similar in other western countries, and it is now increasingly recognized that even late preterm children have a greater risk of morbidity and developmental disorders than their term-born counterparts, but receive little if any developmental follow-up (Raju et al. 2006; Woythaler et al. 2011). Nevertheless, the few studies examining developmental outcomes in late preterm children all show an increased risk of motor dysfunction and learning difficulties at school age (Huddy et al. 2001; Pietz et al. 2004; Kirkegaard et al. 2006; Chyi et al. 2008; Morse et al. 2009; Woythaler et al. 2011).

Preterm birth between 20 and 37 weeks GA corresponds to a time of rapid cortical growth, particularly in the motor and sensorimotor areas (Eyre et al. 2000; Kapellou et al. 2006; Dubois et al. 2008a). The premature transition from the intrauterine to the extrauterine environment alters the trajectory and temporal characteristics of brain development, and the earlier the birth, the greater the perturbation from normal growth trajectories (Kapellou et al. 2006). The corticomotor system (including the brain motor areas and corticospinal tract) is the most common site of brain damage in the prenatal and neonatal periods (Eyre, 2003). An intact corticomotor system is indispensable for learning and performing skilled and dextrous movements, particularly of the hands. Even in the absence of focal brain lesions, many preterm children have reduced regional brain volumes, compromised development of both white and grey matter and reduced cortical surface area and gyrification, evident using magnetic resonance imaging (MRI; Ajayi-Obe et al. 2000; Peterson et al. 2000, 2003; Nosarti et al. 2002, 2008; Nagy et al. 2003, 2009; Peterson, 2003; Counsell & Boardman, 2005). Critically, these abnormalities remain at least into late adolescence and are present in the major motor control areas (Peterson et al. 2000, 2003; Nosarti et al. 2002, 2008; Nagy et al. 2003, 2009; Peterson, 2003; van Soelen et al. 2010).

There is increasing evidence that microstructural abnormalities in cortical development, in particular, underlie many of the cognitive and the more subtle motor sequelae that commonly manifest in preterm children (Ajayi-Obe et al. 2000; Kesler et al. 2004, 2008; Counsell et al. 2008). Diffusion tensor imaging (DTI) studies have shown that, compared with their term-born peers, preterm children and adolescents have altered neural connectivity between brain regions that is associated with reduced white matter integrity (Constable et al. 2008; Miller & Ferriero, 2009; Lubsen et al. 2011). Functionally, these microstructural abnormalities are associated with motor and cognitive impairments (Constable et al. 2008; Counsell et al. 2008; Ludeman et al. 2008; Mullen et al. 2011). However, up to 25% of preterm children who have apparently normal results on MRI/DTI go on to develop motor and cognitive dysfunction (Miller et al. 2005; Ludeman et al. 2008), suggesting that most current clinical imaging modalities have limitations in detecting these microstructural abnormalities.

An alternative approach to investigating the excitability and connectivity of cortex is to use transcranial magnetic stimulation (TMS), a painless technique that allows non-invasive stimulation of the conscious human cortex. When applied at the scalp over the motor cortex region at sufficient intensity, TMS activates corticospinal output neurons and results in a response in, for example, the hand muscles on the opposite, or contralateral, side of the body. This twitch in the muscle, called the motor-evoked potential (MEP), can be measured using surface electrodes on the skin overlying the muscle. The size and latency of the MEP reflect the excitability of neurons and impulse conduction characteristics of the motor cortex, corticospinal tract and spinal motor centres in real time.

In this study, we used TMS and surface electromyography techniques to examine corticomotor excitability in 10- to 13-year-old children born at 25–41 weeks GA. We hypothesized that hitherto undetected alterations in the normal development of the motor areas of the brain and the corticomotor pathways underpin at least some of the motor dysfunction seen in preterm children and are evident as reduced corticomotor excitability that is associated with poorer development of age-appropriate motor skills. To discriminate the effects of preterm birth from those of suboptimal intrauterine growth, we also assessed the effect of birth weight corrected for gestational age at birth on these outcomes. Preliminary results have been presented elsewhere (Pitcher et al. 2011a,b).

Methods

Ethical approval, participants and recruitment

One hundred and fifty-one children (82 males) and their parents/primary care giver gave informed written consent to participate in the study. Gestational age ranged from 25 to 41 weeks (34.44 ± 3.52 weeks), and mean uncorrected age at assessment was 155.4 ± 11.15 months (approximately 12 years and 11 months). All children born at ≤37 weeks GA were born at the Women's and Children's Hospital, North Adelaide, Australia between January 1996 and December 1997. This participant age group was chosen because prevailing evidence indicates it is the youngest age at which MEPs can be consistently evoked in relaxed muscle (Eyre et al. 2000; Fietzek et al. 2000; Garvey & Gilbert, 2004). Neonatal intensive care unit follow-up and labour ward databases were used to identify and approach prospective children and their families. In addition to the exclusion criteria recommended for the safe use of TMS (Wassermann, 1998; Quintana, 2005), children with known abnormality on perinatal cranial ultrasound, known chromosomal or genetic disorders, clearly identified familial or non-familial syndromes (without known chromosomal or genetic defect) and children with a physical or intellectual disability and unable to follow simple instructions were excluded. Twenty-nine children were excluded on the basis of the TMS safety screen (not included in the sample of 151). Four children were excluded before protocol completion owing to the presence of ipsilateral responses to transcranial magnetic stimulation, which can be developmental but may also indicate the presence of a corticospinal tract lesion (Eyre et al. 2001). Five children (two born at term) were taking prescribed methylphenidate (i.e. Ritalin), but all were assessed ‘off’ medication. No children were taking other medications known to alter cortical excitability (e.g. antidepressants). Term-born children matched for age, sex and ethnicity were recruited from a given preterm child's school or classroom. In addition, some control children were also recruited from community newspaper advertisements. Ethical clearances and approval for the study were provided by the Child Youth and Women's Health Service, University of Adelaide, the South Australian Government Department of Education and Children's Services Human Research Ethics Committees, and Catholic Education South Australia. All procedures were performed in accordance with the Declaration of Helsinki (2008 revision).

Separate written consent was obtained to access data on children's birth characteristics, including GA and neonatal morbidities, from the Women's and Children's Hospital Perinatal Statistics collection. Maternal anthropometry (height and weight at first antenatal visit), parity and gravity were also recorded. Gestation Related Optimal Weight software (Gardosi & Francis, 2006) was used to calculate each child's birth weight relative to their predicted term weight, i.e. their birth weight centile (BW%), and adjusted for maternal size, ethnicity and parity. Socio-economic Indexes for Areas scores were calculated for the address each child went home to following their birth (1996 National Census), as well as for their current address (2006 National Census).

