Perinatal stroke, defined as stroke occurring between the 20th week of fetal life and the 28th postnatal day, affects up to one in 2300 term and seven in 1000 preterm infants. The majority of cases are due to ischaemic events. These may be arterial ischaemic strokes, or periventricular venous infarction, due to compression of the medullary veins following germinal matrix haemorrhage prior to 34 weeks' gestation. A smaller number of cases are due to intracerebral haemorrhage or venous sinus thrombosis. Up to 60% of perinatal strokes result in neurological deficits, with hemiplegic cerebral palsy (CP) being a frequent adverse motor outcome. In fact, perinatal stroke is the most common cause of hemiplegic CP. This lifelong condition has implications for performance in activities of daily living, quality of life, and self-esteem. Adults with hemiplegic CP are less likely than their peers to live independently or be in full-time employment. Hemiplegia is not the only reported adverse outcome after perinatal stroke. Other adverse outcomes include cognitive, language and visual deficits, seizures, and behavioural problems.[5, 9] Not surprisingly, there is a substantial long-term financial burden associated with perinatal stroke.
Perinatal stroke is the most common cause of hemiplegic cerebral palsy. No standardized early intervention exists despite evidence for a critical time window for activity-dependent plasticity to mould corticospinal tract development in the first few years of life. Intervention during this unique period of plasticity could mitigate the consequences of perinatal stroke to an extent not possible with later intervention, by preserving the normal pattern of development of descending motor pathways. This article outlines the broad range of approaches currently under investigation. Despite significant progress in this area, improved early detection and outcome prediction remain important goals.
Repetitive transcranial magnetic stimulation
Constraint-induced movement therapy
Why is Perinatal Stroke Problematic to Diagnose and Manage?
In contrast to the significant progress that has been made in the prevention, diagnosis, and management of stroke in adults, management of perinatal stroke remains problematic. The aetiology of perinatal stroke is multifactorial and incompletely understood, limiting preventative options. Whereas stroke in adults is quickly detected, perinatal stroke often presents non-specifically in the first days of life with seizures, lethargy, and/or poor feeding. Moreover, around 40% of cases (‘presumed perinatal strokes’) are first detected outside the neonatal period. For this group, presentation occurs on average 5 months after birth, often with asymmetrical movement difficulties. Most patients with presumed perinatal stroke genuinely appear to have been asymptomatic during the neonatal period and early infancy, with only a minority representing ‘missed symptomatic perinatal strokes’. However, there are often significant delays between the onset of parental concern and the final diagnosis of presumed perinatal stroke, the latter reported to occur at a mean age of 12.6 months.
In contrast to adult stroke, there is no established immediate treatment for perinatal stroke except symptomatic measures (e.g. stabilization and management of bleeding diathesis, thrombophilia, hydrocephalus, etc.). What evidence-based recommendations there are have been summarized by Roach et al. in an article on the management of stroke in infants and children.
To complicate things further, although most cases presenting as presumed perinatal stroke will become confirmed cases of hemiplegic CP, this is not true for symptomatic perinatal stroke. Of the patients presenting in the neonatal period with symptomatic stroke, around 50% develop CP, around 35% may have normal outcome, and the remainder have relatively minor motor abnormalities. Some inroads have been made regarding outcome prediction, although this remains problematic, as will be discussed below. Clearly, this is important when considering who to target with early intervention. At present, however, rehabilitative management for perinatal stroke, though variable, is largely structured around observation and referral to therapy services as and when emerging focal neurological signs are observed. Early identification and referral of at-risk infants for intervention, rather than referral of infants with established CP, has been called for. This approach, however, requires both more widespread use of appropriate tools for early diagnosis and more evidence for effective early interventions.
There is a large and expanding literature base covering the established and experimental approaches to the management of hemiplegic CP. (For the upper limb, this has been recently reviewed and summarized by Sakzewski et al.) ‘Late intervention' approaches for children with established hemiplegia, ranging from orthoses and botulinum toxin injections to orthopaedic surgery, will not be covered here as they are well known. This article aims to explain the rationale for early intervention in perinatal stroke and provide an overview of current experimental approaches addressing this treatment regimen.
Why Aim to Intervene Early After Perinatal Stroke?
