OUP user menu

Limitations on the developing preterm brain: impact of periventricular white matter lesions on brain connectivity and cognition

Marina A. Pavlova, Ingeborg Krägeloh-Mann
DOI: http://dx.doi.org/10.1093/brain/aws334 998-1011 First published online: 1 April 2013


Brain lesions to the white matter in peritrigonal regions, periventricular leukomalacia, in children who were born prematurely represent an important model for studying limitations on brain development. The lesional pattern is of early origin and bilateral, that constrains the compensatory potential of the brain. We suggest that (i) topography and severity of periventricular lesions may have a long-term predictive value for cognitive and social capabilities in preterm birth survivors; and (ii) periventricular lesions may impact cognitive and social functions by affecting brain connectivity, and thereby, the dissociable neural networks underpinning these functions. A further pathway to explore is the relationship between cerebral palsy and cognitive outcome. Restrictions caused by motor disability may affect active exploration of surrounding and social participation that may in turn differentially impinge on cognitive development and social cognition. As an outline for future research, we underscore sex differences, as the sex of a preterm newborn may shape the mechanisms by which the developing brain is affected.

  • preterm birth
  • brain development
  • cognitive functions
  • motor disability
  • brain connectivity


Current progress in pre- and neonatal intensive care leads to an increase in the survival rate of very premature infants (Platt et al., 2007; Ment et al., 2009; Volpe, 2009; Baron and Rey-Casserly, 2010; Doyle and Anderson, 2010; Sellier et al., 2010). Quality of survival, therefore, has become a major concern with clear social relevance (Allin, 2010). Brain imaging indicates that cognitive deficits related to structural and functional brain abnormalities as sequelae of preterm birth persist into adolescence and adulthood (Stewart et al., 1999; Nosarti et al., 2008, 2011; Parker et al., 2008; Hallin et al., 2010; Eikenes et al., 2011; Mullen et al., 2011; Tanskanen et al., 2011). The prevalence of children born preterm (<37 weeks of gestation) is ∼6–11% in western European countries reaching 12–13% in North America (Allen, 2008; Goldenberg et al., 2008; Beck et al., 2010). This means that up to one in seven children in a classroom of an ordinary mainstream school may have a history of prematurity and exhibit learning disabilities and daily life cognitive and behavioural deficits. This prevalence is much higher compared with the prevalence of other neurodevelopmental disorders. For example, the prevalence of autistic spectrum disorders ranges from 0.6% (Newschaffer et al., 2007) to 1.9% in most recent studies (Kim et al., 2011). Among preterm children, up to 30–50% may suffer specific early damage to peritrigonal brain regions of the cerebral white matter, periventricular leukomalacia, the dominant form of brain injury in survivors of premature birth (Olsén et al., 1997; Krägeloh-Mann et al., 1999; Skranes et al., 2005). To date, there are only few population-based studies on the prevalence of periventricular leukomalacia. This lesional pattern is commonly defined by a focal and diffuse component (Volpe, 2001; Back, 2006). The focal component as a hallmark of the disease is best identified post-natally by MRI (Barkovich, 2005). Work in the past decade has shown that periventricular leukomalacia may be accompanied by neuronal abnormalities affecting not only the cerebral white matter but also the thalamus, basal ganglia, cerebral cortex, brainstem and cerebellum that together constitute a complex amalgam of ‘encephalopathy of prematurity’ (Volpe, 2009). In the present analysis, we focus on periventricular leukomalacia as a distinctive form of cerebral white matter injury. Being a result of necrosis of fibres around the lateral ventricles in the peritrigonal area, periventricular leukomalacia is characterized by periventricular gliosis in the white matter with or without tissue loss and secondary ventricular dilatation (Fig. 1; Baker et al., 1988; Krägeloh-Mann et al., 1999). In individuals born preterm, the pathways that connect subcortical structures with cortical areas may be affected, in particular, the posterior thalamic radiation, and cortico-cortical connectivity as suggested by diffusion tensor imaging (DTI, Fig. 2; Hoon et al., 2002; Nagae et al., 2007; de Bruїne et al., 2011; Nagasunder et al., 2011) and resting state functional connectivity MRI (Smyser et al., 2010; Dinomais et al., 2012).

Figure 1

Periventricular leukomalacia as depicted by MRI at a later stage after completed myelination in a 6-year-old male born at 32 weeks of gestation. Axial T2-weighted (T2w), FLAIR and T1-weighted (T1w) images at the level of the centrum semiovale cutting the upper part of the ventricles (top row) and at the level of the lateral ventricles (bottom row) show the characteristic gliosis as hyperintense high signal on the T2-weighted image and hypointense low signal on the T1-weighted image. The gliosis is best seen on FLAIR images, where the periventricular hyperintensities contrast with the hypointense CSF signal. The top row illustrates gliosis, and the bottom row additional tissue loss (mild ventricular dilatation with irregularly extended borders).

