Intraventricular Hemorrhage is a Complex Disorder

IVH has been attributed to alterations in cerebral blood flow to the immature germinal matrix microvascular network [Shalak and Perlman, 2002; Whitelaw, 2001], as shown in Figure 19-2, and studies addressing the etiology of IVH have identified numerous environmental and medical risk factors (Box 19-1). These include low gestational age, absence of antenatal steroid exposure, maternal antenatal hemorrhage, chorioamnionitis/infection/inflammation and/or fertility treatment, neonatal transport to a tertiary care center following birth, maternal sepsis, seizures, hypotension requiring medical therapy, hypoxemia, hypercapnia, and acidosis [du Plessis, 2008; Fabres et al., 2007; Limperopoulos et al., 2008; McCrea and Ment, 2008; O’Leary et al., 2009]. As this lesion is unique to preterm infants, examination of those developmental, environmental, and genetic events that result in IVH is critical. These include anatomic, physiologic, and genetic factors.

Fig. 19-2 Model for the development of intraventricular hemorrhage in the preterm infant.

CBF, cerebral blood flow; IVH, intraventricular hemorrhage; PVL, periventricular leukomalacia.

Box 19-1 Clinical Risk Factors for Intraventricular Hemorrhage

Early Onset

Low birth weight, low gestational age

Maternal fertility treatment

No antenatal steroid exposure

Maternal antenatal hemorrhage

No maternal pre-eclampsia

Maternal chorioamnionitis and/or infection

No tertiary care delivery

Vigorous resuscitation

Neonatal transport following delivery

Onset after 6–8 Postnatal Hours

Hypotension requiring medical therapy

Respiratory distress syndrome

Acidosis

Hypoxemia

Extremes of PCO₂ levels (both high and low)

Sodium bicarbonate exposure

Pneumothoraces

Seizures

Anatomic Factors are Permissive for Hemorrhage

During the late second and third trimesters, the developing brain almost triples in volume, and there are multiple developmental events that occur during this time interval. The germinal matrix and the adjacent subventricular zone (SVZ) are the sites of proliferation of both glial and neuronal precursors. The SVZ generates predominately GABAergic cortical interneurons, although recent data suggest that the germinal matrix also produces GABAergic neurons that migrate to the dorsal thalamus during the risk period for IVH. (For review, please see Volpe [2009a].)

By 24 weeks of gestation, almost but not all of the cortical neurons have migrated from the germinal zones and axonal ingrowth begins [Kostovic, 1990; Kostovic and Rakic, 1990; Kostovic et al., 2002, 1989]. The elaboration of dendritic arbors is at an active stage and synaptic contacts are beginning to form [Huttenlocher and de Courten, 1987; Huttenlocher and Dabholkar, 1997; Huttenlocher et al., 1982]. The germinal matrix remains relatively robust through 32–34 weeks of gestation but almost completely involutes by term [Donat et al., 1978; Rorke, 1982; Whitelaw, 2001].

During this period of rapid cell genesis, the metabolically active germinal matrix requires a rich blood supply, as shown in Figure 19-3, yet its vessels are neuropathologically immature and present numerous risk factors for hemorrhage (Table 19-2). The developing blood–brain barrier is characterized by basement membrane proteins, endothelial tight junctions, capillary pericytes, and astrocytic endfeet [Ballabh et al., 2004]. Immunohistologic and electron microscopic examinations comparing germinal matrix microvessels with those in the periventricular white matter and cortical mantle demonstrate a paucity of fibronectin, fewer pericytes, and decreased perivascular coverage by glial fibrillary acidic protein (GFAP)-positive astrocytic endfeet [Ballabh et al., 2005; Braun et al., 2007; El-Khoury et al., 2006; Wei et al., 2000; Xu et al., 2008]. Since GFAP provides both shape and mechanical strength to the endfeet of astrocytes, its decrease, in combination with alterations in basement membrane proteins and pericyte investiture, suggests a structural weakening of the germinal matrix blood–brain barrier. In contrast, studies of fetuses and preterm neonates during the second and third trimesters of gestation have demonstrated no difference in the primary endothelial tight junction molecules, claudin-5, occludin, and JAM-1, in the germinal matrix compared to cortical gray and white matter regions [Ballabh et al., 2005; Xu et al., 2008].

Fig. 19-3 Morphology of the germinal matrix.

Immunofluorescence of cryosection from the germinal matrix of a 24 week preterm infant labeled with DAPI (blue), GFAP (green), and CD34 (red). The germinal matrix is highly vascular and contains many glial cells.

(Courtesy of Praveen Ballabh, M.D., Department of Pediatrics, Anatomy, and Cell Biology, New York Medical College-Westchester Medical Center, Valhalla, NY.)

Table 19-2 Germinal Matrix Risk Factors for Intraventricular Hemorrhage

GA, gestational age; GM, germinal matrix; N/a, not available; VEGF, vascular endothelial growth factor; WM, white matter.

The germinal matrix has a greater vascular density than other developing brain regions, and recent reports suggest rapid angiogenesis in the germinal matrix, perhaps to support its greater metabolic rate and oxygen requirement [Ballabh et al., 2004]. This angiogenesis is induced by the high levels of vascular endothelial growth factor (VEGF) and angiopoietin (ANGPT)-2 found selectively in the germinal matrix compared to both the cortex and periventricular white matter [Carmeliet, 2003; Ferrara et al., 2003]. Since preclinical studies demonstrate that an increase in the expression of ANGPT-2 in the presence of VEGF promotes the sprouting of immature vessels lacking both basement membrane proteins and pericyte investiture [Carmeliet, 2003; Yancopoulos et al., 2000], the increased presence of these factors in the highly vascular germinal matrix provides further evidence for microvessels lacking a competent blood–brain barrier [Ballabh et al., 2007].

Finally, the venous circulation of the preterm brain may also contribute to the susceptibility to intraparenchymal hemorrhage. Venous blood from the periventricular white matter flows through a fan-shaped array of both short and long medullary veins into the veins of the germinal matrix, and subsequently into the terminal veins found inferior to this highly proliferative zone. The somewhat tortuous anatomy of these veins permits venous stasis, with subsequent hemorrhagic venous infarction of the periventricular white matter [Gould et al., 1987; Takashima et al., 1986; Taylor, 1995].

Alterations in Cerebral Blood Flow Contribute to IVH

Both experimental animal models (i.e., “preclinical data”) and clinical data suggest that alterations in cerebral blood flow may play an important role in the pathophysiology of IVH. (For review see du Plessis [2008].) Cerebral blood flow was reported to be pressure-passive in the asphyxiated preterm infant over four decades ago [Lou et al., 1979], yet the factors that contribute to cerebral autoregulatory systems in the prematurely born are just beginning to be explained. Further, the definition of the cerebral pressure-flow autoregulatory plateau, or that range of cerebral perfusion pressures (mean arterial pressure − cerebral venous pressure) over which cerebral blood flow is normally maintained, is poorly defined for the critically ill, extremely low birth weight preterm neonate. None the less, experimental animal data suggest that at perfusion pressures above this reportedly relatively narrow plateau, cerebral blood flow becomes pressure-passive.

