Flaring and Periventricular Leukomalacia (PVL)
The concept of flaring to describe preterm leukomalacia
The problem of (transient) increased echogenicity of periventricular white matter in preterms has been examined by many since the suggestion by Trounce et al. in 1986 to use the term flaring if these changes were not as bright as clot, subsided only after two weeks and did not always regress with cavitation. A necessity of standardised gain and energy output setting was claimed. The majority of PVL in their experience had associated haemorrhage, suggesting they included some venous infarcts. Spongiosis and microcalcification were typical histopathological findings, adding haemorrhage in only 18 %. No doubt flaring was mostly pathological in this initial study. There is a disturbing lack of cohesion in the subsequent CUS description of preterm white matter, as many assumed flaring to be a straightforward concept (Fawer et al. 1987, de Vries et al. 1988, Appleton et al. 1990, Levene et al. 1992, Jongmans et al. 1993, de Vries et al. 1993, Ringelberg and van de Bor 1993, Fazzi et al. 1994, Lai et al. 1999, Sie et al. 2000). Although critics of the concept claim that machine settings affect appreciation, this has rarely been documented. Grant et al. 1983 stated that echogenicity did not vanish with change in power or gain setting. Contrary to this, Paneth et al. 1990 thought assessment was influenced by minor changes in gain setting. Few mentioned texture of white matter. DiPietro et al. 1986 described a flame-shaped fine blush of echogenicity, not visible from the posterior fontanel due to different orientation of insonated fibers. Pidcock et al. 1990 referred to retrotrigonal white matter in a parasagittal section as homogeneous and striated. For Paneth et al. 1990 the normal was an alternation of interlaced gracile echogenic and echo-poor lines. Whether higher resolution affected subjective interpretation of hyperechogenicity was not examined, but more detail might enhance lesion detection but may also add shades between normal and abnormal. Echogenicity increases by the presence of cells, vessels, calcium and fat. Not only leukomalacia but also normal premyelination gliosis explains this periodic increase of echogenicity around 28-34 w PMA.Historical description
In the early eighties of the 20th century the value of ultrasound in the description of GMH-IVH was recognized. It became clear that haemorrhage was not always found at autopsy in preterms with hyperechoic change in white matter suggestive of bleeding. The early stages of haemorrhagic and purely ischaemic white matter injury seemed sonographically indistinguishable at the time (Hill et al. 1982, Levene et al. 1983, Nwaesei et al. 1984, Dolfin et al. 1984, Delaporte et al. 1985, Rushton et al. 1985, Tamisari et al. 1986). The hyperechoic aspect could be due to clot, but also to necrosis with cellular reaction (astroglial scars and macrophages holding fat and/or hemosiderin) from the subacute stage on. Venous infarction due to GMH/IVH is invariably haemorrhagic, rarely bilateral and if so most likely asymmetrical (Gould et al. 1987, Govaert and de Vries 2010). It affects defined drainage areas: terminal, longitudinal caudate, septal, inferior ventricle vein. In the hours preceding infarction one observes hyperechoic change in white matter, probably due to severe congestion of medullary veins. This is one of the reasons why mild WMD cannot be differentiated with certainty from congestion only. Transience within hours to days possibly attests of congestion, and even some low grade cellular response in white matter might also disappear in days. Because PVL and GMH/IVH co-occur in a substantial number of preterms it is incorrect to dismiss hyperechoic white matter as just congested. Conversely, congestion exists in the absence of GMH/IVH. We have since learnt that preterm leukomalacia is a bilateral and fairly symmetrical disease with predilection for the posterior frontal to parietal peritrigonal white matter, involving anterior frontal, temporal or occipital regions if more extensive (Volpe 1995, Levene 2000, Govaert and de Vries 1997, Agut et al. 2020). There is some evidence that isolated anterior frontal PVL exists but only rarely (Shuman and Selednik 1986, Fawer et al. 1987). Frank bleeding into PVL is uncommon, punctate, short linear or serpiginous haemorrhage is not.Glial fatty metamorphosis: premyelination gliosis or PVL
