Child Neurology
7th Edition

Because of its relatively low energy demands, the neonatal brain has a considerable resistance to hypoxia, and most hypoxic injuries to neonates result from a combination of hypoxia and ischemia. Alterations in cerebral blood flow induced by asphyxia are therefore of primary importance in understanding the genesis of birth injuries (Fig. 6.2). Following the onset of asphyxia, cardiac output is redistributed so that a larger proportion enters the brain. This results in a 30% to 175% increase in cerebral blood flow. The increase in cerebral blood flow is induced locally by a reduction in cerebrovascular resistance and systemically by hypertension. The severity and the speed of onset of the asphyxial insult determine the cerebrovascular response (40). When asphyxia is severe and develops rapidly, cerebral blood flow decreases rather than increases, probably due to increased cerebrovascular resistance. When the hypoxic-ischemic insult is prolonged, these homeostatic mechanisms fail, cerebral vascular autoregulation is lost, cardiac output decreases, and systemic hypotension develops with reduced cerebral blood flow (41) (see Fig. 6.2).

Normal brain vasculature can compensate for the decreased cerebral perfusion by rapid dilatation of the smaller vessels, so that cerebral blood flow is maintained relatively constant as long as blood pressure is kept within the normal range. The constancy of cerebral blood flow in the face of fluctuations in systemic blood pressure is termed autoregulation. The large cerebral blood vessels are believed to be more important for cerebral autoregulation in the neonate than the arterioles, with the response to changes in blood pressure being endothelium dependent (42). A number of chemicals have been implicated in the control of cerebral arterial tone (43). Nitric oxide, by
acting on the calcium-activated potassium channel of vascular endothelium, induces vascular dilatation. Adenosine also mediates vasodilatation, whereas endothelin-1 and prostanoids mediate vasoconstriction (42,44). Hypoxia, hypercarbia, and hypoglycemia all impair cerebral autoregulation. When autoregulation becomes defective as a result of hypoxia, cerebral arterioles fail to respond to changes in perfusion pressure and carbon dioxide concentrations, resulting in a pressure-passive cerebral blood flow.

It is clear that in a clinical setting multiple factors can act in concert to cause cerebral vessels to become unresponsive to systemic blood pressure (45,46). Although in the preterm infant the lower limits of autoregulation are very close to the mean systemic arterial pressure, adequate cerebral perfusion can be maintained as long as the mean arterial blood pressure ranges between 24 and 39 mm Hg (47). When hypotension exceeds these lower limits, the preterm infant is unable to compensate for the drop in blood pressure. With the arteriolar system unable to respond to decreased perfusion pressure by vasodilatation, there is a striking reduction in cerebral blood flow (47).

After termination of the ischemic insult there is a marked increase in cerebral blood flow, probably the result of the various vasodilating factors already cited. This early increase in cerebral perfusion is followed by a decline and a second, delayed increase in cerebral blood flow, probably the consequence of an increased synthesis of nitric oxide. It is during this second phase that most of the deleterious events occur that lead to cell death within the brain.

Asphyxial brain injury is similar regardless of whether the brain has incurred a global asphyxial insult as occurs in perinatal asphyxia, hypoperfusion as after cardiac arrest (see Chapter 17), or focal ischemia as after a vascular occlusion (see Chapter 13). Some authors (and lawyers) attribute great significance to meconium as a cause of intrapartum or neonatal asphyxia, but the expulsion of meconium into the amniotic fluid in the first place often is due to a generalized parasympathetic discharge with increased peristalsis that results from fetal distress or an hypoxic-ischemic insult in utero. Aspirated meconium at delivery can complicate the problem further by obstructing the infant’s airway, but meconium per se is not a cause of asphyxia.

The mechanisms for brain damage in asphyxial brain injury are not completely clear. Volpe, in reviewing the physiologic aspects of asphyxial injury, suggested that the loss of vascular autoregulation coupled with hypotension reduces cerebral blood flow to the point of producing tissue necrosis and subsequent cerebral edema (4). Combined clinical and imaging studies by Lupton and associates in which intracranial pressure of asphyxiated term infants was correlated with their CT scan corroborate Volpe’s view that tissue necrosis precedes cerebral edema rather than vice versa, with maximum abnormalities being seen between 36 and 72 hours after the insult (48). Nevertheless, it
is still likely that tissue swelling can to some extent further restrict cerebral blood flow and cause secondary edema. The earliest phase of cerebral edema probably reflects a cytotoxic component, whereas a vasogenic component characterizes the edema that accompanies extensive tissue injury (49). In asphyxiated newborns, increased intracranial pressure after perinatal asphyxia is a relatively uncommon complication; in the series of Lupton and coworkers it was encountered in only 22% of asphyxiated infants (48). It is important that the clinician recognize that a bulging fontanelle and split sutures in the neonate after an asphyxiating or severe ischemic insult does not signify potentially reversible cerebral edema after obstructive hydrocephalus, mass lesions such as subdural hematomas, and meningitis have been excluded, but rather irreversible encephalomalacia with massive necrosis in both gray and white matter, and hence is a very bad prognostic sign.

Over the last few years increasing attention has been paid to the molecular and cellular aspects of cell death within the nervous system. Gluckman et al. distinguished two phases (50). The first phase occurs during the insult and the immediate period of reperfusion and reoxygenation. The second phase evolves after a period of some hours and extends for at least 72 hours. During the first phase, asphyxia rapidly results in the conversion of nicotinamide adenine dinucleotide (NAD) to reduced NADH. As the energy demands fail to be met, there is a shift from aerobic to anaerobic metabolism, causing acceleration of glycolysis and increased lactate production. In experimental animals, brain lactate increases within 3 minutes of induction of asphyxia (51). At the same time, the concentration of tricarboxylic acid cycle intermediates decreases and the production of high-energy phosphates diminishes. These changes result in a rapid fall in phosphocreatine and a slower reduction in brain adenosine triphosphate (ATP) concentrations. With reduction of ATP levels the various ion pumps, notably the Na⁺–K⁺ pump, the most important transporter for maintaining high intracellular concentrations of potassium and low intracellular concentrations of sodium, becomes inoperative. As ion pump function is lost, the neuronal membrane begins to change. Some neurons, such as the CA1 and the CA3 hippocampal neurons, hyperpolarize, whereas others, such as the dentate granule cells, depolarize. If anoxia persists, all cells undergo a rapid and marked depolarization with complete loss of membrane potential.

The aforementioned changes in lactate and high-energy phosphates can be documented in the asphyxiated infant by proton and ³¹P magnetic resonance spectroscopy. These studies show an early increase in lactate (52). A decrease in phosphocreatine during the initial insult is reversed on resuscitation but is followed by a slow, secondary decline some 24 hours later. Intracellular pH and other indices of cellular energy status frequently remain normal for the first day of life (53,54).

Decreases in intracellular and extracellular pH precede changes in membrane potential as hypoxia induces production of lactate and intracellular acidosis. The decrease in extracellular pH is believed to be the consequence of extrusion of intracellular hydrogen ions, intracellular lactate, or both. At the time when the neuronal membrane potential is abolished, there are a number of striking ionic changes. They include an efflux of potassium and an influx into cells of sodium, chloride, and calcium. The increase in intracellular calcium appears to play a critical role in cellular injury. It results from a failure of energy-dependent calcium pumping mechanisms and an opening of voltage-dependent calcium channels. The accumulation of intracellular calcium in turn initiates a cascade of deleterious events:

Inflammatory reactions involving a variety of cytokines may also contribute to hypoxic-ischemic cell death. The increase in cytokines could stem from an infection, notably a chorioamnionitis that predates the hypoxic-ischemic insult, or it could result from the activation of microglia by asphyxia (63,64).

The late phase of asphyxial injury and cell death is marked by an inappropriate induction of apoptosis. Apoptosis refers to the activation of genetically determined cell-suicide programs. Choi proposed that a single insult might trigger both excitotoxic necrosis and apoptosis, with the severity and duration of the insult determining which death pathway predominates (61). In regions containing large amounts of apoptosis inhibitors, necrosis predominates; conversely, in the absence of endogenous apoptosis inhibitors, hypoxia-ischemia induces apoptosis (39a). Nakajima and coworkers found that apoptosis is more prevalent than necrosis in the hypoxic-ischemic newborn brain (65), suggesting that the impact of apoptosis-executing caspases, key effectors of apoptotic death, is much greater in the immature than in the mature brain (66). The factors that promote postasphyxial apoptosis are under intense investigation. They are believed to include free radicals, increased expression and enhanced concentrations of inflammatory cytokines, and alterations in the concentrations or the response to endogenous growth factors (67). The observation that neurotropins, such as brain-derived neurotrophic factor, act as neuroprotectors after a hypoxic insult provides evidence for the importance of neurotropins in mediating postasphyxial brain injury (68). Asphyxia also induces both rapid and delayed changes in the transcription of several genes, notably c-fos, c-jun, and some of the heat-shock proteins (69, 70,71). These substances are believed to have a significant influence on the extent of apoptosis (70). The contributions of hypoglycemia and intracellular acidosis, whether caused by accumulation of lactic acid or products of ATP hydrolysis, to the extent and severity of asphyxial brain damage have not been resolved (72).

A large number of strategies aimed at blocking the postasphyxial events that lead to neuronal damage have been proposed. These involve inhibition of glutamate release, blockade of glutamate receptors, inhibition of the cytokine effects, and blockade of apoptotic cell death. None has had any significant clinical application (39,73,74). They are reviewed by Berger and coworkers (58). More recently, the effectiveness of growth factors such as erythropoietin and brain-derived neurotrophic factor in improving outcome after an asphyxial insult has been encouraging (39a).