Changes in perfusion pressure, alterations in circulating oxygen and carbon dioxide levels, and the magnitude of these respiratory fluctuations are perhaps the most important regulators of cerebral blood flow in the preterm neonate. In response to these physiologic alterations, cerebral blood flow rises, hemorrhage begins within the germinal matrix, and blood may rupture into the ventricular system. After ventricular distension by an acute hemorrhage event, cerebral blood flow falls. Venous stasis within the periventricular white matter ensues, and parenchymal venous infarction may soon develop [McCrea and Ment, 2008; Volpe, 2001].

Prostaglandins, and particularly the cyclo-oxygenase 2 (COX-2) system, are important modulators of cerebral blood flow in the developing brain [Leffler et al., 1985, 1986]. Hypoxia, hypotension, growth factors, including transforming growth factor-β and epidermal growth factor, and inflammatory modulators, such as interleukin (IL)-6, IL-1β, tumor necrosis factor α (TNFα), and nuclear factor κβ, may all induce COX-2 [Baier, 2006; Dammann and Leviton, 2004; Heep et al., 2003, 2005; Ribeiro et al., 2004]. The resultant increase in prostanoids results in the release of VEGF, the potent angiogenic factor associated in preclinical and clinical studies with IVH.

Following hemorrhage, these same triggers set into motion a cascade of events leading to disruption of tight junction proteins, increased blood–brain barrier permeability, and microglial activation within the developing white matter. The damaged endothelium releases IL-1β and TNFα, and these cytokines promote transmigration of leukocytes across the emerging blood–brain barrier. The activated microglia release reactive oxygen species (ROS), which in turn promote endothelial damage, alter hemostasis, and increase metabolism. (For review, please see du Plessis [2008] and Chua et al. [2009].) The preterm brain is very sensitive to ROS because of the immaturity of those enzyme systems designed to detoxify them, and experimental animal studies suggest that ROS play a significant role in the generation of periventricular parenchymal infarction [Zia et al., 2009].

Clinical events believed to be associated with increases in cerebral blood flow in the preterm infant and the presumed postnatal onset of IVH include endogenous and exogenous events. Vigorous resuscitation, rapid volume re-expansion, endotracheal tube repositioning, recurrent suctioning, complex nursing care procedures, and the administration of sodium bicarbonate, as well as respiratory distress syndrome, hypoxemia, extremes of PCO₂ levels (i.e., both hypo- and hypercapnia), and acidosis have all been associated with IVH [Ancel et al., 2005; Hall et al., 2005; Kaiser et al., 2006; Kluckow, 2005; Kuint et al., 2009; Linder et al., 2003; O’Leary et al., 2009; Osborn and Evans, 2004; Synnes et al., 2001; Whitelaw, 2001] (Box 19-2). Routine screening by bedside cranial ultrasonography has demonstrated IVH after seizures and pneumothoraces in infants who were previously known to have no evidence of hemorrhage [Cooke, 1981; Hill and Volpe, 1981; Hill et al., 1982]. Finally, cerebrovasoactive cytokines are released in association with sepsis and may contribute to IVH.

Box 19-2 Events Associated with Increased Cerebral Blood Flow

Vigorous resuscitation

Endotracheal tube repositioning

Recurrent suctioning

Complex medical and nursing procedures

Acute hypoxemia

Acute hypercarbia

Apnea

Pneumothorax

Rapid volume re-expansion

Cerebral blood flow is markedly diminished after hemorrhage [Del Toro et al., 1991; Volpe, 1997]. Both xenon-133 inhalation and positron emission tomography (PET) have demonstrated significant hypoperfusion for longer than the first postnatal week in preterm infants with IVH [Ment et al., 1984; Volpe et al., 1983]. The cerebral hypoxemia and ischemia that accompany this depression in flow may result in damage not only to the periventricular white matter but also to those neural and glial precursors still residing in the germinal matrix and SVZ. Oligodendroglia and their precursors are among the cells in the developing cortex most sensitive to hypoxia, and their loss is postulated to result in secondary, gestational age-dependent alterations in connectivity in the developing brain [Gimenez et al., 2006; Miller et al., 2002; Nosarti et al., 2009]. Consistent with this hypothesis are reports of significantly decreased cortical gray matter volumes at term equivalent in VLBW preterm infants with grades 1–3 IVH compared to those with no IVH [Vasileiadis et al., 2004], and of decreases in cortical gray and white matter volumes, as well as callosal areas, in preterm subjects at 8–15 years of age [Kesler et al., 2004; Nosarti et al., 2008].

Candidate Genes for IVH

Although the environmental risk factors for IVH have been well studied, not all infants of a given birth weight range or those who suffer a known hypoxemic event experience IVH. Further, the incidence of high-grade IVH has not changed during the past 10 years despite improvements in care (Table 19-3) [Fanaroff et al., 2003, 2007]. Finally, twin analyses show that intraventricular hemorrhage is familial in origin, and over 40 percent of the variance in liability for IVH can be accounted for by shared genetic and environmental factors [Bhandari et al., 2006]. For these reasons, several investigators have tested the hypothesis that the etiology of IVH is multifactorial, involving both environmental and genetic events.

The description of the thrombophilias associated with the factor V Leiden and prothrombin G2021A mutations, and the implication of both abnormalities in fetal and neonatal stroke, suggested to several investigators that these might also be candidate genes for IVH [Debus et al., 1998; Gopel et al., 2001, 2002; Petaja et al., 2001]. Examination of other target genes in the presumptive pathway to IVH shortly followed. As shown in Table 19-4, these include genes contributing to hemostasis, vascular stability, and inflammation [Baier, 2006].

Table 19-4 Susceptibility Genes for IVH

Intraventricular Hemorrhage

Routine cranial ultrasonography via the anterior fontanel is performed in neonatal intensive and special care units around the world, and this modality represents the method of choice for the diagnosis and monitoring of IVH and its complications in preterm infants. Most nurseries use the standard grading system found in Table 19-1, which was originally developed for computed tomography (CT) scanning but fairly quickly applied to ultrasonography (Figure 19-5) [Papile et al., 1978]. Although grade 4 hemorrhages (shown in Figure 19-6) have been noted within the first 6 postnatal hours, most infants initially experience low-grade hemorrhage, which can be seen to progress over the course of time by ultrasonography. The most common sites for parenchymal hemorrhage are the frontal and parietal regions; approximately 50 percent of such hemorrhages occur bilaterally [Shalak and Perlman, 2002].

Fig. 19-5 Germinal matrix hemorrhage (GMH).

A, Coronal view demonstrates a focus of increased echogenicity (arrow) between the head of the caudate and the thalamus in the caudothalamic groove. B, Parasagittal view shows the GMH (arrow) at its usual location along the posterior aspect of the caudothalamic groove.

(Courtesy of Walter C Allan, M.D., Maine Medical Center, Portland, ME.)

Fig. 19-6 Intraparenchymal plus intraventricular hemorrhage.

Large hemorrhage emanating from the left and right germinal matrix destroys the ventricular anatomy bilaterally and merges with parenchymal hemorrhage in the left frontoparietal region (arrow).