Protocol

Data for this study were collected as part of the broader PREMOCODE (Preterm Motor and Cognitive Development) study, for which each child attended at least two assessment visits. Given that the order of assessment components was randomized within and between subjects, some children did not undergo the Movement Assessment Battery for Children, version 2 (MABC2; Henderson et al. 2007) on the same day as the corticomotor neurophysiology assessment. Of the children who underwent both assessments on the same day, the corticomotor assessment was always performed first. The characteristics recorded included the child's height, weight and percentage body fat (%BF; using bio-impedance). Handedness was determined by calculating the laterality quotient using the Edinburgh Handedness Inventory (Oldfield, 1971). The inventory provides a measure of an individual's preferred hand in performing common tasks of daily living. The resultant scale score is between +1 (strongly right hand dominant) and –1 (strongly left hand dominant), with a score of zero indicating no hand preference. A score ≥0.4 was considered right handed, ≤−0.4 was left handed, and scores between −0.39 and +0.39 indicated no hand preference. The MABC2 assessments were administered by a physiotherapist with extensive paediatric experience (J.L.D.). Significant to moderate impairment is defined as a total score less than or equal to the fifth centile and mild to moderate impairment as a total score less than or equal to the 15th centile but greater than the fifth centile (Williams et al. 2010; Roberts et al. 2011). All assessors were blinded to each child's GA.

Cortical neurophysiology assessment

Single-pulse TMS and surface electromyography were used to assess corticomotor excitability in both left and right hemispheres and the respective contralateral corticospinal tracts. In 95 children, short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF) were also measured. Children sat in an armchair with their forearms and hands supported. Motor-evoked potentials were recorded from the index finger abductor muscle, first dorsal interosseous, of both hands. Adhesive silver–silver chloride electrodes were applied, with the active electrode over the muscle belly and the inactive electrode over the metacarpophalangeal joint. Electromyograms (EMGs) were sampled at 5.1 kHz with a laboratory interface (CED 1401; Cambridge Electronic Design, Cambridge, UK), filtered (20 Hz to 1 kHz) and analysed offline using commercially available software (Signal version 4; Cambridge Electronic Design). Transcranial magnetic simulation was applied with a 70 mm figure-of-eight stimulating coil, connected to a monophasic Magstim 200² magnetic stimulator (Magstim Co., Whitland, UK), at the optimal scalp site, which was determined functionally using a ‘hunting’ procedure to find the motor hotspot, i.e. the scalp site where single suprathreshold TMS pulses produced the largest amplitude MEPs in the contralateral first dorsal interosseous muscle (cf. Rossini et al. 1994). The coil was oriented approximately 45 deg to the sagittal mid-line, so that the current induced in the cortex flowed in a plane perpendicular to the estimated alignment of the central sulcus in a posterior-to-anterior direction. The resting motor threshold (rMT) was determined as the lowest TMS intensity that evoked MEPs of at least 50 μV peak-to-peak amplitude in the resting first dorsal interosseous in five of 10 trials.

Corticomotor stimulus–response curves

Stimulus–response curves were constructed for the corticospinal projection from the motor cortex of each hemisphere to the contralateral first dorsal interosseous muscle using the procedure detailed previously (Pitcher et al. 2003, 2009). Briefly, MEP responses to TMS were recorded at stimulus intensities ranging from below rMT and increasing in 3% steps to either 100% of stimulator output, or to where MEP amplitude plateaued. Ten TMS pulses were delivered at each intensity, at an interstimulus interval (ISI) varying between 7 and 8 s, and the mean peak-to-peak amplitude for each block of 10 MEPs was plotted against stimulus intensity for each subject. The data were best fitted using the Marquardt–Levenberg algorithm for least squares convergence using commercially available software (Sigmaplot for Windows 11.0, © 2008 Systat Software, Inc., Chicago, IL, USA). The resultant five-parameter sigmoid is described the following equation:

where the five parameters are the difference between the smallest and the largest MEP amplitudes (a), the minimal MEP amplitude (y₀), the difference in stimulus intensities at 75 and 25% of the maximal MEP amplitude (b), the stimulus intensity required to obtain 50% of the maximal MEP amplitude (x₀) and the slope constant (c). Solutions to the equation to derive the predicted rMT, slope and rate of change of the slope at each stimulus intensity are detailed elsewhere (Pitcher et al. 2003). The area under the curve was calculated with Sigmaplot 9.0 software (Systat Software Inc.) algorithm, as follows:

where y is the stimulus intensity and x is the MEP amplitude at a given intensity. The order in which the hemispheres were tested was randomized between subjects.

Short-interval intracortical inhibition and facilitation

The SICI at 3 ms ISI and ICF at 10 ms ISI were measured using the paired-pulse protocol described previously (Kujirai et al. 1993; Smith et al. 2009). The conditioning stimulus intensity was set at 5% of stimulator output below active motor threshold and the test stimulus adjusted to give an MEP of approximately 1.0 mV peak-to-peak amplitude (Ridding et al. 1995; Ziemann et al. 1996). Left and right motor cortices were assessed separately, and the order was randomized between subjects. Thirty stimuli were applied in pseudorandom order, consisting of 10 single pulse test stimuli, 10 paired pulses given at an inhibitory ISI of 3 ms, and 10 paired pulses given at a facilitatory ISI of 10 ms. The mean peak-to-peak amplitude was measured for each block of 10 MEPs at each ISI and for the test response alone. Inhibited and facilitated MEPs were expressed for analysis as a percentage of the test response alone for a given set of conditions.

Data and statistical analysis

Motor-evoked potentials were recorded at high gain, and any with obvious prestimulus EMG were discarded online. Normal distribution and homogeneity of variance of the data were assessed using the Kolmogorov–Smirnov test and Levene's statistic, respectively, with GA group as the factor.

Data were analysed using R statistical analysis freeware (version 2.13.1; http://www.r-project.org/). Measures of corticomotor excitability (i.e. right hand rMT and left hand rMT, the corticomotor curve slope and area) were examined both as dependent variables in the primary analysis and as explanatory variables. As the rMTs for both hemispheres were highly correlated (r= 0.83, P≤ 0.0001, n= 129), only the right hand rMT is reported in the regression analyses of factors influencing MABC2 scores, because this was also the dominant hand for 83% of the children. In addition, the models did not significantly differ if the left hand rMT was included. Statistical significance was accepted at α= 0.05. All tests were two tailed.