Figure 2

3D reconstruction of the posterior thalamic radiation (PTR) as revealed by DTI. Fibre tracts are shown in yellow for the typically developing control (left), and in red for the patient with periventricular leukomalacia (right). The posterior thalamic radiation is identified in a coronal slice at the level of the splenium of corpus callosum. The posterior thalamic radiation, which projects through the retrolenticular division of internal capsule, is well visible in the control but is severely impaired in the patient with periventricular leukomalacia affecting connectivity between the pulvinar and the lateral geniculate nucleus to posterior cortices and reciprocal corticothalamic projections. From Hoon et al. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology 2002; 59: 752–6. Copyright © 2002 by AAN Enterprises, Inc. with permission of Wolters Kluwer/Lippincott, Williams & Wilkins.

Children born preterm may exhibit subtle to profound neurological and cognitive impairments. Common and well-recognized deficits are motor disability in the form of cerebral palsy (Krägeloh-Mann et al., 1999; Bax et al., 2006; Platt et al., 2007; Sellier et al., 2010; de Vries et al., 2011; Lee et al., 2011) and cognitive impairments ranging from visual perception and attention to social cognitive functions (e.g. Atkinson and Braddick, 2007; Birtles et al., 2007; Fazzi et al., 2009, 2012; Boot et al., 2010; Bottcher, 2010; Cioni et al., 2011; Ortibus et al., 2011). It is not well understood whether—and, if so, how—cognitive impairments in former preterms are related to the nature and distribution of the brain injury. Periventricular leukomalacia extent and topography are associated with neurodevelopmental outcome (Fazzi et al., 1994; Serdaroglu et al., 2004). Visuoperceptual deficits are reported to be associated with ventricular enlargement and white matter reduction in the parieto-occipital regions (e.g. Koeda ans Takeshita, 1992; Goto et al., 1994; Ito et al., 1996; Skranes et al., 1997; Fazzi et al., 2004; van den Hout et al., 2004), whereas deficits in attention and execitive functions are associated with more anterior frontal injury (Schatz et al., 2001). Regional white matter volumes in the sensorimotor and midtemporal regions correlate with measures of neurodevelopmental outcome (Peterson et al., 2003). Current progress in neuroimaging helps to define predictive biomarkers of neurodevelopmental outcome in the preterm brain (Dammann and Leviton, 2006; Woodward et al., 2006; Ment et al., 2009). Within standard medical practice, if preterm children do not exhibit noticeable impairments in motor behaviour and low-level visual functions, structural MRI examinations are not performed. As such, lesions may remain unknown to parents and health care professionals.

The clinical model of periventricular leukomalacia as a distinctive form of cerebral white matter injury is important for understanding cognitive and social functioning in typical and atypical development because (i) compared with lesions acquired later in life, the model deals with brain damage of early origin (early-to-middle third trimester of pregnancy) that prevents proper functioning of the brain networks from the very beginning in life; (ii) periventricular leukomalacia is a lesional pattern of high homogeneity in terms of pathogenesis, timing and structural topography; and (iii) periventricular leukomalacia represents a bilateral (often symmetrical) pattern of lesions that constrains the compensatory capability of the brain, because it usually affects both hemispheres and, therefore, as compared with unilateral lesions, limits the developing brain for reorganization to preserve functions at risk.

Another important deficit of individuals born preterm is bilateral leg-dominated spastic cerebral palsy (upper extremities are also often affected), although the incidence and severity of cerebral palsy in preterm infants has been decreasing (Platt et al., 2007; Sellier et al., 2010; van Haastert et al., 2011). Motor disorders occur from the very beginning in life and prevent normal experience in locomotion. This provides a unique opportunity to examine the limitations in active motor exploration on cognitive and social development.

Recent analyses of the neurodevelopmental outcome in preterm birth survivors do not always disentangle two of the major factors that might affect proper development of particular cognitive functions, namely (i) limitations related to occurrence, topography and extent of lesions; and (ii) restrictions caused by motor disability affecting active exploration of surrounding and social participation that may in turn shape cognitive development. As experimental data have often been collapsed across preterms with and without brain lesions, impact of distinctive white matter injury on cognitive functions is difficult to evaluate. The aim of this work is twofold. First, we analyse findings on cognitive functions that are of importance for daily life adaptive behaviour in individuals who were born preterm. Second, we consider interrelations between motor disability and cognition. Our analysis underscores further ways in investigation of long-lasting consequences of preterm delivery and limitations on the developing brain.

Impact of periventricular lesions on cognition

This part of the review is aimed at analysis of findings uncovering the impact of periventricular leukomalacia in preterm birth survivors on cognitive functions that are of importance for daily life adaptive behaviour. In particular, we focus on visual perception of body motion and social cognition. We ask whether topography and severity of lesions may serve as long-term predictors for cognitive and social outcome, and suggest that periventricular leukomalacia may impinge on cognitive and social functions by affecting structural and functional brain connectivity of the dissociable neural networks underpinning these functions.

