Académique Documents
Professionnel Documents
Culture Documents
E-Mail karger@karger.com
550 16th Street, Fourth Floor, San Francisco, CA 94158-2549 (USA)
www.karger.com/neo
E-Mail donna.ferriero@ucsf.edu
to acquired brain injury that has a unique distribution However, babies born at early-term gestation can have
and pattern depending on the inciting event. In addition, WMI. Li et al. [5] reported 11 babies with WMI out of a
external influences such as stress, drug exposure and nu- cohort of 48 term newborns with encephalopathy. Ten
trition can all affect how the brain adapts to these insults. were diffusion positive, 2 had strokes and 1 had basal gan-
The developing brain faces unique challenges from glia injury as well. Abnormalities in the white matter were
oxidative stress, excitotoxicity, inflammation and pro- associated with lower gestational age at birth, while those
grammed cell death that are different from the mature born at 39 weeks gestation or greater had no evidence of
nervous system. The injury response occurs over hours to WMI.
days to weeks. The incidence of perinatal stroke is high, similar to
In deciding whom to image we must ask why. We need that in the elderly, and produces significant morbidity
to know when the injury occurred, when it stops evolv- and severe long-term neurologic and cognitive deficits,
ing and which MRI modalities are best to identify the including cerebral palsy, epilepsy, neuropsychological
primary problem. Diffusion-weighted imaging enables impairments and behavioral disorders.
identification of when the injury occurred. Magnetic res- The pathophysiology of perinatal brain damage is
onance spectroscopy can detect if the injury is still evolv- multifactorial, including changes in blood flow, blood-
ing by the presence of lactate peaks and the slow decline brain barrier function, neuroinflammation and variable
of N-acetylaspartate (NAA). types of cell death and recovery [6]. Neonatal stroke can
be ischemic, thrombotic or hemorrhagic. MRI should be
performed as soon as possible after suspected stroke (e.g.
Term Brain Injury focal seizures in a well newborn) and should include MR
angiography and MR venography to guide therapy. The
Prior to therapeutic hypothermia, MRI revealed pat- presence of diffusion changes in the cerebral peduncles
terns of hypoxic-ischemic injury and evolution both by indicates that hemiplegia will occur [7]. Fifty percent of
structural MRI and magnetic resonance spectroscopy [1]. all sinovenous thromboses of childhood occur in the
Axial T1-weighted images are normal on day 1, but re- newborn period and are often associated with extracor-
duced average diffusivity (Dav) can be seen in ventrolat- poreal membrane oxygenation, congenital heart disease,
eral thalami in about 10% of cases. Lactate peaks are min- thrombophilias and chorioamnionitis [8]. Some new-
imal at this time. By day 4, normal hyperintensity of the borns have no clinical signs of stroke at birth and present
posterior limb of the internal capsule is gone and poste- later with hemiparesis during infancy. This is called pre-
rior putamina are abnormal. Reduced Dav extends to pos- sumed perinatal ischemic stroke and can be arterial or
terior putamina, now reduced by 50%. Lactate peaks are venous. Stroke in preterm infants has a different pattern
elevated at this time. At 8 days, T1 shortening occurs in a of injury, affecting predominantly the lenticulostriate ter-
globular pattern in all basal ganglia. Reduced Dav now ritory rather than the cortex [9].
shows loss of thalamic abnormality a process known as Other acute metabolic disturbances such as hypogly-
pseudonormalization. At this time lactate peaks are de- cemia and hyperbilirubinemia have distinct patterns of
creasing, but now the NAA peak is reduced suggesting injury on MRI. Hypoglycemia affects the occipital lobes
neuronal cell loss (fig.1). With the use of hypothermia, predominantly, while hyperbilirubinemia resulting in
patterns have changed such that Dav, fractional anisot- kernicterus leaves its mark in the globus pallidus. Infec-
ropy (FA) and NAA/choline are tightly correlated on tions can also cause distinct patterns of injury. Bacterial
days 1 and 3, but pseudonormalization occurs later at day infections cause stroke from arteritis or cystic lesions and
10 in hypothermia-treated patients [2, 3]. The two dis- trapped ventricles (Escherichia coli), while viral infec-
tinct patterns of injury (deep gray and watershed) are also tions, especially enteroviruses and parechovirus, have
affected by hypothermia with a decrease in the severe distinct signatures of WMI [10]. Congenital heart disease
deep gray injury, but no significant change in watershed can cause acute events like stroke, but more importantly
injury [4]. The lack of change in watershed pattern with congenital heart disease also affects brain development
hypothermia most likely reflects the fact that this type of which may explain why the damage in this population is
injury occurred prenatally. predominantly in the white matter [11]. When compared
There is a continuum of injury with the most mature to control newborns, those with congenital heart disease
babies manifesting gray matter changes and the most im- had a decrease of 10% in the ratio of NAA to choline (p =
mature babies manifesting white matter injury (WMI). 0.003), an increase of 28% in the ratio of lactate to choline
132.77.150.148 - 6/10/2016 6:22:16 PM
Weizmann Inst. of Science
a b c
d4
d e f
d8
g h i
Fig. 1. Axial T1 is normal on day 1 (d1), but reduced Dav is seen in ed. At 8 days (d8) T1 shortening is globular in the basal ganglia.