The neonatal brain responds to infarction differently than the mature brain. Rather than dense gliotic scars, pseudocysts are the usual long-term residual lesions. The reasons for the formation of cysts in the newborn brain are that areas of infarction tend to be relatively larger than in the adult because collateral circulation is less well developed and because the ability of neonatal brain to mobilize reactive gliosis is limited. The number of astrocytes per volume of neonatal brain is approximately one-sixth that of the adult brain in both gray and white matter; hence the response to injury is not nearly as effective and the glial cells present are only able to form thin septa without neurons, compartmentalizing the empty space after macrophages have cleared away the necrotic tissue. These glial septa create the multiple pseudocysts of multicystic encephalomalacia. They are pseudocysts rather than true cysts because they are not lined by an epithelium, as are ependymal cysts. Multicystic encephalomalacia is therefore the end result of extensive cerebral infarcts.

When the primate fetus is subjected to acute total asphyxia, a reproducible pattern of brain disorders ensues (77,78). This pattern includes bilaterally symmetric lesions in the thalamus and in a number of brainstem nuclei, notably the nuclei of the inferior colliculi, superior olive, and lateral lemniscus. The neurons of the cerebral cortex, particularly the hippocampus, are especially vulnerable, as are the Purkinje cells of the cerebellum (4,79).

Soon after the initial insult, the first changes observed using electron microscopy are in the neuronal mitochondria, the internal structure of which becomes swollen and disrupted (80). Gradual widespread transneuronal degeneration follows. With progressively longer periods of total asphyxia, the destructive changes in the thalamus become more extensive, and damage begins to appear in the putamen and in the deeper layers of the cortex. (Fig. 6.3). In its extreme form, asphyxiated animals show an extensive cystic degeneration of both cortex and white matter. Connective tissue replaces the damaged areas in the forebrain, but a relative lack of cellular reaction occurs in the central nuclear areas (78).

This experimentally produced picture resembles cystic encephalomalacia (central porencephaly, cystic degeneration) of the human brain, a condition characterized by the formation of cystic cavities in white matter (Fig. 6.4). When small, the cysts are trabeculated and do not communicate with the ventricular system. In their most extensive
form they can involve both hemispheres, leaving only small remnants of cortical tissue. The cavities are generally believed to be the products of insufficient glial reaction, perhaps the result of cerebral immaturity, or to reflect the sudden and massive tissue damage caused by the aforementioned circulatory or anoxic events. This pathologic picture is seen not only as a consequence of severe perinatal asphyxia, as first established by Little (2), but also in twin pregnancies after intrauterine fetal death and in fetal viral encephalitides such as herpes simplex (81,82,83). Infants surviving this type of insult usually develop a severe form of spastic quadriparesis.

The pathologic differentiation between cystic degeneration and hydranencephaly is discussed in Chapter 5.

Occasionally, the major structural alterations resulting from perinatal injury are localized to the cerebellum. In the majority of instances, the involvement is diffuse, with widespread disappearance of the cellular elements of the cerebellar cortex, notably the Purkinje cells, and the dentate nucleus (3,151). As with the periventricular germinal matrix around the lateral ventricles, the external granular layer of the cerebellum is vulnerable to spontaneous hemorrhage, especially in preterm infants of young gestational age. However, in the vast majority of infants, selective cerebellar involvement is not the consequence of asphyxia (152).

In general, the human neonatal brainstem appears to be more resistant to ischemic and hypoxic insults than the cerebral cortex, but it is not invulnerable, and sometimes lesions are more prominent in the brainstem than in supratentorial structures. The lesions are usually symmetric and involve both gray matter nuclei and adjacent white matter tracts, but the gray matter is more focally involved, perhaps because of its higher metabolic rate. Lesions may involve the inferior or superior colliculi almost selectively (3), or infarction may occur in the central core of the brainstem or selectively in the periaqueductal gray matter (3,136,153,154). Such multiple deep microinfarcts at times extend rostrally to involve deep supratentorial structures such as the thalamus and corpus striatum (155). Bilateral tegmental infarcts of the pons and medulla oblongata in
particular are frequent sequelae of transient fetal hypotension because the tegmentum is a watershed zone, futher discussed later (133a). Because of the microscopic size of many of the deep infarcts, particularly those of the brainstem, they are difficult to identify on neuroimaging. Nevertheless, some older lesions that occurred in fetal life several weeks before delivery may be visualized in CT by their calcification in the floor of the fourth ventricle (133a,155a). MRI occasionally demonstrates infarcts in the tegmentum or floor of the fourth ventricle if they are large enough (155b).

Infarction in the nuclei of the brainstem and thalami, induced in experimental animals by total asphyxia of 10 to 25 minutes’ duration, also can be seen in asphyxiated human infants with a history of acute profound asphyxial damage (156). Additionally, transient compression of the vertebral arteries in the course of rotation or hyperextension of the infant’s head during delivery can be a cause for circulatory lesions of the brainstem (157,158).

Involvement of the nucleus ambiguous, just ventrolateral to the tractus solitarius, can result in dysphagia because this nucleus provides motor neurons for the muscles of deglutition. Involvement of the trigeminal motor nucleus by tegmental infarcts may damage motor neurons to the masticatory muscles, such as the masseter and pterygoids, and, in late gestation, the lack of function of these muscles leads to a lack of stimulus to growth of the mandible, so that micrognathia and even ankylosis of the temporomandibular joint is present by the time of birth (133a). Because the paired triads of vessels from the basial artery are 25 to 30 in number, from the midbrain to the caudal end of the medulla oblongata, the watershed tegmental infarcts are columnar and may extend posteriorly to even involve the hypoglossal nucleus so that the infant has atrophy and fasciculations of the tongue that may be mistaken for spinal muscular atrophy (133a,155c).

Neurosensory hearing loss after basilar artery insufficiency has both central and peripheral bases: The lateral lemniscus is at the margin of the tegmental watershed zone and the inferior colliculus is one of the most constant regions infarcted; the stria vascularis and hair cells of the cochlea receive their arterial supply from the basilar artery via the inferior anterior cerebellar artery (IACA). The labyrinth (semicircular canals) also receives arterial blood from the IACA, but its cells seem more resistant to irreversible injury from ischemia.

Table 6.3 lists the lesions and clinical expression seen with tegmental infarcts, but only rarely does an infant exhibit all deficits. Though much more frequent in fetuses and neonates than at any other time in life, tegmental watershed infarcts also can occur in older children and adults.

Unilateral tegmental infarcts also can involve the fetal and neonatal brainstem but are rare; such asymmetric lesions are more likely the result of a vascular anomaly of the basilar artery or birth trauma to the veretebral arteries than of generalized hypoperfusion of the brainstem due to transient shock in the fetus or neonate (133a,158).

Pontosubicular degeneration in isolation or accompanied by widespread cerebral damage has been described in premature and term infants (3,156,157). It is not a rare entity. It has been demonstrated in up to 59% of infants born before 38 weeks’ gestation who died in the first month of postnatal life, which suggests that it is the most common cerebral lesion of preterm neonates, exceeding even germinal matrix hemorrhages (4,159,160,161).

The condition represents a unique topology of pathologic neuronal apoptosis in the fetal and neonatal brain following hypoxia or ischemia. As its descriptive name implies, it selectively involves relay nuclei of the corticopontocerebellar pathway in the basis pontis and the subiculum, a transitional cortex between the three-layered hippocampus and the six-layered hippocampal gyrus. Although it may coexist with other hypoxic lesions in the cortex, thalamus, and cerebellum, these other regions are disproportionately less severely involved than the pontine nuclei and subiculum (136,159,162). Usually, there are no lesions in the tegmentum of the pons, although in rare instances symmetric microinfarcts have been described (163,164). Focal white matter infarcts also occur occasionally.

This distribution of infarcts is generally seen in infants older than 29 weeks’ gestation, most commonly in infants of 32 to 36 weeks’ gestation. The pattern can also occur in stillborns (165) and term infants and has occasionally been encountered in adult brains (166). The reason that pontosubicular degeneration is not better recognized by clinicians is that it is difficult to demonstrate during life and remains essentially a postmortem neuropathologic diagnosis.

The combination of infarcts in the basis pontis and the subiculum is difficult to explain because these regions do not share common afferent or efferent fiber connections, use different neurotransmitters, have morphologically
different types of neurons, arise from different embryologic primordia, and belong to different functional systems of the brain.

The pathogenesis of pontosubicular degeneration is poorly understood. Hyperoxemia in the presence of acidosis and hypoxia is described in some infants, suggesting that oxygen toxicity plays a role (160), but these cases represent only a small minority. Pontine neurons exhibiting karyorrhexis are immunoreactive to ferritin if accompanied by spongy changes and gliosis, suggesting that iron may be released to the damaged pontine neurons (167). In situ DNA fragmentation studies indicate apoptosis rather than frank necrosis as the mechanism of cellular death (168,169).

Pontosubicular degeneration has been described in stillborn fetuses, indicating that the lesions may result from intrauterine fetal distress and are not necessarily acquired intrapartum or postpartum (165,170,171). Sarnat provided details of the histologic progression of changes in the degenerating neurons (136).

In addition to direct trauma, asphyxia, and circulatory disturbances, malformations of the CNS play an important part in the genesis of the lesions of perinatal asphyxia and trauma. Little doubt exists that in the premature infant, for instance, both faulty maturation of the nervous system and a greater vulnerability to perinatal trauma and asphyxia are responsible for the high incidence of neurologic deficits (Table 6.6) (219). The relative frequency of prenatal and perinatal brain lesions in individuals with moderate or severe mental retardation can be determined from autopsy studies such as those by Freytag and Lindenberg (133) (Table 6.7) or from MRI evaluation of children with mental retardation and/or cerebral palsy (31).