In the absence of any obvious primary preventative strategies, the next best option is ‘damage control’. The neuroprotective strategies discussed below could help minimize tissue damage from the initial insult. However, there are also opportunities for intervention outside this immediate postinjury period. Major developmental changes occur in the nervous system in the first year of life, with long-term implications for function. One of the most striking of these, highly relevant to motor outcome after stroke, is the ongoing activity-dependent plasticity within descending motor pathways. The corticospinal tract is the major descending pathway from the brain to the spinal cord controlling voluntary movement. At term, corticospinal fibres from each hemisphere project to each side of the spinal cord (bilateral system). Neurophysiological evidence indicates that in healthy infants, gradual progression to a predominantly crossed projection occurs within the first 2 years of life, with the most marked changes concentrated within the first year. Following unilateral perinatal stroke, corticospinal projections from the affected hemisphere are at a competitive disadvantage and, in patients with poor motor outcome, are gradually downregulated whereas uncrossed projections from the undamaged hemisphere are pathologically retained.[22, 23] Patients with the most favourable functional outcome retain crossed projections from the affected hemisphere.
This is true for a variety of lesion types, though lesions acquired early in pregnancy (such as malformations of cortical development), typically lead to less severe motor dysfunction than those acquired later (e.g. middle cerebral artery territory infarctions) where both the early and late lesions have led to pathologically retained uncrossed projections from the undamaged hemisphere. This may reflect a greater capacity for intrahemispheric reorganizational potential earlier in fetal development. Animal models demonstrate similar findings. Interventions within an early time window after perinatal stroke may have the potential to steer the course of corticospinal tract development towards a more normal pattern, providing a unique opportunity to influence outcome.
Predictions that the course of corticospinal tract development can be altered during this critical time window have been confirmed in animal models by seminal studies by Martin et al. Electrical stimulation of a unilaterally inactivated corticospinal tract re-established normal connectivity and partially restored motor function in a neonatal cat model. Constraint of the unaffected side, plus encouragement of use of the affected limb, improved motor outcome and restored corticospinal tract connections, spinal cord circuitry, and the motor cortical map. Outcome was best when training began early, and training plus constraint was more effective than constraint alone. Preventing normal limb use very early in development, however, could adversely affect function of that limb in the long term. Immediate and total immobilization of the unimpaired forelimb, producing forced overuse of the impaired limb (whilst obviously an extreme example), also worsened the neuronal injury in adult rodent stroke models. Thus the nature and risk benefit ratios of potential intervention strategies must be very carefully explored.
Notwithstanding these dramatic developmental changes during infancy, when the greatest potential effects of intervention on outcome may be expected, there is also the option for later intervention that exploits ongoing neuronal plasticity. Neurophysiological approaches modulating cortical excitability and therapy approaches (e.g. constraint) have been shown to influence motor outcome after stroke even in elderly adults. However, in hemiplegic CP, morbidity can increase over time, examples being developmental disregard of the affected side and contractures. The earlier the onset of an effective intervention, the more secondary morbidity could hopefully be prevented.
What Types of Approach are Being Explored?
Although as yet no standard early interventional approach exists for perinatal stroke, there is a very broad range of experimental approaches that can only be highlighted within the scope of this review. The relative merits of each approach remain to be established.
Minimizing damage from the initial insult: neuroprotective strategies
Therapeutic hypothermia reduces the risk of major neurodevelopmental disability and mortality by around 25% in term and late preterm neonates with hypoxic-ischaemic encephalopathy. Side effects include sinus bradycardia and thrombocytopenia. In a systematic review and meta-analysis of animal models of acute ischaemic stroke, hypothermia produced a 44% reduction in lesion size. The mechanism appears to be multifactorial but is broadly related to reduction in excitotoxicity and inflammation. Hypothermia has not been systematically investigated as an approach to perinatal stroke. However, in one study of neonatal encephalopathy, none of five patients with focal stroke receiving cooling developed neonatal seizures, in contrast to seven of 10 similar patients who did not receive cooling.
It has been suggested that induced hypothermia should be investigated as a therapeutic option in infants with arterial ischaemic perinatal stroke. The evidence from animal models suggests that treatment should begin within 6 hours of stroke in order to be effective. This poses a problem for patients with stroke who do not present with very early onset encephalopathy or seizures.
