Neurologic Emergencies in Infancy and Childhood

*Deceased

Preface

This second edition of Neurologic Emergencies in Infancy and Childhood is written in response to colleagues who have requested an updated version of the book because of numerous advances in child neurology and medicine in general since our first offering. In preparing this text, we and our contributors designed chapters for easy use in emergency and general medical situations. We have followed a similar format to our prior edition, and all chapters have been revised, rewritten, or replaced. To date, it has been used by multiple physicians in its original English version and translation into both Russian and Portuguese.

On the advice of colleagues and those who have reviewed the first edition, chapters have been expanded or added, whereas others have been deleted. As the incidence of Reye syndrome has declined, perhaps secondary to the decrease of aspirin administration to children, that chapter has been deleted, and discussions are found in the chapters concerning coma and metabolic disease. A separate section on spinal cord disorders has greatly expanded the previous chapter on myelomeningocele. It includes an important discussion on spinal cord trauma and other causes of paraplegia. Lastly, a separate chapter on psychiatric emergencies possibly presenting with neurologic signs or symptoms has been added in this edition. With the great number of advances in neuroradiology, this chapter includes discussions of now commonly used magnetic resonance imaging as well as computed tomography and ultrasound.

We have requested that authors, whenever possible, continue to give flow charts and graphs more in-depth explanations in the text, maximizing the usefulness of the book in urgent situations. Information concerning drug dosages and the use of newer agents, including antibiotics, antiepileptics, and other drugs, can be found throughout all chapters.

But with all these advances that have occurred over time, other changes also come about. We mourn the loss of two friends and colleagues who authored chapters for this edition. Dr. James F. Schwartz of Atlanta was also a contributor to our first edition. Dr. Robert Hussey of Richmond added his expertise in spinal cord injury to this edition.

As the second edition is completed, we again thank all the authors for their commitment to this project. We thank our families, colleagues, and residents for the continued support and encouragement.

John M. Pellock, M.D. and Edwin C. Myer, M.D.

Chapter 1

Neonatal Intracranial Hemorrhage and Hypoxia

JAMES F. SCHWARTZ*, PETER A. AHMANN, FRANCINE D. DYKES and ALFRED W. BRANN

Publisher Summary

This chapter discusses the neonatal intracranial hemorrhage and hypoxia. The primary location of these hemorrhages is the subependymal and periventricular region, in the germinal matrix tissue over the head of the caudate nucleus at the level of the foramen of Monro. It is found that before 26 to 28 weeks of gestation, the germinal matrix is a highly vascularized zone of the developing brain, a site of proliferating neuron and glial precursors. Most of these risk factors presumably influence the development of periventricular-intraventricular hemorrhage (PVH-IVH) through their varied effects on cerebral blood flow, especially to the germinal matrix vessels. A number of clinical studies suggest that the fundamental or common thread for many of these risk factors associated with the development of PVH-IVH was a loss of cerebral autoregulation. PVH-IVH may not be the final determinant of the cerebral injury but rather the hemorrhage is followed by significant decreases in cerebral blood flow in the affected hemisphere.

Significant advances in obstetrical and neonatal intensive care and the development of sophisticated technology have resulted in marked improvement in preterm infant mortality and changing patterns of neonatal neurologic disorders. Coincident with the decline in obstetrical birth injury, periventricular-intraventricular hemorrhage (PVH-IVH) has been recognized as one of the most common and serious acute perinatal neurologic disorders of infants, particularly those of <1250 g birth weight or less than 32 weeks gestational age. In all studies, the magnitude of the hemorrhage, the mortality rate, and the subsequent neurologic-developmental deficits have been highest in the smallest infants, particularly those of <1000 g and under 30 weeks gestational age (187,188).

Incidence, Timing, and Grading of Hemorrhage

Before the use of noninvasive imaging techniques, retrospective autopsy studies established an incidence of PVH-IVH of 25% to 59% in live-born preterm infants, primarily those weighing about 1000 g at birth and suffering from the respiratory distress syndrome (8). In contrast, the incidence of PVH-IVH in stillborns was about 5%, implying that PVH-IVH is usually a postnatal event. Subsequently, using computed tomography (CT) scanning to diagnose IVH in living newborns, two large, prospective clinical studies from the late 1970s established an incidence of approximately 40% in infants of <1500 g birth weight or under 35 weeks gestational age with very few exceptions. Subsequent studies have documented similar incidence figures, varying from 32% to as high as 45% (4,32,114,115,142). A more recent report describes a declining incidence of IVH in infants under 35 weeks gestational age or <1500 g birth weight of 34% and 39% in 1980 to 1981 to 19% and 25% in 1986 to 1987 despite the fact that no planned changes in neonatal care had occurred (154). It is interesting to note that although the number of very immature infants of less than 1000 g birth weight is increasing, the incidence of PVH-IVH is not rising concomitantly (personal unpublished data).

The time of hemorrhage has been found by various studies to be as early as within the first 6 hours of life to as late as 14 days. Partridge et al. (146) found hemorrhage detectable by ultrasound on day 1 in 65.7%, by day 4 in 91.4%, by day 7 in 97%, and by day 14 in all 35 patients studied sequentially. Dolfin et al. (43) found that all hemorrhages in their series of 64 study infants, 20 of whom had hemorrhages, occurred in the first 62 hours of life. Approximately 50% of all infants with PVH-IVH have the hemorrhage on the first postnatal day, and only a small percent of patients experience hemorrhage after the fourth postnatal day (152,186). PVH-IVH is not necessarily a single event, but rather there may be progression of hemorrhage in 10% to 20% of infants (48). Serial ultrasound studies have demonstrated that the maximum extent of the bleeding is usually attained within 3 to 5 days after the initial diagnosis (118,163).

Hemorrhage quantitation and grading vary somewhat from one institution to another, making interpretation and comparison of studies of incidence, mortality, morbidity, outcome, and prevention confusing. The current most widely used grading system in the United States is based on Papile’s classification, which was introduced for CT scan images of the hemorrhage and subsequently modified slightly for ultrasound diagnosis (142). Grade 1 PVH-IVH denotes small hemorrhages in the germinal matrix, limited to the subependymal regions. Grade 2 hemorrhages have blood within the lateral ventricles without ventricular dilatation. Grade 3 refers to blood within the lateral ventricles with ventricular dilatation. Grade 4 is indicative of hemorrhage within the lateral ventricles, which are dilated, as well as intraparenchymal hemorrhage, which can be of variable magnitude and may be unilateral or bilateral. Because the PVH-IVH may be asymmetrical, may differ in magnitude within a single grade, and may also be associated with other pathologic lesions, the grade of hemorrhage by itself is insufficient for adequate determination of the magnitude of the clinical insult. Other parameters that must be considered include the degree of ventricular enlargement, the lateral or anterior or posterior extent of hemorrhage, and the presence or absence of other echogenic abnormalities (70).