Visual perception of body motion

Visual perception of body motion of other people is of value for successful daily-life activities such as safe self-locomotion in a crowd and, in particular, for adaptive social behaviour and non-verbal communication. Body (or biological) motion perception has been studied by using ‘point-light displays’ in which dots are placed on the main joints and head of an otherwise invisible human body (Fig. 3). The main advantage of this approach is that it helps to separate information revealed by motion from other cues (shape, colour). The visual sensitivity to point-light body motion is reported to emerge early in development (Fox and McDaniel, 1982; Simion et al., 2008; Klin et al., 2009; Bardi et al., 2011; Falck-Ytter et al., 2011) and is relatively spared in the elderly (Norman et al., 2004; Billino et al., 2008; Pilz et al., 2010; Legault et al., 2012). Alterations in biological motion processing with age appears to be unaffected by gender (Billino et al., 2008). Young children increasingly improve in their ability to spontaneously recognize point-light body motion, with ceiling levels of performance achieved by the age of 5 (Pavlova et al., 2001). In typically developing children, the visual perception of body motion and daily life social cognitive abilities (understanding of intentions, dispositions and emotions of others, body language reading) appear to be tightly linked and, therefore, the perceptual system for analysing body motion might also be functionally integrated with development of social functions (Pavlova, 2012).

Figure 3

Static representation of point-light body motion. Left: The locations of dots on the main joints of a walking figure. For illustrative purposes, an outline of the walking figure is shown. Participants are presented with a set of bright dots against a black background. Four consecutive static frames taken from a point-light biological motion display with the action progressing from left to right. From Pavlova et al. Recognition of point-light biological motion displays by young children. Perception 2001; 30: 925–33, with permission of the publisher, Pion Ltd., London, www.envplan.com.

In a series of psychophysical and brain imaging studies, visual perception of body motion and other cognitive functions were investigated in a cohort of adolescents, who were born prematurely with signs of periventricular leukomalacia on an MRI scan (without additional detectable cortical or subcortical brain abnormalities). Receiver operating characteristic analysis reveals that patients with severe and mild periventricular leukomalacia exhibit compromised ability for visual processing of body motion; their visual sensitivity is lower than in term-born and low-risk preterm-born peers with MRI without detectable brain abnormalities (Pavlova et al., 2006b). The severity of this impairment is related to topography and volumetric lesion extent. Neither frontal nor temporal periventricular leukomalacia relates to deficient body motion processing, whereas the visual sensitivity to body motion decreases with an increase in the volumetric periventricular leukomalacia extent in the parieto-occipital region. The specificity of this relationship is also confirmed by the lack of a link between the overall white matter volume and the visual sensitivity to body motion.

In preterm children aged 5–9 years, visual sensitivity to camouflaged body motion is not associated with gestational age and birth weight (Taylor et al., 2009). In these patients, different vulnerability of biological motion, global motion and global form perception suggests that impairments on these tasks are connected with dissociable neural networks. Detection thresholds for masked body motion are also not related to performance of preterm children on a representational momentum task (Taylor and Jakobson, 2010). It remains unclear whether deficits in body motion perception are related to the occurrence of lesions (detected with neonatal cranial ultrasound that may not be sensitive enough) and to type and topography of brain injury. Signal-to-noise thresholds for perceiving the direction of locomotion of a camouflaged point-light walking figure are reported to be higher in children with periventricular leukomalacia (Guzzetta et al., 2007). In low birth-weight preterm children aged 5–6 years, deficits in motion-defined form recognition relate to the presence of brain injury and retinopathy of prematurity, i.e. to premature birth complications, rather than to a history of prematurity per se (Jakobson et al., 2006). On a segmented motion (with form information) and static form recognition task, 10-year-old full-term and low-risk preterm subjects without lesions do not differ in performance, but visual sensitivity of preterms with periventricular leukomalacia is lower (Guzzetta et al., 2009). Yet, on translational and rotational motion (pure motion) tasks, sensitivity of both preterms with and without lesions is reduced. This suggests that occurrence of lesions and prematurity may differentially affect the networks engaged in visual processing of form and motion.

Numerous imaging, mainly functional MRI, studies in the typically developing adult brain and neuropsychological lesion data indicate that the network specialized for body motion processing involves mainly the right parietotemporal junction and fusiform gyrus, portions of the parietal and frontal cortices and subcortical structures such as the amygdala (Grossman et al., 2000; Vaina et al., 2001; Puce and Perrett, 2003; Saygin et al., 2004; Saygin, 2007; Krakowski et al., 2011; van Kemenade et al., 2012). The right posterior superior temporal sulcus is considered a hub for visual processing of meaningful and expressive human body motion (Grossman and Blake, 2002; Grossman et al., 2004; Saygin et al., 2004, Saygin 2007; Pelphrey et al., 2005; Gobbini et al., 2007; Herrington et al., 2011; Kaiser et al., 2011). Veridical perception of body motion depends on intact communication within this distributed network. The neural circuitry subserving visual processing of biological motion is not only limited to cortico-cortical connections but also engages brain structures beyond the cerebral cortex such as the cerebellum (Grossman et al., 2000; Vaina et al., 2001; Sokolov et al., 2012). Furthermore, for understanding proper functioning of neural circuits underpinning biological motion processing and its pathology, one has to consider the changes in brain activation unfolding over time.