Pathologists (Virchow in 1867 early on, suggesting this was infectious) have extensively reported pathological fatty change of white matter, mainly in preterm infants. The areas affected harbor macrophages and swollen mineralised axons (stalactite shaped) (Parrot 1873: "steatosis, infarction and haemorrhage in periventricular white matter" – "diffuse interstitial steatosis")(references in Larroche 1977). The condition is further developed as periventricular leukomalacia (PVL) by Banker and Larroche 1962. In acute PVL specimens, coagulation necrosis is seen, with invading macrophages. Axons are affected. Severe variants lead to infarction which results in cystic PVL. Focal and diffuse forms exist (Schwartz 1961). Border zone hypoperfusion and oxygen toxicity were early suspected cofactors. Within the corpus callosum as it normally develops there is a complex relation between the physiological accumulation of sudanophilic lipids and postmenstrual age (Leech and Alvord 1974, Larroche 1977), on the basis of normal premyelin lipogenesis. The presence of fat in and around glial cells is therefore to some extent normal. Similar findings (e.g. for the stratum sagittale) were done elsewhere, where in general this physiological fat deposition precedes formation of mature myelin. But glia may also demonstrate abnormalities in the absence of definitely necrotic lesions. Glial cells respond in two ways, by an increase in the number of cells containing lipid droplets and by an increase in the total amount of histologically demonstrable lipid. Such excessive glial fatty metamorphosis is therefore also a parameter of early injury to the immature premyelin glial cell at a moment of particular susceptibility. This combination of normal and abnormal cell reactions explains the challenge of flaring.The problem of hyperechogenicity
The duration of flaring is important. Transient hyperechogenicities disappear within a week without cyst formation or ventricular dilatation. Dammann and Leviton 1997 proposed a classification of hyperechogenicities into brief (1 to 6 days), intermediate (7 to 13 days) or prolonged (14 days or more). It has been reported that the duration of periventricular hyperechogenicities correlates with outcome, even without cyst formation (Cooke et al. 1985, Appleton et al. 1990, Jongmans et al. 1993, Aziz et al. 1995). Prolonged hyperechogenicities have also been found to predict white matter abnormalities on MRI (Maalouf et al. 2001). Data regarding prognostic significance of persisting hyperechogenicities remain conflicting (Bennett et al. 1990, Horsch et al. 1992). Therefore, the use of the terms flaring and 'grade I PVL' can be problematic, as it may falsely suggest an association with motor or cognitive impairment. We recommend (Agut et al. 2020) the descriptive term 'persistent hyperechogenicities' or 'pathological flaring'. It is important to try to look for signs of pathological "flaring" as described above, note the duration of the hyperechoic changes and monitor for subsequent signs of white matter volume loss. Homogeneity versus heterogeneity of white matter echogenicities is important to assess. Inhomogeneous or patchy hyperechogenicities are likely to represent non-cystic WMI and commonly correlate with MRI abnormalities (Sie et al. 2000). Increasingly higher resolution probes are likely to detect subtler white matter abnormalities. Linear hyperechoic changes perpendicular to the ventricle margin often follow deep venous anatomy and have a haemorrhagic component, globular and coalescing nodules under the central groove above the sulcus circularis superior of the insula correlate more with gliotic changes.Complete classification: objective analysis
The sequential CUS interpretation of preterm WMI, including both cystic and non-cystic forms, is biased by a lack of objectivity in interpretation of the early changes. Generally the presenting feature is increased periventricular echogenicity (often named flaring). These hyperechoic changes can disappear within days or persist for several weeks. When they disappear, they can do so without leaving any abnormality (premyelination gliosis, congestion, low grade cell response) or they can evolve into gliotic and even cystic changes and/or ventriculomegaly with other features of brain volume loss and dysmaturation. In the early eighties it became possible to diagnose cystic PVL with CUS (Hill et al. 1982). Ten years later de Vries and collaborators described a grading system for PVL that has been used until now (de Vries et al. 1992). There is evidence of the correlation between this system and prognosis (Pierrat et al. 2001). However, several limitations in using it are encountered in clinical practice. Assessment of mild (grade I) WMI remains difficult because periventricular hyperechogenicity is a subjective finding. The interobserver agreement on this has been remarkably low in some studies (Pinto et al. 1988, Hintz et al. 2007). Overdiagnosis is problematic. Choroid plexus echogenicity was used as a reference for differentiating between grade I PVL and normal periventricular blush. However cPVL can be observed following flaring that never exceeded the brightness of plexus. CUS findings, in addition to periventricular hyperechogenicity, that support grade I PVL include patchy appearance, extension of echogenicity beyond the peritrigonal area and bilateral but asymmetric periventricular distribution (Govaert and de Vries 2007). Homogeneous symmetrical hyperechogenicities are normal around the anterior frontal horns and the parieto-occipital junction of the lateral ventricles, representing parts of the anterior limb of the internal capsule and the optic radiation, respectively (Boxma et al. 2005, Leijser et al. 2009).Mechanisms of leukomalacia