Neuropathologic and Anatomic Features

The primary location of these hemorrhages is the subependymal and periventricular region, in the germinal matrix tissue over the head of the caudate nucleus at the level of the foramen of Monro (93). Before 26 to 28 weeks of gestation, the germinal matrix is a highly vascularized zone of the developing brain, a site of proliferating neuron and glial precursors. Until resorption of the germinal matrix occurs at approximately 35 weeks of gestation, the vessels in this region, which have limited glial support, are particularly vulnerable to mechanical forces generated by changes in intravascular and intracranial pressure. In the rare instance of hemorrhage in the full-term infant, the choroid plexus is the primary site (44). In addition to the characteristics of the poorly supported tissue, the frequency of hemorrhage in the area appears to be related in part to several unique aspects of the vascular anatomy at this particular stage of development. First, there is an abundant but poorly organized capillary bed with irregular-sized capillaries, irregular capillary-vein junctions, and veins with walls of only a single endothelial cell thickness, resembling capillaries at this stage of venous development. Quantitative analysis of cerebral vessels in germinal matrix of the newborn beagle puppy, a well-studied model system of PVH-IVH, revealed that the germinal matrix vessels were distinguished structurally from vessels in hemorrhage-resistant areas by their relatively thin walls, larger diameter, and discontinuous thinner basement membranes (192). Arterial walls in the very small human premature (<30 weeks) neonate lack smooth muscle, collagen, and elastin and resemble capillaries rather than arteries and therefore are likely to be particularly susceptible to rupture during a surge of systemic blood pressure (73). A second determinant of hemorrhage frequency is the anatomic organization of the deep venous system in this region. Venous drainage of the central portions of the cerebral hemispheres is by way of the anterior and posterior terminal, choroidal, and thalamostriate veins, all of which drain into the internal cerebral vein at a sharp angle, resulting in an abrupt change in direction of venous flow. A third anatomic feature is the extensive arterial circulation to the basal ganglia area, adjacent to the germinal matrix. Within the germinal zone, there exists a vascular rete composed of thin-walled sinusoidal vessels indistinguishable as arterioles, capillaries, or venules (141). It is conceivable that such immature blood channels would be especially prone to rupture and hemorrhage. In contrast to the theory of a venous origin for PVH, Hambleton and Wigglesworth (72) directed attention to the arterial side of the cerebral vasculature (95). At this stage of fetal development, basal branches of anterior and middle cerebral arteries are large, relative to the vessels supplying the rest of the hemisphere in contrast to the arterial circulation at term, when cerebral cortical arterial circulation predominates. Stereomicroscopic examinations of serial sections of injected and uninjected immature human brains demonstrated hemorrhages originating from the capillary bed of the germinal layer supplied mainly by Huebner’s artery and failed to substantiate either germinal layer infarction or venous rupture (72).

The major neuropathologic complications of PVH-IVH are hemorrhagic intracerebral involvement, hydrocephalus, germinal matrix destruction, cyst formation, and associated accompanying hypoxic-ischemic lesions. Associated areas of selective neuronal necrosis including brain stem and cerebellum are not unusual, and periventricular leukomalacia has also been found commonly accompanying IVH (10). Thus IVH is uncommonly an isolated abnormality in the brains of those infants with IVH who die.

The hydrocephalus that may follow PVH-IVH is mainly a result of arachnoiditis, caused by blood in the posterior fossa, sealing off the fourth ventricular exit foramina or interfering with cerebrospinal fluid (CSF) circulation in subarachnoid cisterns. Less commonly, the sylvian aqueduct can be blocked by blood clot with ensuing hydrocephalus (95). In infants who survive parenchymal hemorrhage and infarction, local parenchymal destruction may result in the development of porencephalic cysts (148).

Pathogenesis and Etiologic Factors

The exact pathogenesis of PVH-IVH is not yet fully understood. Many risk factors presumed to be implicated in the development of these hemorrhages have been derived from retrospective clinical and autopsy studies. These include the following:*

Bruising at birth.

Apgar scores of less than 5 at 5 minutes.

Hypercarbia.

Severe matabolic acidosis.

Hyaline membrane disease.

Pneumothorax.

Hypoxemia.

Hypercarbia.

Intermittent mandatory ventilation.

High peak pulmonary inflating pressures.

Prolonged inspiratory/expiratory ratio settings on ventilators.

Requirements for bicarbonate administration after the first 24 hours of life.

Systemic hypotension and rapid volume expansion in the first 24 hours of life.

Intrauterine growth retardation.

Hypernatremia and hyperosmolality.

Liberal bicarbonate administration.

Rapid colloid infusions.

Volume expansion.

Extreme blood pressure fluctuations or sudden rises in systemic blood pressure after relative hypotension.

Maternal dexamethasone.

Coagulation factor deficiencies.

Dysfunction of platelet-capillary interactions.

Patent ductus arteriosus (PDA).

Sudden early closure of the patent ductus or PDA ligation.

Fluctuating patterns of cerebral blood flow velocity.

Most of these risk factors presumably influence the development of PVH-IVH through their varied effects on cerebral blood flow, especially to the germinal matrix vessels. Wigglesworth and Pape (196) initially proposed that hypoxia, hypercarbia, or sudden volume expansion would cause vasodilation and sudden increases in cerebral blood flow that might rupture the germinal matrix vessels exposed to systemic blood pressure. Goddard-Finegold (63–66), in studies of a model system for the development of PVH-IVH in newborn beagle puppies, showed that rapid increases in arterial PO2 and systemic blood pressure or rapid volume expansion following hemorrhagic hypotension produced subependymal and intraventricular hemorrhages neuropathologically similar to the lesion seen in the human infant, experimentally confirming the clinical observations regarding hypercarbia, blood pressure changes, and volume expansion. Decreased blood flow to periventricular white matter and germinal matrix occurred during the phase of hemorrhagic hypotension, and this was followed by marked increases in flow to the same areas after volume reexpansion (63–66,120). The beagle puppy, the germinal matrix being an area of low baseline blood flow, might be particularly vulnerable to decreases in total brain blood flow and ischemic injury (147). Following hypotension-induced ischemic injury, hypertension, resulting from rapid transfusions, produced increased blood flow to all germinal matrix tissues but especially to the germinal matrix rostral and inferior to the caudate nucleus. This increased flow might be sufficient to rupture the germinal matrix vessels damaged by the prior ischemic insult.

Coagulation abnormalities and their role in PVH-IVH are emphasized in the series of 35 patients treated with extracorporeal membrane oxygenation (ECMO) (30). Each of the eight preterm infants (<35 weeks gestational age) treated with ECMO had significant intracranial hemorrhages; six of the eight died acutely, whereas the remaining two infants died before one year of age. Presumably infants who receive this treatment have suffered some degree of asphyxia; with institution of ECMO, they experience systemic heparinization, thrombocytopenia, and increased cerebral venous pressure with the ligation of the right internal jugular vein. An increased risk of PVH-IVH is also seen in term infants treated with ECMO (30,175).

A number of clinical studies suggest that the fundamental or common thread for many of these risk factors associated with the development of PVH-IVH was a loss of cerebral autoregulation (51,76103106107124149151). Other investigators, however, concluded that although loss of autoregulation might contribute to the development of PVH-IVH, it was not a necessary condition for the pathogenesis of this lesion, and that other factors unrelated to the pressure passive state were important determinants (2). At present, it is impossible to weigh each of the many factors that have been implicated in the pathogenesis of PVH-IVH. Presumably PVH-IVH results from combinations of these factors that affect both arterial and venous pressures in the vascular bed of the germinal matrix, already rendered susceptible to injury by prior hypoxic-ischemic injury and impaired cerebral autoregulation, which may play an important part in the pathogenetic scheme.