Initial analysis

Gestational age was examined as a continuous variable (i.e. GA in weeks) and as a categorical variable with GA groups as defined by the World Health Organization and others (Tucker & McGuire, 2004; Alexander, 2007), i.e. extremely preterm (≤27 weeks GA), very preterm (28–32 weeks GA), late preterm (33–36 weeks) and term (37–41 weeks GA). However, as there were only seven eligible children in the extremely preterm group, this group was combined with the very preterm group to form the early preterm group (≤32 weeks GA). Linear regression modelling was used to determine the influence of GA (in weeks), BW% and sex on measures of cortical excitability (including the right hand rMT, left hand rMT, slope and area), intracortical inhibition and facilitation, and to determine the influence of GA (in weeks), BW% and sex on each of the five dependent variables, i.e. three MABC2 subscores (manual dexterity, aiming and catching, and balance), the MABC2 total score (MABC2_total) and right hand rMT. Likewise, this approach was used to determine the influence of cortical variables on MABC2_total and MABC2 subscores. Univariate analysis with GA group as the between-subjects factor was used to examine any differences between the GA groups.

Explanatory variables analysis

Gestational age, BW% and sex were included in all analyses except where specifically stated. Explanatory variables were included in the analysis only if they showed an initial correlation with the dependent variables. The birth variables included ponderal index, head circumference, Apgar score at 1 min, Apgar score at 5 min, singleton or multiple birth, parity, maternal age at child's birth, maternal body mass index at first antenatal visit, maternal smoking status during pregnancy (non-smoker, stopped during pregnancy or smoked during pregnancy), and the Index of Relative Social Disadvantage score of their home at birth (IRSD_birth). The current variables included the child's current height and weight, %BF, laterality quotient, the IRSD score for their current home (IRSD_current) and the right hand rMT.

The influences of explanatory variables on MABC2 and corticomotor measures were explored using relative importance regression modelling. This method allows quantification of the contribution of each individual regressor (i.e. explanatory variable) to the overall R² in a multiple regression model (Grömping, 2006). We used Lindeman, Merenda and Gold's heuristic approach of averaging over all orderings of regressors in the model, which allows a more causal interpretation of the variance allocations of each regressor to the total R² (Lindeman et al. 1980; Grömping, 2007). Variable groups were regressed against each of the dependent variables. Variables that accounted for less than around 10% of the variance explained by the model (i.e. by R²) were removed and the regression repeated until either no variable left accounted for <10% of the variance or when removing a minor/non-significant variable did not alter the R². The birth and current variables remaining from these processes were included in the final regression to determine the explanatory variables with the most influence on the dependent variable, i.e. the best model.

Results

At birth, both preterm groups had lower BW%, poorer skeletal growth (length and head circumference) and lower ponderal indices and were more likely to be a multiple birth than term-born infants (Table 1). At assessment, the early preterm children had shorter stature than the late preterm and term-born groups and tended to have lower body weight. Interestingly, the mean BW% in those children born at ≤27 weeks GA was 83.7 ± 8.2%, compared with 26.8 ± 27.0% in those born at 28–32 weeks. Apart from BW% and GA, these two subgroups did not differ significantly when their outcomes were compared. Child age at assessment was not different between the GA groups.

Table 1.

Child and maternal characteristics

		Early preterm, ≤32 weeks GA	Late preterm, 33–36 weeks GA	Term, 37–41 weeks GA
n	Males	21	37	24
	Females	19	28	22
	Total	40	65	46
Birth variables
Gestational age (weeks)	Males	29.7 ± 2.5*†	34.6 ± 1.1*	37.9 ± 1.6
	Females	29.8 ± 1.9*†	35.1 ± 1.0*	38.2 ± 1.4
	Total	29.7 ± 2.2*†	34.8 ± 1.1*	38.0 ± 1.5
Birth weight centile (% term	Males	42.2 ± 36.0*	36.0 ± 29.5*	52.7 ± 31.1
optimal weight)	Females	33.9 ± 30.7*	38.4 ± 34.4*	63.9 ± 29.0
	Total	37.7 ± 33.0*	37.0 ± 31.5*	57.9 ± 30.3
Length (cm)	Males	39.4 ± 3.7*†	46.5 ± 2.3*	48.1 ± 5.3
	Females	38.8 ± 3.4*†	44.8 ± 2.8*	49.3 ± 3.1
	Total	39.1 ± 3.6*†	45.8 ± 2.6*	48.6 ± 4.4
Head circumference (cm)	Males	28.1 ± 2.5*†	32.7 ± 1.7*	35.3 ± 4.3
	Females	27.7 ± 2.4*†	31.8 ± 2.0*	34.2 ± 1.5
	Total	27.9 ± 2.4*†	32.3 ± 1.9*	34.8 ± 3.4
Ponderal index	Males	2.3 ± 0.2*†	2.4 ± 0.2*	3.0 ± 1.3
	Females	2.2± 0.2*†	2.5 ± 0.2*	2.8 ± 0.3
	Total	2.2 ± 0.2*†	2.5 ± 0.2*	2.9 ± 1.0
Apgar score at 1 min (total score)	Males	7.0 ± 1.8	7.9± 1.4	8.0 ± 1.3
	Females	6.1 ± 1.8*†	7.8 ± 1.8	8.2 ± 1.0
	Total	6.6 ± 1.8*†	7.8 ± 1.5	8.1 ± 1.1
Apgar score at 5 min (total score)	Males	9.0 ± 1.3	9.1 ± 0.6	9.0 ± 0.7
	Females	8.3 ± 1.4*†	9.1 ± 0.8	9.4 ± 0.5
	Total	8.6 ± 1.4*†	9.1 ± 0.7	9.2 ± 0.6
Total postnatal days in hospital (n)	Males	50.2 ± 33.2*†	14.0 ± 10.1*	5.7 ± 4.3
	Females	43.6 ± 26.3*†	9.1 ± 6.2	6.5 ± 6.9
	Total	47.3 ± 29.9*†	11.9 ± 9.0	6.1 ± 5.6
Singleton births (n)	Males	16	27	22
	Females	13	18	22
	%n	72.5%*	69.2%*	95.7%
Multiple births	Males	5	10	2
	Females	6	10	0
	%n	27.5%*	30.8%*	4.3%
Maternal age at time of birth (years)	Males	29.8 ± 7.2	31.0 ± 5.8	29.9 ± 4.2
	Females	28.8 ± 5.4	28.5 ± 5.6	31.4 ± 5.1
	Total	29.4 ± 6.3	30.0. ± 4.6	30.6 ± 4.6
Maternal height (cm) (prepregnancy	Males	1.65 ± 0.07	1.68 ± 0.09	1.64 ± 0.08
or first antenatal visit)	Females	1.65 ± 0.05*†	1.60 ± 0.06*	1.63 ± 0.06
	Total	1.65 ± 0.06	1.64 ± 0.09	1.63 ± 0.07
Maternal weight (kg) (prepregnancy	Males	61.07 ± 10.3	73.0 ± 23.7	68.1 ± 20.5
or first antenatal visit)	Females	68.9 ± 10.8	64.5 ± 17.3	70.3 ± 16.3
	Total	65.1 ± 11.1	69.1 ± 21.3	69.2 ± 18.4
Parity	Males	0.90 ± 1.2	0.81 ± 0.9	1.04 ± 0.91
	Females	0.79 ± 1.4	0.64 ± 0.8	0.90 ± 1.0
	Total	0.85 ± 1.3	0.74 ± 0.9	0.98 ± 0.94
Gravida	Males	2.3 ± 1.4	2.1 ± 1.2	2.7 ± 1.3
	Females	2.2 ± 1.6	2.2 ± 1.1	2.5 ± 1.0
	Total	2.3 ± 1.5	2.1 ± 1.2	2.6 ± 1.2
Current variables
Uncorrected age (months)	Males	154.5 ± 10.7	156.2 ± 10.6	155.6 ± 11.9
	Females	152.6 ± 10.9	156.7 ± 11.5	158.6 ± 11.1
	Total	153.6 ± 10.7	156.4 ± 10.9	157.0 ± 11.6
Child's height (cm)	Males	1.45 ± 0.07†	1.53 ± 0.09	1.48 ± 0.06
	Females	1.48 ± 0.09*	1.48 ± 0.09*	1.54 ± 0.09
	Total	1.46 ± 0.08*†	1.51 ± 0.09	1.51 ± 0.09
Child's weight (kg)	Males	37.2 ± 8.8†	45.9 ± 10.5	42.3 ± 10.0
	Females	44.8 ± 11.4	42.3 ± 10.1	49.5 ± 11.7
	Total	40.8 ± 10.7	44.4 ± 10.4	45.7 ± 11.3
Child's total body fat (%)	Males	15.5 ± 6.3	18.9 ± 6.2	17.6 ± 7.4
	Females	25.5 ± 8.1	22.4 ± 6.7	24.9 ± 6.6
	Total	20.2 ± 8.7	20.4 ± 6.6	21.1 ± 7.9
Laterality quotient (–1 = left	Males	0.6 ± 0.7	0.7 ± 0.5	0.8 ± 0.4
handed, 0 = no preference and +1 = right handed)	Females	0.6 ± 0.7	0.5 ± 0.7	0.8 ± 0.2
	Total	0.6 ± 0.7	0.6 ± 0.6	0.8 ± 0.4

