Reorganization after pre- and perinatal brain lesions

Abstract

The developing human brain can compensate for pre- and perinatally acquired focal lesions more effectively than the adult brain. The mechanisms by which this effective reorganization is achieved vary considerably between different functional systems, reflecting differences in the normal maturation of these systems. In the motor system, descending cortico-spinal motor projections have already reached their spinal target zones at the beginning of the third trimester of pregnancy, with initially bilateral projections from each hemisphere. During normal development, the ipsilateral projections are gradually withdrawn, whereas the contralateral projections persist. When, during this period, a unilateral brain lesion disrupts the cortico-spinal projections of one hemisphere, the ipsilateral projections from the contralesional hemisphere will persist. This allows the contralesional hemisphere to take over motor control over the paretic extremities. Although this mechanism of reorganization is available throughout the pre- and perinatal period, the efficacy of this ipsilateral takeover of motor functions decreases with increasing age at the time of the insult. In the somatosensory system, ascending thalamo-cortical somatosensory projections have not yet reached their cortical target zones at the beginning of the third trimester of pregnancy. Therefore, these projections can still ‘react’ to brain lesions acquired during this period, and can form ‘axonal bypasses’ around periventricular white matter lesions to reach their original cortical target areas in the postcentral gyrus. Thus, somatosensory functions can be well preserved even in cases of large periventricular lesions. In contrast, when the postcentral gyrus itself is affected, no signs for reorganization have been observed. Accordingly, somatosensory functions are often poor in these patients. Language functions can be normal even in patients with extensive early left-hemispheric brain lesions. This is achieved by language organization in the right hemisphere, which takes place in brain regions homotopic to the classical left-hemispheric language areas in normal subjects. In patients with periventricular lesions, the degree of right-hemispheric takeover of language functions correlates with the severity of structural damage to facial (and, thus, articulatory) motor projections.

Introduction

The developing human brain possesses superior capabilities to compensate for focal brain lesions. The mechanisms of this postlesional (re)organization are not only relevant in the clinical context but can also yield insights into the processes of normal and abnormal development of the human neocortex and its projections. This review summarizes recent results concerning the mechanisms of this postlesional reorganization in the developing human brain, with a special emphasis on the development of motor and somatosensory pathways in children with early brain lesions.

Reorganization in the motor system

During normal development, cortico-spinal motor projections sprout from the motor cortex and grow in a cortico-fugal manner. By the 20th week of gestation, these descending cortico-spinal projections have reached the spinal cord (Eyre et al. 2000) and enter a process of synaptogenesis with target cells, especially with alpha motor neurons, at the spinal segmental level. During this phase, each hemisphere initially develops bilateral projections, i.e. projections to both the contralateral and ipsilateral extremities. This results in a situation of ‘competition’ between ipsilateral and contralateral projections for motor neurons in the spinal cord. Ongoing normal development is characterized by a gradual weakening of ipsilateral projections, paralleled by strengthening of contralateral projections (Eyre et al. 2001). Experimental data from the macaque monkey suggest that this process is part of a more widespread elimination of cortico-spinal axons, most of which apparently never synapse on spinal neurons (Galea & Darian-Smith, 1995). Furthermore, experiments performed in a neonatal cat model suggest that, during this process, neuronal activity is a crucial factor in determining which projections are strengthened and which are weakened or even withdrawn (Martin et al. 1999).

This normally transient existence of ipsilateral cortico-spinal projections provides the basis for a peculiar type of motor (re)organization. Unilateral brain damage occurring before or during the time of synaptogenesis of cortico-spinal motor projections with spinal alpha motor neurons can reduce or abolish neuronal activity in projections originating in the affected hemisphere, so that the ipsilateral projections from the contralesional hemisphere can exceed the contralateral projections in their neuronal activity. Subsequently the ipsilateral projections persist and are strengthened during further development, whereas any (now weaker) contralateral projections are withdrawn. Eventually, the contralesional hemisphere can become equipped with fast-conducting ipsilateral projections to the paretic extremities (Eyre et al. 2001; Staudt et al. 2002a). This type of cortico-spinal (re)organization can occur throughout the pre- and perinatal period (Staudt et al. 2004), during the first months of life (Eyre et al. 2007) and, in one case report, even up to the age of 2 years (Maegaki et al. 1997). In children beyond this age and in adult stroke patients, such fast-conducting projections have, to date, never been reported.