(Courtesy of Walter C Allan, M.D., Maine Medical Center, Portland, ME.)

Intraparenchymal Echodensities

Cranial ultrasonography and MRI are also used for the diagnoses of IPE and porencephaly. IPE has been described as an asymmetric white matter lesion that may accompany the onset of IVH or follow it. Rarely found without GMH or IVH, IPE is believed to represent hemorrhagic venous infarction of the periventricular white matter and generally occurs in infants of the youngest gestational ages during the first postnatal week [de Vries et al., 2001]. The ultrasound appearance, found in Figure 19-7, is an echogenic juxtaventricular white matter lesion superior and lateral to the lateral ventricle and extending toward the cortex in a flare pattern without evidence for mass effect on the surrounding brain. Most often, IPE eventually cavitates or atrophies to form either a porencephaly or adjacent ex vacuo ventriculomegaly [De Vries et al., 2004].

Fig. 19-7 Intraparenchymal echodensity.

A sizable amount of blood has ruptured from both germinal matrices into the lateral ventricles and the third ventricle, creating acute hemorrhagic distension. A, Coronal view. B, Parasagittal view. C, Follow-up study demonstrates that, in the deep peritrigonal white matter on the left side, a cavity with an echogenic rim has developed in the deep white matter as a result of secondary infarction. Residual clot in the trigone (arrow) has evolved in the typical way, demonstrated by a hypoechoic center surrounded by an echogenic rim.

Improvements in MRI technology and the use of special MRI-compatible incubators and ventilators have permitted MR imaging of hemorrhage in the developing brain. In the newborn period, MR studies demonstrate a single hemorrhagic lesion adjacent to or communicating with the lateral ventricle in addition to IVH [Dyet et al., 2006] (Figure 19-8).

Fig. 19-8 MRI of intraventricular hemorrhage.

Preterm neonate born at 34 weeks gestational age: MRI performed at 35.5 weeks postmenstrual age. (A) T1-weighted images demonstrate significant IVH with ventriculomegaly and intraparenchymal hemorrhage, consistent with grade IV IVH. (B, C) T2-weighted images confirm the IVH, now seen as hypointense, as well as several periventricular foci of intraparenchymal hemorrhage.

(Courtesy of Gordon Sze, MD, Department of Diagnostic Imaging, Yale University School of Medicine, New Haven, CT.)

Porencephaly

Porencephalies (Greek for “holes in the brain”) are hemispheric cavitary lesions. In neonates, porencephalies follow grade 4 hemorrhages, IPEs, or periventricular venous infarctions. Although most porencephalies are freely communicating with the ventricular system, a porencephaly may present as a fluid-filled cyst that causes increased intracranial pressure in the preterm infant. A typical porencephaly is shown by ultrasound and MRI in Figure 19-9.

Fig. 19-9 Neuroradiologic demonstration of the development of a porencephaly.

A, Coronal cranial ultrasound of a 26-week-gestation male triplet at 2 weeks of age demonstrates a large left-sided parenchymal hemorrhage. B, Axial MR image at term equivalent shows a left-sided porencephaly with mild ventriculomegaly. At the corrected age of 18 months, this child has hemiplegia but normal cognitive function.

(Courtesy of Linda S. De Vries, MD, Wilhelmina Children’s Hospital, UMC, Utrecht, The Netherlands.)

Incidence

While the incidence of IVH depends on a number of factors previously discussed, most centers report that 20–25 percent of VLBW neonates experience IVH, and 6 to more than 20 percent experience grades 3–4 IVH [Fanaroff et al., 2007; Zeitlin et al., 2009]. The incidence of IVH is inversely related to birth weight and gestational age [Sarkar et al., 2009], and data for 18,153 neonates of 501–1500 g birth weight from the National Institute of Child Health and Human Development (NICHD) Neonatal Network between 1 January 1997 and 12 December 2002 suggest that 24 percent of infants of 501–750 g birth weight, 14 percent of infants of 751–1000 g birth weight, 9 percent of those with 1001–1250 g birth weight, and 5 percent of infants with birth weights between 1251 and 1500 g experienced grades 3–4 IVH [Fanaroff et al., 2003, 2007]. Grades 3–4 IVH are more common in male neonates [Leijser et al., 2009]; when Synnes evaluated 3772 neonates of <33 weeks gestational age, 317 (8.3 percent) were found to have grade 3–4 IVH. In this cohort, male gender was an independent and important predictor of IVH (OR = 1.5, 95 percent CI = 1.1–1.9) [Synnes et al., 2006].

Timing of IVH

The risk period for IVH for preterm infants is the first 4–5 postnatal days [Ment et al., 2002; Perlman and Rollins, 2000]. Over 50 percent of all hemorrhages are detectable within the first 6 postnatal hours, and the work of Shaver suggests that almost all of these early hemorrhages are in fact detectable within the first postnatal hour [Shaver et al., 1992]. Less than 5 percent of hemorrhages occur after the fourth or fifth postnatal day (see Figure 19-9). The risk period for IVH is independent of gestational age, and between 10 and 65 percent of infants experience progression, or extension, of their hemorrhages over the first several postnatal days [Whitelaw, 2001]. Extension of IVH and the development of high-grade IVH are twice as common in those infants with the earliest onset of IVH. This progression of hemorrhage has been linked to known clinical events, such as pneumothoraces and seizures, which have been demonstrated to increase cerebral blood flow [du Plessis, 2008].

Clinical Manifestations

The clinical manifestations of IVH are variable, and in one large series of postmortem examinations, over 75 percent of hemorrhages were clinically unrecognized [Allan and Sobel, 2004; Perlman and Rollins, 2000; Volpe, 1997]. None the less, as shown in Table 19-5, infants with large hemorrhages generally are noted to suffer a profound decrease in peripheral hematocrit and may experience coma, seizures, abnormal eye findings, including dilated and nonreactive pupils and loss of eye movements, and changes in tone and reflexes [Dubowitz et al., 1998]. Apneic and bradycardic spells may be attributable to either increased intracranial pressure or changes in cerebral blood flow to brainstem respiratory control centers. Of note, infants with even low-grade IVH may have significantly elevated serum glucose values and evidence for inappropriate secretion of antidiuretic hormone [Moylan et al., 1978; Shalak and Perlman, 2002]. Finally, persistent, unremitting metabolic acidosis unresponsive to alkali therapy or pressor agents is the hallmark of high-grade hemorrhage.

Table 19-5 Clinical Symptoms of High-Grade Intraventricular Hemorrhage

Ment unpublished data, NS27116 Randomized Indomethacin IVH Prevention Trial, 1997.