The pathophysiology of WMI of prematurity is extensively reviewed (Khwaja and Volpe 2008, Back 2017, Schneider and Miller 2019). WMI is related to a confluence of maturational factors that render preterm white matter susceptible to injury. Ischaemia and infection/inflammation are upstream mechanisms that activate two critical downstream mechanisms - excitotoxicity and free radical attack by reactive oxygen and nitrogen species - which in turn lead to death of the vulnerable pre-oligodendrocyte (pre-OL). The maturational state of the cerebral vessels adds to the susceptibility to ischaemia. Hypoperfusion of the brain with ischaemia in the zones between the most important arterial irrigation areas ('border zones') was initially considered to be the common denominator. For the third trimester this area is mainly found between ventriculofugal (striatal) and ventriculopetal (cortical) branches of the middle cerebral artery, showing necrosis in a band of white matter located in the coronal axis of the lateral ventricle at a distance of about 1 cm from its lateral wall (at least 3 mm from the ependyma)(Takashima and Tanaka 1978, Rorke 1992). Often the most damaged areas are found at the junction between the parietal and the occipital lobe and within the frontal lobe in front of the foramen of Monro because those are border zones between major cerebral arteries. But the existence of ventriculofugal and ventriculopetal arteries has been disproved (Nelson et al 1991), as high power stereomicroscopy suggested that ventriculofugal arteries were in fact transcerebral venous channels. This challenged the ischaemia theory. The juxtaventricular area is nevertheless an area devoid of small arteries, a border zone between distal branches from perforators and pial arteries (penetrating from the surface in the direction of the ventricle). In addition brain perfusion of the sick preterm infant often has impaired autoregulation (Greisen and Borch 2001, Soul et al. 2007). The consequential inability to maintain flow in the presence of even minor falls in systemic blood pressure can induce ischaemia. Hypotension and hypocarbia (through cerebral vasoconstriction) are the main preventable risk factors for decreased cerebral blood flow, as there is an association between severe hypotension or hypocarbia and WMI (Okumura et al. 2001, Martens et al. 2003, Shankaran et al. 2006).PVL and arterial watershed zones
The border zone theory of periventricular leukomalacia. The interarterial or border zone hypothesis claims that periventricular leukomalacia is the result of infarction in the watershed area between ventriculopetal cortical branches of the middle cerebral artery and ventriculofugal striatal branches of the anterior, middle cerebral and anterior choroidal arteries. Top: power doppler images of pre-atrial white matter, demonstrating a vessel-poor area halfway between pia and ependyma; observe the highest level of a perforator artery at the lateral ventricle wall. Middle: arterial barium-injected specimen of a coronal brain section at 24 weeks (Pape and Wigglesworth 1979). Notice poor perfusion of the periventricular white matter compared with the deep nuclei and subcortical area. Bottom: the difference between a genuine stroke within the distal area of a perforator artery and PVL can be subtle. Strokes are very rarely bilateral and nearly symmetrical. Bottom: parafrontal germinolysis is found in a plane below the lateral ventricle point, contrary to PVL.Possibility and problem of the concept of flaring
In practice, white matter hyperechogenicities are considered important if they match or exceed plexus density. Soft symmetrical radial echodensities are normal around the frontal horns and the parieto-occipital junction of the lateral ventricle. Perinatal white matter is relatively echogenic due to an increased water content, active late glial migration and early myelination (fat laden oligodendroglia). In case of doubt, insonation through the posterior fontanelle allows for easier differentiation from ischaemia or haemorrhage because the sound beam runs parallel to the axons in the 'blush'. This assessment is always a surrogate of the histopathological reality; cellularity with cell death and congestion may mislead into predicting cystic PVL; this is the main reason for deferring interpretation of flaring into the third week after birth