Ischemic, hypoxic, and traumatic insults of the fetal brain in the second and third trimesters can induce malformations that are not primary defects of genetic
programming. Because development is incomplete, lesions that interrupt or alter radial glial fibers, for example, can prevent further neuroblast and glioblast migration before the process is complete and may result in focal dysplasias of cortical lamination and in deep heterotopia of neurons arrested in migration (Fig. 6.19). The abnormal synaptic relations that result from the abnormal anatomic positions of neurons may become a basis for later epilepsy (138).

The degree of a newborn’s functional abnormality secondary to asphyxia incurred during labor and delivery depends on the severity, timing, and duration of the insult. (See the Introduction chapter for a description of the essentials of a neurologic examination of the infant or small child.)

After birth, the infant subjected to perinatal asphyxia shows certain alterations in alertness, muscle tone, and respiration. These important clinical features of HIE were first graded by Sarnat and Sarnat (222) (Table 6.8). Several
other grading schemes have been devised. In essence, they are similar to that of Sarnat and Sarnat; however, some schemes place infants with repetitive or prolonged seizures into grade 3 rather than grade 2 (22). Sarnat and Sarnat specifically excluded seizures as a criterion for grading acute encephalopathy because of poor correlation with outcome in their experience and because some of the most severely involved infants in stage 3 do not have seizures because the cerebral cortex is so severely impaired that it can no longer generate epileptic activity. They did include electroencephalographic (EEG) criteria, however, as a measure of cerebral function rather than of paroxysmal activity. They also emphasized the importance of autonomic features, with sympathomimetic effects in stage 1 and strong parasympathetic (i.e., vagal) effects in stage 2.

In the experience of Levene and coworkers, 3.9 in 1,000 newborn term infants develop grade 1 HIE, 1.1 in 1,000 develop grade 2 HIE, and 1.0 in 1,000 develop grade 3 HIE (223).

Infants with grade 1 HIE are irritable with some degree of feeding difficulty and are in a hyperalert state in which their eyes are open with a “worried” facial appearance and a decreased frequency of blinking. They seem hungry and respond excessively to stimulation. Tremulousness, especially when provoked by abrupt changes of limb position or tactile stimulation, can resemble seizures. Mild degrees of hypotonia can be documented by a head lag and a lack of the normal biceps flexor tone in the traction response from the supine position. The Landau reflex is often abnormal, in that the infant’s body tends to collapse into an inverted U shape.

A greater hypoxic insult results in the evolution of grade 2 HIE. Infants are lethargic or obtunded with delayed or incomplete responses to stimuli. Focal or multifocal seizures are common. Severely asphyxiated infants develop clinical signs of grade 3 HIE. The infant is markedly hypotonic. The sucking and swallowing reflexes are absent, producing difficulties in feeding. Palmar and plantar grasps are weak, the Moro reflex can be absent, and the placing and stepping reactions are impossible to elicit.

A variety of respiratory abnormalities can be encountered. These include a failure to initiate breathing after
birth, which suggests a hypoxic depression of the respiratory reflex within the brainstem. Tachypnea or dyspnea in the absence of pulmonary or cardiac disease also suggests a neurologic abnormality. Periodic bouts of apnea are normal in the smaller premature infant. In larger infants, periodic apnea can result from a depression of the respiratory reflex or can indicate a seizure disorder (see Chapter 14).

The Apgar score has been used to measure the severity of the initial insult. Although in the past the score was used to determine the presence of a hypoxic insult, to do so is a misapplication. For instance, in the experience of Sykes and coworkers, only 20% of infants with a 5-minute Apgar score of less than 7 had an umbilical artery pH below 7.10 (224). Conversely, virtually all of the infants of 35 to 36 weeks’ gestation with a cord pH below 7.25 had a 5-minute Apgar score of 7 or more. In preterm infants the Apgar score is of even less value, and the more premature the infant, the more likely it is that the Apgar score will be low in the presence of a normal cord pH (225). Instead, the value of the Apgar score lies in being a simple and valuable predictor for survival during the neonatal period and for the ultimate outcome. In the series of Casey and coworkers, the neonatal mortality for term infants with a 5-minute Apgar score of 0 to 3 was 244 per 1,000, as contrasted with a mortality of 0.2 per 1,000 for intants with a 5-minute Apgar score of 7 to 10 (226). The extended Apgar scores (Apgar scores taken at 10 and 20 minutes of life), however, are even more valuable in predicting neurologic outcome, in that the likelihood of ensuing cerebral palsy increases significantly once the Apgar score remains under 3 for 10 minutes or longer (227). Moster and colleagues (228) noted that infants with 5-minute Apgar scores of 0 to 3 had an 81-fold increased risk for cerebral palsy compared with infants who had scores of 7 to 10. Term infants who had 5-minute Apgar scores of 0 to 3 and who also had signs consistent with neonatal encephalopathy had a significantly increased risk for impaired minor motor skills, reduced performances in reading and mathematics, and a variety of behavioral problems (229).

After 12 to 48 hours, the clinical picture of the previously hypotonic or flaccid infant (grade 3) can change to that of grade 2 or 1.5 (26). The infant becomes jittery, the cry is shrill and monotonous, the Moro reflex becomes exaggerated, and the infant has an increased startle response to sound. The deep tendon reflexes become hyperactive, and an increased extensor tone develops. Seizures can appear at this time. These signs of cerebral irritation also are noted in an infant who has experienced a major intracranial hemorrhage. In the series of Brown and associates (26), 24% of infants who were judged to have been subjected to perinatal hypoxia demonstrated hypotonia progressing to extensor hypertonus. In the experience of DeSouza and Richards, this clinical course has an ominous prognosis, for none of the infants following it were ultimately free of neurologic deficits (230). In our experience, the greater the delay in emerging from grade 3 HIE, the worse is the ultimate prognosis.

In other instances [24% in the series of Brown and associates (26)], an infant who was judged to have sustained perinatal asphyxia exhibits hypertonia and rigidity during the neonatal period. The clinical picture of spasticity in the neonate is modified by the immaturity of some of the higher centers. In the spastic infant, the deep tendon reflexes are not exaggerated but can be depressed as a result of muscular rigidity. Hyperreflexia becomes evident only during the second half of the first year of life. A more reliable physical sign indicating spasticity is the presence of a sustained tonic neck response, which indicates a tonic neck pattern that can be imposed on the infant for an almost indefinite time and that the infant cannot break down. Such a response is never normal (see Introduction chapter, Fig. I.5). Spastic hemiparesis is manifest during the neonatal period in only 10% of infants (231), usually by a reduction of spontaneous movements or by excessive fisting in the upper extremity. Obvious paralyses during the neonatal period are rarely caused by cerebral damage; instead, they suggest a peripheral nerve or spinal cord lesion.

The evolution of IVH can go unrecognized clinically in more than 50% of infants (232,233). The remainder can have a sudden, sometimes catastrophic deterioration highlighted by alterations in consciousness, abnormalities of eye movements, and respiratory irregularities. Deterioration can continue over several hours, then stop, only to resume hours or days later (4). The presence of a full fontanelle is noted in approximately one-third of asphyxiated infants (26). It can be the consequence of a massive intracranial hemorrhage, cerebral edema, or, less often, an acute subdural hemorrhage, the result of concomitant cerebral trauma (234).

Seizures secondary to perinatal asphyxia usually occur after 12 hours of age. However, when asphyxia is acute and profound, as can happen when a cord prolapse occurs, or when there is significant perinatal trauma, seizures can begin much earlier, and as a rule their onset cannot be used to time the asphyxial episode. The characterization and classification of seizures has been facilitated by the development of time-synchronized video and EEG/polygraphic monitoring. Generalized tonic-clonic convulsions are rare in the newborn. More often, one observes unifocal or multifocal clonic movements that tend to move from one part of the body to another. Generalized slow myoclonic jerks are another common neonatal seizure. EEG abnormalities do not accompany some behaviors previously considered to be neonatal seizures. These include a high percentage of tonic seizures, particularly those manifest by transient opisthotonos, and the various unusual forms of seizure activity (“subtle seizures”). Subtle seizures include paroxysmal blinking, changes in vasomotor tone,
nystagmus, chewing, swallowing, or pedaling or swimming movements. These seizures probably represent primitive brainstem and spinal motor patterns released from the normal tonic inhibition of forebrain structures (235). Apnea is not observed as the sole seizure manifestation. Seizures resulting from birth trauma or perinatal asphyxia often cease spontaneously within a few days or weeks or become relatively easy to control with adequate doses of anticonvulsants. The topic of neonatal seizures is taken up in greater detail in Chapter 14.

Infants who have suffered perinatal asphyxia experience various sequential changes of muscle tone and an abnormal evolution of postural reflexes.