By affecting structural and functional brain connectivity, periventricular leukomalacia may lead to disintegration of neural communication supporting visual body motion processing. In patients with periventricular leukomalacia, stimulus-specific alterations are found in the evoked magnetoencephalographic (MEG) response to unmasked and camouflaged body motion (Pavlova et al., 2006a, 2009a). In the adult brain, visually perceived body motion elicits consecutive peaks of oscillatory MEG activity over the left occipital (100 ms), bilateral parietal (130 ms) and right temporal (170 ms) cortices (Pavlova et al., 2004). In typically developing adolescents, the spectral amplitude of the oscillatory MEG response to body motion peaks at a latency of 170 ms over the right parietotemporal cortex (Fig. 4; Pavlova et al., 2007a). This increase is absent in patients with periventricular leukomalacia. Instead, peaks in the spectral amplitude of the oscillatory response of lower frequency occur later, at a latency of 290 ms over the left temporal region. The posterior thalamocortical fibres are reported to be longer, thinner and less numerous in patients with periventricular leukomalacia, presumably because of ventricular enlargement along with gliosis in the white matter (Hoon et al., 2002; Nagae et al., 2007; Nagasunder et al., 2011). This might explain a delayed latency of the MEG response to body motion in patients with periventricular leukomalacia. More numerous and widely distributed posterior thalamocortical fibres on the left side reported in the healthy brain (Thomas et al., 2005) and larger left-sided white matter volumes in the preterm brain (Peterson et al., 2003) might account for left-hemispheric lateralization of this peak. This assumption, however, needs additional experimental proof.

Figure 4

Grand average time-frequency representations of the difference in the MEG oscillatory response to point-light body motion in adolescents. Separate plots are given for the left hemispheric response (left), and for the right hemispheric response (right), and for (A) healthy control subjects (HC) and (B) patients with periventricular leukomalacia (PVL); (C) difference between control subjects and patients with periventricular leukomalacia (diff.). In control subjects, the peak in time-frequency amplitude of the oscillatory response was observed at a latency of 170 ms over the right parietotemporal cortex. In patients with periventricular leukomalacia, the peak occurred later, at a latency of 290 ms over the left temporal cortex. Reprinted from Pavlova et al. Oscillatory MEG response to human locomotion is modulated by periventricular lesions. Neuroimage 2007a; 35: 1256–63. Copyright © 2007 Elsevier Inc. with permission of the puiblisher, Elsevier.

In a nutshell, the findings suggest that disturbances in structural brain connectivity caused by periventricular brain injury may lead to reduced visual sensitivity to body motion with alterations in functional MEG cortical activity. Despite lifelong daily visual experience with human locomotion, preterm-born adolescents with lesions exhibit deficient processing of body motion. In contrast, long-term visual deprivation does not appear to affect body motion processing. After 40 years of blindness acquired at 3.5 years of age, patient M.M. is reported to show normal visual sensitivity to body motion (Fine et al., 2003). Furthermore, individuals (aged 8–29 years) deprived of visual experience during infancy by dense bilateral congenital cataracts exhibit substantial deficits in the perception of global motion and form, but have intact point-light biological motion processing (Hadad et al., 2012). This outcome points to specific boundaries on the spontaneous functional compensatory potential of the developing brain.

Social cognition

One may extend the analysis to ask to what extent periventricular lesions may affect the ability to perceive and understand social interaction and characteristics of others. There is the lack of experimental data in this domain. Performance level of patients with periventricular leukomalacia is lower than that of healthy control subjects on the event arrangement task administered in the course of neuropsychological examination (WISC III—Wechsler Intelligence Scale for Children, Third edition), whereas former preterms without periventricular leukomalacia perform much like children born at term (Pavlova et al., 2008). On this task, participants have to rearrange a set of cards into a sequence depicting an event in a comic-strip fashion. For successful performance, one needs to reflect the core of the story, which is often based on veridical perception of intentions of the characters involved in a particular event (Baron-Cohen et al., 1986; Sarfati et al., 1997). The extent of lesions over the right temporal region may serve as a predictor of the severity of this impairment, although performance on the event arrangement task is also inversely related to the extent of parieto-occipital lesions in both hemispheres. Bearing in mind that a wealth of brain imaging and neuropsychological data point to the right temporal lobe as a contributor to both visual processing of body motion and social cognition (Allison et al., 2000; Puce and Perrett, 2003; Pelphrey et al., 2004; Gobbini et al., 2007), one can assume that lesions disrupt subcortico-cortical and cortico-cortical connections with the right temporal cortex leading to difficulties in body language reading (perception and understanding of social properties such as intentions, desires, emotions and dispositions of others). This suggests that the neural system for analysing body motion might be functionally integrated with development of social cognition. This assumption, however, calls for further research.

