Central nervous system injury in preterm infants usually creates different neuroanatomical sequelae compared with that in term infants (Chap. 33, Neonatal Encephalopathy). In preterm infants, cerebral lesions detected by neuroimaging include intraventricular hemorrhage, periventricular hemorrhagic infarction, cystic periventricular leukomalacia, and diffuse white matter injury. All of these are strongly associated with adverse neurodevelopmental outcome. Locatelli and colleagues (2010) found a significantly increased incidence of neurological damage in preterm infants who had periventricular hemorrhage, periventricular leukomalacia, or both.
Cranial sonography remains the preferred approach for detecting frequently occurring brain abnormalities and acute events. It is readily available and reliable for detecting common abnormalities and monitoring brain growth. Because cystic injuries may take 2 to 5 weeks to evolve, serial scans are obtained during this time. In infants whose findings are transient and resolve in the neonatal period, prognosis is improved compared with those whose lesions remain and evolve. At the same time, however, between 4 and 10 percent of prematurely born children may develop cerebral palsy in the absence of lesions. Put another way, 90 to 96 percent of preterm infants with cerebral palsy have cerebral lesions that are detectable using cranial sonography.
There are four major categories of intracranial hemorrhage in the neonate (Volpe, 1995). Subdural hemorrhage is usually the result of trauma. Subarachnoid hemorrhage and intracerebellar hemorrhage usually result from trauma in term infants and hypoxia in preterm infants. Periventricular-intraventricular hemorrhage results from either trauma or asphyxia in half of term infants and has no discernible cause in a fourth. In preterm neonates, the pathogenesis of periventricular hemorrhage is multifactorial and includes hypoxic-ischemic events, anatomical factors, coagulopathy, and many others. The prognosis after hemorrhage depends on its location and extent. For example, subdural and subarachnoid hemorrhage often result in minimal, if any, neurological abnormalities. Bleeding into the parenchyma, however, can cause serious permanent damage.
When the fragile capillaries in the germinal matrix rupture, there is bleeding into surrounding tissues that may extend into the ventricular system and brain parenchyma. This type of hemorrhage is common in preterm neonates, especially those born before 32 weeks. However, it can also develop at later gestational ages and even in term neonates. Most hemorrhages develop within 72 hours of birth, but they have been observed as late as 24 days (Perlman, 1986). Because intraventricular hemorrhage usually is recognized within 3 days of delivery, its genesis is often erroneously attributed to birth events. It is important to realize that prelabor intraventricular hemorrhage also can occur (Achiron, 1993; Nores, 1996).
Almost half of hemorrhages are clinically silent, and most small germinal matrix hemorrhages and those confined to the cerebral ventricles resolve without impairment (Weindling, 1995). Large lesions can result in hydrocephalus or in degenerated cystic areas termed periventricular leukomalacia (Cerebral Palsy). Importantly, the extent of periventricular leukomalacia correlates with cerebral palsy risk.
Damage to the germinal matrix capillary network predisposes to subsequent extravasation of blood into the surrounding tissue. In preterm infants, this capillary network is especially fragile for several reasons. First, the subependymal germinal matrix provides poor support for the vessels coursing through it. Second, venous anatomy in this region causes stasis and congestion, which makes vessels susceptible to bursting with increased intravascular pressure. Third, vascular autoregulation is impaired before 32 weeks (Matsuda, 2006; Volpe, 1987). Even if extensive hemorrhage or other complications of preterm birth do not cause death, survivors can have major neurodevelopmental handicaps. DeVries and associates (1985) attribute most long-term sequelae of intraventricular-periventricular hemorrhage to periventricular leukomalacia. These degenerated cystic areas develop most commonly as a result of ischemia and least commonly in direct response to hemorrhage.
The incidence of ventricular hemorrhage depends on gestational age at birth. Approximately half of all neonates born before 34 weeks, but only 4 percent of those born at term, will have some evidence of hemorrhage (Hayden, 1985). Very-low-birthweight infants have the earliest onset of hemorrhage, the greatest likelihood of parenchymal tissue involvement, and thus the highest mortality rate (Perlman, 1986). Preterm black infants are at disparate risk for intraventricular hemorrhage (Reddick, 2008).
The severity of intraventricular hemorrhage can be assessed by neuroimaging studies. Papile and coworkers (1978) devised the most widely used grading scheme to quantify the extent of a lesion and estimate prognosis.
Grade I—hemorrhage limited to the germinal matrix
Grade II—intraventricular hemorrhage
Grade III—hemorrhage with ventricular dilatation
Grade IV—parenchymal extension of hemorrhage
Data from the Neonatal Research Network indicate that 30 percent of infants born weighing 501 to 1500 g develop intracranial hemorrhage, and 12 percent are grade III or IV (Fanaroff, 2007). Jakobi and associates (1992) showed that infants with grade I or II intraventricular hemorrhage had a greater than 90-percent survival rate and a 3-percent rate of handicap—similar to control infants without hemorrhage of the same age. The survival rate for infants with grade III or IV hemorrhage, however, was only 50 percent. Extremely-low-birthweight infants with grade I or II hemorrhage have poorer neurodevelopmental outcomes at 20 months than controls (Patra, 2006).