Other neuroprotective factors are also being explored in the context of neonatal brain injury – this has been recently reviewed by Gonzalez and Ferriero. Some of these are surprisingly familiar from other clinical contexts, such as growth factors (e.g. erythropoietin), antioxidants (including melatonin), anti-inflammatory agents (including minocycline), and a range of agents aimed at reducing excitotoxic damage (including topiramate). Combinations of hypothermia and other neuroprotective agents are now being investigated through clinical trials for neonates with hypoxic-ischaemic encephalopathy.
Cell replacement therapy: stem cell transplantation
Stem cells are cells with the potential for both self-renewal and differentiation into a wide range of cell types. There is much interest in their potential to reduce the burden of morbidity following ischaemic insults to the developing brain. Investigations are being pursued with a range of stem cell types, which are likely to act in different ways. For example, neural precursor stem cells have been shown to migrate to the lesion site following unilateral hypoxic-ischaemic injury in neonatal rats; some of the cells then differentiated into glial subtypes, with a small minority differentiating into neuronal subtypes. In contrast, mesenchymal stem cell transplants from human umbilical cord blood, while also migrating to the lesion site, showed little evidence of differentiation into either neural or glial subtypes in a neonatal rat ischaemic stroke model. The reduced infarct volume and improved functional outcome in the transplanted group were attributed to anti-inflammatory effects through the release of trophic factors.
The potential risks from stem cell transplantation are high, including tumour development and transplant rejection, and there remain many unanswered questions regarding mode of delivery, mode of action, optimal use, timing, etc. However, autologous umbilical cord blood-derived cells, which pose a number of obvious advantages from ethical and safety viewpoints, have reached the stage of phase I clinical trials following perinatal arterial ischaemic stroke (ClinicalTrials.gov Identifier NCT01700166) and acutely in neonatal hypoxic-ischaemic encephalopathy (ClinicalTrials.gov Identifier NCT01506258).
Modulating cortical excitability: neurophysiological approaches
There has been recent interest in the possibility of modulating cortical excitability using non-invasive brain stimulation to improve outcome after stroke. Cortical excitability is altered following stroke, in infants as well as adults, although the precise pattern of alteration differs by lesion location. One of the observed effects in adults following stroke is a high level of interhemispheric inhibitory drive from the intact hemisphere to the affected hemisphere, correlating with poor motor performance of the paretic hand. This finding has driven studies aiming to increase excitability of the affected motor cortex and/or decrease excitability of the unaffected motor cortex. This has been done non-invasively using transcranial direct current stimulation (with the anode being excitatory and the cathode being inhibitory), or using repetitive transcranial magnetic stimulation ([rTMS]; with low stimulation rates being inhibitory and high stimulation rates being excitatory).
Low rate rTMS to the unaffected hemisphere has also been studied in older children with subcortical stroke, demonstrating safety and feasibility of the approach. There is an ongoing trial comparing the effect of rTMS and constraint-induced movement therapy in combination with each treatment separately (or neither) on motor outcomes in school-age children with hemiplegia due to perinatal ischaemic stroke (ClinicalTrials.gov Identifier NCT01189058). It is unclear at present as to whether such non-invasive brain stimulation approaches are applicable earlier in life after perinatal stroke, when interhemispheric inhibition is still immature.
Therapy-based approaches with onset in infancy
The current emphasis on therapy-based interventions remains focussed on children with established hemiplegic CP, not on young infants. One approach is constraint-induced movement therapy (CIMT), in which use of the ipsilesional hand is prevented for a period of time (e.g. using a restraining glove). During this time, intensive movement practice is achieved with the contralesional (more affected) arm and hand. An alternative approach is intensive bimanual therapy. While a number of unanswered questions remain, including the optimal duration and mode of delivery and the effect of the nature and severity of the underlying lesion, there is evidence for the effectiveness of both CIMT and bimanual therapy in this older age group.
In the first few years of life, the principle of activity-dependent competition shaping corticospinal tract development might suggest that constraint of the ipsilesional hand would be the more appropriate approach. However, along with an increased potential for benefit in this age group comes increased risk from aberrant plasticity. For example, constraint of the ipsilesional hand in infants at risk of hemiplegia may have adverse effects on development of movement control of that hand. The correct balance between promoting use of the contralesional arm and hand versus hindering development of the ipsilesional arm and hand must be struck if constraint is to be used. In a non-randomized trial, constraint improved the use of the hemiplegic hand in bimanual play compared with standard therapy approaches in children age 18 months to 4 years. More commonly, infants undergoing constraint have been studied as part of a larger group with a broad age range. There is increasing recognition that therapy interventions in an even earlier time window need to be explored. A randomized single blind trial comparing modified CIMT with ‘baby massage’ for infants aged 3 to 8 months with unilateral brain lesions and emerging motor asymmetries is currently recruiting (ClinicalTrials.gov Identifier NCT01864811).