Abbreviation: GA, completed weeks gestational age. Data are GA group means ± SD. * Denotes P < 0.05 when compared with the term group (Tukey's HSD). † Denotes P < 0.05 when compared with the late preterm group (Tukey's HSD).

Effect of sex, GA and BW% on corticomotor excitability characteristics

Motor thresholds were obtained for at least one hemisphere in 138 children and in both hemispheres from 129 children. Twelve children chose not to participate in this part of the study. In seven children, thresholds could be obtained only with target muscle contraction, but this was not related to either GA or BW%. Responses could not be obtained from one male (30 weeks GA, BW% < 1). This is reflected in the different sample sizes for the various motor threshold analyses.

Overall, rMTs were not different when males and females were compared (Table 2). Preterm birth was linearly associated with higher resting motor thresholds for the right (R²= 0.09, F_(1,137)= 13.6, P≤ 0.0001, n= 137; Fig. 1A) and left hands (R²= 0.06, F_(1,131)= 8.3, P= 0.005, n= 131; Fig. 1B). Univariate analyses showed a significant effect of GA group on both right hand rMT (F_(2,136)= 4.58, P= 0.01) and left hand rMT (F_(2,130)= 3.43, P= 0.04), because rMTs in the early preterm group were higher than those of the late preterm and term-born groups.

Table 2.

Resting motor thresholds to transcranial magnetic stimulation

		Early preterm, ≤32 weeks GA (n)	Late preterm, 33–36 weeks GA (n)	Term, 37–41 weeks GA (n)
Right hand
Resting motor threshold (% stimulator output)	Males	55.3 ± 13.3* (20)	52.5 ± 11.2 (35)	49.2 ± 11.4 (22)
	Females	56.7 ± 9.5 (15)	52.8 ± 12.7 (26)	49.0 ± 8.3 (20)
	Total	56.0 ± 11.4* (35)	52.6 ± 11.8 (51)	49.1 ± 10.0 (42)
Active motor threshold (% stimulator output)	Males	42.9 ± 9.2 (19)	40.4 ± 8.7 (34)	38.5 ± 10.2 (22)
	Females	45.5 ± 7.4 (15)	42.7 ± 10.9 (27)	38.7 ± 8.9 (20)
	Total	44.0 ± 8.4* (34)	41.5 ± 9.8 (61)	38.6 ± 9.5 (42)
Area under corticomotor stimulus–response curve	Males	17.2 ± 8.4 (12)	20.3 ± 9.5 (19)	25.1 ± 9.6 (15)
	Females	28.5 ± 10.8 (8)	33.8 ± 21.4 (11)	24.3 ± 9.2 (12)
	Total	21.7 ± 10.7 (20)	25.2 ± 16.1 (30)	24.7 ± 9.3 (27)
Curve slope	Males	0.22 ± 0.20 (12)	0.17 ± 0.14 (19)	0.23 ± 0.14 (15)
	Females	0.30 ± 0.19 (8)	0.29 ± 0.17 (11)	0.25 ± 0.19 (12)
	Total	0.25 ± 0.18 (20)	0.21 ± 0.16 (30)	0.23 ± 0.16 (27)
SICI at 3 ms ISI (% test MEP)	Males	49.8 ± 41.7 (14)	42.3 ± 21.3 (23)	34.6 ± 26.8 (14)
	Females	63.6 ± 61.7 (11)	38.1 ± 30.7 (18)	36.9 ± 23.3 (15)
	Total	55.9 ± 50.8 (25)	40.5 ± 25.6 (41)	35.8 ± 24.6 (29)
ICF at 10 ms ISI (% test MEP)	Males	115.7 ± 48.8 (14)	139.3 ± 78.8 (23)	134.9 ± 67.2 (14)
	Females	164.1 ± 100.7 (11)	120.6 ± 36.6 (18)	151.5 ± 76.7 (15)
	Total	137.0 ± 78.2 (25)	131.1 ± 63.8 (41)	143.5 ± 71.5 (29)
Left hand
Resting motor threshold (% stimulator output)	Males	55.6 ± 12.1 (20)	55.3 ± 13.8 (35)	51.9 ± 12.1 (22)
	Females	58.9 ± 8.7* (15)	53.0 ± 12.1 (26)	49.7 ± 6.4 (20)
	Total	57.2 ± 10.6* (35)	54.3 ± 13.1 (61)	50.9 ± 9.8 (42)
Active motor threshold (% stimulator output)	Males	43.2 ± 7.9 (17)	43.6 ± 9.9 (33)	38.5 ± 9.7 (21)
	Females	47.1 ± 7.0 (14)	42.3 ± 9.3 (25)	39.4 ± 6.8 (20)
	Total	45.0 ± 7.6* (35)	43.1 ± 9.6* (61)	38.9 ± 8.3 (42)
Area under corticomotor stimulus–response curve	Males	19.1 ± 9.2 (7)	17.8 ± 9.3 (12)	21.7 ± 11.3 (13)
	Females	21.3 ± 15.3 (5)	35.3 ± 17.5 (10)	16.4 ± 8.5 (10)
	Total	20.0 ± 11.5 (12)	25.8 ± 16.0 (22)	19.4 ± 10.3 (23)
Curve slope	Males	1.03 ± 1.6 (7)	0.19 ± 0.8 (12)	0.17 ± 0.1 (13)
	Females	0.21 ± 0.1 (5)	0.79 ± 1.3 (10)	0.18 ± 0.2 (10)
	Total	0.69 ± 1.3 (12)	0.46 ± 0.9 (22)	0.17 ± 0.1 (23)
SICI at 3 ms ISI (% test MEP)	Males	27.0 ± 12.4 (9)	46.1 ± 32.1 (21)	47.0 ± 43.0 (11)
	Females	39.3 ±23.3 (7)	28.7 ± 16.9 (18)	51.0 ± 30.9 (12)
	Total	32.3 ± 18.4 (16)	38.1 ± 27.3 (39)	49.1 ± 36.4 (23)
ICF at 10 ms ISI (% test MEP)	Males	136.3 ± 57.3 (9)	158.4 ± 98.1 (21)	104.6 ± 57.5 (11)
	Females	119.1 ± 73.5 (7)	133.3 ± 84.9 (18)	147.5 ± 71.0 (12)
	Total	128.8 ± 63.2 (16)	146.8 ± 92.0 (39)	127.0 ± 67.1 (23)