ventrolateral thalami. Lactate is minimal at this time. At day 4 (d4) Reduced diffusivity now shows loss of thalamic abnormality
the normal hyperintensity of the posterior limb of the internal cap- (pseudonormalization). Lactate is down, as is NAA. Reprinted
sule is gone and the posterior putamina are abnormal. Reduced with permission from Barkovich et al. [1].
diffusivity extends to the posterior putamina and lactate is elevat-
132.77.150.148 - 6/10/2016 6:22:16 PM
Weizmann Inst. of Science
Fig. 2. Depending on severity and stage, WMI can appear as small single dots in white matter (a), as a string
of dots with signs of cavitation around them (b, arrow) or (at a later stage) as a cavity with small residual dots
(c, arrows) on the periphery.
(p = 0.08), an increase of 4% in Dav (p < 0.001) and a de- rent infection was independent of chronic lung disease
crease of 12% in white matter fractional anisotropy (p < with an OR of 6.2 (95% CI: 1.037.0, p = 0.043) [17].
0.001). Acute periventricular leukomalacia (PVL) can be seen
as multiple white matter lesions on MRI, but as time goes
on, the cysts coalesce leaving only ventricular dilatation
Preterm Brain Injury with some evidence of scarring around the ependyma. In-
deed, cystic PVL is now less common than heretofore. A
While head ultrasound is the best screening device in cohort study demonstrated a decreased prevalence of ul-
the preterm newborn, consecutive MRI scans with at trasound-detected cystic PVL from 1992 to 2002 among
least one at near-term equivalent age is best for under- preterm newborns, partially explained by a reduced dura-
standing how myelination and cortical maturation have tion of mechanical ventilation over the same period [18].
progressed. Longitudinal studies can also show the ex- This was confirmed by a more recent study suggesting
tent of WMI and evaluate cerebellar growth and injury that not only is cystic PVL declining, but severe noncystic
[12]. The term encephalopathy of prematurity has been WMI is as well, with odds of decreasing 0.9-fold per birth
proposed by Volpe [13] to include the findings of corti- year [19]. There was again an association of duration of
cal and subcortical gray matter abnormalities associated mechanical ventilation with severe WMI in this cohort.
with WMI. This study also suggested a protective effect of indometh-
A number of years ago, it was observed that the diffuse acin.
gliotic lesions of WMI affect the very immature oligoden- Bleeding into the subependymal germinal matrix leads
drocyte (preoligodendrocyte) specifically [14]. Initially it to intraventricular hemorrhage (IVH). These hemor-
was thought that these cells died, but later it was found rhages have been categorized based on severity by Papile
that rather than die they undergo a maturational arrest et al. [20] such that grade I refers to germinal matrix hem-
[15]. This arrest prevents them from becoming full oligo- orrhage only, grade II to IVH without ventricular dilata-
dendrocytes with the capacity to make myelin. tion, grade III to IVH with acute ventricular dilatation
There is a spectrum, from mild to severe, of WMI in and grade IV to a venous infarction resulting in an intra-
the preterm newborn brain [16] (fig.2). Many factors can parenchymal hemorrhage or periventricular hemorrhag-
affect the degree of injury seen, especially postnatal infec- ic infarction (PVHI; fig.3). The term PVHI is preferred
tion. In one study, postnatal infection was associated with to grade IV [21]. Bilateral IVH significantly affects neu-
progression of WMI even when adjusting for gestational rodevelopmental outcome only in infants with PVHI
age at birth and chronic lung disease. The effect of recur- [22]. Infants with bilateral grades IIII IVH have similar
132.77.150.148 - 6/10/2016 6:22:16 PM
Weizmann Inst. of Science
rates of neurodevelopmental impairment and similar term-equivalent age. No association was found between
mean Bayley scores as infants with unilateral grades IIII IVH grades IIIII and brain volumes or apparent diffu-
IVH. Surprisingly, infants with grade III IVH had similar sion coefficient values, confirming the observations from
neurodevelopmental outcomes as those with grades III. the study by Brouwer et al. [24].