Among numerous open questions concerning deficits in social cognition is whether children born preterm exhibit difficulties in perception and understanding of social properties through observing body movements and dynamic faces, though some difficulties are reported in recognition of static faces (Fazzi et al., 2009). Preterm infants seem to process faces differently than do their full-term peers (Field, 1979; Rose et al., 2002).

The other intriguing question is whether brain mechanisms responsible for deficient social cognition are similar in preterm children and individuals with autistic spectrum disorders. Compared with children born at term, those born preterm appear to be at greater risk of being considered autistic (Indredavik et al., 2008; Limperopoulos et al., 2008; Schendel and Bhasin, 2008; Moore et al., 2012; see also meta-analysis by Gardener et al., 2011). Deficits in visual social perception in autism are associated with the right superior temporal sulcus (Saitovitch et al., 2012), and abnormalities are characterized by decreased grey matter volume, rest hypoperfusion and abnormal functional activation during performance of social cognition tasks. It appears extremely unlikely that deficiencies in social cognition can be explained by abnormalities in a single brain region, but rather by alterations in structural and functional brain connectivity. In accord with this, structural MRI provides evidence for poorly organized white matter in autism (McAlonan et al., 2005; Cheng et al., 2010). DTI in autistic individuals indicates disrupted white matter connections between regions implicated in social cognition including the right temporal cortex (Barnea-Goraly et al., 2004, 2010; Noriuchi et al., 2010; Jou et al., 2011). Recent data show that compared with typically developing children and unaffected siblings, autistic children exhibit reduced functional MRI response to body motion in several brain areas including the right posterior superior temporal sulcus and bilateral fusiform gyri (Kaiser et al., 2010; see also Koldewyn et al., 2011). Moreover, in agreement with the assumption that biological motion processing may serve as a hallmark of social cognition (Pavlova, 2012), the ability for body language reading (revealing emotions from point-light body motion) is related to a more basic capability for discrimination between biological and scrambled motion in individuals with autistic disorders and typically developing adults (Nackaerts et al., 2012). In very low birth-weight preterm-born adolescents, social deficits, as assessed by the autism spectrum screening questionnaire, are associated with disruptions of several white matter tracts (Skranes et al., 2007). It appears, however, that preterm children may exhibit considerably milder forms of social cognition deficits than individuals with autistic spectrum disorders. Clarification of this issue requires future work.

Way Finding

A further question is whether such an essential ingredient of adaptive daily-life behaviour as visual navigation that reguires intact visuospatial abilities may be affected in former preterms. Patients with periventricular leukomalacia are often reported by their care providers to have specific difficulties in way finding, and they easily get lost even in familiar surroundings and crowded places (Jacobson et al., 1996, 1998; Dutton et al., 2006; Dutton, 2009). Most families report specific problems in way finding when asked about the patients’ difficulties in daily life and in school. For example, the father of one of the patients (16-year-old at examination), a mainstream school principal, reported that for many years his daughter with verbal IQ in the normal range had been unable to find the way from one classroom to another without external support and had also had serious difficulties in way finding in their local place. Visual navigation ability is inversely linked to the volumetric periventricular leukomalacia extent over the parietal and, in particular, frontal regions of the right hemisphere, and as indicated by a stepwise multiple regression analysis, the extent of right frontal lesions serves as the best predictor of deficits in a paper-and-pencil labyrinth test as a part of the WISC III (Pavlova et al., 2007b). Extremely preterm birth (less than 25 weeks of gestation) results in route finding deficits in 6-years-olds as assessed by NEPSY, a developmental NEuroPSYchological Assessment (Marlow et al., 2007). Very low-birth-weight preterm adolescents (performance and normal IQ within the normal range) with hippocampal volume reduction fail in retracing a route around the room and in recalling cues critical for navigation as tested by the Rivermead Behavioral Memory Test (Isaacs et al., 2003). Given that the brain network underpinning visual navigation involves the right prefrontal lobe and hippocampus (Maguire et al., 1998; Grön et al., 2000; Wolbers and Hegarty, 2010), one can assume that more anterior periventricular lesions affect connectivity between the right hippocampus and frontal cortices leading to disintegration of the neural networks engaged in visual navigation. In accord with this, in healthy adults, high performance in a virtual navigation task is associated with a larger volume of prefrontal grey and white matter (Moffat et al., 2007). Further steps towards uncovering the neuropathology of visual navigation would be investigation in natural environments and an analysis of functional brain connectivity during way finding in virtual mazes.