Data are GA group means ± SD (number of subjects). Abbreviations: GA, completed weeks gestational age; ICF, intracortical facilitation; ISI, interstimulus interval; MEP, motor-evoked potential; SICI, short-interval intracortical inhibition. * Denotes P < 0.05 when compared with the term group (Tukey's HSD).

Figure 1. Factors influencing corticomotor excitability.

Preterm birth was associated with reduced corticomotor excitability, i.e. higher resting motor threshold (rMT), in both the right (A) and left hands (B). Data are individual resting motor thresholds for males and females, although there were no differences in rMT due to sex. The regression equation describes the relationship between completed weeks gestational age (GA) and rMT. The best relative importance regression model of the explanatory variables influencing the rMT in the right and left hands is shown in C. *P < 0.05 for a given predictor in the model.

Figure 1C shows the best relative importance regression model of variables explaining right hand rMT (R²= 0.17, F_(5,104)= 4.16, P= 0.002), expressed as the percentage of the total variability explained by the model, i.e. R²= 0.17 = 100%. A lower right hand rMT was associated with a longer gestation (P= 0.02), being more right handed (i.e. having a laterality quotient closer to +1; P= 0.04), being female, of higher BW% and having a higher Apgar score at 5 min. Only GA and laterality quotient were significant at P≤ 0.05, whereas BW% was not (P= 0.09).

Low left hand rMT was associated with being more right than left handed (P= 0.01), of longer gestation (P= 0.02) and having a higher BW% (P= 0.02); (Fig. 1C). Sex remained in the final model, but was not significant.

Complete stimulus–response curves could be obtained from the right hands of 76 children and from the left hands of 57 children. Of these, curves for both hands could be obtained from 51 children (30 males). Neither the slope nor the area under the curve was influenced by GA or BW%. A steeper slope and a greater area under the right hand curve was associated with being a female and having a lower right hand (i.e. left hemisphere) rMT (slope, R²= 0.11, F_(2,76)= 4.72, P= 0.01; and area, R²= 0.17, F_(2,76)= 7.31, P= 0.001; Fig. 2). However, there were no similar findings for the left hand (i.e. right hemisphere). No associations were found to account for differences in the maximal MEP that could be evoked in either hemisphere.

Figure 2. Corticomotor stimulus–response curves for males and females.

Data are group mean ± SD motor-evoked potential (MEP) amplitudes for the early preterm (A and B), late preterm (C and D) and term-born groups (E and F). Panels on the left (A, C and E) show data from the left hemisphere and right hand, while panels on the right (B, D and F) show data from the right hemisphere and left hand.

Effect of sex, GA and BW% on intracortical excitability characteristics

For 95 children SICI/ICF data were obtained from the left hemisphere and right hand, for 72 children from the right hemisphere and left hand, and from both hemispheres in 70 of these children (Table 2). Data were not normally distributed, so were logarithmically transformed for analysis. Unblinding of the data revealed a negative correlation between active motor threshold and GA (in weeks; r=−0.31, P= 0.002, n= 95). In order to ensure that subjects were being stimulated at equivalent intensities relative to their active motor threshold (Orth et al. 2003), the individual conditioning stimulus intensities were expressed as a percentage of active motor threshold. The conditioning stimulus intensity was 87.9 ± 2.3% of active motor threshold in early preterm children, 86.6 ± 2.8% of active motor threshold in late preterm, and 86.7 ± 2.7% of active motor threshold in term-born children. One-way ANOVA showed that there was no difference in the conditioning stimulus intensity when the GA groups were compared. Neither covariate analysis with GA (in weeks) nor factor analysis of GA groups showed an effect of preterm birth on SICI. Likewise, ICF was not influenced by either active motor threshold or GA. However, the data were highly variable, and the statistical power of the sample was less than 30%.

Effect of sex, GA, BW% and corticomotor excitability on age-appropriate motor skills development (MABC2)

One hundred and forty-eight children completed the MABC2 assessment (Table 3). There was no difference in scores due to age (corrected or uncorrected for GA). Eleven children (seven boys) had significant impairment and five (two boys) had mild impairment for MABC2_total. All were born at ≤37 weeks GA. One-way ANOVA showed sex differences in the MABC2 subscores, so sex was included in all regression models. The BW% had no effect on either MABC2_total or subscores. With the exception of rMT, there was no association between any of the corticomotor stimulus–response curve characteristics or SICI/ICF and MABC2_total or subscores.

Table 3.