or after a severe inflammatory or ischaemic event that occurs before PMA 34 w. There can be a clear impression of white matter pathology in preterm infants of similar gestational age, when one has built up experience with similar appropriate machine settings. Echoreflections that are (nearly) as echodense as choroid plexus, a tissue often used as a reference for diagnosing leukomalacia, are abnormal. This approach is not without problems: (i) plexus is highly perfused in this age group and furthermore there are varying degrees of congestion or haemorrhage; and (ii) the more preterm the child is, the more prominent and echoic the plexus appears to be, and consequently its property as a calibrant is questionable in the very immature baby. When evaluating periventricular white matter echogenicities it is important to look for signs of pathological flaring (persistent hyperechogenicities), note the duration of the changes and monitor for subsequent signs of white matter volume loss. Homogeneity versus heterogeneity of white matter echogenicities is important to assess: inhomogeneous or patchy hyperechogenicities are likely to represent non-cystic WMI and commonly correlate with MRI abnormalities. High-resolution ultrasound probes and wider angles of insonation are likely to detect subtler white matter abnormalities. The best way to detect white matter lesions is with a wide sector angle (≥ 90 degrees) and high frequency transducer (≥ 7.5 MHz, preferably ≥ 9 MHz). Appropriate time gain compensation should be set so subcortical white matter appears isoechoic throughout, both near and distant to the fontanel. Coronal images from the anterior fontanel are better than parasagittal for studying WMI, because the most vulnerable area (which is lateral and superior to the lateral ventricle at the fronto-parietal transition) is more easily seen on coronal views. Linear hyperechoic changes perpendicular to the ventricle margin often follow deep venous anatomy and have a haemorrhagic component, globular and coalescing nodules under the central groove correlate more with gliotic changes.De Vries classification / leukomalacia grading system
Leukomalacia grading system based on ultrasound (de Vries et al.) versus outcome:- cystic leukomalacia is the best predictor of major handicap, usually cerebral palsy (Rennie and Roberton 2000)
- normal neonatal sonograms: major handicap in up to 5 % (often mental retardation and visual handicap)
- uncomplicated subependymal/intraventricular haemorrhage (no venous infarction, no hydrocephalus): major handicap in 10-15 %
- moderate to severe ventriculomegaly without parenchymal infarction: major handicap in 34 %
- shunted hydrocephalus: major handicap in 60 %
- cystic leukomalacia: major handicap in 90 %
Ultrasound analysis of secondary white matter volume loss or dysmaturation of the central nervous system at term equivalent age (Leijser et al. 2007, Hagmann et al. 2011)
A. Size of frontal horns of lateral ventricles in a coronal section at the foramen of Monro1. Lateral ventricle index > 13 mm
2. Short axis of the frontal horn > 3 mm B. Size of ventricle midbody (> 10 mm) and thalamo-occipital distance (> 24 mm) as indirect sign of parietal (peritrigonal) and occipital white matter volume loss, measured in the parasagittal view C. Width of the interhemispheric fissure, measured in a coronal view at the level of the foramen of Monro as a mean distance between hemispheres >3 mm (between gyral crests above sulcus cinguli) D. Sino-cortical width in the coronal view > 4 mm at the level of the foramen of Monro E. Thickness of the body of corpus callosum in a mid-sagittal view < 1.5 mm F. Complexity of gyral folding inappropriate for gestational age (Inder et al. 2003, Woodward et al. 2006) To depict cysts, serial ultrasound scans should be performed at least weekly due to the relatively late appearance and change over time of lesions. Serial ultrasound imaging is crucial in preterm infants with antenatal, intrapartum or postnatal risk factors for WMI. If information from CUS performed beyond the first weeks of life is not taken into account, the sensitivity of this neuroimaging tool has been proven to be low compared to conventional MRI for predicting CP and other long-term neurological morbidities. A CUS scan should be obtained at TEA (late CUS), looking for signs of cerebral atrophy. Signs of impaired brain growth or brain atrophy are related to adverse neurodevelopmental outcome. In recent years, the de Vries et al. I-IV classification has been adapted, to make sure some entities are well described and included, to stress the rarity of subcortical cystic leukomalacia, and to accomodate the general decrease in prevalence of cystic PVL. This adaptation needs validation with outcome, although mild (I), moderate (II) and severe (III and IV) are similar to the de Vries classification.