Therefore, evaluating PVHI as a separate entity, i.e. The WMI not only creates focal scarring, but also dis-
stroke, is warranted. turbs brain development. Using measures of microstruc-
Typically, PVHI occurs between 6 and 96 h of life in a tural brain development, Dav and FA, Dav decreases with
preterm baby (fig.4a). Mechanical ventilation, chorioam- age and FA increases with age in white matter. However,
nionitis and funisitis are associated with typical PVHI. when there is WMI, Dav is actually increased in wide-
Atypical PVHI, occurring <6 or >96 h, is more often as- spread areas and there is no observed increase in FA, sug-
sociated with fetal heart rate abnormalities, planned ce- gesting that the brain is not developing normally [25]. We
sarian section, hypertensive disorders, and fetal throm- now know that extreme prematurity itself is not the driv-
bosis and infarction [23]. Periventricular hemorrhagic er of poor brain development. Brain microstructure mat-
dilatation is independently associated with reduced vol- uration as measured serially in preterm infants is inde-
umes of deep gray matter and the cerebellum, and posi- pendent of extremely preterm birth [4]. Brain injury and
tively associated with apparent diffusion coefficient in comorbid conditions may be the important determinants
occipital, parieto-occipital and parietal white matter at of microstructure maturation, so environmental influ-
132.77.150.148 - 6/10/2016 6:22:16 PM
Weizmann Inst. of Science
ences and existing WMI must be considered when evalu- coids were excluded, the association of morphine was
ating brain development in preterm newborns. even more pronounced, with an 8.2% decrease in cerebel-
Extrinsic factors can also affect brain growth. For ex- lar volume. Morphine did not affect cerebral volumes in
ample, when preterm newborns experience pain-related this study [32].
stress, there is significantly thinner cortex in a variety of Newer techniques of advanced MRI such as suscepti-
cerebral regions, but predominately in the frontal and pa- bility-weighted imaging have allowed increased recogni-
rietal lobes [26]. tion of cerebellar hemorrhage (fig.4b, 5). In one center,
Perhaps the most significant impact of preterm birth cerebellar hemorrhage occurred in 37% of preterm in-
and its complications is manifested in abnormalities of fants and interestingly was associated with intubation at
the cerebellum. Using a variety of volumetric techniques, birth [33]. Prenatal exposure to magnesium sulfate was
it has been shown that cerebellar growth is affected by associated with a reduced number of cerebellar hemor-
supratentorial injury such as IVH, whereas WMI showed rhages (p < 0.008).
no association [27]. Evaluating the growth of the entire It is clear that advanced imaging has enabled us to
hindbrain, Kim et al. [28] found that moderate-to-severe evaluate old and newer lesions and their effects on brain
IVH did indeed slow growth and this was also seen with development and neurodevelopmental outcomes. How-
severe cerebellar hemorrhage. ever, we are still unsure of when much of the injury occurs
It has long been known that prenatal steroids are im- in these populations of newborns, and when the injury
portant to reduce respiratory distress and potentially pro- might stop evolving. Which MRI modalities are the best
tect the fetal brain. However, more recently it has been to identify the primary problem is still unknown. It is rec-
shown that postnatal steroids such as hydrocortisone and ommended that all newborns with encephalopathy
dexamethasone can slow cerebellar growth independent should be imaged with MRI to aid in diagnosis, and the
of gestational age, IVH and other clinical factors [29]. In timing of this imaging will depend on the suspected diag-
addition, both procedural pain and neonatal infection re- nosis. MRI is also helpful in determining the effects of
sult in impaired cerebellar growth [30, 31]. Indeed, small- preterm birth on brain development. When future neu-
er cerebellar volumes were related to poorer cognition rotherapeutic trials commence, MRI will be a useful
and motor/visual integration at 7 years [30]. Other drugs marker of early brain injury to assess efficacy.
also affect cerebellar growth. A recent study showed that
a 10-fold increase in morphine exposure was associated
with a 5.5% decrease in cerebellar volume after adjust- Disclosure Statement
ment for multiple clinical confounders and total brain
volume (p = 0.04). When infants exposed to glucocorti- The author has nothing to disclose.
132.77.150.148 - 6/10/2016 6:22:16 PM
Weizmann Inst. of Science