PVH-IVH may not be the final determinant of the cerebral injury, but rather the hemorrhage is followed by significant decreases in cerebral blood flow in the affected hemisphere (118,119). These decreases in cerebral blood flow, which appear to be secondary phenomena, may be more important determinants of mortality and neurodevelopmental outcome than the hemorrhage itself. Positron emission tomography (PET) studies have demonstrated significant reduction in cerebral blood flow and marked ischemia in areas of intracerebral hemorrhage involvement (grade 4 hemorrhages), indicating that there is marked impairment of regional cerebral blood flow in a hemisphere containing a restricted parenchymal hemorrhage, evidence that the hemorrhage itself is only one component of a much larger ischemic lesion (189).

Diagnosis

The classic clinical description of PVH-IVH in the preterm infant in the pre–CT scan era was that of a sudden catastrophic onset of changes in tone; diminished activity and apathy; signs of shock, apnea, or sudden increases in oxygen requirements; and a bulging fontanelle. A shrill cry, fixed stare with absent pupillary reactions, seizures, and tonic posturing were also described as part of this rapidly evolving clinical syndrome. Associated laboratory findings included a severe metabolic acidosis, dramatic declines in hematocrit, and hypoxemia. This constellation of clinical and laboratory findings was based on cases verified by autopsy or lumbar puncture in infants for whom the diagnosis was sought because of the sudden clinical deterioration. A second less dramatic and much less frequent clinical syndrome has been described by Volpe (185), consisting of similar clinical signs but with a saltatory or stuttering course with sudden deterioration followed by improvement or stabilization. A cycle of alternating deterioration and improvement may be repeated several times until improvement is sustained or death occurs, usually within 48 hours after the ictus.

With the advent of noninvasive imaging techniques, it is evident that many hemorrhages are clinically inapparent (40,88) and that many infants without hemorrhage have symptoms that might have been incorrectly attributed to PVH-IVH (96,142). It appears that in many infants with intracerebral hemorrhage, the quantity of blood or the pathophysiologic disruption is insufficient to cause clinical symptoms, and further, very sick premature infants have a limited repertoire of responses, so even those infants without hemorrhage may be mistakenly considered to have suffered PVH-IVH on the basis of similar clinical manifestations.

It has been suggested that using careful clinical observation with the additional help of lumbar puncture findings, including CSF xanthochromia or increased numbers of erythrocytes, one could reliably detect or screen all the high-risk preterm infants for hemorrhage. Multiple studies, however, have shown that CSF findings have not been reliable or useful diagnostically for PVH-IVH (8,39,142,169). CSF glucose levels have similarly been shown to have little value as an early indicator of PVH-IVH (112,135,183).

CT scanning was introduced and gained acceptance as a sensitive reliable method for detecting and quantitating hemorrhage in the high-risk premature infant. The expense of the procedure, hazards of radiation exposure, distance of many neonatal units from CT scanners, and difficulties with transporting critically ill, small premature infants to these CT units restricted the use of CT scanning, however, at least as a general screening procedure for PVH-IVH.

The prime method of diagnosis of PVH-IVH is cranial ultrasound, which has now replaced CT. Because of ready availability, relatively low cost, ease of performance at the bedside, and ability to repeat examinations on even the sickest infant without detriment, cranial ultrasound is now a routine screening study for all infants under 35 weeks gestational age or weighing less than 1500 g in most nurseries (84,140,168). Ultrasound has been extremely useful for timing of hemorrhage, following progression of hemorrhages into ventricles or brain parenchyma and resolution of hemorrhages; for detecting parenchymal white matter lesions; and for detecting and quantitating posthemorrhagic hydrocephalus (43,148,163). Ultrasound images can be used as indicators of mortality, morbidity, and long-term prognosis (177). All infants of 35 weeks gestational age or weighing less than 1500 g are now routinely subjected to an ultrasound examination, as a screening procedure. The proper use of cranial ultrasound requires knowledge of the problems, pitfalls, and instrument factors that can affect interpretation (53).

Other imaging modalities such as PET and magnetic resonance imaging (MRI) appear to have limited and primarily investigative roles in the diagnosis of PVH-IVH. PET appears to be of great benefit in the determination of associated changes in cerebral metabolism and altered blood flow, which may prove important to our understanding of the pathogenesis of the hemorrhage and the associated lesions (189).

Other noninvasive measures that have been used for early diagnosis of PVH-IVH include transillumination with a fiberoptic light source, transcephalic impedance, Doppler ultrasound, and electroencephalography (EEG) (12,47,167).

An association between the presence of rolandic positive sharp waves in the EEG and IVH, initially reported by Cukier (35), has been confirmed by others (21,31). Positive rolandic sharp waves, however, have also been identified in infants with periventricular leukomalacia and meningitis, so this EEG abnormality is not specific for IVH (136). Positive rolandic sharp waves may be evident in more than one-third of all newborn, very low birth weight infants regardless of the presence or absence of IVH (93). The EEG may be of clinical value in helping to determine the severity of a variety of neonatal neurologic disorders and the prognosis of infants with PVH-IVH (62).

Short-Term Prognosis, Complications, and Management

Mortality of PVH-IVH has changed since the original autopsy reports (88,142,186). The decline in mortality rates since the late 1970s appears to reflect improved diagnosis as well as improved neonatal resuscitation and intensive care practices. Overall mortality rates for all PVH-IVH infants, reported to vary from 27% (Ahmann) (3) to 35% (Papile) (142) to 52% (Krishnamoorthy) (88), obscured the fact that mortality rates are significantly related to magnitude of hemorrhage. Mortality rates are considerably higher for the infants with large hemorrhages and very low for the grade 1 or 2 hemorrhages. Shinnar (165), in a study of 115 consecutive premature infants who had a 35% incidence of PVH-IVH, reported only one death related specifically to the PVH-IVH, this being one of eight patients with grade 4 hemorrhage. Mortality rates of 5% to 15% in patients with grade 2 hemorrhage were reported in 1981, whereas current data suggest that grade 1 and grade 2 hemorrhage should never be the primary cause of death. Further, mortality rates vary significantly by birth weight. In one of the largest PVH-IVH series, infants with birth weights of <1000 g with grade 4 IVH had a 94% mortality, whereas the overall mortality for all of the grade 4 hemorrhages in this study was 50%, indicating the importance of very low birth weight (115). Also, survivors of the largest hemorrhages are much more significantly at risk for neurologic sequelae related to hypoxic-ischemic events antedating the hemorrhage, brain tissue destruction produced by the hemorrhage itself, ischemic injury occurring as the result of the hemorrhage, and posthemorrhagic hydrocephalus.