| Grade | Description |
|---|---|
| Mild | bilateral hyperechoic change in WM dorsal and lateral to the external angle of the lateral ventricle, most pronounced in posterior frontal to occipital areas; gradually disappearing over days |
| Moderate | - hyperechoic changes in WM persisting beyond the second week with heterogeneous appearance (patchy), typically subrolandic - localized cystic change adjacent to the external angle of the lateral ventricle, typically subrolandic - homogeneous hyperechoic changes in WM followed by one or two of signs of white matter volume loss in serial scans |
| Severe | - persistent and widespread periventricular density (patchy) associated with more than two signs of white matter volume loss in serial scans - extensive cysts in frontoparietal and occipital periventricular white matter or extensive cysts in subcortical white matter (cystic periventricular leukomalacia) |
Three major players for preterm PVL / the fragile oligodendrocyte precursor
The pathogenesis of pre-OL injury relates to operation of two upstream mechanisms, hypoxia-ischaemia and systemic infection/inflammation, both of which are common in premature infants. Cofactors are partial-pressure passivity of the cerebral circulation and postnatal acute hypocarbia. In animal models inflammation (e.g. induced by LPS) potentiates the effects of hypoperfusion. The potentiation has caused subthreshold ischaemic insults to still lead to white matter injury. Chronic hypoxia reduces sirtuin 2 (Sirt2) function in mice: this increases proliferation but limits differentiation of the OL precursor -> rarefied neuropil in white matter, vacuoles in OLs (Wei et al. 2025). Dysmaturation is not an optimal term for this, as it is not specific for alteration in oligodendroglial development. In a postmortem report the presence of NMDA receptors in human control brains was compared with 10 instances of white matter injury (some with typical PVL around 30w PMA) (Jantzie et al. 2015). Increased NMDAR-mediated vulnerability was suggested during early brain development due to an upregulation of receptor subunits, in particular highly calcium permeable NR2B-containing and magnesium-insensitive NR3A NMDARs. Preterm fetal sheep (0.6 gestation) received either sham-asphyxia or asphyxia by cord occlusion for 30 min, and recovered for either 3 or 35 days. The 35 day recovery groups received either intraventricular insulin-like growth factor-1 (1 µg/24 h) or vehicle, from 3 to 14 days after asphyxia (Wassink et al. 2024). Asphyxia was associated with loss of frontal and parietal white matter, reduced myelin basic protein and oligodendrocyte transcription factor 2, persistent inflammation and caspase-3 activation. Four of eight fetuses developed cystic lesions in white matter. Prolonged infusion with insulin-like growth factor-1 restored frontal white matter, improved numbers of oligodendrocyte transcription factor 2-positive and mature oligodendrocytes, reduced astrogliosis and microgliosis after 35 days recovery. Elucidation of these factors lead to experimental (and limited clinical) study of early treatment of white matter injury (Butts et al. 2008, Volpe et al. 2011, Murugan et al. 2013, Kuhn et al. 2019). Although these lost pre-oligodendrocytes may be replaced in focal circumscript lesions by new cells from the subventricular zone and/or by abundant local proliferation, the latter seem not to follow the normal trajectory to produce mature myelin.Risk factors for PVL
Schneider and Miller 2019, Leijser and de Vries 2019, Agut et al. 2020Hypocarbia