The ability for mental calculation represents a fundamental prerequisite for development of intelligence, which is predictive for educational and occupational success in life (Dehaene et al., 2004; Butterworth, 2005). Many individuals with calculation difficulties are premature birth survivors. There are indications that children born prematurely may exhibit specific deficits in mental arithmetic (Isaacs et al., 2001; Skranes et al., 2007; Keller-Margulis et al., 2011). DTI in very low birth-weight preterm adolescents reveals that raw arithmetic scores as measured by WISC III are correlated with fibre tract integrity (as reflected by fractional anisotropy) in the right superior, and the left middle and inferior fascicles (Skranes et al., 2007). The brain mechanisms of calculation deficit and, in particular, the networks that can be affected by premature birth are largely unknown. It is important to clarify whether calculation difficulties are associated with periventricular injury or with the factor of prematurity per se. Mental calculation abilities in patients with periventricular leukomalacia are unrelated to volumetric extent and topography of lesions (Pavlova et al., 2009b). The lack of relationship between calculation abilities and periventricular leukomalacia extent and topography suggests the existence of a topographically restricted neural substrate that may serve as a keystone for mental calculation. Brain imaging findings reveal specific brain structures engaged in mental calculation. In typically developing adults and children, the horizontal segment of the bilateral intraparietal sulcus mesial to the supramarginal gyrus is systematically activated in calculation tasks and is believed to host a central amodal and language-independent representation of numerical quantity (Simon et al., 2002; Cantlon et al., 2006; Piazza et al., 2007). The parietal cortex becomes more specialized for arithmetic tasks with age (Davis et al., 2009) and training (Grabner et al., 2007). Voxel-based morphometry indicates that preterm adolescents with and without calculation deficits differ in grey matter volume in a relatively small brain area within the left intraparietal sulcus (Isaacs et al., 2001). Presumably, periventricular brain injury does not substantially affect the connectivity of this region with other brain structures engaged in mental calculation. Future functional brain imaging research, in particular, by using techniques providing for high temporal resolution, should shed light on limitations affecting the complex process of mental calculation in the preterm brain.

Overall, it is becoming apparent that periventricular leukomalacia topography and severity specifically affect structural and functional brain connectivity later in life, and thereby, the distinct and dissociable neural networks underpinning particular cognitive functions (Table 1). The specificity of the relationship between lesion extent and topography and particular cognitive functions is proved by the absence of a link between the whole-brain white matter volume and particular cognitive deficits. Impact of lesions topography on distinct cognitive and social abilities may also explain the absence of a link between overall cerebral white matter reduction and specific cognitive outcome in former preterms (Skranes et al., 2008; but cf. Soria-Pastor et al., 2008 for association between processing speed with overall white matter reduction), although the white matter reduction is reported to be associated with full-scale and performance IQ, general neurodevelopmental and cognitive outcome, and educational and behavioural difficulties (Yung et al., 2007; Soria-Pastor et al., 2008; Boardman et al., 2010; Northam et al., 2011).

View this table:
Table 1

Topography of periventricular leukomalacia and cerebral palsy associated with cognitive deficits

Cognitive functionNeural networkPVL topographyMotor disorder
Visual perception of body motionPortions of parietal and frontal cortices, right parietotemporal junction, right or bilateral pSTS, right or bilateral fusiform gyrus, left lateral cerebellum (e.g. Grossman and Blake, 2002; Puce and Perrett, 2003; Saygin et al., 2004; Pelphrey et al., 2005; Gobbini et al., 2007; Saygin, 2007; Kaiser et al., 2010; Herrington et al., 2011; Krakowski et al., 2011; Sokolov et al., 2012)Right parieto-occipitalNo reported association
Visual social cognitionRight posterior temporal cortex, in particular pSTS, parietotemporal junction, fusiform gyri, amygdala, orbitofrontal cortices (e.g. Allison et al., 2000; Pelphrey et al., 2004; Gobbini et al., 2007; Pavlova et al., 2010)Right temporalNo reported association
Way findingRight prefrontal cortex, right hippocampus (e.g. Maguire et al., 1998; Grön et al., 2000; Wolbers and Hegarty, 2010)Right frontal and parieto-occipitalLeg-dominated cerebral palsy
CalculationBilateral or left intraparietal sulcus (e.g. Isaacs et al., 2001; Simon et al., 2002; Dehaene et al., 2004; Cantlon et al., 2006; Grabner et al., 2007; Piazza et al., 2007; Davis et al., 2009)No reported associationControversial
  • pSTS = posterior superior temporal sulcus; PVL = periventricular leukomalacia.