Cerebrospinal Fluid Studies

Several authors have demonstrated that examination of the CSF in preterm infants undergoing unremarkable sepsis evaluations routinely reveals 0–20 white blood cells (predominantly lymphocytes and monocytes per high-power filed), the absence of red blood cells, and protein values of 100–200 mg%. In patients with new-onset IVH, CSF findings include large numbers of red blood cells, white cells in direct proportion to the peripheral red blood cell/white blood cell ratio, elevations of protein, and low glucose values. In addition, CSF protein levels and red blood cell counts have been reported to correlate with the grade of IVH. Subjects with grades 3–4 IVH generally are found to have CSF protein levels of 500–1000 mg% and red blood cell values in excess of 1,000,000/mm³. In contrast, patients with grade 2 IVH generally demonstrate CSF protein values of less than 500 mg% and red blood cell values of 10,000–250,000/mm³. Patients with GMH, or grade 1 IVH, generally have no evidence for red blood cells within the CSF unless they have also experienced a subarachnoid bleed; similarly, infants with grade 1 IVH are usually found to have normal values for CSF protein. Profound hypoglycorrhachia, with CSF glucose values below 10 mg%, may be found in infants with all grades of hemorrhage, although this is also more common in patients with high-grade hemorrhage.

Days to weeks after the primary hemorrhage, examination of the CSF demonstrates decreasing red blood cell numbers but elevated white blood cell counts in excess of 500–1000/mm³ and persistent hypoglycorrhachia. The latter two findings may lead the clinician to suspect bacterial meningitis, a diagnosis that can only be determined by cerebrospinal culture studies.

Posthemorrhagic hydrocephalus

Posthemorrhagic hydrocephalus, shown by ultrasound in Figure 19-10, is diagnosed in the infant who meets all of the following criteria: presence of intraventricular blood, increasing ventriculomegaly (most commonly diagnosed by cranial ultrasonography), and evidence of increased intracranial pressure (defined as an opening pressure of greater than 140 mm H₂O by either lumbar puncture or, if indicated, cerebral ventricular tap) [McCrea and Ment, 2008; Whitelaw, 2001; Whitelaw et al., 2004]. In addition, in infants with PHH, the ventricular width at the intraventricular foramen by sonographic measurement exceeds 4 mm over the 97th percentile for gestational age [Levene, 1981]. PHH is most common in those infants with high grades of hemorrhage and less frequent in infants with the lowest gestational ages [Kazan et al., 2005]. Because of the compliance of the neonatal brain, the signs and symptoms of hydrocephalus may not be evident for several weeks following hemorrhage, and routine ultrasonographic monitoring is recommended in those infants at risk for PHH.

Fig. 19-10 Intraventricular hemorrhage leading to posthemorrhagic hydrocephalus.

A, Coronal view of a 3-day-old, 29-week preterm neonate with intraventricular blood and right-sided parenchymal hemorrhage. B, Ultrasound performed at age 10 days to follow grade 4 IVH demonstrates significant ventriculomegaly; the patient developed clinical signs of increased intracranial pressure at postnatal day 14–16.

(Courtesy of Walter C Allan, MD, Maine Medical Center, Portland, ME.)

In most instances, the sequence of events resulting in PHH begins with multiple small clots throughout the ventricles. Although PHH is traditionally thought to represent a communicating hydrocephalus with obstruction to the passage of CSF at the level of the arachnoid villi, it is also probable that the CSF is reabsorbed across the ependyma into small penetrating vessels within the developing white matter. Obstruction to flow may thus also occur at the level of these vessels, or less commonly, at the foramina of Luschka and Magendie in the posterior fossa [Cherian et al., 2004]. In the presence of an obstruction in the normal CSF flow pathways, ventriculomegaly develops and periventricular white matter damage may ensue. Suggested mechanisms for periventricular white matter injury following PHH include raised pressure and edema, damaging effects of the free iron released into the CSF during clot lysis, and/or proinflammatory cytokines [Savman et al., 2001]. In addition, a small percentage of infants with IVH may develop acute obstruction at either the foramen of Monro, as shown in Figure 19-11, or the aqueduct secondary to the presence of clots obstructing CSF flow; these infants present with signs of acutely increased intracranial pressure (i.e., apnea, bulging fontanel, split sutures, and lethargy or coma). These infants require immediate neurosurgical attention, as shown in Video 19-1.

Fig. 19-11 Endoscopic view of the foramen of Monro.

Endoscopic view of the foramen of Monro in a preterm neonate with IVH. The third ventricle is central in the image and is filled with clot. To the right of the third ventricle, the ependyma underlying the fornix contains residual clot and hemosiderin, while the thalamus can be seen in the lower left corner and the choroid plexus is visualized in the superior left quadrant.

(Courtesy of Charles C Duncan, MD, Department of Neurosurgery, Yale University School of Medicine, New Haven, CT.)

Randomized controlled trials have evaluated several treatment strategies either to prevent or to reduce the extent of PHH. These include intraventricular streptokinase, repeated lumbar or ventricular punctures, and DRIFT (drainage, irrigation and fibrinolytic therapy), but none of these has yet proved effective [Shooman et al., 2009; Whitelaw, 2000a, 2000b; Whitelaw and Odd, 2007; Whitelaw et al., 2001, 2007].

When treating infants with PHH, the physician must both protect the neonate from damage secondary to increased intracranial pressure and avoid the need for permanent CSF drainage procedures. (For review see Shooman et al. [2009].) Infants with intraventricular and/or intraparenchymal hemorrhage should undergo frequent determinations of occipitofrontal circumference and ultrasonographic determinations of ventricular size. Because prolonged increased intracranial pressure may result in poor feeding, lethargy, apnea, and, ultimately, optic atrophy, if ventricular size increases, neurosurgical consultation is recommended. Further, although some investigators have recommended ventricular puncture with the acute removal of 10–20 mL/kg of CSF for symptomatic infants with PHH, others have explored temporizing measures, such as subgaleal shunt placement, ventricular reservoir placement for intermittent tapping, or third ventriculostomy (Video 19-2). These strategies aim to avoid permanent ventriculoperitoneal shunt placement, and a small retrospective review of these interventions in neonates with IVH reported that 91 percent of those infants with subgaleal shunt placement and 62 percent of those with reservoir taps required permanent shunt placement. Of importance, infection rates were similar in the two populations, but future randomized trials are required to explore these treatment options.

Finally, those infants with acute signs of increased intracranial pressure and ventriculomegaly with blockage at the aqueduct will require immediate neurosurgical intervention.

Cerebral Palsy

Cerebral palsy (CP) is an important and well-established outcome of preterm infants that, in some centers, has been monitored for more than 45 years [Saigal and Doyle, 2008]. Recent data suggesting that the prevalence of neurodevelopmental handicap is increasing in the United States have attributed this finding to the increasing survival rates of many preterm and VLBW infants [Allen, 2008; Behrman and Stith Butler, 2007; Larroque et al., 2008].

Several recent studies have assessed the incidence of CP in children solely with IVH. In all, the incidence of CP is reported to increase with the grade of IVH. In pre-surfactant era studies, Pinto Martin reported that grade 4 IVH was strongly associated with CP (odds ratio = 15.4; 95 percent CI = 7.6–31.1), and in the study of Hansen and colleagues, the odds ratio for CP with grades 3 or 4 IVH was 19.9 (CI = 6.1–64.8) [Hansen et al., 2004; Pinto-Martin et al., 1999]. Further, in Pinto-Martin’s study, any grade IVH alone was associated with CP (odds ratio = 3.14; 95 percent CI = 1.5–6.5).