Cortico-spinal motor pathways pass through the periventricular white matter on their way from the primary motor cortex (precentral gyrus/central sulcus) to the internal capsule. Thus, damage to these projections is frequent in patients with periventricular white matter lesions (‘early third trimester lesions’; Staudt et al. 2002a) but more rare in patients with cortico-subcortical ‘infarct-type’ lesions (‘late third trimester lesions’; Staudt et al. 2004). Cortico-subcortical ‘infarct-type’ lesions often do not extend so far medially to also affect the periventricular white matter. Thus, crossed cortico-spinal projections from the lesioned hemisphere are often at least partially intact in such patients even with surprisingly large cortico-subcortical lesions (see Fig. 1 for example; Staudt et al. 2006a).

Fig. 1.

Diffusion tensor tractography in a patient with a pre- or perinatally acquired cortico-subcortical infarction in the territory of the middle cerebral artery (from Staudt et al. 2006a; reprinted with permission from Lippincott Williams & Wilkins © 2006). Left: Coronal T1-weighted image depicting the cystic lesion, leaving only a small bridge of preserved white matter between the lesion and the enlarged lateral ventricle. Nevertheless, transcranial magnetic stimulation and magnetoencephalography indicated preserved crossed cortico-spinal motor projections (red) and preserved crossed spino-thalamo-cortical somatosensory projections (blue). Right: Diffusion tensor tractography (in random colors) visualizes the extensive connectivity mediated by this small bridge of preserved tissue (seed area for fiber tracking). Fiber tracking performed using DTI studio software.

Many patients depending on such ipsilateral cortico-spinal projections show a useful grasp function with their paretic hand, some even with preserved individual finger movements (Staudt et al. 2002a, 2004). A normal (or near to normal) function of the paretic hand has, however, never been reported for patients depending on ipsilateral cortico-spinal projections. However, many patients cannot use their paretic hand for active grasping, although the existence of such fast-conducting pathways can be demonstrated. This variability in the quality of paretic hand function can be explained, at least partially, by different stages of development at the time of the insult. The earlier during development that a brain lesion occurs, the better the paretic hand function will be (Staudt et al. 2004). Consequently, in many children with brain damage acquired around term birth or postnatally, no useful hand function can be observed although ipsilateral tracts are present (Staudt et al. 2004; Eyre et al. 2007).

All patients controlling both the paretic and non-paretic hand with the contralesional hemisphere show a distinct clinical feature; during voluntary one-handed movements both with the paretic and non-paretic hand, the other hand shows involuntary ‘mirror movements’, which persist beyond the age of 10 years (the period during which mirror movements can be observed in normal children; Müller et al. 1997).

Reorganization in the somatosensory system

An ipsilateral representation of hand functions – as in the motor system – apparently never occurs for the primary somatosensory (S1) hand representation, neither transiently during normal development nor as a consequence of an early unilateral lesion (Guzzetta et al. 2007; Wilke et al. 2009). In the somatosensory system, however, a different mechanism of postlesional reorganization can be observed.

During normal development, outgrowing thalamo-cortical afferent projections reach their cortical destination sites over a prolonged period of time, which starts at the beginning of the third trimester of pregnancy (Kostovic & Judas, 2002). This explains why developing thalamo-cortical somatosensory projections can still ‘bypass’ even large periventricular brain lesions acquired during this phase to reach their original cortical destination areas in the postcentral gyrus (Staudt et al. 2006b) (Fig. 2).

Fig. 2.