Percentile scores for the Movement Assessment Battery for Children, version 2

		Early preterm, ≤32 weeks GA	Late preterm, 33–36 weeks GA	Term, 37–41 weeks GA
n	Males	20	35	24
	Females	18	27	22
	Total	38	62	46
Manual dexterity	Males	36.8 ± 26.1	34.4 ± 22.3	41.6 ± 24.3
	Females	32.3 ± 21.5†	53.1 ± 27.4	42.0 ± 25.5
	Total	34.7 ± 23.8	42.5 ± 26.2	41.8 ± 24.6
Aiming and catching	Males	60.8 ± 28.4	74.5 ± 23.8	70.5 ± 28.2
	Females	49.1 ± 31.4	60.0 ± 27.1	63.4 ± 28.3
	Total	55.3 ± 30.0	68.2 ± 26.1	67.1 ± 28.1
Balance	Males	49.3 ± 34.6	45.5 ± 30.6	62.7 ± 26.7
	Females	62.5 ± 31.6	67.6 ± 31.2	71.4 ± 28.5
	Total	55.6 ± 33.5	55.1 ± 32.5	66.9 ± 27.6
Total score	Males	46.2 ± 34.1	50.4 ± 26.0	59.8 ± 23.8
	Females	45.4 ± 26.2	59.4 ± 29.9	60.8 ± 26.8
	Total	45.8 ± 30.2*	54.3 ± 27.9	60.2 ± 25.0

Data are GA group means ± SD. *Denotes P < 0.05 when compared with the term group (Tukey's HSD). † Denotes P < 0.05 when compared with the late preterm group (Tukey's HSD).

The MABC2_total increased with GA (R²= 0.06, F_(1,146)= 7.66, P= 0.006; Fig. 3A). When each sex was analysed separately, this association remained statistically significant for males (males only, R²= 0.06, F_(1,79)= 4.83, P= 0.03) but not for females (P= 0.1). Children with lower rMTs had better MABC2_total scores (R²= 0.07, F_(2,128)= 4.86, P= 0.009; Fig. 3B). Aiming and catching skills were better in males than in females, and improved with increased GA (both sexes, R²= 0.07, F_(2,145)= 5.03, P= 0.008). Conversely, balance skills were better in females than in males, and improved with increased GA (both sexes, R²= 0.09, F_(2,145)= 7.03, P= 0.001).

Figure 3. Factors influencing motor skill abilities.

A higher total Movement Assessment Battery for Children, version 2 (MABC2) score was associated with a longer gestation (A) and greater corticomotor excitability (B). Data are individual MABC2 percentile scores, GAs and rMTs (right hand). The best relative importance regression model of the explanatory variables influencing total MABC2 score and the component subscores is illustrated in C. Abbreviations: BMI, body mass index; IRSD, Index of Relative Social Disadvantage. *P < 0.05, **P≤ 0.0001 for a given predictor in the model.

Explanatory variables analyses of age-appropriate motor skills development (MABC2)

Total MABC2

Figure 3C shows the best model of explanatory variables influencing MABC2_total when right hand rMT was included (R²= 0.26, F_(5,126)= 9.02, P≤ 0.0001). The MABC2_total was highest in children who had lower adiposity (i.e. %BF) at assessment (%BF = 40.7%, P≤ 0.0001), greater corticomotor excitability (i.e. lower right hand rMT = 21.4%, P= 0.02), a higher GA (18.3%, P= 0.02), went home to a less socially disadvantaged home at birth (IRSD_birth= 14.1%, P= 0.05) and were female (5.54%, P= 0.03).

Manual dexterity

Eighteen children (13 boys) had significant impairment of dexterity, and 11 (five boys) had mild impairment. All were born at ≤37 weeks GA. Figure 3C shows the best model of explanatory variables influencing manual dexterity when right hand rMT was included (R²= 0.125, F_(4,131)= 4.67, P= 0.001). Better manual dexterity was associated with having greater corticomotor excitability (i.e. lower right hand rMT = 44.73%, P= 0.01), being female (24.8%, P= 0.01) and having lower %BF (24.0%, P= 0.02).

Aiming and catching skills

Four children (one boy) had significant impairment of aiming and catching skills, and three (two boys) had mild impairment. All, except one girl (40 weeks GA) with a significant impairment, were born at ≤37 weeks GA. Figure 3C shows the best model for aiming and catching skills (R²= 0.09, F_(3,118)= 3.76, P= 0.01). Better aiming and catching skills were associated with having a mother with a lower body mass index at the first antenatal visit (50.7%, P= 0.02) and a higher GA (35.6%, P= 0.04). Sex was not a significant predictor (13.7%, P= 0.24).

Balance skills

As with aiming and catching skills, corticomotor excitability had little influence on balance. Ten children (six boys) had significant impairment, and nine (eight boys) had mild impairment. All were born at ≤37 weeks GA. Figure 3C shows the best model for balance skill development (R²= 0.22, F_(3,144)= 13.25, P≤ 0.0001). Superior balance skills were associated with being female (43.8%, P≤ 0.0001), having lower adiposity (39.0%, P≤ 0.0001) and having a higher GA (17.2%, P= 0.006).

Handedness

Most children were right handed (n= 123, 82.6%), with only 20 (13%) being left handed and six children (4%) showing no hand preference. While there was only a weak trend for preterm birth to be associated with being less strongly right handed (r= 0.15, P= 0.07, n= 149), all children who were either left handed or had no hand preference were born at ≤38 weeks GA. Lateralization was influenced by a child's right hand rMT (i.e. left hemisphere; R²= 0.09, F_(1,128)= 12.50, P= 0.001) but not the left hand rMT. Children with low rMTs in both hands were more likely to have a higher laterality quotient, i.e. to be more right handed. Including predictors such as prenatal exposure to corticosteroids and/or magnesium sulphate, BW%, being a singleton or multiple birth, and the child's sex failed to account for this finding.

Discussion

The main novel finding of this study is that preterm birth is associated with reduced corticomotor excitability that persists in late childhood. Suboptimal fetal growth, evident as a lower BW%, further contributes independently to reduce corticomotor excitability although, unlike GA, this appears to be confined to the right motor cortex. While the most preterm children have the greatest decrements in corticomotor excitability, there does not appear to be a ‘critical’ GA after which no further gains in corticomotor excitability are made (i.e. every week of GA up to approximately 39–40 weeks is important). While corticomotor excitability is positively associated with functional motor skill development, particularly hand dexterity, other postnatal factors appear to play an important role in determining the level of motor skills attained.

High corticomotor thresholds in preterm children may reflect reduced white matter integrity and functional connectivity

We speculate that the higher rMTs seen in preterm children may reflect, at least in part, persistent reductions in white matter integrity and functional connectivity in cortico-cortical and/or cortico-thalamic projections involving the primary motor cortex. In this study, the primary measure of corticomotor excitability was the rMT, which is the lowest TMS intensity required to excite corticospinal fibres indirectly via excitatory input from cortico-cortical projections and to evoke a small MEP response in the target muscle (Ziemann, 2004). Excitation of these projections by TMS is also influenced by membrane voltage-gated sodium channels, because drugs that inhibit the activity of these channels and modify membrane excitability increase the rMT (Chen et al. 1997; Ziemann, 2003, 2004). In neurologically healthy adults, the rMT is negatively correlated with the maturation, myelination and structural integrity of the white matter of the primary, premotor and prefrontal cortices, the internal capsule, corpus callosum, corona radiata and cerebral peduncles (Kloppel et al. 2008). The maturation of white matter is associated with increasing fractional anisotropy (FA) and a reduction in the apparent diffusion coefficient on DTI (Hüppi et al. 1998; for review see Mathur et al. 2010); that is, individuals with low rMTs have better white matter microstructural integrity (i.e. higher FA) in the neuronal networks subserving motor control.