Before the late 1970s, posthemorrhagic hydrocephalus was considered an almost inevitable outcome for infants who survived IVH (27,40, 95). Progressive ventricular enlargement following hemorrhage has been reported to develop in 37% to 60% of survivors (4,6,118,144,146,163). Allan and Volpe (7) reported a progressive decline in the incidence of posthemorrhagic hydrocephalus over the 5-year period between 1980 and 1986, presumably related to the decline of incidence of grade 3 and 4 hemorrhages. The degree of posthemorrhagic ventricular dilatation, like mortality rates, is related to birth weight. The smallest infants, those under 1000 g birth weight, who have the highest incidence of grade 3 and 4 hemorrhage are most likely to develop posthemorrhagic progressive ventricular dilatation, whereas this complication is less frequent in those infants with blood limited to the germinal matrix or with small amounts of intraventricular blood. It has become increasingly apparent to all observers that in the majority of infants who develop ventricular dilatation after IVH, ventricular size stabilizes and then spontaneously diminishes in the weeks after hemorrhages; that is, the ventricular dilatation is transient. In a prospective study of the natural history of posthemorrhagic hydrocephalus, of 409 inborn preterm infants who had had subependymal hemorrhage or IVH, posthemorrhagic hydrocephalus was said to be present if progressive ventricular dilatation was evident on two successive follow-up cranial ultrasound examinations (50). Only 53 infants (13%) of those with IVH developed progressive posthemorrhagic hydrocephalus; even in these 53, ventricular enlargement then arrested in 23, with spontaneous regression in ventricular size in 12 others. Only 18 infants neither stabilized nor resolved but had further progression of ventricular dilatation to a stage of severe hydrocephalus.

Presumably some of the infants with dilated ventricles, referred to by some authors as ventriculomegaly have transient enlargement concomitant with the hemorrhage, and as the blood resolves, so also does the ventricular dilatation (5,7,186). Increased intracranial pressure is not necessarily present despite marked ventricular dilatation (78). Presumably hydrocephalus develops with normal pressure because of increased compliance of periventricular white matter, related to an immature stage of myelin development and lack of glial support. Only when the limits of tissue compliance of brain and cranium have been exceeded is there clinical evidence of increased intracranial pressure. Hydrocephalus is readily detectable by ultrasound 2 to 4 weeks before it becomes clinically manifest, and it has been shown that in those infants who develop posthemorrhagic hydrocephalus, the progressive ventricular dilatation is evident by 30 days of age (49). Thus ventricular enlargement is commonly far advanced before the infants develop the classic signs of hydrocephalus, including excessive rates of head enlargement, bulging fontanelle, separation of cranial sutures, scalp vein distention, ocular palsies, and sunsetting of the eyes (191).

Because of the natural course toward stabilization and regression of ventricular dilatation after hemorrhage makes intervention unnecessary in most instances, various attempts have been made to define indications for treatment and rational treatments for hydrocephalus. Allan et al. (5) adopted a conservative approach to therapy, using two criteria, an increase of head circumference at a rate greater than 2 cm a week or symptoms of increased intracranial pressure, as indications for intervention, with periodic weekly cranial ultrasound examinations and careful clinical examinations. Others have used cortical mantle measurement in the occipital region of less than 1 cm as an indication for intervention. Just as indications for treatment have been debated, therapy is also a controversial issue. Because of dismal results and complications from conventional ventriculoperitoneal shunts in these very small preterm infants, serial lumbar puncture treatment has been advocated (27,68). The reasons for avoiding shunts are many, including the technical problems of shunt placement in a very small infant, the greater surgical risk, the fragility of the skin overlying the shunt with the possibility of skin tissue breakdown, the high incidence of shunt infections, and the problem of acute or recurrent shunt malfunctions and requirement for further surgery.

Serial daily lumbar punctures have been recommended by many as a simple, relatively noninvasive, and readily available medical treatment for those infants with continuing or increasing ventricular dilatation, with numerous reports documenting the utility of serial daily lumbar punctures, although negative reports were given by others (9,33,50,68,87,104,108144). Hill et al. (77) and Lazarra et al. (97) reported that if daily lumbar punctures were done, with removal of sufficient CSF to reduce intracranial pressure by 50% and produce significant reduction in ventricular size, some infants whose ventricles had initially continued to enlarge would stabilize and respond to this medical treatment. Allan and Volpe have recommended that if there is a definite decrease in ventricular size and occipitofrontal head circumference after one lumbar puncture with removal of more than 10 ml of CSF, daily lumbar punctures should be done. If improvement occurs, daily lumbar punctures are continued. If the procedure fails and the posthemorrhagic hydrocephalus increases, external drainage of CSF is performed for 7 days, using a 21-gauge, 3.2-cm intravenous catheter placed in the right lateral ventricle percutaneously. After 1 week of ventricular drainage, the catheter is removed. Ventricular drainage is again performed if progressive ventricular dilatation recurs. If following this second period of ventriculostomy drainage the hydrocephalus is still progressive and severe, shunt surgery is performed, but only if the infant weighs more than 2 kg. Others have advocated a similar course action, resorting to ventriculo-peritoneal shunt surgery when the infant reaches 1800 g weight. Using this treatment plan, Allan and Volpe (7) found that only one-third of their patients with severe posthemorrhagic hydrocephalus required shunts. Dykes et al. (50) also found that shunts could be avoided in half of their patients with asymptomatic severe posthemorrhagic hydrocephalus who were managed by a study protocol, which randomized these infants to close observation or repeat lumbar puncture groups. Although it is generally agreed that shunt surgery is inevitable for a small proportion of patients with progressive posthemorrhagic hydrocephalus that is symptomatic, the role of lumbar puncture treatment is still unclear; lumbar puncture treatment may be no more effective than no treatment at all and may be a temporizing measure, simply delaying the inevitable need for ventricular drainage and shunts for a small number of symptomatic infants.

Other recommended nonsurgical treatments for progressive posthemmorhagic hydrocephalus have included close observation (50), administration of acetazolamide and furosemide, and administration of isosorbide and glycerol (164, 168). No single therapy appears to have higher utility than another, and no well-controlled studies of these alternative therapies have been carried out.

The long-term outcome of infants with IVH who suffer progressive posthemorrhagic hydrocephalus may not be related to the degree of the hydrocephalus but rather to the severity of the original insult to brain parenchyma by both the hemorrhage and the associated anoxic-ischemic injury. Thus a conservative approach to the management of these infants seems warranted.

Long-Term Prognosis

The long-term outcome of children who suffered PVH-IVH as infants has been the subject of a number of studies, with varying conclusions. Because of limited numbers of patients in any single study, variable but generally short periods (1 to 2 years) of follow-up in most studies, lack of matched controls, incomplete documentation of hemorrhages by CT or ultrasound, or even inclusion of only clinically symptomatic infants, comparisons of these studies are difficult. The incidence of major handicaps, including the motor deficits of hemiparesis, diplegia, quadriparesis, or dystonic limb movements as well as variable degrees of intellectual impairment, in infants who had IVH must be compared with the incidence of these same handicaps in comparable low birth weight infants who did not suffer IVH to get a clearer picture of the role of IVH in the genesis of the deficits. Studies to date have not considered school-aged children, so significant learning problems, attention deficits, organically determined behavior problems, and deficiencies in fine motor performance may yet be identified as significant, albeit lesser deficits in these groups of patients.