Resistance index in the pericallosal artery was significantly lower (higher relative diastole) in infants with PVL and hypocarbia during the first 72 h of life (Okumura et al. 2002), which is suggestive of vasoparalysis at the level of large arteries. This seems at odds with the idea that hypocarbia causes vasoconstriction as a mechanism leading to PVL. The role of autoregulation disturbances in promoting WMI is partly elucidated (Pryds et al. 1989, Greisen 2009). Most preterms have periods of pressure passive perfusion by impaired autoregulation (Soul et al. 2007, Verhagen et al. 2014) or conditions like sepsis or necrotizing enterocolitis (Schat et al. 2016, van der Laan et al. 2016). In children going on to develop severe GMH-IVH pressure-flow autoregulation is seen to be impaired (Pryds et al. 1989). In these children it was found that the CBF-CO2-reactivity was reduced. This may express maximal vasodilatation prior to haemorrhage, related to abnormal changes in setpoint for CBF-CO2 reactivity. Certainly acute prolonged hypocarbia (as for instance during oscillation ventilation (Wiswell et al. 1996, Greisen and Vannucci 2001, Resch et al. 2012) can contribute to the onset of cystic PVL. On the other hand infants with PVL also seem to hyperventilate themselves. So timing of the hypocarbia in relation to imaging is relevant for retaining a low CO2 as a causal factor (Kubota et al. 1996, Stenzel et al. 2020).The fragile OL precursor: infection and inflammation
Infection/inflammation together represent the other major upstream mechanism of preterm WMI (Khwaja and Volpe 2008). Systemic upregulation of pro-inflammatory cytokines and diffuse activation of microglia within white matter lead to pre-OL injury or death. Thus, both intrauterine and postnatal infections as well as necrotizing enterocolitis are known risk factors for preterm WMI and adverse outcome (Wu and Colford 2000, Yoon et al. 2000, Shah et al. 2008). Back and colleagues described the late progenitor OL in human brains between 23 and 32 weeks of gestation and believe that cell type to be in a "developmental window of vulnerability" in the OL lineage. In tissue culture late progenitor OLs are exquisitely vulnerable to toxic effects (Back et al. 2002), far more than early OL progenitors and more mature OLs, which are resistant to damage. It was shown in a culture model that glutamate is highly toxic to differentiating oligodendroglia (Oka et al 1993). Back et al. demonstrated in human postmortem material that injury to oligodendrocyte precursors is mainly due to selective lipid peroxidation and not due to protein nitration (Back et al. 2005). In that hypothetic frame ischaemia is still the most accurate model of generating these lesions, with glutamate excitotoxicity further down the cascade. The culprit is the oligodendroglial cell in diffuse and focal white matter injury, with focal gliosis as one of the consequences.Major maturation-dependent factors underlying the three downstream mechanisms in PVL (Volpe et al. 2011, Schneider and Miller 2019)
Free radical attack: abundant production of both ROS and RNS in PVL (by pre-OLs, microglia, astrocytes); delayed development of antioxidant defenses in pre-OLs; acquisition of Fe2+ by pre-OLs Excitotoxicity: exuberant expression of major glutamate transporter (source of glutamate) by pre-OLs; exuberant expression on soma of pre-OLs of AMPA receptors, which are deficient in the GluR2 subunit and therefore are Ca2+-permeable; exuberant expression on neurites of pre-OLs of NMDA receptors, which also are Ca2+-permeable; likely mechanism of excitotoxicity is receptor dependent generation of ROS/RNS and non-receptor dependent glutathion depletion Microglial activation: microglia are especially abundant in PVL; potent sources of ROS/RNS; presence of TLRs on microglia; activation results in release of free radicals; maturation-dependent concentration of microglia in normal cerebral white matter; microglial activation releases potentially injurious cytokines; TNFalpha derived from microglia in diffuse PVL, potentiates the maturation-dependent toxicity to pre-OLs by interferon gamma (interferon gamma expressed in astrocytes in diffuse PVL and its receptor expressed in pre-OLs); microglial activation impairs glutamate transport and accentuates excitotoxicity The presence of high-level proinflammatory cytokines in preterms with WMI on MRI is consistent with animal models (Hagberg et al. 2002 and 2015; Ellison et al. 2005; Favrais et al. 2011). Microglial cells accumulate in diffuse WMI and their activation releases cytokines, such as TNF-alpha and IFN-gamma, that are toxic to pre-OLs (Verney et al. 2012). Activated microglia also appear in the cortex above such lesions (Penn et al. 2016). Persistent inflammation sensitizes the developing brain to subsequent episodes, and the damage may become apparent only after a second hit (Fleiss and Gressens, 2012).Diffuse white matter disease in the preterm infant
