Edward Neuhauser Lecture Current Concepts of Brain the Premature Infant

243 Edward B1 Neuhauser Lecture Current Concepts of Brain the Premature Infant Injury in ...- Joseph J. Volpe1 Brain injury in the premature is an enormous problem. infant and, particularly, approximately Currently, year in the United States with a birth weight the prevention of that injury infants are born each less than or equal to 1500 g. With 42,000 modern neonatal intensive care, approximately 85% of these infants survive [1], and of the survivors, approximately 5-15% exhibit major spastic motor deficits, grouped under the rubric, “cerebral palsy,” and an additional 25-50% exhibit less prominent developmental disabilities, particularly school failure [2-10]. Moreover, data from Sweden [11] and England [12] show that in recent years the prevalence of cerebral palsy in infants with birth weight less than or equal to 1500 g has increased, fragile small probably infants. largely a result of the ever-increasing survival rates for these Major Neurologic Manifestations and Neuropathology of survivors. The major neuropathology for the spastic motor deficits, with or without accompanying intellectual deficits, comprises periventricular leukomalacia and periventricular hemorrhagic infarction (previously termed by us hemorrhagic intracerebral involvement) [13]. Other neuropathologic substrates constitute some of the brain injury in the premature infant, such as posthemorrhagic hydrocephalus and pontosubicular necrosis, but on the basis of current data their roles appear to be small when compared with the two neuropathologic states under discussion. We will The major spastic motor with greater diplegia), and accompaniments. noted earlier, neurologic manifestations of brain injury in the premature infant are deficits consisting primarily of spastic quadriparesis, characteristically affliction of lower than upper extremities (thus, the term spastic spastic hemiparesis. severe Prominent disturbances intellectual of motility deficits and are not infrequent occur, as Less in 25-50% cognition discuss the neuropathology, prevention of periventricular tion. Received January vision April 18, 1989. Presented at the Society for Pediatric April 1988. 30, 1989; 31st annual Radiology, accepted after re- diagnosis, leukomalacia probable causes, and periventricular and possibilities for hemorrhagic infarc- meeting of The San Diego, CA, Periventncular Leukomalacia 1 Departments of Pediatrics, Neurology, and Biological Chemistry, Washington University School of Medicine, St. Louis, MO 63110. Address reprint requests to J. J. Volpe, St. Louis Children’s Hospital, 400 5. Kingshighway Blvd., St. Louis, MO 63110. AJR 153:243-251, August 1989 0361-803X/89/1532-0243 0 American Roentgen Ray Society of cerebral white matter dorsal and lateral [13]. Virchow [14] described the lesion over a century ago; several years later Parrot [15] noted that the injury often affected the premature infant; approximately 60 years later, Rydberg [16] suggested that the injury was related in some way to circulatory insufficiency at delivery; and in 1961, Schwartz [17] postulated that venous stasis caused by parturient events played a role in pathogenesis. The most lucid and complete to the external angle of the lateral ventricles Periventricular leukomalacia is necrosis 244 VOLPE AJR:153, August 1989 description of periventricular leukomalacia is that of Banker and Larroche [18], who in 1962 described the characteristic topography of the lesion and its cellular characteristics and suggested a relation to arterial border zones. Subsequent work has refined further the pathogenesis of periventricular leukomalacia. later. Su,bsequently, with major degrees of periventricular leukomalacia, cystic cavities may develop (Fig. 1), and with less severe degrees of leukomalacia, only diminished cerebral myelin with dilated lateral ventricles is seen. Finally, less marked examples of cystic periventricular leukomalacia may be followed by disappearance of cysts (apparently a result of gliotic scarring) and by diminished volume of cerebral myelin Neuropathology Periventricular leukomalacia is characterized by focal ne- and dilated lateral ventricles Diagnosis in the Neonatal [24]. Period crosis of periventricular white matter dorsal and lateral to the external angles of the lateral ventricles. The two most common sites for periventncular leukomalacia are at the level of the optic radiation adjacent to the trigone of the lateral yentricles and at the level of the frontal cerebral white matter near the foramen of Monro [1 8, 1 9]. These sites are distinctly similar as loci for arterial border zones and/or end zones. The prevalence of such lesions in autopsies of premature infants increases as a function of duration of postnatal survival and of frequency and severity of cardiorespiratory disturbances [19, 20]. In recent years, the prevalence in very-low-birthweight infants has been approximately 25-40% [21-24]. Less severe examples of periventricular leukomalacia consist of the appearance in the periventncular region of acutely damaged glial cells and astrocytosis [25]. Gilles et al. [25] used the term perinatal telencephalic leukoencephalopathy to The two principal periventricular sonography diagnostic procedures used to identify infant are cranial is preferable be- leukomalacia in the living and CT. Cranial sonography cause of its high resolution, portable instrumentation, lack of ionizing radiation, and lower cost. In the coronal projection the lesions appear on sonograms as bilateral, often linear echodensities adjacent to the external angles of the lateral ventricles (Fig. 2) [1 3]. On parasagittal projections the echodensities may be diffusely distributed in periventricular white matter or localized to the sites of predilection for