It is not yet established which is the principal factor, or if there is a principal factor, determining outcome. The neurologic and developmental deficits have been variously ascribed to the magnitude of the hemorrhage, the degree of posthemorrhagic hydrocephalus, the effects of increased intracranial pressure secondary to the hemorrhage and the ventricular dilatation, or the relative importance of periventricular-intraparenchymal hemorrhage or hemorrhagic necrosis of periventricular tissue.

Papile et al. (145), in the largest study to date, of 94 infants with varying degrees of PVH-IVH, evaluated at 12 to 24 months of age, found that the outcome of infants with grade 1 and 2 hemorrhage was similar to that of a control group without hemorrhage; a major handicap was present in 9% of infants with grade 1 and 2% of those with grade 2 hemorrhages, compared with a 10% incidence of major handicaps in at-risk infants who did not suffer hemorrhages. None of the grade 1 or 2 IVH patients had posthemorrhagic hydrocephalus. Of 14 patients with grade 3 hemorrhage who survived and could be evaluated, 5 or 36% had a major handicap, and four of these six were multihandicapped; only two of the 14 were normal. Seventy-six percent of the grade 3 and 4 hemorrhage survivors had a much poorer outcome than controls. Infants with posthemorrhagic hydrocephalus had the same incidence of major handicaps (59%) as did comparable infants with grade 3 to 4 hemorrhage without posthemorrhagic hydrocephalus, suggesting that posthemorrhagic hydrocephalus did not increase an infant’s risk for major handicaps. They related the poor outcome in this series of patients to the magnitude of the underlying hemorrhage. Similar conclusions were reached by others (89,163,197). In a follow-up study extending to 30 months of age, of all neonates of less than 1250 g birth weight, both with and without PVH-IVH, Ment et al. (121) noted a low incidence of major neurologic abnormalities in both groups, and although there were few survivors of the more severe grades of IVH, no difference between developmental outcomes of subependymal (SEH)-IVH and nonhemorrhage groups could be detected.

Dykes et al. (50), in a long-term (greater than 3 years) outcome study of all IVH patients who developed posthemorrhagic hydrocephalus, confirmed Papile’s observation that the presence of hydrocephalus was not the critical factor with regard to major handicaps or neurodevelopmental deficits; rather, the poor outcome was related to the magnitude of the hemorrhage: Infants with grade 3 IVH fared better than did those with grade 4 hemorrhages. Infants of less than 30 weeks gestational age at birth fared more poorly than did those of greater gestational age at birth.

The role of the ischemic component of the grade 4 hemorrhage in determination of outcome has been amplified by McMenamin et al. (115) and Guzzetta et al. (70). The parenchymal lesion in grade 4 hemorrhage, which had been considered to be simply extension of blood from lateral ventricle or germinal matrix into brain parenchyma, referred to as intraparenchymal echodensities, on the basis of their ultrasound appearances, in limited autopsy studies were determined to be hemorrhagic infarctions of white matter. The immediate outcome of the infants could be related to the magnitude of these intraparenchymal echodensities on ultrasound; 29% of infants with small intraparenchymal echodensities died, whereas 76% of those with large intraparenchymal echodensities died. Thirty infants with these intraparenchymal echodensities were followed for 8 to 24 months (mean 17 months): Two of 22 surviving infants with small echodensities were moderately handicapped; surviving infants with large echodensities had a 100% incidence of moderate to severe handicaps. Thus the size of the intraparenchymal lesion bore a striking relation to outcome, the large lesions predicting either death or significant neurologic impairment. With extensive intraparenchymal hemorrhage and necrosis, there appeared to be little or no chance for survival with normal neurologic and cognitive function. This study stressed the importance of a careful quantitative assessment of ultrasound examination findings, especially of periventricular-intraparenchymal echodense lesions, as a guide to long-term outcome. The matter is not yet settled because others have continued to relate the neurologic and developmental deficits more closely to the presence of posthemorrhagic ventricular dilatation than to the size of the hemorrhage itself (138).

The importance of long-term longitudinal follow-up studies becomes even more evident from Shinnar’s finding (165). Their overall conclusion, based on a 12- to 22-month follow-up of 11 patients with grade 3 or 4 hemorrhage, was optimisitic, with only two showing serious motor and intellectual impairment at that time. An important note in that study is the observation that although six of their infants had variable degrees of hemiparesis when examined at less than a year of age, when examined at 18 months, the motor deficit had resolved almost completely in all but one patient.

In a longitudinal, prospective, long-term follow-up study of infants with PVH-IVH by Schub et al. (159), the outcome of PVH-IVH infants, compared with a group of simultaneously selected matched control infants of low birth weight who also required neonatal intensive care, was not significantly worse for either neurologic deficit or developmental test scores. Twenty-one of 33 PVH-IVH infants, compared with 19 of 30 nonhemorrhage infants, showed a good outcome without major deficits and developmental index greater than 90 at a mean age of follow-up of 39 months. As has been shown repeatedly in numerous studies, patients who had grade 1 or 2 hemorrhages had predominantly good outcomes with no differences between the IVH patients and their matched controls. Even with grade 3 hemorrhage, infants who survived the neonatal period had an approximately 50% chance of a good outcome. Thus IVH even with ventricular dilatation does not preclude a good long-range prognosis. Clearly infants with minor or small hemorrhage do better than infants with major hemorrhage. Those aspects of illness most predictive (although not invariably) of a poor prognosis for infants with PVH-IVH include the following:

1. A major or marked IVH with intraparenchymal echodensities of moderate or larger size.

2. A catastrophic clinical presentation with full fontanelle, seizures, apnea, and other classic clinical signs.

3. Posthemorrhagic hydrocephalus that does not spontaneously arrest or stabilize and does not respond to lumbar puncture or ventriculostomy (52).

4. Gestational age at birth of <30 weeks with major hemorrhage and posthemorrhagic ventricular dilatation (50).

Prevention

The most significant means of preventing IVH is the prevention of prematurity. Alteration or elimination of some of the already identified risk factors, including preventing or decreasing the severity of hyaline membrane disease; improving methods of neonatal resuscitation and ventilation to prevent hypoxia, hypercarbia, and acidosis; maintaining systemic blood pressure and preventing abrupt decreases and increases in blood pressure; and controlling rates and volumes of fluid and blood infusions, are established means of altering the incidence of IVH.

Various pharmacologic interventions have also been studied. Phenobarbital administration, as a prophylactic measure, has been evaluated. Although several reports showed a decrease in the overall incidence or severity of IVH in infants given phenobarbital in loading doses followed by maintenance therapy, most studies have failed to show a beneficial effect (10,14,45,46,82,91,128). Antenatal administration of phenobarbital to women in imminent danger of delivery of preterm infants, however, has shown promising results (85,126,162).

Ethamsylate, a capillary stabilizing agent that also inhibits prostaglandin synthesis, has been evaluated in two studies, which showed a reduction in the incidence of IVH and improved outcome for treated infants (17,31,121). Further controlled studies of this drug are needed to confirm these reports.

Vitamin E, which appears to exert its effects by acting as a scavenger of free radicals, thereby protecting capillary endothelial cells from hypoxic injury, has also been reported to be useful as a protective agent, limiting IVH but not preventing subependymal hemorrhages (171).