Most often, there is a gradual change from the generalized hypotonia in the newborn period to spasticity in later life. In these patients, the earliest sign of spasticity is the presence of increased resistance on passive supination of the forearm or on flexion and extension of the ankle or knee. In spastic diplegia, this abnormal stretch reflex is first evident in the lower extremities and is often accompanied by the appearance of extension and scissoring in vertical suspension (see Introduction chapter, Fig. I.6), the late appearance or asymmetry of the placing response, a crossed adductor reflex that persists beyond 8 months of age, and the increased mobilization of extensor tone in the supporting reaction (221). In spastic hemiplegia, abnormalities first become apparent in the upper extremity. When five infants with unilateral hemispheric lesions detected by routine imaging studies during the first week of life were subjected to regular neurologic examinations, no abnormalities could be detected until 3 months of age, when one of the infants showed an asymmetric popliteal angle. Between 3 and 6 months of age the signs were subtle and usually consisted of asymmetric kicking in vertical suspension, which was seen in three infants by 6 months of age. Hand preference became apparent between 3 and 9 months of age (236). In some instances asymmetries of generalized movements and “fidgety” movements can be detected as early as 3 weeks of age (237). The absence of “fidgety” movements is also a good predictor for the development of dyskinetic and spastic quadriparetic cerebral palsy (238). As a rule, the more severe the hemiplegia, the earlier do the abnormalities make their appearance. Other signs of hemiparesis include inequalities of muscle tone, asymmetry of fisting, and inequalities of the parachute reaction (see Introduction chapter, Fig. I.7). In many instances, parents also note poor feeding and frequent regurgitation.

Ingram observed a remarkably constant sequence of neurologic manifestations in the progression from hypotonia to spasticity (239). The hypotonic stage lasts from 6 weeks to 17 months or longer. In general, the longer its duration, the more severely handicapped is the child.

In a significant percentage of children, 1.3% in the series of Skatvedt (240), but 20% of the group with cerebral palsy at the Southbury Training School, Southbury, Connecticut (241), the hypotonic state persists beyond the second or third year of life, and, accordingly, the condition is designated as hypotonic (atonic) cerebral palsy, a term first proposed by Förster in 1910 (242). The differential diagnosis between hypotonic cerebral palsy and abnormalities in muscle function is discussed more fully in Chapter 16.

A stage of intermittent dystonia often becomes apparent when the infant is first able to hold up his or her head. At that time, abrupt changes in position, particularly extension of the head, elicit a response that is similar to extensor decerebrate rigidity. The frequency with which this intermediate dystonic stage is observed is probably a function of the care with which neurologic observations are performed. In the majority of children, dystonic episodes are present from 2 to 12 months of age. Ultimately, as rigidity appears, episodes become less frequent and more difficult to elicit. Transient dystonic posturing, notably torticollis or opisthotonos, has been associated with maternal use of cocaine (243).

In a smaller number of children with cerebral palsy, a transition occurs from the diffuse hypotonia seen in the neonatal period to an extrapyramidal form of cerebral palsy. Although a characteristic feature of the motor activity of the healthy premature and full-term infant is the presence of choreoathetoid movements of the hands and feet, the fully developed clinical picture of dyskinesia is not usually apparent until the second year of life (Table 6.9) (221). Until then, the neurologic picture is marked
by persistent hypotonia accompanied by retention of the immature postural reflexes. In particular, the tonic neck reflex, the righting response, and the Moro reflex are retained for longer periods in infants with extrapyramidal cerebral palsy than in those in whom a spastic picture predominates (221). In general, the earliest specific evidence for extrapyramidal disease is observed in the posturing of the fingers when the infant reaches for an object (Fig. 6.20). This can be noted as early as 9 months of age, and, as a rule, the early appearance of extrapyramidal movements indicates that the ultimate disability will be mild. In the child with dyskinesia, dystonic posturing can be elicited by sudden changes in the position of the trunk or limbs, particularly by extension of the head. Characteristically, when an infant with early extrapyramidal disease is placed with support in the sitting position, the infant resists passive flexion of the neck and tends to retroflex the back and shoulders. The assessment of the general movement patterns was used by Prechtl to accurately predict the ultimate evolution of cerebral palsy (244).

Every physician examining infants suspected of having sustained a cerebral birth injury has encountered a group of patients who appear to have clear-cut neurologic signs in early infancy but who, on subsequent examinations, have lost all of their motor dysfunction (245). Many of these have not escaped brain damage; follow-up studies show them to have delayed milestones, a high incidence of mental retardation (22%), abnormalities of extraocular movements (22%), and afebrile seizures (4.4%) (245). Still, approximately one-third appear normal, or, at worst, demonstrate mild perceptual handicaps or hyperkinetic behavior patterns (246).

With the decline in neonatal mortality and improved documentation, the prevalence of cerebral palsy rose significantly in most countries during the 1970s and 1980s but has remained fairly stable in the last 10 years. A recent estimate is 2.4 per 1,000 children (247). This contrasts with in prevalence of 1.23 in 1,000 3-year-old children born in Northern California between 1983 and 1985 (248). In this group of children, 53% of children with cerebral palsy had birth weights of 2,500 g or less and 28% had birth weights less than 1,500 g. In Sweden, 2.12 in 1,000 children born between 1991 and 1994 were diagnosed as suffering from cerebral palsy; 38% of these children were preterm (31a). The increase in the incidence of cerebral palsy as a consequence of a decrease in neonatal mortality and the increased frequency of multiple births also has been reported from Australia, England, and Ireland (249).

In older children, the manifestations of cerebral birth injuries are so varied that it is difficult to devise an adequate scheme of classification. Yet the differences in cause, clinical picture, and prognosis require that cerebral palsy be subdivided into various entities based on the clinical picture. In this chapter, the following system is used:

Table 6.10 shows the incidence of some of these forms of cerebral palsy. As many as one-seventh of children show a mixture of clear-cut pyramidal and extrapyramidal signs, and almost every child with spastic diplegia is found to
have spastic quadriparesis on careful examination of the upper extremities. A number of authors have distinguished an ataxic form of cerebral palsy (239,240). In Skatvedt’s series, published in 1958, this condition accounted for approximately 7% of children with cerebral palsy (240). Subsequent series do not distinguish this form of cerebral palsy. In most instances, neuropathologic studies on patients with cerebellar signs have revealed malformations of the cerebellum, often accompanied by even more conspicuous malformations of the cerebral hemispheres (151,152,157). Conversely, most patients with histologically verified lesions of the cerebellum attributable to perinatal trauma or asphyxia did not show cerebellar signs during their lifetime (139). Although Crothers and Paine (231) distinguished a condition termed spastic monoplegia (Table 6.11), this entity is probably rare. Most children fitting this designation have spastic hemiparesis revealed on subsequent examinations (231).

Evidence of intrauterine distress includes an alteration of fetal heart rate pattern, the passage of meconium, and abnormalities in fetal acid–base status as determined by scalp or cord blood sampling. Although beat-to-beat variability and deceleration beginning after the start of a contraction and peaking well after the peak of the uterine contraction (late decelerations) are ominous in terms of fetal well-being, they do not seem to predict ultimate neurologic or intellectual deficits (308). In the experience of Valentin and coworkers, infants who demonstrated a pattern of reduced variability, reduced variability with late decelerations, or bradycardia with late decelerations had the most abnormal cord arterial pHs (309) and would therefore be at greatest risk for neurologic abnormalities. However, Nelson and coworkers in a retrospective study found that 37% of children with cerebral palsy showed none of the risk factors on prenatal fetal heart monitoring (310). In view of the heterogeneity of the cerebral palsy population this is not a surprising finding. Passage of large amounts of meconium after rupture of the fetal membranes also correlates with the subsequent presence of neurologic deficits (311). However, infants with developmental abnormalities as well as those who have sustained neurologic injuries before the onset of labor are likely to demonstrate abnormal fetal heart rate patterns during labor (308). Mothers of infants whose neurologic handicaps were secondary to cerebral malformations have a statistically higher incidence of vaginal bleeding during pregnancy and a suggestively higher incidence of prenatal infections. Additionally, a breech presentation and the application of midforceps are significantly associated with both mechanical and asphyxial birth injury.

The relationship between arterial cord blood values and ultimate neurologic deficits has been the subject of several investigations. Infants with a complicated neonatal course tend to have a lower cord arterial pH and a significantly greater mean base deficit than those with an uncomplicated neonatal course (312,313). However, fewer than one-half of infants with an arterial cord pH of 7.00 or less have neonatal complications (314), and, conversly, infants born with acute birth asphyxia do not invariably demonstrate umbilical artery acidemia at birth (315). Hermansen pointed out that an infant may produce acid in the tissues without developing acidemia, with different mechanisms being responsible for a lack of acidosis in the presence of acidemia. Thus in the presence of complete circulatory arrest, acid may accumulate in fetal tissues but not enter the bloodstream until after resuscitation and after circulation has been restored (315). Concomitant hypoglycemia could result in an inability to produce lactate. Because lactate is a crucial energy substrate in the brain during recovery from ischemia, a reduction in lactate would put the neonate at an increased risk for neurologic damage (316,317). Sehdev and coworkers found the combined presence of a high arterial base deficit and a low 5-minute Apgar score to be good predictors for neonatal morbidity (314). An arteriovenous pCO₂ difference of greater than 25 torr has been found to be a highly
specific parameter for identifying asphyxiated infants who go on to develop HIE (318). Hypoglycemia (blood sugar of less than 40 mg/dL on the initial specimen) is another important risk factor for perinatal brain injury, particularly in infants who require resuscitation and have severe fetal acidemia (318a).

An infant small for gestational age (SGA) (i.e., one whose birth weight is more than two standard deviations below the mean for any given week of gestation) is at an increased risk for perinatal asphyxia, the meconium aspiration syndrome, hypoglycemia, and the complications of polycythemia. In a survey of children with spastic cerebral palsy, 30.4% were below the 10th percentile for birth weight and 15.5% were below the 3rd percentile for birth weight. SGA infants between 34 and 37 weeks’ gestational age were at highest risk of developing spastic cerebral palsy (319). Amiel-Tison and Pettigrew suggested that intrauterine growth retardation is an adaptive response to placental insufficiency and that as judged from their physical examinations and their brainstem-evoked responses, SGA infants are neurologically more mature than infants whose birth weights are appropriate for their gestational age (AGA) (320). Generally, the infant who is both underweight and short experienced an insult in early fetal life, whereas the infant who is underweight but of normal growth suffered an insult that was briefer and occurred later during gestation. Under both circumstances, brain growth, as inferred from head circumference, is less affected than body weight or length.