periventricular leukomalacia, such as the regions adjacent to the trigone of the lateral ventricles (Fig. 2) and/or adjacent to the ventricles near the level of the foramina of Monro (Fig. 3). Interestingly, the pathologic correlate of the echodensities has been primarily nonhemorrhagic periventricular leukomalacia [27, 28]. characterize this lesion. Some cases of periventricular leukomalacia may be complicated by petechial hemorrhages within the area of leukomalacia [26]. This secondary hemorrhage into areas of periventricular leukomalacia is more common the more premature the infant [24]. We believe that hemorrhagic periventricular leukomalacia should be distinguished from the periventricular hemorrhagic infarction discussed The characteristic tricular leukomalacia evolution of the echodensities of multiple of perivensmall cysts is the formation rendering a “swiss cheese” appearance (Figs. 2 and 3) [24, 27, 29-31 ]. In some infants, a decrease in echodensities precedes visualization of the cysts [29-32]. Cyst formation occurs 2-3 weeks after the appearance of the echodensities. With relatively small, circumscribed cysts, it is common for the cystic lesions to disappear, at least sonographically, after 1-3 months, leaving enlarged ventricles with decreased cerebral myelin (Fig. 2E). A correlative sonographic-neuropathologic study showed that only 28% of periventricular white detected in vivo with cranial sonography matter [33]. lesions Diffuse were astro- cytic gliosis with or without myelin loss and/or focal necrosis [33] were undetected by sonography. The clinical significance of these missed lesions is unknown. It is difficult periventricular to differentiate lucencies caused by acute leukomalacia from those normally seen on CT scans of neonates [34, 35]. Moreover, small cystic lesions are demonstrable only with sonography [36]. CT shows the - extent of the major lesions well, especially after several weeks or more when the degree of white matter atrophy can be assessed [37]. The need to transport a critically ill, respiratordependent premature infant to the CT scanner has limited its clinical utility. MR imaging also is of limited value early in the neonatal period cause Fig. 1.-Cavitated region of tion of a cerebral hemisphere several weeks. Cavitated infarct descending to internal capsule. perlventrlcular leukomalacia. Coronal secfrom a premature infant who survived for is located directly within myelinating fibers (Courtesy of Dr. Margaret Norman.) for the diagnosis of periventricular the relatively leukomalacia long duration beof of the need for transport, the study, and the difficulty of monitoring the sick infant while in the scanner. However, MR imaging effectively shows the chronic pathologic consequences of periventncular leukomalacia [38, 39]. AJR:153, August 1989 BRAIN INJURY IN PREMATURE INFANTS 245 Fig. 2.-Sonographic findings in a premature infant with presumed penventricular leukomalacia. A, Coronal sonogram at 6 days of age. Note subtle, linear, symmetric echodensitles adjacent to external angles of lateral ventricles. B, Coronal sonogram, transducer angled posterioriy, same study as In A. Note more prominent, symmetric echodensities adjacent to lateral ventricles. C, Coronal sonogram, transducer angled posteriorly, at 5 weeks of age. Note perlventrlcular echolucent regions, presumed cystic periventrlcular leukomalacla. D, Parasagittal sonogram, same study as In C. Note multiple echolucent regions and presumed cystic periventricular Ieukomalacia in posterior parietal region, adjacent to trigones of lateral ventricles. E, Parasagittal sonogram at 4 months of age. Note that cysts are no longer visible and posterior portions of lateral ventricle are moderately dilated. Fig. 3.-Parasaglttal sonogram in a premature Infant 3 weeks of age. Note echolucent regions, presumed cystic pailventricular leukomalacla, antedoily, adjacent to frontal horn of lateral ventricle. Sonograms In first week of life had shown echodensity at this site. The extent genesis of peniventnicular deficits white spastic matter Certainly injury infants in the with of intellectual is unclear. the largest intellectually lesions deficient. and marked diplegia that are often of pre- It is noteworthy the sites dilection for periventricular leukomalacia include fibers subserving the association of visual, auditory, and somatesthetic functions, which are critical for learning. The fact that these fibers are posteriorly located may underlie the observation that neurologic-cognitive outcome in infants with apparent periventricular leukomalacia shown by sonography is worse Clinicopathologic Correlations LEG LEG TRUNK The major long-term clinical correlates of periventricular to a lesser extent, is a type of spastic MOUTH leukomalacia are intellectual deficits spastic diplegia and, [1 3]. Spastic diplegia quadriparesis in which lower extremities are affected more than upper extremities. Three major lines of evidence indicate that periventricular leukomalacia results in spastic diplegia. First, the topography cerebral white matter motor cortex, and those of the lesion includes the region of traversed by descending fibers from fibers subserving the function of the lower extremities are more likely to be affected by the penventricular locus of the necrosis (Fig. 4). Second, the peniventricular echodensities visualized by cranial sonography in the neonatal period ular leukomalacia of spastic and shown on autopsy to reflect periventrichave been documented repeatedly by this author and by others [29] to be followed diplegia. Third, both by the development leukomalacia periventricular infant. and spastic characteristic diplegia have been known for many years to be of the premature Fig. 4.