More recently, Sherlock evaluated 270 neonates of <1000 g birth weight or <28 weeks gestational age at 8 years of age and found CP in 100 percent of those children with grade 4 IVH [Sherlock et al., 2005, 2008]. In contrast, approximately one-fifth of neonates with grades 2 or 3 IVH developed CP, and those with grade 1 hemorrhage experienced no increase in incidence of motor handicap when compared to children with no IVH. Similarly, Patra examined 362 children with birth weights <1000 g at 20 months of age; 104 had experienced grade 1–2 IVH [Patra et al., 2006]. Abnormal neurological examinations, including CP, were significantly more common in the IVH group (odds ratio = 2.60, 95 percent CI = 1.06–6.36).

Although most investigators include the results of all scans performed during the infant’s hospitalization, Allan has reported that CP can be predicted from postnatal day 3 onward in preterm infants in whom serial ultrasonography is performed [Allan et al., 1997]. Further, although preterm infants with CP have traditionally been found to suffer spastic diplegia, concomitant with the rising rate of disabling spastic motor handicap, several investigators have also noted that CP among preterm infants is almost equally divided among the three classical forms: spastic diplegia, hemiplegia, and quadriplegia [Allan et al., 1997; Saigal and Doyle, 2008]. It is believed that this change in distribution reflects more focal and diffuse CNS injuries producing proportional increases in hemiplegias and quadriplegia.

Cognitive Outcome in Neonates with Intraventricular Hemorrhage

IVH is a significant cause of disability in the prematurely born. As shown in Figure 19-12, preterm subjects with grade 4 IVH have significantly worse cognitive abilities and educational performance when compared to those without IVH, as well as to those with lesser grades of hemorrhage [Luu et al., 2009a; Neubauer et al., 2008]. In Sherlock’s study of 298 infants with birth weight <1000 g or gestational age <28 weeks, 100 percent of children with grade 4 IVH had full-scale IQ scores more than 1 SD below the population mean, compared to 60 percent of subjects with grade 3 IVH and approximately 40 percent of all other study participants (no IVH, grade 1 IVH, and grade 2 hemorrhage) [Sherlock et al., 2005, 2008]. Similarly, in Patra’s study of 362 preterm neonates of the same birth weight, 45 percent of children with grades 1–2 IVH had mental scores >2 SD below the population mean, compared to 25 percent for those with no evidence of hemorrhage in the newborn period (odds ratio = 2.0, 95 percent CI = 1.20–3.60) [Patra et al., 2006].

Fig. 19-12 Cognitive outcome of intraventricular hemorrhage at adolescence.

Cognitive outcomes at 16 years of 315 subjects with birth weights between 600 and 1250 g, who participated in the Multicenter Randomized Indomethacin IVH Prevention Trial (NS 27116). The graph demonstrates the percentage of subjects with normal full-scale IQ scores (>80), borderline scores (70–80), and scores in the mentally retarded range (<70); p = 0.005 for differences among the groups. Gr, grade; IVH, intraventricular hemorrhage.

Alterations in Brain Development

The advent of MRI has permitted assessment of the long-term influence of a neonatal injury such as IVH on the developing brain [Mewes et al., 2006; Nosarti et al., 2002; Reiss et al., 2004; Woodward et al., 2006]. Preterm children have smaller brain volumes at adolescence and beyond, and lower cortical gray, cortical white, deep gray, and cerebellar volumes when compared to term controls. In addition, IVH has been associated with significant additional reductions in cerebral gray volumes in preterm subjects during childhood and adolescence [Kesler et al., 2008; Nosarti et al., 2008; Reiss et al., 2004; Vasileiadis et al., 2004]. Studying the largest cohort of subjects born preterm at age 14 years and comparing them with age- matched and socioeconomically matched term adolescents, Nosarti also noted widespread alterations in white matter volumes throughout the developing hemispheres [Nosarti et al., 2008]. All areas of differential volumes between the preterm and term subjects were functionally related, and subjects with IVH and ventriculomegaly had the greatest white matter changes.

Environmental Prevention Strategies

The risk factors for IVH have been well described [du Plessis, 2008; Leijser et al., 2009; Shalak and Perlman, 2002; Soraisham et al., 2009; Synnes et al., 2001, 2006]. When preterm birth is certain, transport of the mother and fetus to a tertiary perinatal center specializing in high-risk obstetric care is mandatory; for over three decades, neonatologists have recognized that those infants who were “outborn” and required transport to the neonatal intensive care unit have consistently higher rates of IVH than “inborn” patients [Hohlagschwandtner et al., 2001].

Abrupt increases in blood pressure, either by volume re-expansion or pharmacologic means, result in the loss of cerebral autoregulation and thus increased cerebral blood flow and are to be avoided. Blood pressure and transcutaneous oxygen pressure should be continuously monitored to prevent hypoxemia, hypotension, and secondary acidosis. Similarly, hypocarbia with PaCO₂ levels below 30 mmHg have also been significantly associated with severe IVH. (For review please see du Plessis [2008].) Depending on the gestational age and the postnatal day, preterm infants are known to have a high incidence of patent ductus arteriosus, and previous studies have suggested that the pharmacologic closure of this lesion is less likely to be associated with abrupt changes in cerebral blood flow than the surgical approach might produce [Weiss et al., 1995].

Pharmacologic Prevention

The well-known sequelae of IVH have prompted the development of pharmacologic prevention strategies. The hallmark of antenatal prevention is corticosteroid therapy, while current potential postnatal strategies include indomethacin, activated factor VIIa, and erythropoietin.

Combined Environmental and Pharmacologic Strategies

Repeated reports from numerous investigators have documented the variations in IVH rate and grade from institution to institution across Europe and the United States [Express Group et al., 2009; Fanaroff et al., 2007; Zeitlin et al., 2009]. Variances in patient populations, availability of prenatal care, and perinatal strategies may all contribute to these findings. Most recently, Obladen reported the results of a prospective monitoring trial for neonates of <1000 g birth weight [Obladen et al., 2008]. Monthly care conferences including all obstetric and neonatology personnel were held, and the hypothesis was that a rigidly adhered-to standard of care would lower the incidence of IVH in this vulnerable patient population. Mandatory care strategies included the following:

Employing this strategy, Obladen noted a decrease in IVH rate from 34 to 13 percent.

Hypoxia-Ischemia

Banker and Larroche originally suggested that WMI occurred in the regions of vascular border zones in the cerebral white matter, and that ischemia would thus be expected to affect these zones preferentially [Banker and Larroche, 1962]. Subsequent neuropathological studies employed postmortem injections of the vessels to demonstrate vascular border and end zones in the periventricular white matter where WMI is typically found [De Reuck, 1984; Takashima and Tanaka, 1978]. These watershed zones are thought to render the periventricular white matter vulnerable to ischemic injury during periods of low cerebral perfusion. This tenuous vascular structure is notable, given that cerebral blood flow is quite low in the white matter of preterm newborns [Borch and Greisen, 1998]. Moreover, there is evidence to suggest that critically ill premature newborns may have at least an intermittently pressure-passive circulation, which puts them at risk for episodes of cerebral ischemia [Lou et al., 1979; Soul et al., 2007]. Indeed, a higher incidence of WMI (as well as IVH) was found in one study of newborns who demonstrated evidence of a pressure-passive circulation [Tsuji et al., 2000]. In a larger prospective study, episodes of pressure-passivity were found in the first 5 days after birth in 97 percent of critically ill newborns with birth weights of <1500 g, with some of the sickest newborns of gestational age <29 weeks showing frequent episodes of pressure-passivity (>50 percent) and hypotension (>40 percent) [Soul et al., 2007]. Notably, these fluctuations in pressure-passivity did not necessarily correlate with periods of hypotension, which occurred less often than episodes of pressure-passivity. These data suggest that a normal blood pressure does not necessarily guarantee adequate cerebral perfusion, particularly given the uncertainty regarding normal or ideal blood pressure ranges for preterm newborns [Laughon et al., 2007].