Whether SGA infants ultimately develop learning disorders and minor cognitive and behavioral abnormalities beyond those caused by socioeconomic factors has not been fully resolved. Long-term follow-up studies indicate that SGA infants have a slightly increased risk for developmental delay when compared to AGA infants (321). Although verbal IQ was not confounded by parental demographic and child-rearing factors, these variables accounted for much of the deficits in nonverbal IQ (322). Sommerfelt and coworkers found maternal smoking to be associated with birth of a SGA infant and a reduction of 5 points in performance IQ at 5 years of age (322). They were, however, unable to determine whether the effect of maternal smoking was due to maternal socioeconomic or child-rearing factors (322).

The prevention of perinatal trauma and asphyxia is largely the task of the obstetrician and is, therefore, outside the scope of this text. Treatment of the asphyxiated neonate is largely outside the domain of the neurologist who will be called on to assist in the management of cerebral edema and neonatal seizures.

Neither fluid restriction nor corticosteroids are effective in the management of postasphyxial cerebral edema. Mannitol will reduce brain edema but does not affect the clinical outcome. The value of hyperventilation or osmotic agents in improving the outcome of infants with cerebral edema is probably also minimal (223). Mild head cooling and mild systemic hypothermia have also been suggested for treating the infant with severe HIE (387). Although two small randomized, controlled trials demonstrated no evidence of harm, evidence is inadequate to assess its efficacy. The treatment of neonatal seizures is considered in Chapter 14.

Experimental work suggests that there is a therapeutic window of some 1 to 2 hours between the time an infant has been resuscitated successfully from perinatal asphyxia and the onset of the cascade of secondary changes that lead to apoptotic and necrotic neuronal death. Several pharmaceutical agents have been proposed that could intervene in the production of secondary brain injury.

N-methyl-D-aspartate (NMDA) antagonists, notably MK-801, have been found to protect the brain from experimentally induced asphyxial injuries, even when administered up to 1 hour after the insult. None of them has been used in the human neonate. Calcium-channel blockers, inhibitors of nitric oxide production, and oxygen-free radical inhibitors and scavengers such as allopurinol and indomethacin also have been suggested but have
not received any clinical trials (388). Under experimental conditions, antisense oligodeoxynucleotides to one of the NMDA-receptor channel complexes inhibit synthesis of the protein component of the channel and selectively reduce the expression of NMDA receptors and the extent of focal ischemic infarctions. Although the use of antisense nucleotides represents a novel and exciting approach to the treatment of asphyxia, it has undergone no clinical trials (389).

Several regimens have been suggested to prevent IVH or to reduce its severity. Antenatal administration of glucocorticoids, usually betamethasone or dexamethasone, clearly reduces the incidence of IVH in premature infants. Our nurseries at Cedars-Sinai Medical Center use prenatal treatment of the fetus with corticosteroids to prevent respiratory distress syndrome and thereby lessen the likelihood for IVH. Antenatal treatment with betamethasone is also associated with a decreased incidence of cystic PVL in very small premature infants (111).

Infants with birth weights lower than 1,250 g are given prophylactic indomethacin. Indomethacin (0.1 mg/kg) given at 6 to 12 hours and every 24 hours for two further doses lowered the incidence of the most extensive hemorrhages (grades III and IV) (390,391). Indomethacin reduces cerebral blood flow velocity and increases cerebral vascular resistance and thus attenuates the adaptive vasodilatory response to asphyxia (392). It also reduces the formation of free radicals and accelerates maturation of the microvasculature in the germinal matrix (393). Ethamsylate administration also is associated with a lowered incidence of IVH, particularly the more extensive hemorrhages (394).

We believe that maintaining a good airway with a good cardiovascular support system and trying to prevent excessive swings in blood pressure and cardiac output reduce the likelihood of a large IVH. Additionally, allowing the pCO₂ to increase, but maintaining it below 55 mm Hg, appears to reduce the incidence of pneumothorax, which contributes to the evolution of IVH. Such factors as hypoxemia, acidemia, and rapid volume expansion all can lead to extension of the hemorrhage and should therefore be avoided.

Other therapeutic modalities that have been investigated include the prenatal administration of phenobarbital, which in combination with vitamin K does not appear to reduce the incidence of severe IVH (395).

Vitamin E supplementation in preterm infants reduced the risk of IVH but increased the risk of sepsis. In very low birth weight infants it increased the risk of sepsis, and reduced the risk of severe retinopathy and blindness among those examined. This evidence does not support the routine use of vitamin E supplementation by intravenous route at high doses (396).

Antenatal treatment with magnesium has not resulted in a statistically significant reduction of IVH or cerebral palsies in very preterm births (397).

In treating progressive ventricular enlargement, many centers suggest that lumbar punctures should be performed first unless ultrasound or CT evidence exists of noncommunication between the ventricles and the spinal subarachnoid space. Acetazolamide or furosemide can be used as an adjunct to lumbar punctures. If these interventions fail to arrest the progressive ventricular enlargement, external ventricular drainage is indicated (398). A ventricular reservoir is inserted, and serial reservoir taps are performed until the ventricular fluid protein falls below 2,000 mg/dL and there has been no resolution of the hydrocephalus, as indicated by serial ultrasounds. At that time, a ventriculo-peritoneal shunt replaces the reservoir. Serial monitoring with ultrasonography is recommended in all infants who have had an IVH, whatever treatment has been adopted. Randomized, controlled trials have found that neither acetazolamide nor furosemide can control posthemorrhagic ventricular dilatation (399).

Only a small proportion of patients undergoing a shunting procedure become neurologically and developmentally healthy adults. Several studies have demonstrated that in these patients nonverbal skills are more impaired than verbal skills. The outlook for infants whose hydrocephalus responds to medical treatment with repeated lumbar punctures, osmotic diuretics, or acetazolamide is significantly better than it is for those who require a shunt (400). It is equally evident that the extent of parenchymal damage that attends IVH determines the ultimate prognosis (4,400).

The treatment of children with cerebral palsy has been the subject of innumerable publications, most of them surprisingly uncritical and devoid of controls. Although cerebral palsy is, by definition, a “static encephalopathy,” the associated musculoskeletal pathology is progressive Tizard proposed that before treatment is initiated, the following points should be considered (401): (a) Does the child need treatment? (b) What are the aims of treatment? (c) Do the family and child have the time required for treatment? (d) Will treatment disrupt family life?

Various means are available for the treatment of spasticity, which represents the most important disorder of motor control in children with cerebral palsy (247). Management involves the use of a continuity of modalities, and most patients require a combination of modalities. Thus, Graham and coworkers found that the combination of
physiotherapy and orthotics increased the benefits from intramuscular botulinum toxin (402).

Physiotherapy is the most traditional and principal nonsurgical form of treatment. Its aims are to prevent additional deformities and to promote functionally useful posture and movements. From the studies of Crothers and Paine (231) and Paine (403), it seems clear that to be most effective, the program must be determined by the nature of the handicap. Fixsen divided orthopedic management into two aspects: the management of the child who will ultimately become ambulatory and hat of the nonambulatory child (404).

Intensive physical therapy (1 hour a day, 5 days a week) has been considered to be helpful for children with spastic hemiparesis and spastic diplegia. If possible, effectiveness of management should be monitored by gait analysis (404). Most children with these forms of cerebral palsy require regular physiotherapy as soon as they begin to walk, with special attention to the presence of contractures in the lower extremity. There is little evidence-based data supporting the effectiveness of physiotherapy. In evaluating effectiveness of the physiotherapy for spastic children, Wright and Nicholson found no evidence that it improved the range of dorsiflexion of the ankle or abduction of the hip or that it affected the retention or loss of reflex automatisms (405). A controlled study of infants with spastic diplegia found that the routine use of physical therapy offered no short-term advantages over an infant stimulation program (406). A randomized, controlled trial also found no statistical difference between intensive and routine physiotherapy (2 to 15 hours per 3 months) when measured by changes in gross motor function or performance (407). Bracing often is necessary. According to Tardieu and his group, bracing for 6 hours a day prevents contractures at the ankle (408). It is difficult to state categorically at what age treatment should be initiated. The controlled studies by Paine indicate that the eventual gait is better and contractures are fewer when physical therapy is begun before the age of 2 years (403).

Much controversy surrounds the management of the hemiplegic hand. No differences in hand function, quality of upper extremity movement, or parents’ perception of the child’s hand function were noted when children received intensive neurodevelopmental therapy plus casting or regular occupational therapy. However, casting appears to be more effective in older children (409). Ultimate hand function is usually not improved by physical therapy or surgery, and when the hemiplegia is complicated by a hemisensory deficit, the affected hand will probably never do more than assist the good hand (410). Forced-use treatment of the hemiparetic upper extremity has shown some benefit, but growth of the affected side is not improved by any mode of treatment (411). When the hemiplegia is mild, children achieve a good gait whether treated or not.

Few objective data indicate that either physiotherapy or orthopedic measures improve the disabilities of the child with a predominantly extrapyramidal disorder. In these patients, hand function and the quality of the ultimate gait depend principally on the original severity of the disorder, and most children with extrapyramidal movements, particularly those with unimpaired intelligence, are able to teach themselves to assume certain positions and perform substitute movements by which involuntary movements are avoided. The substitute movements are often so complex that they could not possibly be invented by therapists. In all instances, coordination and function tend to improve with age. Various drugs have been tried for relief of the extrapyramidal movements. None of these has resulted in long-term benefit sufficient to outweigh the side effects (401).