Fresh frozen plasma infusions had a beneficial effect on the prevention of IVH without any noticeable effect on mortality, although specific coagulation deficits were corrected (19). Coagulation abnormalities have been shown to be associated with the occurrence of IVH (113,161).

Pancuronium bromide (Pavulon) administration to premature newborns who required ventilatory support, which prevented wide swings or the fluctuating pattern of cerebral blood flow velocity in infants on ventilators, was shown in one study to decrease the incidence of PVH-IVH markedly in this group of infants (149). Further studies are needed to corroborate this protective effect.

Indomethacin, an inhibitor of prostaglandin synthesis, was shown in the beagle puppy model of IVH produced by hemorrhagic hypotension followed by volume reexpansion to have a protective effect (123). A similar protective effect, decreasing the incidence of PVH-IVH, was determined in a randomized, prospective indomethacin or placebo study of 48 preterm human infants (116). In a subsequent study, a similar protective effect was shown, and in a very limited study, low-dose indomethacin therapy was shown to have an effect in preventing extension of IVH in infants with grade 1 and 2 hemorrhage (117). Further studies on this drug and other therapies are warranted as means of preventing IVH.

The effect of the use of surfactant for the prevention of or treatment of respiratory distress syndrome on the incidence of PVH-IVH is as yet unclear (102).

Intraventricular Hemorrhage in the Full-Term Infant

IVH may also occur in the full-term infant but does so in fewer than one of 100 full-term infants (25,44,54). IVH in the full-term infant most commonly originates from the choroid plexus rather than from the subependymal germinal matrix, little of which remains at term. Many of these infants have evidence at birth of intrapartum asphyxia. Cranial birth trauma is also occasionally reported, but no identifiable precipitating cause is found in the majority. A report of intracranial hemorrhage in an infant born to a cocaine-abusing mother implicates prenatal or postnatal hypertension in the cause of the PVH-IVH (28). Again, term infants who are treated with ECMO for severe respiratory failure have an increased risk of PVH-IVH (175).

Onset of symptoms of IVH in the full-term infant may occur acutely within the first 4 days of life (137). In one study, 90% of affected full-term infants had onset of neurologic symptoms within the first 24 hours (92). Delayed onset of illness has been described, however, occurring as late as 10 days to 29 days of age (18,125). Seizures are a very prominent symptom in almost all cases of PVH-IVH in the full-term infant. Parenchymal extension of hemorrhage has been commonly found on CT or ultrasound. The epidemiology and possible associated etiologic factors are even less well understood with IVH in the full-term than in the preterm infant. Because ultrasound is not part of the routine evaluation of every full-term infant, no data regarding incidence of the disorder are available, nor is there information to support or negate the possibility that IVH in the full-term infant could be a clinically silent event and therefore remain undiagnosed or be a cause of later unexplained hydrocephalus or neurologic deficits.

Primary Subarachnoid Hemorrhage

Primary subarachnoid hemorrhage, hemorrhage within the subarachnoid space that is not unrelated to a primary source of bleeding in the subdural or epidural spaces or to extension from intracerebral or intracerebellar hemorrhage, is perhaps the most frequent type of intracranial bleeding in the full-term newborn. A high percentage of neonates have some red blood cells (RBCs) in their CSF: Median numbers of RBCs were 180 mm³ in healthy term infants and 112 mm³ in healthy preterm infants in the normative study of Sarff (157). Small amounts of subarachnoid blood may be detected in postmortem examinations of newborn infants who were neither clinically symptomatic nor suspected of having suffered intracranial bleeding. Most commonly, RBCs in CSF are an almost incidental finding, perhaps a result of mild trauma during delivery.

Findings in CSF indicative of subarachnoid bleeding are increased numbers of RBCs, elevated CSF protein, and xanthochromia. In the evaluation of bloody CSF specimens, to distinguish subarachnoid bleeding from a traumatic lumbar puncture, it is necessary to count RBCs in first and third (or fourth) tubes of CSF as well as examine centrifuged CSF for xanthochromia; the latter is indicative of subarachnoid blood and is not the result of a traumatic lumbar puncture.

Pathogenesis

Significant subarachnoid hemorrhage can occur with hypoxia, particularly in preterm infants; the bleeding is related to asphyxiai injury to endothelium of congested capillaries and small veins. There may be an additive effect of trauma in the course of delivery (55). Clinically significant bleeding, particularly in the full-term infant, can result from trauma alone.

Subarachnoid hemorrhage related to trauma may occur as an isolated process or may be associated with subdural bleeding, with or without cerebral contusion. When subarachnoid hemorrhage is associated with other types of intracranial hemorrhage or signs of physical injury, as further documented by CT or ultrasound, caused by difficult delivery often with mid or high forceps application, the outcome is frequently poor. There may be demise in the early neonatal period or significant brain damage in survivors.

Clinical Presentation

The clinical presentation of the full-term infant with isolated subarachnoid hemorrhage may vary considerably. Most commonly there are no specific clinical signs, and the diagnostic lumbar puncture is performed as part of a more complete evaluation for suspected sepsis, meningitis, or convulsions. Infants with primary subarachnoid hemorrhage may appear normal initially and then develop convulsions on the first or second day of life as the only clinical manifestation. The infant may appear remarkably well between seizures and have a benign subsequent course, without sequelae. In contrast, the initial clinical manifestations of neonatal subarachnoid hemorrhage may be the early onset of alternating depression and irritability, with seizures that are difficult to control. It may be impossible to establish the primary role of birth trauma or asphyxia in the pathogenesis of the subarachnoid bleeding and its manifestations. Infants may develop hydrocephalus as a late complication of subarachnoid bleeding, with obstruction of CSF circulation in the subarachnoid cisterns, as well as other later signs of neurologic impairment and delayed development.

Treatment

Most infants with subarachnoid hemorrhage require no specific treatment, other than anticonvulsant medication if seizures occur. Repeated lumbar punctures, to drain the blood from the subarachnoid space and thereby prevent adhesions and secondary hydrocephalus, have been advocated by some as a useful measure, although there have been no controlled studies of the efficacy of this form of treatment. Except for those rare infants with the catastrophic clinical syndrome, the prognosis for isolated subarachnoid hemorrhage of the newborn is excellent.

Subdural Hemorrhage

Subdural hemorrhage, which is most commonly seen with difficult and at times clearly traumatic deliveries, has become a rare neurologic disorder of the newborn. The two principal locations of subdural hemorrhage are over the cerebral hemispheres and in the posterior fossa. Common historical events associated with this type of bleeding include the following: The mother is primiparous, with a total labor and delivery time of 2 to 3 hours or less; the delivery is difficult and may involve mid or even high forceps application; the infant is large for gestational age; and the birth canal is relatively small or rigid. The events of labor produce excessive molding of the infant’s calvarium, either vertically or in the fronto-occipital plane, with relatively sudden stretching and tearing of superficial venous channels over the hemispheres or of the venous sinuses in the posterior fossa. The clinical presentation depends on the quantity of blood and its location as well as the presence of any associated cerebral injury.