With the increasing use of MRI attention was drawn to a diffuse form of white matter disease, not well identified with ultrasonography. This diffuse white matter involvement, injury without infarction or focal gliosis, was first reported by Gilles and Murphy (1969) as telencephalic leuko-encephalopathy. It is a histological diagnosis, based on abundance and hypertrophy of astroglial cells and/or glial cell injury (karyorrhexis). Other features are glial fatty metamorphosis (increase of glial cells carrying fat droplets in an increased amount) and the presence of amphophilic globules (perivascular exsudates with mineral deposition). A limited amount of sudanophilic staining (oil red O positive) is present in the normal preterm brain within small, irregular, often spindle-shaped cells where it probably reflects an immature stage of myelin precursor turn-over (myelination gliosis). The normal accumulation of this lipid starts around 25 weeks of gestation and is found up to 3 months post term; premyelin lipid would not be expected in corpus callosum in the late second trimester. This entity is clinically less well defined than cystic PVL, but is possibly similar to the so-called DEHSI (diffuse excessive high signal intensity) seen on T2 MRI (Counsel 2003). This diffuse degree of white matter disease may lead to ventriculomegaly without preceding cystic lesions (Kuban 1999) and may have a relation with impaired growth of the corpus callosum in the early postnatal period (Anderson et al. 2005). Selective neuronal necrosis in cortex or brainstem is also common in infants with astroglial cell injury. That pulvinar is often co-affected (Yokochi et al. 1997) may help to predict mental retardation and oculomotor disturbances, though this needs confirmation with MRI and histopathological examination.Diffuse telencephalic leuko-encephalopathy
Focal frontal cPVL is very uncommon, usually the focal cystic variant affects the subrolandic area at the fronto-parieto-occipital transition. The diffuse type of white matter injury without cyst formation has been originally referred to as telencephalic leukoencephalopathy. Cystic lesions are characteristically located in the periventricular area in the preterm infant; with maturation, the sensitive area moves to the cortex. Mixed periventricular and subcortical variants are not common (Sie 2005); fluid within white matter cysts is eventually absorbed by surrounding white matter, giving rise after some time to a gliotic scar zone with locally disturbed myelination.Typical patterns of preterm white matter injury
Typical "white spots against an ivory background" = focal PVL areas in preterm infants with abnormal fatty metamorphosis and coagulation necrosis (Banker and Larroche 1962): lesions are PAS positive, the areas carry granular or sticklike disintegrating axons (Cajal's retraction balls), calcium deposition and iron in microglia; the end result is "sclerosis of the white matter" with loss of volume and decreased myelin content; an area of 2 mm separates the lesions from the ependyma, the temporal lobe is least affected; the ventricle wall is intact; the lesions are bilateral but not strictly symmetrical.PVL versus venous infarction
Necrotic white matter, even if it has not been superimposed on GMH, may suffer haemorrhagic conversion due to hyperperfusion, microvascular damage due to free radical formation and disturbed haemostasis. Differentiation is therefore necessary between venous infarction of white matter associated with GMH/IVH (right hemisphere) and PVL (left hemisphere). Near the frontal horn, venous infarction creates a triangular hyperechoic area pointing to the compressed vein in the lateral ganglionic eminence. This triangular appearance is not seen with PVL.| IPL (venous infarction) | cPVL (cystic leukomalacia) |
|---|---|