Various pharmacologic approaches have been tried for relief of spasticity. The most commonly used medications are baclofen, diazepam, and dantrolene. Doses are started at low levels and are increased until spasticity improves or side effects, notably drowsiness, occur. As a rule, these medications only have a small effect. The intrathecal administration of baclofen by continuous infusion is proving to be an effective means for treating spasticity in the lower extremities in a selected group of patients who have responded favorably to a trial dose of intrathecal baclofen (412,413,414). Complications of this procedure are seen in about one-fourth of patients. In the experience of Gooch and coworkers, they mainly involved mechanical failures of the pump or the catheter (415). Side effects from the drug are usually temporary and can be managed by reducing the rate of infusion.

The injection of botulinum toxin A (Botox) into the gastrosoleus and hamstring group of muscles appears to improve gait for up to 6 months and may help to delay or obviate the need for serial casting or orthopedic surgery (416). The effect of the toxin becomes evident within 12 hours to 7 days, and the effect can last between 2 and 10 months (416,417). Graham and coworkers presented a consensus statement with respect to patient selection and assessment, drug dosage, injection technique, and outcome measurements (418). Forssberg and Tedroff believed that treatment should be initiated at a time when children are still developing their motor control apparatus. This might prevent them from entering a vicious cycle in which CNS lesions affect the musculoskeletal system, thereby preventing the development of motor functions (419). Graham confirmed their experience (420). In addition, experimental data on the formation of a cortical somatotopic map during early life indicate that the periphery plays an instructional role on the formation of central neuronal structures (421). The
economics and effectiveness of this type of treatment also require further study.

No evidence indicates that treatment programs that attempt to modify sensory input, inhibit primitive reflexes, or modify or inhibit abnormal movement patterns are ever successfully incorporated into the maturing nervous system with resulting improvement in motor function. Several authorities and parent groups have expressed their enthusiasm for infant stimulation programs. Even lacking objective evidence of their usefulness, they are apparently beneficial in controlling behavior and are here to stay (422,423).

Underlying the prognosis of the asphyxiated infant is the unanswered question about whether asphyxia produces a continuum of brain damage (i.e., whether mild asphyxia causes a small amount of damage and more severe asphyxia causes more severe damage) or a threshold exists beyond which the brain is damaged in an all-or-none manner. It is our opinion that there is a threshold of asphyxia beyond which there is a continuum of brain injury, and below this threshold there is no damage.

It is clear that most infants who experience perinatal asphyxia do not exhibit abnormal neurologic signs or subsequent evidence of brain injury. It is also clear that unless clinical signs of neurologic abnormality are present during the neonatal period, the outcome of perinatal asphyxia is entirely favorable. In the experience of Dubowitz and Dubowitz, all term infants who were healthy on neonatal examination were healthy at 1 year of age (450). The best predictor of outcome from a perinatal asphyxia is the clinical status of the infant during the neonatal period as assessed by the Sarnat scale (see Table 6.8). Numerous studies have confirmed the prognostic value of such an assessment (222,451). In terms of outcome, staging is best done according to the most severe signs and according to the most severe stage of encephalopathy between 1 hour and 7 days of life (451). In the experience of Robertson and
Finer, all infants with stage 1 encephalopathy had a normal outcome, and their cognitive performance at 8 years of age did not differ from that of controls. The results of Handley-Derry and coworkers are similar (452). Infants with stage 3 encephalopathy fare poorly. Robertson and Finer found that 82% died by school age and 18% were severely handicapped (451). The outcome of infants with moderate (stage 2) HIE is less certain. Robertson and Finer found that 80% were neurologically normal, 15% were neurologically disabled, and 5% died (451). Of the nondisabled survivors in this series as well as in several others a large proportion had significant cognitive dysfunction and poor school performance (451). In the study of Robertson and Finer, the presence of neonatal seizures did not appear to affect the prognosis. Temple and coworkers, however, found that infants who developed neonatal seizures after an asphyxial injury demonstrated significant cognitive deficits, even when the neurologic examination was normal (453). Infants with subtle seizures and infants who demonstrate more than two seizure types tend to have a worse outcome in terms of continued epileptic manifestations and mental retardation (454). Renal injury, particularly prolonged oliguria, when associated with asphyxia also presages a poor neurologic outcome (455).

Other factors that influence the infant’s response to an asphyxial insult, and thus his or her outcome, include preexisting cerebral anomalies, fetal maturity, cerebral energy stores at the time of asphyxia, the adequacy of uteroplacental blood flow, and the fetal adaptive response to asphyxia. This adaptive response involves a redistribution of cardiac output so that blood flow to brain and heart is maintained at the expense of blood flow to kidney, the gastrointestinal tract, and the musculoskeletal system. Finally, experimental data indicate that not only can asphyxial injury cause neuronal loss, but it also can interfere with the normal developmental processes including cortical reorganization, formation of synaptic connections, and programmed cell death that may have been in progress at the time of injury (138,363,456).

In view of the interaction of these several factors, it should come as no surprise that many infants with severe and prolonged asphyxia recover without any neurodevelopmental deficits, and an arterial cord pH as low as 6.60 is compatible with normal neurologic and cognitive examinations at a mean age of 47 months (313). Nevertheless, aggressive resuscitation of the newborn infant is not in order. Levene recommends that if an infant has no cardiac output after 10 minutes of effective resuscitation, treatment should be abandoned (223). The outcome of infants with good cardiac output who do not breathe spontaneously by 20 minutes is poor, and only 25% were left without significant neurologic deficits (457). These considerations are important in a time of medical cost control.

Several other methods have been used in an attempt to identify infants with a poor prognosis. The most traditional of these has been the Apgar score, even though it is now evident that a low Apgar score does not indicate the presence of asphyxia in either term or premature infants (225,227). Whereas the predictive value of the 1- and 5-minute Apgar scores in terms of subsequent neurologic deficits is limited, term infants with 5-minute Apgar scores of 6 or less are three times as likely to be neurologically abnormal at 1 year of age as those with scores of 6 to 10 (458). The likelihood of permanent brain damage increases even more significantly when depressed Apgar scores persist. Of infants with scores of 3 or less at 10 minutes of age, 68% die during the first year of life, and 12.5% of survivors are neurologically damaged. The prognosis is even worse when an Apgar score of 3 or less persists for 20 minutes. Of those infants, 87% die, and 36% of survivors have cerebral palsy (227). A poor outcome was seen in 30% of infants with a low Apgar score but no significant acidosis (459). In some instances, this is due to acidosis paradox, discussed in another portion of this chapter (315,316).

Fetal blood sampling can provide somewhat better prognostic information, although there still are no absolute values of blood pO₂, pCO₂, or pH beyond which irreparable brain damage is certain to ensue. Studies designed to correlate the severity of acidosis in term infants at birth with outcome have failed to show any consistent relationship (460,461). When infants with umbilical cord pH of 7.0 or less are selected to be followed, the outcome is still unpredictable; however, when severe acidemia is associated with persistent bradycardia or seizures, the outcome is poor in 85% of cases (210). Because most of the asphyxial insults were believed to occur in utero during labor, electronic monitoring of the fetal heart rate was considered to be of prognostic value. Prospective studies have indicated otherwise. Although one-fourth of high-risk infants whose fetal heart rate patterns showed severe variable decelerations or late decelerations were neurologically abnormal at 1 year of age, these abnormalities did not persist into later childhood (308,462). Ekert and coworkers presented a model that can predict within 4 hours of birth a poor outcome from perinatal asphyxia (463). The three major adverse factors singled out by them were onset of spontaneous respirations after 10 minutes of life, administration of chest compression, and onset of seizures before 4 hours of age.

Because a significant proportion of infants with a neurologic examination consistent with Sarnat stage 2 fare well, prognosis in this group should be influenced by the results of neurodiagnostic tests (464).

The EEG is one of the commonly used means of evaluating the neonate with a history of perinatal asphyxia and clinical evidence for moderate or severe HIE. Recovery is more likely to occur if the tracing is normal or if it demonstrates a single focus rather than showing multifocal paroxysmal discharges or a burst-suppression pattern (465,466). Serial EEGs first obtained between 12 and
36 hours of life and then between 7 and 9 days postpartum provide a better indication of ultimate outcome (467).

CT scanning and MRI, particularly when coupled with EEG, provide the best prognostic tools. The presence of areas of decreased density in brain parenchyma seen on CT predicts a major neurologic handicap. This likelihood is particularly true when two or more focal areas of hypodensity exist or when the density of the basal ganglia is reduced (468). Asphyxiated infants with normal CT rarely exhibit major neurologic residua.

MRI provides more detail about the anatomic localization and severity of asphyxial injury and is a better means of predicting the ultimate outcome than the Apgar score or staging the severity of neonatal encephalopathy (469,470). Lesions of the basal ganglia are generally predictive of a poor outcome. Parasagittal lesions, focal areas of necrosis, have a better outlook. Asphyxiated infants with a normal MRI do well developmentally and neurologically.

MR spectroscopy is a potentially valuable prognostic tool. Neonates with poor outcomes had significantly lower N-acetyl aspartate (NAA)/choline ratios and significantly higher choline/creatine ratios in the occipital region compared with patients with good to moderate outcomes. In addition, the persistence of lactic acids predicts a poor outcome (471,472). The practical problems associated with obtaining such a study preclude it from being widely used in the early evaluation of asphyxiated infants.