Subdural hemorrhage over the cerebral hemispheres occurring at birth may be clinically silent, may be clinically apparent in the first few days after birth, or may not be apparent until as late as the sixth week of life. Early manifestations include the signs of increasing intracranial pressure, jaundice, and anemia. There may be a bulging fontanelle, suture separation, and gradual or abrupt decline in the infant’s level of consciousness. Seizures may occur, possibly accompanied by the appearance of focal neurologic deficits, including hemiparesis or a dilated pupil, following third cranial nerve compression by temporal lobe herniation. When a convexity subdural hematoma presents as late as the fourth to sixth week of life, the manifestations are usually those of increasing head size, poor feeding, vomiting, failure to thrive, irritability, and altered level of consciousness. Even though an increasing head size in the infant is more commonly caused by hydrocephalus, subdural hematoma must always be considered in the differential diagnosis, especially following difficult labor and delivery.

CT scanning is a rapid and reliable method of diagnosing subdural hemorrhage and quantitating the amount of blood present as well as shift of cerebral ventricles and is more reliable than cranial ultrasound. Management of subdural collections of blood or fluid usually involve watchful waiting, with careful sequential observations and recording of head circumference. In the acutely ill infant, with signs of rapid deterioration, emergency surgical intervention is necessary. Repeated subdural taps, to remove small amounts of residual fluid or hygroma, in the asymptomatic infant, although recommended in the past to prevent the development of constricting membranes, is rarely necessary.

Subdural hemorrhage in the posterior fossa potentially poses more of a threat to the infant than does subdural bleeding over the convexities of the cerebral hemispheres. This type of bleeding can occur in either full-term or preterm infants and in many cases has been associated with tentorial laceration. There may be associated cerebellar contusion or intracerebellar hematoma, particularly in the preterm infant. Because of the limited space around the brain stem for blood to accumulate before brain stem function is compromised, this type of hemorrhage may be lethal in a short period of time (i.e., hours). Clinical manifestations include full or bulging fontanelle, lethargy and hypotonia, irritability, progressive impairment of ocular movements, other cranial nerve palsies, and irregular respirations or apneic episodes. Rapid confirmation of the diagnosis, especially with evolution of the clinical syndrome and fall in hematocrit, should be made by CT scan, and neurosurgical intervention is mandatory (20, 160,193).

Cerebellar Hemorrhage

Massive hemorrhagic destruction of the cerebellum, although infrequent, may occur in low birth weight infants in association with perinatal trauma and asphyxia (110). A substantial number of the cerebellar hemorrhages seem to be direct consequences of massive supratentorial IVH, according to autopsy studies (70), or are associated with germinal matrix hemorrhages in very small preterm infants. Intracerebellar hemorrhages have also been reported in term newborns treated with ECMO (175) and in term newborns with inborn metabolic errors, propionic, methylmalonic, and isovaleric acidemias, which have associated metabolic acidosis and thrombocytopenia (36).

The clinical course of such infants may be characterized as a rapid deterioration, similar to the course of posterior fossa subdural hematoma, with progressive apnea, falling hematocrit, and death, although there may be a less dramatic course, with opisthotonic posturing, lethargy, and progressive unresponsiveness. Although successful nonoperative management has been reported, surgical intervention is usually recommended with early surgical decompression (52). Termination of ECMO may be indicated in infants with significant hemorrhage. Operative intervention would be contraindicated in an infant on ECMO.

Hypoxic-Ischemic Encephalopathy

Hypoxic-ischemic encephalopathy (HIE) is the leading cause of perinatal neurologic morbidity and the major cause of static encephalopathies in children, including mental retardation, seizures, and a variety of motor deficits (79). Estimates of the incidence of asphyxiai injury range from two to four per 1000 live births in term infants to as high as 60% in preterm infants, with high morbidity and mortality (181). Because of the difficulty of separating the specific effects of hypoxia from those resulting from reduced cerebral blood flow, the term hypoxic-ischemic encephalopathy has been accepted as a broader, more meaningful term than postasphyxiai injury. Even in experimental studies, it is difficult to produce asphyxia without hypotension and secondary or simultaneous changes in cerebral perfusion.

Early Prediction and Recognition

Knowledge of the effects of maternal disease states on blood pressure and circulation, placental blood flow, oxygen delivery to the fetus, and information concerning intrapartum events, all gained by biophysical monitoring of fetal heart rate and rhythm and biochemical monitoring of fetal blood gases and pH, has made it possible to identify fetal distress that may not be detected on the basis of previously used criteria of intrapartum asphyxia, such as meconium in amniotic fluid, fetal bradycardia, or low Apgar scores at birth. Some infants who suffer intrapartum asphyxia may appear normal initially and may not require immediate resuscitation and yet within the first 24 hours of life develop clinical signs of HIE. Criteria for detection of asphyxia that are both sensitive and reliable are necessary because early prediction or recognition of hypoxic-ischemic insults may lead to earlier intervention and termination of the asphyxiai event and to alleviation of anatomic or physiologic abnormalities that may potentiate hypoxic-ischemic injury and treatment before the full range of pathologic events has occurred. At present, the reliability, significance, and predictability of intrapartum-monitoring data with regard to asphyxiai injury are still under study and not fully proved.

Pathophysiology

Deprivation of oxygen supply to the brain, either because of decreased availability of oxygen in arterial blood (i.e., hypoxemia) or because of reduced cerebral blood flow (i.e., ischemia), may produce a variety of physiologic and pathologic changes in the infant’s brain. Magnitude of injury, distribution of lesions, and degree of reversibility of recovery, with or without treatment, may vary considerably depending on the maturity of the infant, the timing and duration of the insult, additive effects of other organ-system failure, and the coexistence of metabolic abnormalities (e.g., hypoglycemia) or adverse physiologic alteration (e.g., hypothermia or hypotension).

Newborns are more resistant to hypoxia than adults; the greater tolerance of the developing brain is said to reflect, in part, lower cerebral energy demands. Further, there are regional differences in susceptibility to hypoxic injury, related to regional differences in circulation and metabolic requirements. Many newborns and particularly premature neonates do not manifest the same rostrocaudal pattern of neuronal sensitivity or vulnerability to hypoxia that is so well described in the adult.

Cerebral perfusion deficits enhance hypoxic injury (38). Under normal circumstances cerebral blood flow is maintained by alterations in cerebrovascular resistance independent of systemic arterial blood pressure changes within a wide range, the phenomenon of cerebrovascular autoregulation. Hypoxia and hypercapnia are potent stimuli, causing marked cerebral vasodilatation, which helps to maintain cerebral blood flow and tissue oxygen supply in the presence of asphyxia and systemic hypotension. In the asphyxiated newborn, however, maximum vasodilatation has already occurred in response to hypoxia, and there is no additional vasodilatation possible as systemic blood pressure falls. Lou et al. (106) have shown that not only are rates of cerebral blood flow in the newborn infant lower than those of the adult, but with asphyxia, cerebrovascular autoregulation is impaired in the distressed newborn. Cerebral blood flow then becomes pressure-passive, and with systematic hypotension, which frequently occurs with moderate or severe asphyxia, cerebral blood flow then becomes markedly reduced. The resulting impairment in cerebral perfusion enhances the likelihood of ischemic injury in areas of the brain that are inherently vulnerable because they are border zones or watershed regions of arterial circulation. Periventricular leukomalacia, a neuropathologic lesion that occurs primarily in premature infants, is thought to be related to the failure of perfusion of arterial boundary zones or watershed areas between terminal distribution of ventriculopetal and ventriculofugal branches of small arteries in periventricular white matter (41). The increased frequency of these lesions in the premature infant, compared with the full-term infant, is not related to any unique anatomic relationships existing in premature infants but rather to the greater frequency of perinatal distress, asphyxia, hypotension, and resulting impaired perfusion (166). A watershed hypothesis has similarly been invoked to explain the hypoxic-ischemic lesions in full-term infants with symmetrical parasagittal cortical injury, involving superior-medial aspects of the cerebral hemispheres, which are border zones between anterior, middle, and posterior cerebral artery circulation (190). Although the watershed hypothesis appears attractive, there is no direct evidence, that is, measurement of regional cerebral blood flow during hypotension, to support it at present.