-Schematic diagram of corticospinal tracts from their origin in motor cortex, with descent through periventricular region and into Internal capsule. Loci of pertventrlculanleukomalacla (black squares) are expected to affect descending fibers for lower extremities more than more laterally placed fibers for upper extremities and face. 246 VOLPE AJR:153, August 1989 in those with lesions in posterior rather than anterior cerebral white matter [3, 40]. Perhaps of major importance is the possibility that the relatively large proportion of premature infants who later exhibit less prominent developmental disabilities and schoolfailure relates at least in part to the presence in these patients of smaller degrees of injury to periventricular white matter, which are apparently frequently missed by neonatal cranial sonography (see earlier discussion). passive circulatory abnormality could result from (1) the hypercapnia or hypoxemia (or both) of perinatal asphyxia, res- piratory disease, and/or “normal” vaginal delivery; (2) the cranial “trauma” to the easily deformed premature head at the time of normal vaginal delivery; (3) an “immature” autoregulatory system related to the deficient muscularis of cerebral arterioles in the premature infant; (4) the occurrence of normal blood pressures that are dangerously close to the down slope of a normal autoregulatory curve; Decreases or (5) a combination in arterial blood pressure of Pat hogenesis Pathogenesis of periventricular principal factors. First, certain tomic factors appear to render premature infant vulnerable to tending the basic observations leukomalacia relates to five periventricular vascular anathis region of the brain of the cerebral ischemia. Thus, exof van den Bergh and vander these factors. cations Whatever the mechanism(s), the clinical impli- are enormous. may lead to seriously lowered cerebral blood flow and, ultimately, ischemic injury to vulnerable regions, such as the periventricular white matter. Decreases in systemic blood Eecken, deReuck and coworkers [41 -45] used a colloidal pressure in the premature infant may result from such events as perinatal asphyxia, patent ductus arteriosus [51 ], myocardial failure, apneic spells with bradycardia [52], sepsis, and even simple handling with caretaking procedures [53]. injection-radiographic of arterial border technique to demonstrate the presence zones and end zones in the periventricular region. “distal These arterial fields,” that border and end zones zones, are essentially would be Coupled with the pressure-passive are the relatively limited vasodilatory circulatory disturbance capacity of periventric- is, watershed which expected to be most vulnerable to a fall in perfusion pressure ular blood vessels and the relatively active anaerobic glycolysis with oxygen deprivation in the periventricular white mat- and cerebral blood flow. Moreover, deReuck [43] and Takashima and Tanaka [46] have shown that the prominence of the border zones in periventricular white matter is inversely related to gestational age. Additionally, it was shown that premature infants with periventricular leukomalacia and no obvious history of circulatory disturbance usually were the most premature infants, whereas those infants with periventricular leukomalacia and a clear history of circulatory disturb- ter. Thus, experimental shown a limited increase studies of neonatal animals have in cerebral blood flow in the periven- tricular white matter, presumably because of limited vasodilatation, in response to such potent stimuli as hypoxemia, hypercapnia, and hypotension, when comparisons are made with other brain regions [54-57]. Moreover, when compared with other brain regions, the periventricular white matter of the neonatal insult exhibits (or fetal) animal subjected relatively active anaerobic to hypoxic-ischemic glycolysis, which ance more often were less premature infants [46]. These data suggest that the degree of ischemia required to produce periventricular leukomalacia depends on the state of development of the periventricular vessels and that this state of development is primarily a function of gestational age. The arterial border zones and end zones in the periventricular region have a characteristic distribution, and it is within these exceeds substrate supplies and leads to accumulation of lactic acid and depletion of high-energy compounds in cerebral white matter [54-58]. The latter two metabolic effects presumably occur because of both the limited vasodilatory Capacity and the relatively active glycolytic capacity of periven- zones that periventricular leukomalacia occurs. Indeed, the tricular glial cells. Finally, and perhaps scribed, it is likely that stage related vulnerable of active to the two factors just dewhite most frequent loci for periventricular leukomalacia are within the two distinctive anterior and posterior periventricular bor- the glial cells in periventricular matter are intrinsically to injury because differentiation they are der zones (see neuropathology Second, a pressure-passive section). cerebral circulation appears to in a developmental to astrocytes exist in the premature infant, particularly the severely ill premature infant and this phenomenon would render the infant susceptible to decreases in cerebral blood flow and injury to periventricular white matter with hypotension. Thus, a direct linear relationship between systolic blood pressure and cerebral blood flow, the latter measured by the 1Xe-clearance technique [47], was documented in the first hours of life in a and to oligodendroglia. Moreover, some of the latter have begun active myelination in the perinatal period, and the distribution of hypoxic-ischemic periventricular white matter injury particularly [25]. includes areas of early myelinating activity Probable Cause(s) and Prevention series of premature infants. With intact cerebrovascular autoregulation, cerebral blood flow