The question of whether asphyxia resulting in mild to moderate neonatal encephalopathy can result in mild mental retardation or learning disabilities in the absence of gross neurologic deficits has not been resolved. Nelson and Ellenberg failed to find a statistically significant increase in mental retardation in children with low Apgar scores who did not also have cerebral palsy (227). They concluded that when mental retardation is a consequence of perinatal asphyxia, it is usually severe and accompanied by evidence of neurologic damage, notably spastic quadriparesis, athetosis, or both. Recent work by Hopkins-Golightly and colleagues however argues against the presence of an all-or-none threshold phenomenon for development of brain injury and cognitive impairment, and suggests that there is a continuum of brain injury in asphyxia (472a). This issue is further discussed in Chapter 18.

Modern methods of perinatal care have brightened the outlook in terms of survival of even the smallest premature infant. However, concurrent with a decrease in neonatal mortality, there has been an increase in the incidence of neurologic handicaps in the smallest group of preterm infants (249). The incidence of neurologic disabilities in infants born from 1978 to 1979 as compared with infants born from 1988 to 1989 is depicted in Table 6.14 (473).

Four factors contribute to the increased incidence of neurologic disability in the smallest preterm infants: the presence and severity of PVL, the presence and severity of periventricular and intraventricular hemorrhages, the effects of premature delivery and neonatal complications on cerebral cortical and white matter organization, and the likelihood that a significant proportion of premature infants are neurologically compromised prior to birth.

The presence of PVL as detected on ultrasound or by neuroimaging studies predicts the evolution of spastic diplegia over the ensuing months. In addition, children with PVL have a significantly curtailed intellect. In the study by Pharoah and colleagues, 68% of children with spastic diplegia had an IQ of 70 or higher, 15% had moderate mental retardation, and 17% had severe mental retardation (474). Visual perceptual deficits and other cognitive disorders are common in the group of children with PVL who appear to have grossly normal intelligence.

Several studies indicate that children who had sustained an uncomplicated IVH (grades I or II) have subtle neurologic deficits. Most of these are not apparent during the first few years of life but can be documented when follow-up examinations are performed between 5 and 7 years of age or later (475,476,477). The risk of developing neurologic deficits is significantly increased for infants who have sustained a major IVH. In the series of Fazzi and colleagues, only 20% of infants who sustained this lesion were considered to have developed normally at 2 years of age. By 5 to 7 years of age, this figure had declined to 8% (477). Other follow-up studies have yielded similar results (475,478).

The cause of the cognitive and attention deficits of premature infants who are grossly normal neurologically is not clear. Based on experimental work of Ghosh and Shatz (479), Volpe postulated that neurons in the subependymal germinative zone, the subplate neurons, which are involved in cerebral cortical organization, are disrupted by the presence of even minor germinal matrix hemorrhages or PVL (480). Their injury could lead to the variety of cortical deficits that are so common in small premature infants. One should also remember that most of the growth of cortical connections and complexity occurs after 25 weeks’ gestation. Whereas neuronal migration is largely complete by 25 weeks, glial migration continues, as does cortical connectivity. The cerebral cortex of extremely premature infants when imaged at gestational age 38 to 42 weeks has less cortical surface area and is less complex than that of the normal term newborn (481). It is likely that these deficits acquired during a critical period of brain development are permanent.

Numerous follow-up studies have been done on small preterm infants. The earlier reports were relatively optimistic; more recent series have noted a significant incidence of cognitive and behavioral abnormalities that lead to learning disabilities. In the United Kingdom series of infants with gestational age of 25 weeks or less, collected in 1995 by Wood and coworkers, 49% of the surviving children had significant disability detected at 30 months of postconceptual age. About half of these infants (23% of surviving infants) had severe disability (482). This study is probably more reliable than those that report outcome according to birth weight categories, which can be confounded by the inclusion of infants that are more gestationally mature but were growth restricted. Using the latter methodology, Hirata and colleagues found that in a group of infants with birth weights between 501 and 750 g, of the 40% who survived, 66% appeared normal, some 20% had borderline or low-average IQs, and approximately 10% had significant neurologic sequelae (483). Kitchen and coworkers examined infants with birth weights between 500 and 999 g at 2 years of age. Their results were similar, with 68% of surviving infants born between 1985 and 1987 having no apparent functional handicap (484). However, as a rule, a large proportion of neurodevelopmental deficits are not apparent during the first 2 to 3 years of life. The experience of Collin and coworkers, who followed infants into their preschool years, underlines this fact. They observed a downward drift in performance between infancy and preschool, with only approximately one-third of children who were deemed developmentally normal when seen between 12 and 25 months of age performing in the normal range when reassessed at approximately 4 years of age (485). The negative effect of extreme prematurity on cognitive and language function appears to be modified in infants from families with a high socioeconomic status (485a). Hack and coworkers believed that perinatal growth failure, as reflected in a subnormal head circumference at 8 months of age, predicts impaired cognitive function and academic achievement. Because most very low birth weight infants had a normal head circumference at birth, postnatal events are responsible for impairment in head growth and, by inference, brain maturation (486). More recently published studies show similar results (487,488).

The ultimate social adjustment of the patient with cerebral palsy is determined primarily by the severity of the physical and mental handicap. Individuals who can be predicted to be unable to work are those with an IQ under 50, are nonambulatory and without communication skills, and need assistance in using their hands. Those predicted for sheltered work are those with IQs between 50 and 79 who walk and have some communication skills (489). Generally, mental handicap is less of a barrier to useful employment than physical handicap. In the series of Sala and Grant, all children with spastic hemiparesis became ambulatory. Children with spastic diplegia became ambulatory in 86% to 91% of cases, with those who were able to sit independently by 2 years of age having the best prognosis for ambulation (490). Patients with hemiplegia are most often able to become competitive, approximately one-third of them being economically productive. One-fourth of patients with extrapyramidal disorders are ultimately able to work competitively (287). The outlook for patients with spastic quadriparesis is far worse. In the series of Crothers and Paine, published in 1959, none of the adult patients with spastic quadriparesis was gainfully employed (231). Also important in the ultimate prognosis is the presence or absence of associated disabilities, including seizures, and impairment of vision or hearing.

Adults with cerebral palsy develop a variety of new functional disabilities. These include lower extremity contractures, scoliosis, cervical pain, and back pain (491). For those who are mobile when they become adults there is a marked decline in ambulation in late adulthood, and few of the 60-year-olds who walked well preserved this skill over the following 15 years (492). Speech and self-feeding appear to be well preserved over the years. Survival rates of ambulatory older adults are only moderately worse than those of the general population (492).

Finally, and most difficult to evaluate when first seeing a patient, is the attitude of the patient and his or her family to the disability and the stability of the home in view of the severe and chronic emotional trauma caused by cerebral palsy (489).

Various estimates of life expectancy of children with cerebral palsy have been published. As a rule, life expectancy in mildly to moderately disabled children is only slightly curtailed (493), and 98% of persons without severe disabilities live to age 35 years (473). Life expectancy for
the profoundly handicapped and immobile child, however, is severely curtailed (494), and more than one-third of those with severe disability l die before age 30 years (495).

As a rule, brachial plexus injuries result from stretching of the plexus owing to turning the head away from the shoulder in a difficult cephalic presentation of a large infant. An injury also can be consequent to stretching of the brachial plexus owing to traction on the shoulder in the course of delivering the aftercoming head in a breech presentation (527,528). The condition has been reported in a few instances after delivery by cesarean section (529). However, whatever scant evidence exists for a classical brachial plexus injury resulting from intrauterine maladaptation is principally based on faulty interpretation of EMG (530,531). When intrauterine palsies do occur, they are characterized by limb atrophy and abnormal dermatoglyphics at birth (532).

In most instances, the brachial plexus is compressed by hemorrhage and edema within the nerve sheath. Less often, there is an actual tear of the nerves, or avulsion of the roots from the spinal cord occurs, with segmental damage to the gray matter of the spinal cord (443). With traction, the fifth cervical root gives way first, then the sixth, and so on down the plexus. Thus, the mildest plexus injuries involve only C5 and C6, and the more severe involve the entire plexus (528).

The incidence of brachial plexus injury in 2000 was 4.6 per 1,000 births (533). This figure should be contrasted with an incidence of 2.2 per 1,000 births in 1994 as determined from the Swedish Medical Birth Registry (534) and an incidence of 1.3 per 1,000 births in 1980, a significant increase over the course of the last 20 years. The incidence of brachial plexus palsy is 45 times greater in infants with birth weights greater than 4,500 g than in those that weigh less than 3,500 g (534). In approximately 80% of infants, Erb-Duchenne paralysis is confined to the upper brachial plexus (535). Involvement was unilateral in approximately 95% of instances in the series of Eng and coworkers (532). One of us (J.H.M.) has seen an infant with what appeared to be bilateral brachial plexus injury who had agenesis of the biceps muscles. The right side is more frequently affected than the left, with the ratio in the series of Eng and coworkers being 55:45 (535). The weakness is recognized soon after delivery, with the involved arm assuming a characteristic posture. The shoulder is adducted and internally rotated, the elbow extended, the forearm pronated, and the wrist occasionally flexed. This position results from paralysis of the deltoid, the supraspinatus and infraspinatus, biceps, and brachioradialis muscles. The Moro reflex is absent or diminished on the affected side, but the grasp reflex remains intact. Unlike in the healthy neonate, the biceps reflex is abolished or is less active than the triceps reflex. In most infants, a sensory loss cannot be demonstrated, although occasionally cutaneous sensation is lost over the deltoid region and the adjacent radial surface of the upper arm.