Impact of cerebral palsy on cognition

Former preterms with periventricular leukomalacia can exhibit signs of bilateral, primarily leg-dominated spastic cerebral palsy, which is conventionally associated with damage to the corticospinal tract. Motor impairments of all four extremities correlate with the severity of damage to the respective portion of the pyramidal tract in the contralateral brain hemisphere (Staudt et al., 2003), and as indicated by diffusion-weighted MRI, the presence of diffusivity in the corticospinal tract is associated with motor impairment (Kidokoro et al., 2008). Most recent DTI data show that fractional anisotropy within the bilateral corticospinal tracts and within the posterior body and isthmus of the corpus callosum significantly correlate with motor dysfunction (Lee et al., 2011). Motor disorders in this population occur from the very beginning in life, largely preventing normal experience in locomotion. It is hypothesized that cerebral palsy represents a specific constraint that might restrict cognitive functioning and social participation (Jenks et al., 2007; Bottcher, 2010; van Rooijen et al., 2011). However, there is a lack of insight into origins of this impact.

Production and perception of body motion

Recent theoretical reasoning and experimental evidence suggest that production and perception of body motion may be closely linked and share a common representational neural network (Rizzolatti and Sinigaglia, 2010). Indeed, common motor programmes between observers and actors as well as motor training are reported to facilitate visual body motion perception (Casile and Giese, 2006; Bidet-Ildei et al., 2010; Calvo-Merino et al., 2010). Hemiplegic patients are compromised in recognition of point-light gestures portrayed by the affected contralesional arm (Serino et al., 2009), and paraplegia patients are impaired in detection and discrimination of direction of point-light human locomotion even a few months after injury (Arrighi et al., 2011). Reduction in sensitivity in these patients, however, may be caused by specific or general impairments of the central nervous system. In adolescents with periventricular leukomalacia, visual sensitivity to body motion is not related to lifelong restrictions in locomotion ranging from mild impairments in walking pattern to a complete walking disability (Pavlova et al., 2003). Therefore, locomotion experience does not appear to be a necessary prerequisite for the visual analysis of body motion. As human locomotion is of fundamental evolutionary and functional significance, a hard-wired schema for perception of body motion might be inherent for the typically developing brain. This is in general agreement with developmental studies indicating that visual tuning to human locomotion either appears to be intrinsic or requires a rapid post-natal period of growth rather than being acquired through or closely connected with motor self-experience (Fox and McDaniel, 1982; Klin et al., 2009; Bardi et al., 2011).

Production of actions is also suggested to be closely tied to understanding of social properties such as intentions and dispositions of others conveyed by their movements (Rizzolatti and Fabbri-Destro, 2008). Actions that are difficult or impossible to perform might also yield deficient social interpretation. In adolescents with periventricular leukomalacia, no relationship has been found between the severity of cerebral palsy and performance on a task requiring understanding of social interaction represented in a series of static snapshots (Pavlova et al., 2008). Self-estimates of the quality of life in 8- to 12-year-old preterm children with cerebral palsy are similar to those of healthy peers in all domains except schooling (Dickinson et al., 2007). Is there a connection between limitations in social participation and deficits in visual social cognition? This question remains open.

Way finding and motor disability

Limitations in active exploration of surroundings in patients with motor disability may prevent proper development of visual navigation. This agrees well with findings demonstrating that the ability to navigate substantially improves with practice and might be compromised in individuals with restrictions in active spatial exploration. Similar to insects and rodents, humans continuously update their spatial representation of the environment as they move (Wang and Spelke, 2000). The ability for visual orientation and navigation appears to be associated with the severity of leg-dominated cerebral palsy, which explains a significant percentage of variance in this ability (Pavlova et al., 2007b). Ex-preterm children aged 6 years without cerebral palsy also may exhibit deficits in route finding (Marlow et al., 2007).

Motor disability and mental arithmetic

Although it is reported that patients with cerebral palsy can exhibit difficulties in numeracy and mental arithmetic (Arp and Fagard, 2005; Jenks et al., 2007, 2009a, b), the issue of whether motor disability or other confounding factors affect calculation ability remains unclear. One can assume that deficits in mental calculation may be explained within the framework of embodied cognition suggesting a close relationship between the mind and a physical body that interacts with the outer world (van Rooijen et al., 2011). Yet, experimental findings indicate that arithmetic difficulties in patients with cerebral palsy are more likely to be mediated by deficits in working memory and other cognitive functions (Arp and Fagard, 2005; Jenks et al., 2007, 2009a, b). The relation of low arithmetic scores to deprived left-hand function is interpreted in terms of involvement of right-hemispheric functions in the complex process of mental calculation (Kiessling et al., 1983; Dellatolas et al., 2005; Jenks et al., 2009a, b). Mental calculation abilities appear to be unrelated to the severity of cerebral palsy of upper or lower extremities in patients with periventricular leukomalacia (Pavlova et al., 2009b). In line with this, the poorer arithmetic performance of 7-year-old children with cerebral palsy in special versus mainstream education is not related to unilaterality/bilaterality and severity of motor impairments. Moreover, in a mainstream school, children with cerebral palsy do not substantially differ from healthy peers on arithmetic ability (Jenks et al., 2009b, 2012). Some findings, however, suggest an association between fine motor skills in children with cerebral palsy and arithmetic abilities (van Rooijen et al., 2012). Future research is required to determine whether intact motor abilities represent a requirement for proper development of mental calculation.