should not be pressurepassive but rather should remain constant over a wide range of blood pressure, because of arteriolar constriction with elevations of blood pressure and arteriolar dilatation with decreases of blood pressure. (Recent data suggest that clinically stable premature infants do not exhibit this pressurepassive cerebral circulation [48-50].) Experimental and hu- The discussion that periventricular temic hypotension of pathogenesis leads to the conclusion leukomalacia is caused primarily by syssufficiently severe to lead to impaired cerebral blood flow. The timing of the insult appears to be mainly postnatal, and the fact that the prevalence of periventricular leukomalacia observed at autopsy increases as a function of duration of survival (see neuropathology that section) man studies (see, for review, [13]) suggest that the pressure- raises the possibility important. (However, that cumulative postnatal it has been documented insults are some AJR:153, August 1989 BRAIN INJURY IN PREMATURE INFANTS 247 cases of periventricular leukomalacia may relate to prenatal events [59-61 ].) Thus, prevention of this entity depends on careful monitoring of circulatory status; prompt correction of sepsis, apneic spells, and other complications that lead to circulatory failure; and prompt therapy of such failure. However, it is likely that in the most immature infants, because of periventricular vascular factors related to maturation-dependent deficiencies of the periventricular microcirculatory network (see earlier), penventricular leukomalacia may occur with sysescape detection by delineation of patho- temic disturbances so slight that they current methods of monitoring. Further genesis, intervention precise to prevent detection of imminent injury, and timely tissue will be to protect vulnerable periventricular needed komalacia. the entire spectrum of periventricular leu- Periventricular Hemorrhagic Infarction Periventricular hemorrhagic infarction refers to hemorrhagic necrosis of periventricular white matter that is usually large and asymmetric. The lesion most often coexists with intraventricular hemorrhage and, indeed, approximately 15% of all infants with intraventricular hemorrhage exhibit periventricular hemorrhagic infarction [1 3]. In contrast to periventricular leukomalacia, this lesion does not have a long history in the medical literature, and its current prominence relates to the recent increase in survival of very small premature infants, in whom the prevalence is highest. Fig. 5.-Perlventricular hemorrhagic Infarction. Coronal section of ccrebrum. Note region of impending infarction in perlventrlcular white matter lateral and dorsal to external angle of lateral ventrIcle and homolateral to a large asymmetric germinal matnix-intraventricular hemorrhage. ventricular angle where these veins become confluent and ultimately join the terminal vein in the subependymal region. Thus, it appears likely that periventricular hemorrhagic necrosis occurring in association with large intraventricular hemorrhage is a venous infarction. This periventricular hemorrhagic Neuropathology The neuropathology of periventricular tion consists of a relatively large region hemorrhagic of hemorrhagic infarc- sis in the periventricular white matter, necrojust dorsal and lateral ventricle (Fig. 5). The the largest series re- necrosis is distinguishable neuropathologically from secondary hemorrhage into periventricular leukomalacia. However, distinction of these two lesions in vivo often is very difficult. to the external angle of the lateral necrosis is strikingly asymmetric-in Diagnosis in the Neonatal Period ported [62], 67% of such lesions were exclusively unilateral and in virtually all of the remaining cases, grossly asymmetric, although bilateral. Approximately half of the lesions are exten- sive and involve the periventricular white matter from frontal to parietooccipital regions; the rest are more localized. Approximately 80% of cases are associated with large intraventricular hemorrhage, and commonly the parenchymal hemorrhagic lesion is incorrectly described as an “extension” of intraventricular hemorrhage. That the simple extension of blood into cerebral white matter from germinal matrix or lateral ventricle does not account for the periventricular hemorrhagic necrosis has been shown clearly by several neuropathologic The two principal diagnostic procedures used to identify periventricular hemorrhagic infarction in the neonate are cranial sonography and CT scanning, with sonography being the method of choice in the ill neonate. Cranial sonography performed in the coronal projection reveals unilateral or clearly asymmetric bilateral lesions that are globular or triangular (“fan-shaped”) echodensities radiating from the external angle of the lateral ventricle (Fig. 6). On parasagittal frontal projections the extent regions) of the lesion or localized is visualized best and may be classified to parietooccipital as extensive (i.e., extending (i.e., from involving studies [62-68]. study of this periventricular hemorrhagic ne- only the frontal, parietal, or parietooccipital The characteristic sonographic evolution densities is to formation of cysts, which, region) (Fig. 7). of the large echounlike the cysts of Microscopic crosis indicates that the lesion is a hemorrhagic infarction [62, 63, 65-68]. The careful studies of Gould et al. [67] and Takashima et al. [69] emphasize that (1) the hemorrhagic component usually consists of perivascular hemorrhages that follow closely the fan-shaped distribution of the medullary veins in periventricular white matter and that (2) the hemorrhagic component tends to be most concentrated near the periventricular leukomalacia, tend to be single and large. Thus, the term porencephaly often is used to describe the lesion. Also unlike the cysts of periventricular leukomalacia, the cysts that form after periventricular hemorrhagic infarction rarely disappear over time. The full extent of the periventricular parenchymal injury in the months after the acute period is delineated well by CT. 248 VOLPE AJR:153, August 1989 Fig. 6.