(Re)organization of motor and somatosensory projections in a patient with a large unilateral periventricular lesion (adapted from Staudt et al. 2006b; reprinted with permission from Lippincott Williams & Wilkins © 2006). (A) Coronal T1-weighted image depicting the extensive lesion, which had disrupted the normal crossed cortico-spinal projections [no response following transcranial magnetic stimulation (TMS) of the affected hemisphere] and induced the formation (or, better, maintenance) of ipsilateral cortico-spinal projections from the contralesional hemisphere to the paretic (P) hand (yellow TMS coil symbol, yellow arrows indicate cortico-spinal projections). (B) Functional magnetic resonance imaging during active movements of the paretic hand reveals bilateral activation of the Rolandic (pericentral) cortices, plus midline activation representing the supplementary motor area. The ipsilateral activation (in the contralesional hemisphere) reflects activation of the reorganized primary motor (M1) representation of the paretic hand. (C) During passive movement of the paretic hand, only the contralateral Rolandic area in the affected hemisphere is activated, indicating a contralaterally preserved primary somatosensory (S1) representation of the paretic hand in the affected hemisphere. Accordingly, the blue dot represents the topography of the magnetoencephalographically determined S1 representation of the paretic hand. (D) Diffusion tensor tractography visualizes trajectories of somatosensory afferent fibers that bypass the lesion on their way to the Rolandic cortex of the affected hemisphere. Seed regions: tegmentum pontis and subcortical Rolandic white matter. Fiber tracking performed using DTI studio software.

Evidence for the existence of these projections comes from the combination of three techniques. First, functional magnetic resonance imaging revealed that passive movements of the paretic hand activate the Rolandic area of the affected hemisphere (contralateral to the paretic hand), which demonstrated that somatosensory information (or, in the case of active hand movement, somatosensory feedback) reaches this area despite the large periventricular lesions. Second, magnetoencephalography recorded the first cortical response to a peripheral tactile stimulus to the paretic thumb to occur with a normal short latency of approximately 20 ms (representing the N20 m response), demonstrating the presence of fast-conducting spino-thalamo-cortical projections. Dipole reconstructions of these responses always colocalized with the functional magnetic resonance imaging activation sites during passive hand movements, demonstrating that these spino-thalamo-cortical projections indeed terminated in the Rolandic area of the affected hemisphere. Finally, magnetic resonance diffusion tensor tractography visualized the course of these projections, forming ‘axonal bypasses’ around the lesion. In this context, it is important to mention that only the multi-modal approach that we applied allowed this functional interpretation. Without the additional evidence from magnetoencephalography, the trajectories detected by magnetic resonance tractography could just as well represent cortico-fugal projections connecting the Rolandic cortex of the affected hemisphere with the pontine tegmentum.

The fact that these outgrowing projections accept such long detours to reach the ‘hand area’ of the postcentral gyrus strongly argues in favor of an early functional specialization of the developing human neocortex – which exists apparently before the ingrowth of specific thalamic afferents.

Recently, a similar observation of developing thalamo-cortical projections finding alternative routes to their original cortical destination sites has been reported (Little et al. 2009). In Sema6A mutant mice lacking the guidance molecule Semaphorin-6A, developing thalamo-cortical axons growing out from the visual part of the thalamus (the dorsal lateral geniculate nucleus) are initially misrouted in the ventral telencephalon. Nevertheless, these misrouted axons are eventually able to find their way to the visual cortex via alternate routes and establish a normal pattern of thalamo-cortical connectivity.

Both observations are not compatible with the so-called ‘tabula rasa’ hypothesis of human neocortical development (Creutzfeld, 1977) but rather argue in favor of the ‘protomap hypothesis’ (Rakic, 1988), which postulates early regional specialization to already exist before the ingrowth of specific thalamic afferents (for review see Rakic, 2007).

However, several recent studies suggest that specific thalamic input does indeed play a major role in cortical field generation. One impressive example is the observation that removal of the caudal parts of the cortical neuroepithelial sheet unilaterally at an early stage of development in marsupials does not abolish cortical fields that normally reside in the removed cortex. Rather, in this reduced cortical sheet, normal spatial relationships between visual, somatosensory and auditory cortical fields are established. This implies that cortical fields can apparently form in a new location on portions of the cortical sheet that would normally be occupied by a different sensory modality (Huffman et al. 1999).