Fractures of the clavicle or humerus, slippage of the capital head of the radius, and subluxation of the shoulder and the cervical spine often accompany a brachial plexus injury (528). When a significant degree of injury to the fourth cervical root is present, phrenic nerve paralysis can accompany injury to the upper brachial plexus (536). An affected infant can show signs of respiratory distress, including tachypnea, cyanosis, and decreased movement of the affected hemithorax. When the phrenic nerve palsy is unaccompanied by injury to the brachial plexus, as occurs occasionally, the condition can mimic congenital pulmonary or heart disease (537,538).

In the more severely involved infants, 12% in the series of Eng and colleagues (532), the entire brachial plexus is damaged and the arm is completely paralyzed. The limb is flaccid, the Moro and grasp reflexes are unelicitable, deep tendon reflexes are absent, and sensory loss occurs over a portion of the extremity.

Isolated paralysis of the lower part of the brachial plexus (Klumpke paralysis) is relatively uncommon. It constituted only 2.5% of brachial plexus birth palsies in the experience of Ford (257), and a spinal cord lesion should be suspected whenever the paralysis involves the intrinsic muscles of the hand (532). The clinical picture of a lower plexus injury includes weakness of the flexors of the wrist and fingers, an absent grasp reflex, and a unilateral Horner syndrome caused by involvement of the cervical sympathetic nerves. Loss of sensation and sudomotor function over the hand can sometimes be demonstrated. Interference with the sympathetic innervation of the eye results in a delay or failure in pigmentation of the iris, and it is one of several causes for heterochromia iridis (539).

The diagnosis of brachial plexus injury is usually readily apparent from the posture of the affected arm and from the absence of voluntary and reflex movements. Radiographic examinations can detect associated fractures, usually of the clavicle, humerus, or both, whereas fluoroscopy can be used to ascertain any limitation of diaphragmatic movement. In severe injuries causing avulsion of the spinal roots and bleeding into the subarachnoid space, the CSF can be bloody. MRI can be used to visualize the brachial plexus and to demonstrate root evulsions (540,541). The presence of pseudomeningoceles is a bad prognostic sign (540). EMG performed 2 to 3 weeks after the injury can confirm the extent of denervation. Because fibrillation potentials are seen in the proximal musculature of some healthy newborns during the first month of life and in the distal musculature during the first 3 months of life, an indication that the muscles are not completely innervated, both sides should always be studied (542). Repeated EMG
examinations at 6-week intervals indicate the degree of recovery (532).

Congenital Horner syndrome can occur in the absence of trauma and can be associated with anomalies of the cervical vertebrae, enterogenous cysts, or congenital nerve deafness (543,544). The association of heterochromia iridis, congenital nerve deafness, white forelock, broad root of the nose, and lateral displacement of the medial canthi of the eyes and inferior lacrimal puncta is well recognized under the term Waardenburg syndrome (see Chapter 5) (543,544).

Treatment and prognosis is a function of the severity of the injury. Hoeksma and coworkers distinguished five grades of injury (533):

In grades 1 and 2, complete recovery can be expected. In grade 1, this occurs within days to 2 to 3 weeks; in grade 2, recovery can take up to several months. In grade 3, axonal regeneration is complicated by intrafascicular scarring and the regeneration of axons in foreign endoneural tubes. In grades 4 and 5, axon regeneration can take several months or never. As a rule, external rotation and supination are the most affected movements and the last to show recovery.

Gentle passive exercises of the affected arm should be instituted at approximately 1 week after birth. The infant’s sleeve should be pinned in a natural position rather than in abduction and external rotation as many texts suggest (257). Follow-up studies indicate that overimmobilization of the affected arm is conducive to contractures and deformities that can persist despite spontaneous recovery of nerve function (545).

There is no consensus with respect to selection criteria for surgical intervention. Grossman reviewed the criteria for early surgical repair and stressed that whether surgery is performed at 3 months of age, as recommended by some surgeons, or at at 6 to 8 months of age makes little difference in terms of ultimate functional outcome (546). His group also has found that even when surgery is performed after one year of age, there can be significant functional improvement in selected cases (546a). Hoeksma and colleagues stated that when there is no sign of biceps by 3 months of age, the outlook is poor (533). Many surgeons use the absence of biceps contraction at 3 months of age as the criterion for placing an infant into the surgical group (546). The good surgical results with early intervention should be tempered by reflecting on the high incidence of improvement with conservative management (547,548). In the series of Michelow and colleagues, 92% of infants recovered spontaneously (549). Hoeksma and coworkers, however, reported only 66% complete recovery. In half of these children recovery was delayed up to 16 months of age (533). The presence of elbow flexion and elbow, wrist, and finger extension at 3 months of age is predictive of a good recovery (533,546). Eng and colleagues believed that if contraction occurs in the biceps and deltoid muscles by 2 months of age, shoulder function will recover almost completely. Mild sequelae, such as winging of the scapula and limited shoulder flexion and abduction, are common, however (532). We believe that if no improvement is noted in 3 to 6 months, surgical exploration of the brachial plexus should be considered, with repair of the damaged segment. In most instances in which poor recovery occurs, nerve damage is caused by avulsion of the spinal roots and surgical repair would, therefore, not be expected to result in any improvement.

In most cases, most of the recovery occurs between 2 and 14 months of age, with the condition becoming essentially stationary thereafter (534). Rossi and colleagues, however, believed that significant improvement is still possible up to the start of school (550).

As a rule, when the entire plexus is involved and an associated Horner syndrome exists, the outlook for full recovery is poor. Almost all children are left with hypoplasia of the limb and well-defined motor deficits, usually more marked in the proximal musculature. Contractures at the shoulder and elbow are common; approximately one-third have a clear-cut sensory deficit, whose location is variable. Trophic changes involving the fingers are unusual (535,550). They were seen in 2% of children in the series of Eng and colleagues (532).

Some infants have an apparently good return of neuromuscular function and sensation yet are unable to use the affected arm (535). Probably, transitory sensory motor deprivation in early life impairs the development of normal movement patterns and the organization of cortical body image (551). This would be in line with what has been observed with respect to the organization of the visual system after early deafferentation (552).

Because idiopathic torticollis, a relatively common condition, is in some instances believed to be related to perinatal trauma, it is considered in this chapter.

Torticollis, meaning twisted neck, is a syndrome characterized by contracture of the sternocleidomastoid muscle accompanied by a tilt of the head to the affected side and a rotation of the neck so that the chin points to the opposite shoulder. Asymmetry of the skull and facial bones can be present.

The condition can be congenital, acquired, spasmodic, or intermittent.

Congenital muscular torticollis is encountered most frequently. It is characterized by the presence of a tumor in the affected sternocleidomastoid muscle. In the majority of cases, the tumor develops during the first few weeks of life, gradually disappearing by age 46 months. In the experience of Coventry and Harris (557), the tumor disappeared at a mean age of 14 weeks.

Considerable debate surrounds the cause of this condition. Although some instances are associated with congenital malformations of the cervical spine, abnormalities in the function of the extraocular muscles, and tumors of the brainstem or spinal cord, the majority of cases are truly idiopathic. The two most likely etiologic factors are restricted movement of the head owing to an unusual fetal posture or amniotic adhesions, and perinatal trauma to the sternocleidomastoid muscle (558). A relatively high incidence of breech deliveries (28% to 40%) is compatible with either cause.

Pathologic examination of the muscle usually reveals extensive fibrosis and nonspecific myopathic changes. When the condition is chronic, evidence exists for denervation and reinnervation, probably secondary to repeated episodes of minor trauma to the muscle, which results in an entrapment neuropathy of the accessory nerve. The innervation of the sternocleidomastoid is unique in the body because the accessory nerve to the clavicular head passes through the sternal head and may become entrapped or compressed by fibrosis, producing denervation of the clavicular head (558). Separate arterial supplies predispose to ischemia in the sternal head, resulting in focal myopathy with fibrosis (558). The sternocleidomastoid tumor is not a neoplasm, either benign or malignant; it is composed of actively proliferating fibroblastic tissue that surrounds fragments of atrophic or degenerating muscle fibers (559,560). The initial tumor is usually a hematoma.

Caputo and coworkers noted that when torticollis is the consequence of disturbed function of the extraocular muscles it disappears when the infant is placed into the supine position or when one eye is occluded (561).

Considerable debate surrounds the treatment of congenital torticollis. However, the majority of clinicians opts for early physical therapy measures such as stretching and massage rather than surgical sectioning of the sternocleidomastoid muscle, removal of the tumor, or a combination of the two procedures (562). With physical therapy, the torticollis resolves in 70% of children by 12 months of age regardless of its severity or the presence or absence of focal fibrosis (563).

Acquired torticollis most commonly is associated with an infratentorial tumor. It also can be seen with colloid cysts of the third ventricle, syringomyelia, and spinal cord tumors. On rare occasions, it develops in the course of a progressive muscular disease.

The association of torticollis with hiatus hernia and gastroesophageal reflux (Sandifer syndrome) was first reported by Kinsbourne (564). Torticollis can be present at birth; generally, no shortening of the sternocleidomastoid muscle occurs.

Spasmodic torticollis, occurring in isolation or progressing to a focal, segmental, or generalized dystonia, is rare during childhood. This condition is discussed to a greater extent in Chapter 3.

Paroxysmal torticollis is characterized by recurrent episodes of head tilting starting in infancy. These episodes are sometimes accompanied by vomiting and ataxia. The EEG and caloric tests during an attack are usually normal. Generally, symptoms subside within a few hours or days. Attacks occur at monthly intervals, and the condition remits spontaneously within a few years or is replaced by vertigo or migraine (565,566). A family history of similar attacks is not uncommon, and in some families there is a linkage to a calcium-channel gene (CACNA1A) (566).