In addition to the intrinsic properties of the cerebral circulation in preterm newborns, the occurrence of other medical illnesses may contribute to episodes of cerebral ischemia. Newborns are often hypotensive in the first hours and days after birth for a variety of reasons, and a patent ductus arteriosus in particular may contribute to systemic and cerebral hypoperfusion. Serious postnatal infections, such as necrotizing enterocolitis, sepsis, and/or meningitis often affect both systemic and cerebral hemodynamics. Hypoxia and hypercarbia from acute and chronic lung disease, and the impact of various ventilation strategies, suctioning, medications, infusions, transfusions, and other interventions may all exert a deleterious effect on cerebral hemodynamics and contribute to the pathogenesis of WMI [Greisen et al., 1987a; Hoecker et al., 2002; Murase and Ishida, 2005; Patel et al., 2000; Perlman and Volpe, 1983; Perlman et al., 1983; van Alfen-van der Velden et al., 2006].

Inflammation/Infection

Both epidemiological and experimental studies suggest a role for inflammation and/or infection in the pathogenesis of WMI [Dammann and Leviton, 1997; Leviton et al., 1999b, 2010; Van Marter et al., 2002; Wu and Colford, 2000; Yoon et al., 1996]. Studies have shown an association between maternal infection, prolonged rupture of membranes, cord blood IL-6 levels, placental inflammation/infection, and WMI, suggesting that maternal infection is an important etiologic factor in the pathogenesis of WMI [Dammann and Leviton, 1997; Yoon et al., 1996]. Similarly, postnatal infection has been found to be associated with WMI [Shah et al., 2008; Van Marter et al., 2002]. Neuropathologic data showing that there is an abundance of activated microglia in the white matter of the preterm brain suggest a potential mechanism by which WMI may be mediated [Billiards et al., 2006]. Indeed, rodent studies show that activated microglia are necessary to mediate impaired development and death of immature oligodendrocytes by lipopolysaccharide [Lehnardt et al., 2002; Pang et al., 2010]. Peroxynitrite produced by inducible nitric oxide synthase and nicotinamide adenine dinucleotide phosphate (NADPH) oxidase in activated microglia likely plays a role in lipopolysaccharide-induced death of oligodendroglia [Li et al., 2005], and excitotoxicity mediated by N-methyl-D-aspartate (NMDA) receptors on microglia appears to mediate astrocyte death [Tahraoui et al., 2001]. Experimental studies suggest a role for cytokines, such as interferon-γ, in the pathogenesis of WMI. For example, apoptotic death of immature oligodendrocytes has been demonstrated in the presence of interferon-γ, a cytokine generated by macrophages and reactive astrocytes that likely acts on the premyelinating oligodendrocytes through receptor-mediated mechanisms [Baerwald and Popko, 1998; Folkerth et al., 2004a]. These experimental data demonstrate mechanisms by which inflammation/infection may result in WMI, given the observed association between infection and WMI in human epidemiological studies. It should be noted that hypoxia-ischemia can result in activation of microglia, providing another mechanism for oligodendrocyte injury and death, and that significant infection may result in hypoxia-ischemia; thus there is likely an interaction or even synergism between these two mechanisms of injury. There are likely to be genetic factors that modify the degree to which inflammation/infection contributes to WMI, as demonstrated by a study where specific IL-6 genotypes were associated with mental retardation in children with WMI [Resch et al., 2009]. Thus there is increasing evidence of the important role of antenatal and postnatal inflammation/infection in the pathogenesis of WMI; further work is needed to elucidate the specific mechanisms involved and potential preventive therapies.

Vulnerability of Immature White Matter, Particularly Immature Oligodendrocytes

The hypothesis that the periventricular white matter of the premature newborn may be more vulnerable to anoxia than the mature brain was also proposed by Banker and Larroche in their seminal paper [Banker and Larroche, 1962]. A maturational vulnerability of the periventricular white matter is suggested by the finding that WMI occurs much more commonly in the preterm newborn than in the term newborn. Specifically, the observation that the diffuse gliotic lesion of WMI affects the immature oligodendrocyte predominantly, with relative preservation of other cellular elements, suggests that the immature (premyelinating) oligodendrocyte is the cell most vulnerable to injury at this age. Immature oligodendrocytes are susceptible to injury and apoptotic cell death by free radical attack [Back et al., 1998; Oka et al., 1993]. Indeed, apoptosis was demonstrated to be the mechanism of free radical-mediated cell death by cystine deprivation [Back et al., 1998], which was an important finding because apoptosis is postulated to be the mechanism of cell death by moderate ischemia, rather than necrosis, which typically results from severe ischemic insults [Ankarcrona et al., 1995]. A relative lack of antioxidant defenses in the immature oligodendrocyte is one reason why this cell population is particularly vulnerable to free radical attack [Folkerth et al., 2004b]. In addition to free radical attack, premyelinating oligodendrocytes are also susceptible to injury or death by excitotoxicity [Deng et al., 2006; Follett et al., 2004; Husson et al., 2005]. This vulnerability to excitotoxicity likely results from high levels of extracellular glutamate due to high expression of glutamate transporter and because of high expression of glutamate receptors (AMPA and NMDA) by immature oligodendrocytes [Deng et al., 2006; Desilva et al., 2007; Follett et al., 2004; Husson et al., 2005; Karadottir and Attwell, 2007; Rosenberg et al., 2003]. Thus there is cellular and biochemical evidence to support the original postulate that the preterm newborn’s white matter displays a maturational vulnerability that contributes to the high incidence of WMI.

Additional Risk Factors

There are numerous risk factors for WMI that have been identified in epidemiologic and prospective studies of preterm newborns, but the precise mechanism by which each risk factor exerts its effect is not always clear. Lower gestational age at birth and birth weight are associated with a higher risk of WMI and associated neuronal abnormalities, but these two factors are likely related to the three major pathogenetic factors listed above, as well as to other genetic and acquired risk factors [Inder et al., 2003b, 2005]. Chronic ventilation for chronic lung disease in preterm newborns has been identified as a risk factor for later neurodevelopmental deficits [Short et al., 2003], which may be mediated by the respiratory, hemodynamic, and inflammatory disturbances known to be associated with chronic lung disease and ventilator support.