Likewise, higher rMT in preterm children may reflect weaker functional connectivity between the secondary and primary motor areas (Kloppel et al. 2008). Functional connectivity MRI studies indicate that preterm birth disrupts the normal development of neural connectivity within and between cortical regions, even in the absence of lesions (Smyser et al. 2010, 2011). In preterm infants, the networks contributing to motor and sensory function, auditory and visual processing and language are less mature and have aberrant connectivity when compared with their term-born peers (Smyser et al. 2010). The mechanisms underlying this are not yet clear, although white matter damage and reduced myelination are known to affect cortico-cortical and cortico-thalamic connections (Smyser et al. 2010).

A limitation of the study is that MRI facilities were not available to assist confirmation of the participants’ lesion status (neonatal or current) or to investigate white matter integrity; however, we think it unlikely that the increase in rMT with reduced GA is mainly due to undiagnosed perinatal lesions. Notwithstanding the inherent limitations, no included participants had any history of abnormal perinatal cranial ultrasound and none had either been diagnosed with, or suspected of having, cerebral palsy. Any children with ipsilateral responses to TMS, which might indicate a corticospinal tract lesion (Eyre et al. 2001), were excluded. As stated earlier, however, up to 25% of preterm children who have apparently normal results on MRI/DTI go on to develop motor and cognitive dysfunction (Miller et al. 2005; Ludeman et al. 2008). This observation is more consistent with microstructural or functional connectivity abnormalities, rather than lesions due to perinatal bleeds per se. It is unlikely that either a greater coil–cortex or scalp–cortex distance in the preterm children explains their higher rMTs (McConnell et al. 2001; Knecht et al. 2005). If anything, the available evidence suggests that the distance is likely to be shorter in preterm children (Backström et al. 2005), and adjustment for this would result in even higher rMTs compared with their term-born peers. There was no evidence that intracortical pathways mediating SICI/ICF are abnormal due to preterm birth, nor responsible for the effects of GA on rMT. However, given the widespread cortical microstructural abnormalities seen by others, it seems unlikely that intracortical and/or corticomotor pathways responsible for rMT would be selectively affected and the cortical networks responsible for mediating SICI/ICF spared. We would therefore caution against overinterpretation of the SICI/ICF data, particularly given the low statistical power in the sample. Lastly, there were no differences in participants’ age that might explain the relationship. We cannot rule out the possibility that the development of adult levels of corticomotor excitability is delayed, rather than permanently reduced, in preterm children. However, the findings clearly show that the reduced corticomotor excitability has functional effects in children well into their school years. While these changes may seem modest, they are larger than those in children with mild cerebral palsy/periventricular leukomalacia, in whom a median increase in rMT of 3% stimulator output compared with control subjects was reported (Koerte et al. 2010). It is unlikely that the motor areas are the only cortical regions affected, and reduced excitability in other cortical regions may impair other functions (and their development) in, for example, the cognitive and behavioural domains.

Corticomotor stimulus–response curves

The corticomotor curve slope reflects the size of the cortical representation and the distribution of excitability within the corticospinal projection (Ridding & Rothwell, 1997; Siebner & Rothwell, 2003), while the area under the curve is generally accepted as a relatively robust overall measure of corticomotor output and projection strength (Talelli et al. 2008; Pitcher et al. 2009). Given the clear effects of reduced GA on rMT, the lack of detectable differences due to GA on these stimulus–response curve characteristics appears anomalous, because they are believed to reflect aspects of the integrity and/or maturation of the cortex and corticomotor projection. However, full curves could be constructed for only about half the children in the sample. This was principally due to increased rMTs in the preterm children, in whom stimulator output reached maximum before the maximal MEP amplitude was reached. Therefore, the curve data are representative of higher GA children whose rMT was sufficiently low to allow a full stimulus–response curve to be constructed, with only limited data from lower GA children with higher rMTs, i.e. the children in whom the curve might be expected to be altered. We would advise future researchers of preterm birth who consider utilizing corticomotor curves to construct them against a background muscle contraction (which reduces the stimulation threshold to TMS), because resting curves appear of little value.

Effects of gestation and corticomotor excitability on age-appropriate development of motor skills

Preterm birth was associated with poorer overall development of age-appropriate motor skills. The most ubiquitous factor in determining motor skill level, however, was the adiposity of the child at assessment and/or their mother's adiposity at the first antenatal visit, which were correlated (r= 0.30, P= 0.004). Neither child nor maternal adiposity correlated with GA or BW% of the child, so the most parsimonious explanation is that, regardless of GA or BW%, high adiposity in the child and/or mother is associated with a lower likelihood of practising motor skills regularly, although we offer no direct evidence to support this.

Reduced corticomotor excitability selectively affects manual dexterity skills

Almost 23% of all preterm children in this study had some degree of manual dexterity impairment. Neither GA nor BW% had a direct influence on manual dexterity (Fig. 3A), but low GA and low BW% were both independently associated with reduced corticomotor excitability (i.e. rMT; Fig. 1), which was the most influential factor in determining dexterity. This is not surprising, given the well-established critical role of the motor cortex and corticospinal tract in the dextrous control of the hands and fingers (Porter & Lemon, 1995; Lemon et al. 1998; Lemon, 2008). In children, rMT has been shown to correlate inversely with finger tapping speed (Garvey et al. 2003). Likewise, Skranes et al. (2007) showed that preterm adolescents with poor manual dexterity (measured with MABC) had reduced white matter development in both hemispheres, particularly in the posterior limb of the internal capsule and the corpus callosum. Diffusion tensor imaging studies of white matter maturation in preterm neonates commonly report decreased FA and increased apparent diffusion coefficient values in various white matter regions, particularly the internal capsule and corpus callosum, and these abnormalities are associated with a suboptimal neuromotor outcome (Nagy et al. 2003; Drobyshevsky et al. 2007; Krishnan et al. 2007; Constable et al. 2008; Counsell et al. 2008). When measured in the brains of preterm adolescents, this decreased FA appears to be due to fewer and smaller diameter axons, as well as poorer myelination (Nagy et al. 2003). This is consistent with our finding that a high rMT was associated with a reduced area under the corticomotor curve and a flatter slope in males, both of which reflect the number, size and degree of myelination of axons in the corticomotor projection. However, we were not able to show a relationship between manual dexterity and either of these corticomotor projection measures.