-Coronal sonogram, presumed periventricular hemorrhagic infarction. Note large, unilateral periventricular echodensity in an infant with bilateral intraventricular hemorrhage. Parenchymal echodensity is on side with larger amount of intraventrlcular blood. Fig. 7.-Periventrlcula parasagittal sonograms. A, Extensive. B, Localized. Similarly, although not practical during the acute period, MR imaging is particularly effective in showing the extent of parenchymal destruction in the months after the neonatal period. Although not generally available, the measurement of regional cerebral blood flow with positron emission tomography (PET) has shown that the extent of the white matter injury in periventricular hemorrhagic infarction may be underestimated by cranial sonography [66]. The entire infarction may not be hemorrhagic and, therefore, may not be demonstrable with cranial sonography. Thus, in approximately 1 5 infants with large, asymmetric intraventricular hemorrhage and associated periventricular parenchymal echodensity, the latter presumably representing the hemorrhagic component of the parenchymal lesion, the PET findings have been similar and dramatic. First, anteriorly, in the region of the presumed hemor- LEG MOUTH Fig. 8.-Schematic diagram of cerebral hemisphere with descending corticospinal tract fihors from their origin in motor contex and periventricular locus (shaded regIon) of periventricular hemorrhagic Infarction. Note that descending fibers for lower cxtremity are likely to be affected as much as, or more than, descending fibers from upper cxtremity. rhagic component flow of the parenchymal lesion, cerebral blood phy as periventricular 59% with [62]. (This should grade is reduced markedly, as expected. Second, and unexpectedly, markedly diminished cerebral blood flow is observed also in the posterior cerebral (parietal and panetooccipital) white matter in the same hemisphere. Indeed, the impairment of cerebral blood flow in the hemisphere containing the periventricular hemorrhagic lesion is much more extensive than could be accounted for by the locus of the sonographically estimated parenchymal blood. This more extensive disturbance has correlated well with the extent of the infarction, demonstrated neuropathologically. echodensity be contrasted greater with than 1 cm, was a mortality hemorrhage, rate of i.e., 8% in the same the severest neonatal unit at the same time for infants of intraventricular grade Ill, but no associated periventricular echodensity [62].) Among the 22 survivors of presumed periventricular hemorrhagic infarction, 86% exhibited major motor deficits, and 64% had cognitive function less than 80% of normal. The motor deficits correlated with the topography of the periven- Clinocopathologic Correlations tricular echodensity, and thus consisted of either spastic hemiparesis or asymmetnc spastic quadriparesis. Prognosis clearly paralleled the severity of the periventricular echodensity sive echodensity (Table 1). Thus, among infants with exten(i.e., echodensity that included frontoparieregions) (Fig. 7), 30 (81%) of 37 died, and of the The major long-term correlates of periventricular hemorrhagic infarction are spastic hemiparesis (or asymmetric quadriparesis) and intellectual deficits. The spastic hemiparesis characteristically affects lower extremities and upper extremities equally, presumably because the periventricular locus of the lesion affects descending fibers from the lower extremity region of motor cortex as much as those from the upper extremity region (Fig. 8). This topography is distinctly different tooccipital seven survivors, all had motor deficits. Overall, only one of the 37 infants with extensive periventricular echodensity survived to have an 10 greater than 80, and none was completely normal (i.e., free of motor deficit and 10 greater than 80). Among infants with localized periventncular echodensity (i.e., echodensity confined to either the frontal, parietal, or parietooccipital regions) (Fig. 7), outcome was more favorable from the more laterally placed middle cerebral which results artery infarct of the asphyxiated full-term extremity involvement. infant, in more upper (Table perivensonograidentified on cranial The overall mortality tricular hemorrhagic rate of infants with presumed infarction, (37%) died. subsequently, 1). Of the 38 infants with localized echodensity, 14 Of the 15 survivors who could be monitored three were free of major motor deficit and AJR:153, August 1989 BRAIN INJURY IN PREMATURE INFANTS 249 TABLE 1: Outcome of Premature Periventricular Echodensity as Infants with a Function of Major Severity Seventy of Penventricular Echodensity1 Outcome Extensive Localized (%) Mortality Major motor deficits Cognitive <80%b Normal survivorc 30/37 7/7 (81) (100) (%) 14/38 12/15 (37) (80) 6/7 0/37 (86) (0) 8/15 (53) 3/29 (10) Severity classified as extensive if echodensity extended from frontal to parietooccipital regions and as localized if echodensity confined to the frontal, parietal, or parietooccipital region, as visualized on parasagittal sonogram. (All echodensities were >1 cm in at least one dimension on sonogram.) Cognitive function <80% of average for age; psychometric tests used were different because they were chosen according to age and best abilities of patient. Surviving infant was free of motor deficit and had cognitive function >80% of average. Fig. 9.