Events that predispose to germinal matrix hemorrhage and subsequent periventricular leukomalacia are multifactorial and complex. As noted, the preterm fetus has fragile intracranial blood vessels that make it particularly susceptible. Moreover, preterm birth is frequently associated with infection, which further predisposes to endothelial activation, platelet adherence, and thrombi (Redline, 2008). RDS and mechanical ventilation are commonly associated factors (Sarkar, 2009).
Prevention with Antenatal Corticosteroids
These agents, given at least 24 hours before delivery, appear to prevent or reduce intraventricular hemorrhage incidence and severity. A Consensus Development Conference of the National Institutes of Health (1994) concluded that such therapy reduced rates of mortality, respiratory distress, and intraventricular hemorrhage in preterm infants born between 24 and 32 weeks and that the benefits were additive with those from surfactant therapy. The consensus panel also concluded that benefits of antenatal corticosteroid therapy probably extend to women with preterm premature membrane rupture. A second consensus statement by the National Institutes of Health (2000) recommended that repeated courses of corticosteroids should not be given. They noted that there were insufficient data to prove benefit or to document the safety of multiple courses (Chap. 42, Corticosteroids for Fetal Lung Maturation).
Subsequently, the Maternal-Fetal Medicine Units Network reported that repeated corticosteroid courses were associated with some improved preterm neonatal outcomes, but also with reduced birthweight and increased risk for fetal-growth restriction (Wapner, 2006). Surveillance of this cohort through age 2 to 3 years found that children exposed to repeated—versus single-dose—steroid courses did not differ significantly in physical or neurocognitive measures (Wapner, 2007). It was worrisome, however, that there was a nonsignificant 5.7-fold relative risk of cerebral palsy in infants exposed to multiple steroid courses. At the same time, the 2-year follow-up of the Australasian Collaborative Trial was reported by Crowther (2007). In more than 1100 infants, the incidence of cerebral palsy was almost identical—4.2 versus 4.8 percent—in those given repeated versus single-course steroids, respectively.
Other Preventative Methods
The efficacy of phenobarbital, vitamin K, vitamin E, or indomethacin in diminishing the frequency and severity of intracranial hemorrhage, when administered either to the neonate or to the mother during labor, remains controversial (Chiswick, 1991; Hanigan, 1988; Thorp, 1995). Data from various sources suggest that magnesium sulfate may prevent the sequelae of periventricular hemorrhage, as discussed on Prevention—Neuroprotection.
It is generally agreed that avoiding significant hypoxia both before and after preterm delivery is paramount (Low, 1995). There is presently no convincing evidence, however, that routine cesarean delivery for the preterm fetus presenting cephalic will decrease the incidence of periventricular hemorrhage. Anderson and colleagues (1992) found no significant difference in the overall frequency of hemorrhage in infants whose birthweights were below 1750 g and who were delivered without labor compared with those delivered during latent or active labor. Infants delivered of mothers in active labor, however, tended to have more grade III or IV hemorrhages.
This pathological description refers to cystic areas deep in brain white matter that develop after hemorrhagic or ischemic infarction. Tissue ischemia leads to regional necrosis. Because brain tissue does not regenerate and the preterm neonate has minimal gliosis, these irreversibly damaged areas appear as echolucent cysts on neuroimaging studies. Generally, they require at least 2 weeks to form but may develop as long as 4 months after the initial insult. Thus, their presence at birth may help to determine the timing of a hemorrhagic event.
This term refers to a group of conditions that are characterized by chronic movement or posture abnormalities that are cerebral in origin, arise early in life, and are nonprogressive (Nelson, 2003). Epilepsy and mental retardation frequently accompany cerebral palsy. The cause(s) of cerebral palsy are different in preterm and term infants (Chap. 33, Contributing Factors).
Cerebral palsy is commonly classified by the type of neurological dysfunction—spastic, dyskinetic, or ataxic—as well as the number and distribution of limbs involved—quadriplegia, diplegia, hemiplegia, or monoplegia. The major types and their frequencies are:
Spastic quadriplegia, which has a strong association with mental retardation and seizure disorders—20 percent
Diplegia, which is common in preterm or low-birthweight infants—30 percent
Choreoathetoid types—15 percent
Mixed varieties (Freeman, 1988; Rosen, 1992).
Incidence and Epidemiological Correlates
According to the Centers for Disease Control and Prevention, the prevalence of cerebral palsy in the United States was 3.1 per 1000 children in 2000 (Bhasin, 2006). Importantly, this rate either has remained essentially unchanged or has increased since the 1950s (Torfs, 1990; Winter, 2002). In some countries, the incidence has risen because advances in the care of very preterm infants have improved their survival, but not their neurological prognosis. For example, Moster and associates (2008) presented long-term follow-up of more than 900,000 births in Norway. Of nonanomalous term infants, the cerebral palsy rate was 0.1 percent compared with 9.1 percent in those born at 23 to 27 weeks. Similarly, O’Callaghan and coworkers (2011) studied the epidemiological associations of cerebral palsy and found preterm birth to be the greatest risk factor.