Aiming, catching and balance skills

Neither aiming and catching nor balance skills were influenced by corticomotor excitability, but 17% of preterm children were classified as at risk or having impaired balance skills, and 5% had aiming and catching skill problems. The reason for this is not clear, although the findings are again consistent with those of Skranes et al. (2007), who found no associations between FA values and either balance or aiming and catching skills. A limitation of the present study is that we did not collect physical activity history data. However, the better balance skills in the girls and the better aiming and catching skills in the boys may reflect that these are the types of motor skills that they prefer and therefore practise the most. Current participation rate statistics for Australian children in this age group support this hypothesis (Sutton, 2009; Pink, 2011). This suggests that, while preterm birth makes acquisition of these skills more difficult, this could potentially be ameliorated if the child is assisted to learn and regularly practises the skills.

Fetal growth effects on corticomotor excitability

Few studies of neurodevelopmental outcome after preterm birth examine GA and BW% separately, although preterm birth is frequently accompanied by varying degrees of intrauterine growth restriction. The finding that suboptimal fetal growth (i.e. BW%) adversely affects only right corticomotor excitability is consistent with previous similar findings by us in young adults (Pitcher et al. 2009) and anatomical findings by others in adolescents (Martinussen et al. 2005). The mechanisms underlying the association between BW% and cortical development of the right hemisphere are not yet clear. However, our results show that GA and BW% can have different effects on neurodevelopment, and failure to differentiate these effects may contribute to the difficulties inherent in predicting the possible neurodevelopmental outcome for a preterm infant. All children born at ≤27 weeks GA who met the study entry criteria had a BW% greater than 70% and were less likely to have significant neurological and/or sensorineural impairments than their lower BW% peers. We were not able to determine whether this related to the reasons for their preterm birth (i.e. cervical insufficiency vs. maternal infection).

Preterm birth and the development of cerebral lateralization

Low GA was associated with a reduction in cerebral lateralization that would normally favour left hemisphere dominance. However, while all of the left handed children (or those who showed no hand preference) were born at ≤38 weeks GA, the prevalence of left handedness/no preference in the study population (17%) was only fractionally higher than that reported in the general Australian population (12–15%; Medland et al. 2009). Our findings indicate that development of corticomotor excitability in both hemispheres is important for the development of strong motor lateralization. The reason for the greater influence of the left hemisphere is not clear. However, it has previously been shown in vivo that the maturation and gyrification of the left hemisphere is delayed with respect to the right, that this right-greater-than-left hemispheric asymmetry is evident in preterm infants at birth and is predictive of their motor and neurobehavioural scores at term-equivalent age (Dubois et al. 2008a,b). Twin studies suggest that left hemisphere development is under less genetic control than the right and is more vulnerable to an adverse environment in utero and in early life (Geschwind & Galaburda, 1985; Geschwind et al. 2002). Our data are suggestive of two groups, one group in which preterm birth has no effect on lateralization and a second group in which reductions in GA (or an associated pathology) are associated with reduced lateralization towards the left hemisphere that persists into late childhood. However, sample sizes in the tens of thousands are required to discriminate between genetic and environmental/pathological influences on handedness (Medland et al. 2009), and the present study was underpowered to explain why some preterm children experience less hemispheric lateralization than others. Hemispheric motor lateralization and language lateralization are correlated (Pujol et al. 1999; Knecht et al. 2000), and the failure to develop a dominant hemisphere significantly increases the risk of difficulties with language, reading and speech (Orton, 1966; Annett, 1975). Given that up to 50% of preterm children will experience problems with motor function, language, reading and/or speech by school age (Marlow et al. 1993; Bracewell & Marlow, 2002; Chyi et al. 2008), determining which factor(s) associated with preterm birth leads to abnormal or reduced development of cerebral lateralization would seem to be a major priority for both early identification and intervention of those children at most risk.

Study limitations

There are a number of limitations of this study. Three have already been mentioned, namely the lack of MRI data to confirm the lesion status of our subjects, the lack of physical activity history data and the limitations of the SICI/ICF data. While detailed perinatal medical histories for the most preterm children were available, there was often little information regarding the late preterm children and the circumstances leading to their preterm birth, as many were never admitted to receive neonatal care. Furthermore, we chose to study across the range of survivable gestations, and the samples size of each of the gestation groups precluded robust within-group analyses of the influence of possible perinatal medical factors affecting neurodevelopmental outcomes, even where these data were available. Future studies that are either larger or that concentrate on a more defined GA range will be required to elucidate the influence of these pregnancy-specific and perinatal factors on postnatal neurophysiological and functional motor outcomes more fully. Lastly, the study participant age group corresponds with the onset and/or early stages of puberty, and changes in sex steroid levels are known to modulate corticomotor excitability and brain maturation (Smith et al. 1999; Herting et al. 2012). We did not determine pubertal stage in the participants or control for females’ cycle stage (if present) at cortical assessment, which may have contributed to data variability.

Conclusions

This study has provided the first neurophysiological evidence that preterm birth is associated with reduced motor cortical excitability that remains evident in late childhood. Functionally, this reduced excitability appears to affect mainly the development of dextrous use of the hands and to reduce hemispheric lateralization. Perhaps most importantly, the findings suggest that, in addition to GA, BW% and other physiological factors, the socio-economic and psychosocial characteristics of the home environment and family a preterm newborn goes home to are critical factors in predicting later motor outcomes. Larger and prospective studies are required to elucidate the influence of specific preterm birth aetiologies on these outcomes, to confirm whether intracortical inhibitory and facilitatory networks are altered and to determine whether the cortical excitability reductions ‘catch up’ later in adolescence or persist into adulthood.

Glossary

%BF

percentage body fat

BW%

birth weight centile

DTI

diffusion tensor imaging

FA

fractional anisotropy

GA

completed weeks gestational age

ICF

intracortical facilitation

IRSD

Index of Relative Social Disadvantage

ISI

interstimulus interval

MABC2

Movement Assessment Battery for Children, version 2

MEP

motor-evoked potential

MRI

magnetic resonance imaging

rMT

resting motor threshold

SICI

short-interval intracortical inhibition

TMS

transcranial magnetic stimulation

Author contributions

J.B.P. conceived the study. J.B.P., N.R.B., T.J.N., M.C.R., R.R.H. and J.S.R. contributed to the design of the experiments. J.B.P., L.A.S., J.L.D., R.D.H., N.R.B. and R.R.H. collected and analysed the data. J.B.P., L.A.S., J.L.D., N.R.B., M.C.R., T.J.N., J.S.R. and R.R.H. contributed to the drafting and revisions of the manuscript.