It is most likely that neither of these two opposing hypotheses is exclusively true; rather, that some combination of early regional differentiation on the one hand and specific thalamic afferent contribution on the other hand is responsible for the differentiation of cortical fields in development (for review, see O’Leary et al. 1994).

One could also argue that thalamo-cortical projections might already have been laid down when the lesion occurred. Admittedly, this possibility cannot be excluded, especially as the exact gestational age when the lesion occurred is not known in the patients of this study (Staudt et al. 2006b). With the magnetoencephalographic evidence of fast-conducting somatosensory pathways terminating in the Rolandic area of the affected hemisphere, one would then have to assume that either additional thalamo-cortical axons had sprouted from the thalamus after the lesion had disrupted the normal, already existing projections, or these connections had formed by rerouting of thalamo-cortical axons that had originally grown towards other cortical areas, so that the lesion did not affect them. A third alternative, that the lesion initially exerted a ‘mass effect’ and had thus only ‘pushed away’ any already present thalamo-cortical projections without disrupting them, seems quite unlikely. This ‘mass effect’ should have also pushed away cortico-spinal axons, which then should still be detectable by transcranial magnetic stimulation – instead of disrupting these cortico-spinal projections, as suggested by the absence of motor evoked potentials in the paretic hand after transcranial magnetic stimulation of the affected hemisphere. Although none of these alternative hypotheses can be completely ruled out, we consider our initial hypothesis of ‘axonal bypasses’ developing around the lesion to be the most parsimonious explanation for our observations.

Functionally, patients with such ‘axonal bypasses’ typically show no or only minor somatosensory deficits, which sometimes contrast with marked motor dysfunctions (Staudt et al. 2006b; Wilke et al. 2009).

Cortico-subcortical lesions acquired as thrombo-embolic infarctions in the vascular territory of the middle cerebral artery often show a direct involvement of the postcentral gyrus. Even in these patients, no clear evidence has yet been found for reorganization of S1. During functional magnetic resonance imaging of passive hand movements, these patients typically activate the intact portions of the postcentral gyrus, with a somewhat more variable topography as determined in group analyses (Wilke et al. 2009). Functionally, many of these patients show severe somatosensory deficits, which sometimes contrast with relatively spared motor abilities (Wilke et al. 2009).

An example of a large cortico-subcortical lesion is given in Fig. 1.

Reorganization in the language system

In the majority of normal subjects, language develops predominantly in the left hemisphere. This is true for almost all right-handers, and also for most left-handers, although bilateral or right-hemispheric language organization occurs more frequently in these subjects (Pujol et al. 1999).

Despite this clear ‘preference’ of normal language development for the left hemisphere, even extensive damage to the left hemisphere can be fully or almost fully compensated when the insult occurs during the pre- or perinatal period. In these subjects, language functions develop in the right hemisphere (Rasmussen & Milner, 1977), in areas homotopic to the classical language zones in the left hemisphere of healthy subjects (Staudt et al. 2002b) (Fig. 3).

Fig. 3.

Architecture of lesion-induced right-hemispheric language organization (from Staudt et al. 2002b; reprinted with permission from Elsevier © 2002). Functional magnetic resonance imaging of speech production (silent generation of word chains) in five healthy right-handers (right) and five patients with predominantly right-hemispheric language representation due to left-sided periventricular brain lesions (left). Statistical Parametric Mapping 99, fixed-effect group analyses.

In patients with left-sided periventricular brain lesions, a similar ‘hemispheric dissociation’ has been observed as in the sensorimotor system. In both systems, the ‘afferent’ components (somatosensory hand representation/perception of speech) can still be located in the affected hemisphere, whereas the ‘efferent’ components (motor hand representation/production of speech) are (re)organized in the contralesional hemisphere (Staudt et al. 2001; Staudt, 2007). Moreover, the degree of right-hemispheric takeover of language production has been demonstrated to be related to structural damage to left-hemispheric facial motor tracts with their well-known relevance for articulation (Staudt et al. 2001, 2008).

Conclusions

Taken together, these observations suggest that, for both the sensorimotor and language systems, the different normal trajectories of development for afferent vs. efferent pathways determine the mechanisms of postlesional reorganization in the developing human brain.