Ultrasound

The evolution of echogenic (or hyperechoic) lesions in the periventricular white matter over the first few weeks after birth, with or without cyst formation (which are echolucent, or hypoechoic), is the classic description of WMI detected by serial cranial ultrasound studies (Figure 19-14). None the less, the detection of WMI by cranial ultrasound remains problematic, in that interobserver variability is high even in rigorous studies of WMI by cranial ultrasound, particularly with regard to the detection of hyperechoic lesions, with less than 50 percent agreement among observers in one study [Kuban et al., 2007]. Furthermore, there is no general agreement regarding the number, size, location, and persistence of hyperechoic lesions defining WMI diagnosed by serial cranial ultrasound [Kuban et al., 2001]. In contrast, hypoechoic or echolucent lesions denoting cysts and the presence of ventriculomegaly are easily and reliably detected by readers of cranial ultrasound [Kuban et al., 2007]. Notably, cystic lesions and significant ventriculomegaly are significantly associated with later cerebral palsy [Kuban et al., 2009]. Ventriculomegaly resulting from atrophy of the periventricular white matter (i.e., volume loss) is usually present within weeks after birth (see Figure 19-14), although it may not be observed by cranial ultrasound until late in the newborn period, suggesting that a cranial ultrasound performed prior to discharge from the neonatal intensive care unit provides additional diagnostic benefit, particularly when previous studies have been normal. Notably, isolated ventriculomegaly is associated with an increased risk of CP [Allan et al., 1997], suggesting that ventriculomegaly without radiologically evident white matter abnormalities likely indicates WMI, particularly in the absence of IVH. Studies correlating ultrasound and autopsy data have demonstrated that the incidence of WMI is underestimated by cranial ultrasound [Carson et al., 1990; Hope et al., 1988], and in the living newborn, MRI has been shown to be superior for the detection of WMI and other brain lesions in the preterm newborn.

Fig. 19-14 Serial ultrasound scans of white matter injury.

A–D, Cranial ultrasound scans performed at 1, 3, and 4 weeks of age show the progressive evolution of WMI, with a progressive increase in echogenicity in the periventricular white matter, an increase in ventricular size (atrophic ventriculomegaly) and extra-axial cerebrospinal fluid volume, and the appearance of a small cyst (arrow) at 4 weeks of age.

Magnetic Resonance Imaging

MRI has been increasingly used to detect WMI in newborns and children born prematurely, although cranial ultrasound remains the preferred imaging modality for most centers because of its portability, safety, and lower cost. Conventional T1- or T2-weighted MRI sequences easily demonstrate cystic lesions that are the hallmark of the focal necrotic lesions of WMI. Noncystic WMI is also detected by MRI, evident as high signal intensity in the cerebral white matter by T2-weighted MRI, and low signal intensity by T1-weighted sequences in the newborn period. Diffusion weighted imaging (DWI) shows restricted diffusion if the MRI is obtained very early in the newborn period [Bozzao et al., 2003], but scans at later ages show increased diffusion in areas of WMI [Counsell et al., 2003]. Punctate lesions have also been detected by MRI in the white matter, often in the corona radiata and posterior periventricular white matter (and occasionally the gray matter) [Dyet et al., 2006; Ramenghi et al., 2007]. The etiology and neuropathology of these punctate lesions are currently unknown, although they have been hypothesized to represent tiny infarctions (possibly venous infarcts), small hemorrhages, or areas of activated microglia or increased cellularity [Rutherford et al., 2010]. Similarly, the clinical significance of these punctate lesions is unknown and follow-up studies are needed to determine if these lesions are associated with neurodevelopmental impairments. MRI performed in infancy or childhood can be useful for the detection of WMI, particularly for children who present with neurodevelopmental impairments but whose neonatal cranial ultrasound studies were normal or only mildly abnormal. MRI studies beyond the newborn period rarely show cysts but typically show increased signal intensity on T2-weighted and fluid-attenuated inversion-recovery (FLAIR) images in areas of gliotic white matter with delayed or decreased myelination (Figure 19-15). Thinning of the corpus callosum usually indicates a decrease in cerebral white matter volume (atrophy) associated with WMI.

Fig. 19-15 MRI of mild and severe white matter injury in young children.

A–C, Axial T2-weighted (A) T2 fluid-attenuated inversion-recovery (FLAIR) (B) and sagittal T1-weighted (C) images demonstrating mild WMI in an 12-month-old born at 30 weeks gestational age; at age 3 years this patient had mild spastic diparesis with accompanying gross and fine motor impairments/delays. D–F, Axial T2-weighted (D), T2 FLAIR (E) and sagittal T1-weighted (F) images demonstrating severe WMI in a 2-year-old boy born at 29 weeks gestational age; this patient had severe diplegia (quadriplegia), infantile spasms, severe cognitive impairment with IQ <70, and cerebral visual impairment. Arrows demonstrate areas of signal abnormality in the white matter on axial sequences and thinning of the corpus callosum on the sagittal sequence.

MRI has been demonstrated to be more sensitive than cranial ultrasound for the detection of WMI in preterm newborns, especially for the noncystic form of WMI [Inder et al., 2003a; Maalouf et al., 2001; Roelants-van Rijn et al., 2001]. As for cranial ultrasound studies, there is no universally accepted measure of the severity or extent of signal abnormality by MRI that defines WMI. While it is clear that greater severity of WMI is correlated with a higher incidence of later neurodevelopmental deficits, there is a broad range of outcomes for mild, moderate, and severe WMI [Woodward et al., 2006], and the threshold for defining clinically significant WMI has not been determined. For example, one study reported diffuse excessive high signal intensity (DEHSI) by T2-weighted MRI within the cerebral white matter at term age in 80 percent of infants born at 23–30 weeks gestational age [Dyet et al., 2006]. It has not yet been demonstrated that this high signal intensity on T2-weighted MRI is correlated with either neuropathological correlates of WMI (such as gliosis) or clinically significant WMI, although the authors reported a correlation between DEHSI and mild developmental delay at 18 months of age [Dyet et al., 2006]. Thus caution is needed when prognosticating outcome based on neuroimaging findings in preterm newborns.