-Schematic diagram of cerebral hemisphere, with medullary veins, which drain cerebral white matter, reaching a point of confluence to form terminal vein In subependymal germinal matrix. seven had psychometric test scores in excess of 80% of normal. Overall, three (1 0%) of the 29 infants with localized echodensity and known outcome had both normal motor and cognitive function. The subset of eight infants with localized echodensity that was unilateral had the most favorable cognitive outcome of all; only one had a psychometric test score less than 70% of normal. Pat hogenesis The pathogenesis of the periventricular hemorrhagic necrosis that appears to be a venous infarction is not entirely established. However, a direct, causative relation to germinal matrix-intraventricular hemorrhage seems likely on the basis of three recently defined facts [62]. First, approximately 80% of the parenchymal lesions were observed in association with large (and usually asymmetric) intraventricular hemorrhage. Second, the parenchymal lesions invariably occurred on the same side as the larger amount of intraventricular blood. Third, the parenchymal lesions developed and progressed after the occurrence of the intraventricular hemorrhage. The peak time of their occurrence was the fourth postnatal day [62], that is, when 90% of cases of intraventricular hemorrhage already have occurred [13]. These data raised the possibility that the intraventricular hemorrhage and/or its associated germinal matrix hemorrhage led to obstruction of the terminal veins and hemorrhagic venous infarction. A similar conclusion has been suggested from a recent neuropathologic study [67]. Nevertheless, experimental studies raise the possibility that the intraventricular blood could contribute to the periventricular necrosis by causing (1) impairment of periventricular blood flow resulting from increased intraventricular pressure [70] and/or local release of K from hemolyzed RBCs [71], or (2) local release by RBCs of lactic acid [72] or perhaps other vasoactive or otherwise injurious compounds. Study of high-energy phosphates by in vivo 31P-nuclear MR spectroscopy in five preterm infants with intraventricular hemorrhage has shown changes suggestive of chronic metabolic disturbance after intraventricular hemorrhage [73]. However, these changes may reflect completed injury to periventricular tissue. On balance, therefore, I consider most probable the pathogenetic notion of obstruction of medullary and terminal veins by intraventricular and germinal matrix blood clot (Fig. 9). The pathogenetic scheme that I consider to account for most examples of periventricular hemorrhagic infarction is shown in Figure 1 0. This scheme should be distinguished from that operative for hemorrhagic periventricular leukomalacia (Fig. 1 1), although clearly the lesions could coexist. The frequency of coexistence of the two lesions is not known. Additionally, the two pathogenetic schemes could operate in sequence, that is, periventricular leukomalacia could become hemorrhagic and perhaps a larger area of injury when germinal matrix or intraventricular hemorrhage subsequently causes venous obstruction. Probable Cause(s) and Prevention The preceding discussion of pathogenesis leads to the conclusion that the major cause of periventricular hemorrhagic infarction is germinal matrix-intraventricular hemorrhage. Thus, prevention of the infarction centers on prevention of intraventricular hemorrhage. Although several approaches have been reported to be at least partially effective for prevention [1 3], we have favored the use of muscle paralysis [74]. As noted earlier, a minority of cases of periventricular hemorrhagic infarction may be related to cerebral ischemia and the initial occurrence of periventricular leukomalacia. Thus, in such cases prevention would center on prevention of impaired cerebral blood flow, and the measures described earlier in relation to causes and prevention of periventricular leukomalacia are relevant in this context. In conclusion, two principal lesions underlie the brain injury and the neurologic manifestations thereof in the premature infant, namely, periventricular leukomalacia and periventricu- 250 VOLPE AJR:153, August 1989 GERMINAL MATRIX - ISHEMIA INTRAVENTRICULAR HEMORRHAGE + PERIVENTRICULAR LEUKONALACIA 6EIINAL MATRIX INJURY Fig. 10.-Pathogenesis of peniventricular hemorrhagic infarction. Formulation indicates a major role for germinal matrix and/or intraventricular hemorrhage in causing periventricular venous infarction. PERIVENTRICULAR I (NON-IwMDRRHAGIc VENOUS INFARCTION) CONGESTION PERIVENTRICULAR PERIVENTRICULAR I I ISCHENIA HEMORRHAGIC HEMORRHAGIc PERIVENTRIcULAR INTRAVENTRICLJLAR HEMORRHAGE Fig. 1 1.-Pathogenesis of hemorrhagic periventricular leukomalacia. INFARCTION LEuKONALAcIA 10 11 TABLE 2: Penventricular White-Matter Lesions in the Premature Infant with Intraventncular Hemorrhage Probable Site of Circulatory Disturbance Arterial Venous Proposed Designation Markedly Asymmetric Uncommon Nearly invariable Grossly Hemorrhagic Uncommon Invariable Periventricular Periventricular leukomalacia hemorrhagic infarction lar hemorrhagic infarction. The basic features of these two lesions differ, as shown in Table 2. Both of these lesions may be largely preventable-periventricular leukomalacia by preventing impaired cerebral blood flow, particularly that caused by systemic hypotension, and periventricular hemorrhagic infarction by preventing germinal matrix-intraventricular hemorrhage. REFERENCES 1 . Horbar JO, McAuliffe TL, Adler SM, et al. Variability in 28-day outcomes for very low birth weight infants: an analysis of 1 1 neonatal intensive care units. Pediatrics 1988;82:554-559 2. Calame A, Fawer CL, Claeys V, Arrazola L, Ducret 5, Jaunin L. Neurodevelopmental outcome and school performance of very-low-birth-weight infants at 8 years of age. Eur J Pediatr 1986;145:461-466 3. Fawer CL, Diebold P. Calame A. Periventricular