Résumé and outline for future research

The outcome of the present analysis suggests that periventricular leukomalacia as a distinctive form of cerebral white matter injury is associated with cognitive and social capabilities. It appears that topography and severity of lesions can be considered as putative long-term predictors for proper development of a particular cognitive function, if they affect connectivity of the dissociable neural network underpinning this function. Specific cognitive and neurodevelopmental deficits associated with disruption of the neuronal networks depend on which particular circuits have been affected. Impact of early brain injury is an ongoing process that may affect structural and functional differentiation of the developing brain. Clarification of possible developmental trajectories of periventricular leukomalacia impact on cognition and brain connectivity by combining structural and functional brain imaging with evaluation of cognitive functions calls for future cross-sectional and, in particular, longitudinal studies in preterm birth survivors. Combination of modern sophisticated neuroimaging uncovering alterations in structural brain connectivity (Ment et al., 2009) with an analysis of functional brain activity in well-defined groups of former preterms represents an important step towards our understanding of limitations on the developing brain. For tapping complex cognitive and social functions underlying adaptive daily life behaviour, future research should implement ecologically valid tools (e.g., using as test moving stimuli instead of static faces and bodies) meeting latest advances in cognitive and social neuroscience. A further pathway to explore is the origins of the relationship between cerebral palsy and cognitive outcome. Restrictions caused by motor disability may affect active exploration of surrounding and social participation that may in turn differently shape cognitive development and social cognition. It appears that in patients with similar lesion extent and topography, and with similar severity of motor disorders, the cognitive and social outcome may be profoundly modulated by attention of care-providers and early training.

Sex differences in cognitive consequences of a premature birth represent one of the essential issues for future research. It is known that males are at a 14–20% higher risk of premature birth (Melamed et al., 2010) and its complications in the brain and cognition (Wood et al., 2005). There is growing evidence for early sexual dimorphism of the brain (Cahill, 2006; Vasileiadis et al., 2009; Nugent and McCarthy, 2011), and recent DTI findings indicate sex differences in the white matter structures of the adult brain (Chou et al., 2011; Menzler et al., 2011). During the first 30 years of life, typically developing females tend to have greater grey matter volume relative to brain size and lower white-to-grey matter ratio than males (Groeschel et al., 2010). In children aged 4.5–18 years, developmental trajectories of brain tissue volumes are predominantly curvilinear in females, but linear in males (Brain Development Cooperative Group, 2012). Cognitive abilities are inversely related to the corpus callosum midsaggital area size in 6- to 18-year-old males, but not in females (Ganjavi et al., 2011).

Preterm males appear to be more vulnerable to white matter injury and more often exhibit signs of cerebral palsy (Johnston and Hagberg, 2007; Platt et al., 2007; Beaino et al., 2010; Bajwa et al., 2011). Male preterm birth survivors show a lower rate of developmental change than girls (van de Weijer-Bergsma et al., 2010). DTI and probabilistic tractography indicate gender differences in language and motor-related fibres in healthy preterm neonates at term-equivalent age (Liu et al., 2011). At 8 years of age, males are particularly vulnerable to the effects of preterm birth on white matter development; preterm males only exhibit significantly reduced white matter compared with term males, whereas white matter volumes are equivalent in preterm and term-born females (Reiss et al., 2004). Compared with term-born male peers, 12-year-old preterm males continue to demonstrate abnormal neurodevelopment in terms of lower white matter volumes in bilateral cingulum, corpus callosum, corticospinal tract, prefrontal cortex, superior and inferior longitudinal fasciculi. Grey matter volumes in the prefrontal cortex, basal ganglia and temporal lobe are also significantly reduced, whereas brain morphology in preterm females does not differ from term-born peers (Kesler et al., 2008). Preterm adolescents exhibit gender-dependent differences in the structural brain correlates of spelling abilities (Scott et al., 2011). In general, it appears that male children exhibit greater vulnerability to preterm birth complications. The findings indicate that sex of a preterm newborn might shape the mechanisms by which the developing brain is affected. However, the neurobiological nature of sex effects on the developing preterm brain remains unknown.


The Else Kröner Fresenius Foundation (Grants P63/2008 and P2010_92); the Werner Reichardt Center for Integrative Neuroscience (CIN, pool project 2009-24), supported by the German Research Council (DFG), the Reinhold Beitlich Foundation to M.A.P., and EU SCPE Net 20081307 to I.K.-M.


The authors acknowledge their collaborators for contributions to work analysed and described here. They appreciate the patients, family members and care providers for their kind cooperation.

diffusion tensor imaging


View Abstract