Various clinical and pathological data link severe intraventricular hemorrhage—grade III or IV—and resulting periventricular leukomalacia to cerebral palsy. As described earlier, grade I or II hemorrhages usually resolve without extensive tissue injury. Luthy (1987) reported a 16-fold increased risk of cerebral palsy for low-birthweight infants who had grade III or IV hemorrhage compared with the risk in infants who had either no or grade I or II hemorrhage.
Preterm infants are most susceptible to brain ischemia and periventricular leukomalacia. Before 32 weeks, the vascular anatomy of the brain is composed of two systems. One penetrates into the cortex—the ventriculopedal system. The other reaches down to the ventricles, but then curves to flow outward—the ventriculofugal system (Weindling, 1995). There are no vascular anastomoses connecting these two systems. As a result, the area between these systems, through which the pyramidal tracts pass near the lateral cerebral ventricles, is a watershed area vulnerable to ischemia. Vascular insufficiency before 32 weeks leading to ischemia would affect this watershed area first. Resulting damage of the pyramidal tracts may cause spastic diplegia. After 32 weeks, vascular flow shifts toward the cortex. Thus, hypoxic injury after this time primarily damages the cortical region.
Periventricular leukomalacia is more strongly linked to infection and inflammation than to intraventricular hemorrhage. Zupan and colleagues (1996) studied 753 infants born between 24 and 32 weeks, 9 percent of whom developed periventricular leukomalacia. Those born before 28 weeks, those who had inflammatory events during the last days to weeks before delivery, or those who had both were at highest risk. Perlman and associates (1996) found that periventricular leukomalacia was strongly associated with prolonged membrane rupture, chorioamnionitis, and neonatal hypotension. Bailis and coworkers (2008) reported that chronic—and not acute—placental inflammation was associated with leukomalacia.
Fetal infection may be the key element in the pathway between preterm birth and cerebral palsy (Burd, 2012; Leviton, 2010). In the pathway proposed in Figure 34-2, antenatal reproductive tract infection evokes the production of cytokines such as tumor necrosis factor and interleukins-1, -6, and -8. These in turn stimulate prostaglandin production and preterm labor (Chap. 42, Infection). Preterm intracranial blood vessels are susceptible to rupture and damage, and the cytokines that stimulate preterm labor also have direct toxic effects on oligodendrocytes and myelin. Vessel rupture, tissue hypoxia, and cytokine-mediated damage result in massive neuronal cell death. Glutamate is released, stimulating membrane receptors to allow excess calcium to enter the neurons. High intracellular calcium levels are toxic to white matter, and glutamate may be directly toxic to oligodendrocytes (Oka, 1993).
Schematic representation of the hypothesized pathway between maternal or intrauterine infection and preterm birth or periventricular leukomalacia. Both potentially lead to cerebral palsy. LPS = lipopolysaccharide; PG = prostaglandin.
Many studies have shown that infection and cytokines can directly damage the immature brain. Inoculation of rabbit embryos with Escherichia coli causes histological damage in white matter (Yoon, 1997a). Moreover, tumor necrosis factor and interleukin-6 were more frequently found in the brains of infants who died with periventricular leukomalacia (Yoon, 1997b). Cytokines are strongly linked to white matter lesions even when organisms cannot be demonstrated (Yoon, 2000).
Andrews and colleagues (2008) provided data that raise questions regarding an increased incidence of adverse neurodevelopmental outcomes related to chorioamnionitis. In a cohort of infants born between 23 and 32 weeks, they studied several surrogate indicators and direct markers of in utero inflammation. These included clinical findings, cytokine levels, histological findings, and microbial culture results. Infants undergoing comprehensive psychoneurological testing had similar incidences of intelligence quotient (IQ) scores < 70, cerebral palsy, or both, regardless of these markers. The researchers interpreted their findings to support current practices that employ efforts to delay delivery with preterm pregnancies in the absence of overt intrauterine infection. This does not apply to preterm pregnancy in which clinical chorioamnionitis is diagnosed (Soraisham, 2009). Of 3094 singletons born < 33 weeks, 15 percent had evidence of clinical chorioamnionitis, which included foul-smelling amnionic fluid, maternal fever, uterine tenderness, fetal tachycardia, and maternal leukocytosis. Compared with noninfected infants, cases complicated by chorioamnionitis had significantly increased rates of early-onset sepsis—4.8 versus 0.9 percent—and intraventricular hemorrhage—22 versus 12 percent.
Because epidemiological evidence suggested that maternal magnesium sulfate therapy had a fetal neuroprotective effect, three large randomized trials have been performed to investigate this hypothesis. These studies are discussed in detail in Chapter 42 (Magnesium Sulfate for Fetal Neuroprotection), as well as other evidence that magnesium sulfate is effective in reducing cerebral palsy in preterm infants.