leucomalacia and neurodevelopmental outcome in preterm infants. Arch Dis Child 1987;62:30-36 4. Blennow G, Pleven H, Lindroth M, Johansson G. Longterm follow-up of ventilator treated low birthweight infants. Acta Paediatr Scand 1986;75: 827-831 5. Stewart A, Hope PL, Hamilton P. et al. Prediction in very preterm infants of satisfactory neurodevelopmental progress at 12 months. Develop Med Child Neurol 1988;30:53-63 6. Weisglas-Kuperus N, Uleman-Vleeschdrager M, Baerts W. Ventricular haemorrhages and hypoxic-ischaemic lesions in preterm infants: neurodevelopmental outcome at 3V2 years. Develop Med Child Neurol 1987:29: 623-629 7. Nwaesei CG, Allen AC, Vincer MJ, et al. Effect of timing of cerebral ultrasonography on the prediction of later neurodevelopmental outcome in high-risk preterm infants. J Pediatr 1988;1 12:970-975 8. Sugita K, lai M, Nakajima H, Ohta A. Consistency and changes in the development of extremely low birthweight infants. Brain Dev 1988;1 0: 231 -235 9. Vohr BR, Coll CG, Oh W. Language development oflow-birthweight infants at two years. Develop Med Child Neurol 1988;30:608-61 5 10. Bozynski MEA, Nelson MN, Genaze 0, et al. Cranial ultrasonography and the prediction of cerebral palsy in infants weighing 1 200 grams at birth. Develop Med Child Neurol 1988;30:342-348 1 1 . Hagberg B, Hagberg G, Olow I. The changing panorama of cerebral palsy in Sweden. Acta Paediatr Scand 1984;73:433-440 12. Pharoah POD, Cooke T, Rosenbbcom I, Cooke RWI. Trends in birth prevalence of cerebral palsy. Arch Dis Child 1987;62:379-384 13. Volpe JJ. Neurology of the newborn, 2nd ed. Philadelphia: Saunders, 1987:160-280 14. Virchow A. Zur pathologischen anatomie des gehims. I. Congenitale encephalitis and myelitis. Virchow Arch Pathol Anat 1867;38 :129-141 1 5. Parrot J. Etude sur le ramollissement de l’encephale chez le nouveau-ne. Arch Physiol Norm Pathol 1873;5:59-75 1 6. Rydberg E. Cerebral injury in newborn children, consequent on birth trauma: with an inquiry into the normal and pathological anatomy of the neuroglia. Acta Pathol MicrobiolScand 1932;19S: 1-31 17. Schwartz P. Birth injuries of the newborn: morphology, pathogenesis, clinicalpathology and prevention. New York: Hafner, 1961 1 8. Banker BO, Larroche JC. Periventricular leukomalacia of infancy. Arch Neurol 1962;7:32-50 19. Shuman RM, Selednik LL. Periventncular leukomalacia. A one-year autopsy study. Arch Neurol 1980;37:231-239 20. Barth PG. Stem FC, Oosterkamp RF, Bezemer PD, Kcopman PA. On the relationship between germinal layer haemorrhage and telencephalic leucoencephalopathy in the preterm infant. Neuropaediatrie 198011:17-24 21 . Pape KE, Armstrong DL, Fitzhardinge PM. Central nervous system pathology associated with mask ventilation in the very low birthweight infant: a new etiology for intracerebellar hemorrhage. Pediatrics 1976;58:473481 22. Skullerud K, Westre B. Frequency and prognostic significance of germinal AJR:153, August 1989 BRAIN INJURY IN PREMATURE INFANTS 251 matrix crosis 23. hemorrhage, periventricular in preterm neonates. Acta leukomalacia, and pontosubicular Neuropathol (Berl) 1986;70:257-261 ne- 50. Greisen G, Trojaborg W. Cerebral blood flow, Pa, changes, and visual 24. 25. 26. 27. 28. 29. 30. 31 . 32. 33. 34. 35. 36. 37. 38. 39. 40. 41 42. 43. 44. 45. 46. . 47. 48. 49. Larroche JC. Hypoxic brain damage in fetus and newbom: morphological characteristics, pathogenesis, prevention. J Perinatal Med 1982;1 0(S)2: 29-31 DeVries LS, Wigglesworth JS, Regev R, Dubowitz LMS. Evolution of periventncular leukomalacia during the neonatal period and infancy: correlation of imaging and postmortem findings. Early Hum Dev 1988;17: 205-219 Gilles FH, Leviton A, Dooling EC. The developing human brain: growth and epidemiologic neuropathology. Boston: John Wright, 1983 Armstrong D, Norman MG. Periventncular leucomalacia in neonates: complications and sequelae. Arch Dis Child 1974;49:367-374 McMenamin JB, Shackelford GD, Volpe JJ. Outcome of neonatal intraventricular hemorrhage with periventricular echodense lesions. Ann Neurol 1984;1 5:285-290 Nwaesei CG, Papa KE, Martin DJ, Becker LE, Fitz CH. Penventricular infarction diagnosed by ultrasound: a postmortem correlation. J Pediatr 1984;105:106-1 10 Bowerman RA, Donn SM, DiPietro MA, Damato CJ, Hicks SP. Penventricular leukomalacia in the pre-term newborn infant: sonographic and clinical features. Radiology 1984;1 51:383-390 Dubowitz LMS, Bydder GM, Muschin J. Developmental sequence of penventricular leukomalacia. Arch Dis Child 1985;60:349-358 Slovis TL, Shankaran S. Ultrasound in the evaluation of hypoxic-ischemic injury and intracranial hemorrhage in neonates: the state of the art. Pediatr Radiol 1984;14:67-75 Tamisari L, Vigi V, Fortini C, Scarpa P. Neonatal periventricular leukomaada: diagnosis and evolution evaluated by real-time ultrasound. Helv Paediatr Acta 1986;41 :399-407 Hope PL, Gould Si, Howard S. Hamilton PA, Costello AM de-L, Reynolds EOR. Precision of ultrasound diagnosis of pathologically verified lesions in the brains of very preterm infants. Develop Med Child Neurol 1988;30: 457-471 DiChiro G, Anmitsu T, Pellock JM, Landes RD. Periventricular leukomalacia related to neonatal anoxia: recognition by computed tomography. J Comput Assist Tomogr 1978;2:352-359 Estrada M, Gammal TE, Dyken PR. Periventricular low attenuations. Arch Neurol 1980;37:754-760 Chow PP. Horgan JG, Taylor KJW. Neonatal periventricular leukomalacia: real-time sonographic diagnosis with CT correlation. AJR 1985;145: 155-1 60 Flodmark 0, Roland EH, Hill A, Whitheld MF. Penventncular leukomalacia: radiologic diagnosis. Radiology 1987;162: 1 19-1 24 Flodmark 0, Lupton B, Li 0, et al. MR imaging of periventricular leukomalacia in childhood. AJNR 1989;1 0:111-118 Baker LL, Stevenson DK, Enzmann DR. End-stage periventricular leukomalacia: MR