Exploitation of the ‘mirror neuron system’ is also under investigation as a method for improving motor outcome in children with hemiplegic CP. Mirror neurons were first identified in animal studies and were so named because they fired when the animal either performed a motor task or observed the same task being performed. Indirect evidence for a human mirror neuron system has subsequently been presented. Cortical responses to action observation resembling those to action performance have been demonstrated in infants and young children as well as adults.[56-58] There is increasingly strong evidence that the mirror neuron system is an important component of motor learning. For example, movement practice combined with observation of the same movements increases motor learning compared with movement practice alone. Therapy based on action observation and imitation was shown to improve motor recovery following stroke in adults. A recently published randomized controlled trial indicated short-term improvements in the use of the affected hand in bimanual tasks with therapy involving repeated action observation and execution compared with repeated practice alone in 24 children with hemiplegia aged 5 to 15 years. Researchers from the same group have taken on the significant challenge of applying similar approaches to infants with predominantly unilateral brain lesions from the age of 9 weeks.
Predicting the Development of Hemiplegic Cerebral Palsy
One of the difficulties with very early interventions following perinatal stroke is the need for early identification of affected infants and early, accurate prediction of outcome so that interventions can be targeted to those at significant risk of developing hemiplegia. For neonates presenting with seizures (for which stroke is the second most common cause) or encephalopathy, early cranial imaging is needed to make the diagnosis. For those infants presenting later with pathological hand preference or motor delay, swifter referral for investigation is an important aim. The two most effective early predictors of motor outcome following symptomatic perinatal stroke are the pattern of magnetic resonance imaging (MRI) abnormalities and the presence of abnormal ‘general movements’; however, neither of these assessments are infallible.
Following symptomatic neonatal unilateral arterial ischaemic stroke at term, early MRI evidence of involvement of both deep (basal ganglia, posterior limb of internal capsule) and superficial (distal, cortical) middle cerebral artery territory structures is associated with a high probability of the development of hemiplegia.[63-66] Conversely, involvement of only superficial cortical areas in the infarct is associated with a low probability of developing hemiplegia, as is pure basal ganglia involvement.[63-66] Extensive corticospinal tract involvement (more precisely, involvement of the posterior limb of the internal capsule and cerebral peduncle) on diffusion-weighted imaging at around age 1 week also has a high risk of hemiplegia;[63, 67] diffusion tensor imaging to study the corticospinal tract at age 3 months is similarly predictive. For infants who have sustained a haemorrhagic parenchymal infarct, either antenatally or following preterm delivery, asymmetrical myelination of the posterior limb of the internal capsule detected on MRI at term-equivalent age is also predictive of hemiplegia.
A trained observer of spontaneous infant movements, using Prechtl's method, can discern a number of patterns. These include complex and variable writhing ‘general movements’ in the first few months of life, and elegant, small, circular ‘fidgety movements’, peaking at between 9 weeks and 13 weeks. Absent fidgety movements, or movements of abnormal quality, are associated with the emergence of subsequent neurological deficits. In a study of 11 term neonates with predominantly unilateral cerebral infarction, the absence of fidgety movements predicted the development of hemiplegia. Similar findings were observed in a group of preterm infants with unilateral periventricular lesions. Minor early movement asymmetries could also be detected with video analysis. Algorithms for automated assessment of general movements are under investigation.
There is active interest in developing early intervention approaches for infants with unilateral perinatal stroke. Improved early detection rates and optimized outcome prediction will be needed in order to implement all but the most low-risk strategies. The degree to which outcome can be changed remains to be seen, especially as early imaging-based findings of corticospinal tract asymmetry currently predict later function, but early intervention is a challenge we must take on.
Dr Basu is funded by a Clinical Trials Fellowship award from the National Institute for Health Research. The views expressed in this publication are those of the author and not necessarily those of the NHS, the National Institute for Health Research, or the Department of Health. Thanks to Gavin Clowry, Nicholas Embleton, Claire Marcroft, and the two anonymous reviewers for comments on an earlier draft. The author states that she had no interests that might be perceived as posing a conflict or bias.