Anatomic characteristics modify response to asphyxiai injury. As previously noted, the small preterm infant of 28 to 35 weeks gestational age is at great risk for PVH-IVH, partly because of the anatomic characteristics of the germinal matrix and because of the arterial capillary and venous circulation adjacent to the periventricular region at this particular developmental stage. Despite this specific vulnerability, these immature infants are relatively protected from suffering hypoxic-ischemic parasagittal cerebral cortical lesions, described earlier, because of rich, arterial meningeal anastomotic connections between anterior, middle, and posterior cerebral artery branches at this particular developmental stage (141). Differences in regional susceptibility to hypoxic injury are also presumably related to regional differences in circulatory response to asphyxia. In some immature animals, asphyxia has been shown to result in selective increases in brain stem blood flow, when flow to cerebral hemispheres was reduced (74). Other studies in animals have shown that when systemic blood pressure was maintained at normal levels, asphyxia produced increased blood flow to all areas of the brain but with the largest increases going to brain stem and deep areas of cerebral hemispheres (83). Insufficient data in the human newborn are available to make definite statements regarding regional flow differences in the asphyxiated infant.

Information is available regarding regional metabolic requirements that may modify pathophysiologic responses to hypoxia and ischemia. Although gray matter of adult cerebral cortex has a much greater metabolic rate than that of underlying white matter, in the newborn, gray and white matter have almost identical metabolic rates, implying that newborn white matter might be relatively more vulnerable to hypoxia than is adult white matter. Nonhomogeneity of metabolic activity and requirements has been used to explain differential susceptibility to hypoxic-ischemic lesions. Metabolic rates of inferior colliculus and cochlear, vestibular, and thalamic nuclei are extremely high; thus a brain stem distribution of hypoxic-ischemic lesions can be partially explained on the basis of increased metabolic activity with increased metabolic demands that outstrip supply. Animal studies show remarkable changes in central nervous system function before there are significant metabolic changes, so it is not possible to explain decreased cerebral function simply on the basis of metabolic depletion (180,184).

Neuropathologic Features

There is no single or uniform location or type of pathologic lesion resulting from hypoxia and ischemia. Periventricular leukomalacia is one characteristic lesion of HIE in preterm infants and results presumably from localized reduced perfusion with resultant areas of necrosis in periventricular white matter, predominantly in the regions adjacent to the external angles of the lateral ventricles (13). The area involved includes white matter through which long descending motor tracts pass, with the leg fibers being closest to the ventricles and therefore more likely to be damaged, most commonly resulting in spastic paresis of the legs. With lateral extension of the lesion, there may also be involvement of the arms, resulting in spastic diplegia or quadriplegia. Other sequelae may include visual deficits, presumably related to involvement of optic radiations. More extensive lesions involving association and commissural fibers would be expected to result in mental retardation and other neurologic sequelae. A striking decrease in the incidence of spastic diplegia in neonates under 2500 g birth weight, mainly because of increased emphasis on maintaining blood pressure and circulation in the high-risk infant, has been noted in population surveys of cerebral palsy (71).

A second type of pathologic injury resulting from hypoxia and ischemia in full-term infants is neuronal necrosis, involving cerebral and cerebellar cortex in a laminar distribution, especially in depths of sulci with particular involvement of the hippocampus, Purkinje cells of the cerebellum, or, in more severe cases, all of the cerebral cortex diffusely. This is similar to the adult form of hypoxic injury. In some full-term infants, these lesions of cortical necrosis, followed by gliosis and shrinkage of cortical tissue with capillary proliferation, may be preceded by the occurrence of generalized brain swelling, with convolutional flattening and hemorrhagic necrosis at the depths of cortical sulci, similar to the animal experimental lesion produced by prolonged partial in utero asphyxia. The areas of necrosis may be primarily in parasagittal cortex, most marked posteriorly and extending anterior into regions of motor cortex. Whether these zones of cortical necrosis occur as a consequence of reduced cerebral blood flow induced by brain swelling and increased intracranial pressure, as in the animal model of prolonged partial asphyxia, or as a result of reduced cerebral perfusion in boundary zones at the end of the distribution of blood supply from anterior, middle, and posterior cerebral arteries, is not known (22,190). The frequency of cerebral edema following hypoxic-ischemic injury in the full-term human infant is unknown. Generalized edema found on CT scan of asphyxiated infants correlates with autopsy findings of diffuse cerebral edema in full-term perinatal ischemic brain damage (61). Brain swelling, defined by enlargement of cerebral hemispheres, gyral flattening, and sulcal narrowing, has been noted as a frequent pathologic finding in severely asphyxiated infants. Clinically brain swelling in the full-term asphyxiated neonate may be manifested by bulging fontanelle and suture separation.

A third neuropathologic topographic pattern of anoxic-ischemic cell change, described by Leech and Alvord (98), is the distribution of ischemic cell change, neuronal loss, and gliosis in lateral geniculate bodies, ventrolateral and dorsomedial thalamic nuclei, brain stem cranial nerve nuclei, and brain stem reticular formation. These brains also had evidence of focal or patchy cortical damage. A fourth but extremely rare lesion of perinatal hypoxic-ischemic injury is status marmoratus, a unique pathologic process involving basal ganglia with neuronal loss, gliosis, and increase in myelinated fibers, giving the basal ganglia a marbled appearance. This lesion is said to be the pathologic lesion responsible for bilateral choreoathetosis. The reason for the rarity of both the clinical disorder and this particular pathologic finding in contemporary neuropathologic studies is totally obscure.

Experimental Observations

Two models of intrauterine asphyxia, acute total asphyxia and prolonged partial asphyxia, have been developed, although neither model is identical with the clinical-pathologic disorders in the human infant. Episodes of total asphyxia resulted in a restricted and reproducible pattern of pathologic abnormality within the brain stem. This may be an uncommon pattern of pathologic injury in the asphyxiated infant, but brain stem damage is common in infants with anoxic accidents (23).

A pattern of brain pathology more closely resembling human perinatal asphyxia injury in the full-term infant was described by Myers (131), in a full-term monkey who had been subjected both to chronic in utero asphyxia and to total aspyxia after delivery. This animal was initially lethargic, unresponsive, and hypotonic and at