evaluation. Radiology 1988;1 68:809-815 Monset-Couchard M, de Bethmann 0, Radvanyi-Bouvet M-F, Papin C, Bordarier C, Relier JP. Neurodevelopmental outcome in cystic periventricular leukomalacia (CPVL) (30 cases). Neuropediatrics 1988;19: 124-1 31 DeReuck J. The human penventricular arterial blood supply and the anatomy of cerebral infarctions. Europ Neurol 1971;5:321-329 DeReuck J, Chattha AS, Richardson EP Jr. Pathogenesis and evolution of periventricular leukomalacia in infancy. Arch Neurol 1972;27 :229-238 DeReuck JL. Cerebral angioarchitecture and perinatal brain lesions in premature and full term infants. Acta Neurol Scand 1984;70:391 -399 Van den Bergh A, vander Eecken H. Anatomy and embryology of the cerebral circulation. Progr Brain Res 1968;30: 1-26 Van den Bergh R. Centrifugal elements in the vascular pattern of the intracerebral blood supply. Angiologica 1969;20:88-98 Takashima S, Tanaka K. Development of cerebrovascular architecture and its relationship to periventricular leukomalacia. Arch Neurol 1978;35: 11-19 Lou HC, Lassen NA, Fnis-Hansen B. Impaired autoregulation of cerebral blood flow in the distressed newborn infant. J Pediatr 1979;94: 118-125 Younkin DP, Reivich M, Jaggi JL, Obrist WD, Delivoria-Papadopoulos M. The effect of hematocrit and systolic blood pressure on cerebral blood flow in newbom infants. J Cereb Blood Flow Metab 1987;7:295-299 Greisen G. Cerebral blood flow in preterm infants during the first week of life. Acta Paediatr Scand 1986;75:43-51 51 . 52. 53. 54. 55. 56. 57. 58. 59. 60. 61 62. . 63. 64. 65. 66. 67. evoked potentials in mechanically ventilated, preterm infants. Acta Paediatr Scand 1987;76:394-400 Periman JM, Hill A, Volpe JJ. The effect of patent ductus arteriosus on flow velocity in the anterior cerebral arteries: ductal steal in the premature newbom infant. J Pediatr 1981;99:767-771 Perlman JM, Volpe JJ. Episodes of apnea and bradycardia in the preterm newbom: impact on the cerebral circulation. Pediatrics 1985;76:333-338 Omar SY, Greisen G, Ibrahim MM, Youssef AM, Friis-Hansen B. Blood pressure responses to care procedures in ventilated preterm infants. Acta Paediatr Scand 1985;74:920-924 Duffy TE, Cavazzutti M, Cruz NF, Sokoloft L. Local cerebral glucose metabolism in newbom dogs. Ann Neurol 1982;1 1:233-239 Cavazzutti M, Duffy TE. Regulation of local cerebral blood flow in normal and hypoxic newbom dogs. Ann Neurol 1982;1 1:247-254 Young ASK, Hemandez MJ, Yagel 5K. Selective reduction of blood flow to white matter during hypotension in newbom dogs: a possible mechanism of periventricular leukomalacia. Ann Neurol 1982;1 2:445-452 Rice JE Ill, Vannucci AC, Brierley JB. The influence of immaturity on hypoxic-ischemic brain damage in the rat. Ann Neurol 1981;9: 131 -1 41 Wagner KR, Ting P. Westfall MV, Yamaguchi 5, Bacher JO, Myers RE. Brain metabolic correlates of hypoxic-ischemic cerebral necrosis in midgestational sheep fetuses: significance of hypotension. J Cerebr Blood Flow Metabol 1986;6:425-434 Skullerud K, Skjaeraasen J. Clinicopathological study of germinal matrix hemorrhage, pontosubicular necrosis, and penventricular leukomalacia in stillbom. Childs Nerv Syst 1988;4:88-91 Monset-Couchard M, de Bethmann 0, lritz N, Aelier JP. Leucomalacies cystiques neonatales Amamnese perinatale chez 30 survivants. J Gynecol Obstet Biol Reprod 1988;17: 183-189 Ellis WG, Goetzman BW, Lindenberg JA. Neuropathologic documentation of prenatal brain damage. Am J Dis Child 1988;142:858-866 Guzzetta F, Shackelford GD, Volpe S, Penman JM, Volpe JJ. Periventricular intraparenchymal echodensities in the premature newbom: critical determinant of neurological outcome. Pediatrics 1986;78:995-1006 McMenamin JB, Shackelford GD, Volpe JJ. Outcome of neonatal intraventricular hemorrhage with penventricular echodense lesions. Ann Neurol 1984;1 5:285-290 Armstrong DL, Sauls CD, Goddard-Finegold J. Neuropathologic findings in short-term survivors of intraventricular hemorrhage. Am J Dis Child 1987;141 :617-621 Flodmark 0, Becker LE, Harwood-Nash DC, Fitzhardinge PM, Fitz CA, Chuang SH. Correlation between computed tomography and autopsy in premature and full-term neonates that have suffered perinatal asphyxia. Radiology 1980;137:93-103 Volpe JJ, Herscovitch P, Periman JM, Aaichle ME. Positron emission tomography in the newbom: extensive impairment of regional cerebral blood flow with intraventricular hemorrhage and hemorrhagic intracerebral involvement. Pediatrics 1983;72:589-601 Gould SJ, Howard S. Hope PL, Reynolds EOR. Periventricular intraparenchymal cerebral haemorrhage in preterm infants: the role of venous infarction. J Pathol 1987;1 51:197-202 Rushton Dl, Preston PA, Durbin GM. Structure and evolution of echo dense lesions in the neonatal brain. Arch Dis Child 1985;60:798-808 69. Takashima S, Mito T, Ando Y. Pathogenesis of periventricular white matter hemorrhages in preterm infants. Brain Dev 1986;8:25-30 70. Batton DG, Nardis EE. The effect of intraventricular blood on cerebral blood flow in newbom dogs. Pediatr Res 1987;21 :511-515 71 . Edvinsson L, Lou HC, Tvede K. On the pathogenesis of regional cerebral ischemia in intracranial hemorrhage: a causal influence of potassium? Pediatr Res 1986;20:478-480 72. Pranzatelli MA, Stumpf DA. The metabolic consequences of experimental intraventricular hemorrhage. Neurology 1985;35:1299-1303 73. Younkin 0, Medoff-Cooper B, Guillet R, Sinwell T, Chance B, DelivonaPapadopoulos M. In vivo 31P nuclear magnetic resonance measurement of chronic changes in cerebral metabolites following neonatal intraventricular hemorrhage. Pediatrics 1988;82:331 -336 74. Periman JM, Goodman 5, Kreusser KL, Volpe JJ. Reduction in intraventricular hemorrhage by elimination of fluctuating cerebral blood-flow velocity in preterm infants with respiratory distress syndrome. N EngI J Med 1985;313: 1353-1357 68.