Numerous MRI research studies have been helpful in elucidating the timing, distribution, and effects of WMI in children born prematurely. MRI scans performed soon after birth may be useful for demonstrating acute evidence of WMI, although later MRI studies (e.g., at term age) are superior for demonstrating the full extent of any WMI, gray matter abnormalities, or other congenital or acquired structural brain abnormalities. Diffusion tensor imaging (DTI) has been useful to demonstrate normal white matter development, as well as acute and chronic white matter abnormalities, showing the disruption of normal white matter architecture with WMI [Counsell et al., 2003; Huppi et al., 2001; Yoo et al., 2005]. These DTI abnormalities also have been shown to correlate with developmental outcome [Krishnan et al., 2007]. Similarly, volumetric analysis of brain MRI data has been used to demonstrate normal brain development and the specific gray and white matter volume changes related to WMI. Volumetric MRI data analysis was key to the discovery of the marked reduction in cortical and subcortical gray matter volumes accompanying WMI that had been underappreciated prior to the availability of these quantitative measurement techniques [Abernethy et al., 2004; Inder et al., 2005; Peterson et al., 2000]. Volumetric MRI studies of older children born preterm have been particularly useful in determining the clinical correlates of these volumetric changes, since the larger, mostly myelinated brain of the older child or adolescent can be analyzed in greater detail and more precisely than that of the newborn, and cognitive, social, and behavioral outcome measures are also more definitive and specific than at the young ages often used in neonatal outcome studies [Abernethy et al., 2002; Nosarti et al., 2008; Peterson et al., 2000]. Such studies have demonstrated that, compared with term-born matched controls, children born preterm showed reduced volumes of the white matter, hippocampus, parieto-occipital and sensorimotor cortex, cerebellum, and basal ganglia, and larger ventricular volumes [Nosarti et al., 2008; Peterson et al., 2000]. There was some variability in the findings among studies, which may be related to differences in study populations and measurement techniques, but volumes of the hippocampus, caudate, and sensorimotor cortex correlated with measures of cognitive outcome such as IQ [Nosarti et al., 2008; Peterson et al., 2000]. There is much more to be learned about the pathogenesis and outcome of WMI by detailed quantitative analysis of volumetric, DTI, and functional MRI data obtained at both newborn and later ages, particularly when correlated with perinatal/neonatal risk factors for WMI and with functional outcome measures.

Recommendations for Imaging the Preterm Neonate and Child Born Preterm

Current recommendations for imaging the preterm neonate suggest that screening cranial ultrasound should be performed on all infants with gestational ages <30 weeks at 7–14 days of age, and should be repeated at 36–40 weeks postmenstrual age [Ment et al., 2002]. This recommendation by the Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society is designed to detect both clinically unsuspected IVH, which may require additional clinical and/or radiologic monitoring and changes in management plans, and evidence for PVL and/or ventriculomegaly, which are useful for prognosis and best seen when the infants are examined at term. Although older published guidelines do not recommend routine MRI scans for all premature newborns [Ment et al., 2002], emerging data suggest that cerebral MRI with DTI at term equivalent age may provide important prognostic information about developmental and motor handicap in the prematurely born. A brain MRI may also be performed in an older infant or child born prematurely to confirm clinically suspected WMI when there are cognitive, motor, and/or sensory impairments of unclear etiology (see Figure 19-15).

Cognitive

Lower overall IQ and many specific types of cognitive impairments have been found in children born prematurely when compared with term-born children. Lower gestational age and birth weight are major risk factors for cognitive impairments, as demonstrated in a study of extremely low birth weight infants (<1000 g) showing that only 30 percent of such children were performing at grade level without extra support at 8 years of age [Bowen et al., 2002]. Chronic lung disease, in particular, has been associated with worse cognitive function and greater need for special education and therapies when comparing children born prematurely with and without lung disease [Short et al., 2003]. Deficits of attention and organization, math skills, and reading comprehension are much more common in children born prematurely than in term-born age-matched controls [Johnson et al., 2009]. Most large prospective studies of the cognitive outcome of older children born prematurely do not associate cognitive measures with specific brain lesions (such as WMI) by neonatal cranial ultrasound studies, but quantitative imaging studies to date suggest that it is indeed WMI and its gray matter correlates that underlie much of these cognitive impairments (see Neuroimaging section above) [Abernethy et al., 2002; Nosarti et al., 2008; Peterson et al., 2000]. Knowledge of neuroanatomy and the distribution of abnormalities with WMI allows for hypotheses regarding the potential neural substrate of various types of cognitive impairments. For example, deficits of attention and working memory may relate, at least in part, to involvement of the mediodorsal and reticular nuclei of the thalamus noted in one neuropathologic study of WMI [Ligam et al., 2008].

Social/Behavioral

There has been increasing recognition that children born prematurely have higher rates of social and behavioral difficulties than do term-born children [Delobel-Ayoub et al., 2006]. These include attention deficit/hyperactivity, emotional, anxiety, autism spectrum, and perhaps conduct disorders, particularly in children born extremely prematurely [Delobel-Ayoub et al., 2006; Johnson et al., 2010]. Of these, attentional problems are the most commonly recognized, with anxiety and other similar emotional disorders being increasingly identified and diagnosed. Various rates of these disorders have been reported, likely related in part to the age at evaluation and the method used to define or test for these disorders.

Motor

Lower gestational age at birth is also a risk factor for motor disability. The incidence of CP is much higher in children born extremely prematurely, occurring in approximately 20 percent of children born at ≤26 weeks gestational age, but in only 4 percent of children born at 32 weeks gestational age [Ancel et al., 2006; Wood et al., 2000]. Notably, one study showed that 25 percent of newborns diagnosed with WMI by neonatal cranial ultrasound had later CP, whereas only 4 percent with normal neonatal cranial ultrasound had CP, supporting the notion that significant motor impairments are typically related to WMI [Ancel et al., 2006]. Spastic diparesis is the most common form of CP in children born prematurely, which may present as quadriparesis when severe [Ancel et al., 2006; Kuban et al., 2008]. Hemiparesis is found less frequently than spastic diparesis, and is more likely to result from periventricular hemorrhagic infarction, but may result from asymmetric WMI [Ancel et al., 2006]. The topography of the cerebral white matter is such that the axons subserving the lower extremities are located closest to the ventricle, the axons of the upper extremities are situated lateral to them, and the axons of the facial musculature are located furthest from the ventricle. Since WMI is typically more severe proximal to the ventricle, WMI results in greater abnormalities of tone and power in the lower extremities than the upper extremities and face, i.e., a spastic diparesis. Dystonia is a frequent component of spastic diparesis caused by PVL, and occasionally hypotonia, ataxia, or some other motor abnormality is observed. In the absence of overt CP, coordination difficulties and other abnormalities of motor function or development may be found in children born prematurely.

Visual

While some children born prematurely have retinopathy of prematurity affecting their vision, WMI and other cerebral lesions alone can result in strabismus, nystagmus and other disorders of ocular motility, decreased acuity, visual field deficits, and perceptual difficulties, some of which may not be recognized until school age or later [Jacobson et al., 2002; Jacobson and Dutton, 2000; Pike et al., 1994]. Inferior visual field deficits may be found as a result of WMI, particularly since the optic radiations subserving the lower visual field and the white matter of the association areas of visual cortical regions are frequently affected by WMI [Jacobson et al., 1998]. Thalamic injury has also been shown to be associated with cerebral visual impairment in infants with WMI, separate from the effect of WMI alone on visual function [Ricci et al., 2006]. Children with WMI may have later visual perceptual defects or other higher-order visual impairments that contribute to their cognitive impairments, or in other words, worsen their cognitive and school function, so these are particularly important to detect [Jacobson et al., 1996].

Epilepsy

Children with severe WMI occasionally develop epilepsy, although epilepsy is more commonly related to gray matter lesions, such as periventricular venous infarction. In one study of 5-year-old children born at <26 weeks gestational age, 6 percent developed recurrent nonfebrile seizures [Wood et al., 2000]. Those premature newborns who develop later epilepsy likely have significant neuronal injury, such as the child whose MRI is shown in Figure 19-14, who developed infantile spasms.