| usually unilateral, together with ipsilateral (large) IVH, fan shaped echodensity, sharply delineated, almost half develop hemiplegia, 15-20% require shunt, focal gliosis on later MRI; posthaemorrhagic ventriculomegaly regular shape, signs of increased pressure (ballooning), no signs of atrophy (no widening of interhemispheric fissure) | mainly bilateral, no/only small IVH or hyperechoic germinolysis, patchy appearance, irregular border, outcome almost invariably CP, even with unilateral cysts bilateral gliosis on later MRI; ventriculomegaly due to leukomalacia irregular shape, no signs of increased pressure (ex vacuo), signs of atrophy (widening of interhemispheric fissure) |
Timing of cystic lesions
As PVL may develop antenatally, for example following TTTS with or without the death of the co-twin, an early, first day ultrasound examination is crucial in timing the injury. In these situations fully developed cysts can be seen on the scan made shortly after birth. Leukomalacia is considered antenatal when the cysts are present at birth, or when cysts are clearly evolving from echogenic areas during the first 10 days after birth. With a perinatal insult, it usually takes two to three weeks to develop extensive cystic PVL, even longer: the median date of cyst formation is 18 days but the range was 10-39 days (Kubota et al. 2001). The shortest event to first sonographic cyst interval we observed was 13 days. There is a relation between the severity of cystic PVL and the time it takes for the cysts to develop, i.e. the smaller the cysts, the more time for them to develop (Pierrat et al 2001, de Vries 2004). Sometimes the initial echogenicity develops several weeks after birth following an acute clinical deterioration like septicaemia or necrotising enterocolitis which is referred to as late-onset PVL (de Vries et al. 1986).Significance of white matter injury
In very low birthweight infants (<1500g) there has been a reported 15–25% overall prevalence of sonographic leukomalacia (before 2000). The cystic variant prevailed in around 3 % of VLBW infants, whereas the non-cystic variant reached three to five times this figure. With ultrasound cystic PVL was more common between 1000 and 1500 g than below 1000g. Cystic leukomalacia was the best predictor of major handicap, usually cerebral palsy [normal neonatal sonograms: major handicap in < 5 %; uncomplicated GMH/IVH (no venous infarction, no hydrocephalus): major handicap in 10-15 %; moderate to severe ventriculomegaly without parenchymal infarction: major handicap in 34 %; shunted hydrocephalus: major handicap in 60 %; cystic leukomalacia: major handicap in 90 %]. In contrast to GMH/IVH, which mostly occurs during the first 72 hours and rarely beyond the first week, in preterm infants leukomalacia can occur until term. Three overlapping neuropathological variants have been described: cWMI with macroscopic focal necrosis evolving to cysts; non-cystic WMI with multiple focal areas of necrosis that evolve into glial scars; and diffuse astrogliosis without focal necrosis (Khwaja and Volpe 2008, Back et al. 2017). To some extent, these variants correlate with distinct clinical outcomes. Hypoxic-ischaemic and toxic injury to the metabolically active oligodendroglia (OL) are now considered to be important. Back and colleagues describe the late progenitor OL in human brains between 23 and 32 weeks of gestation and believe that cell type to be in a "developmental window of vulnerability" in the OL lineage. Several groups found an association between ascending intrauterine infection, production of proinflammatory cytokines and white matter injury (Leviton & Gilles 1984, Perlman, Risser, & Broyles 1996, Verma et al. 1997)(measuring proinflammatory cytokines in amniotic fluid, fetal plasma and cord blood). TNF alpha, interleukin (IL)-1 beta and interleukin-6 have been shown to be raised in the amniotic fluid of pregnant women with chorioamnionitis. Others studied the in situ expression of proinflammatory cytokines and found a high expression of TNF alpha, a myelinotoxic factor, in the brains of newborns with PVL (Kadhim et al. 2001). The relationship between chorioamnionitis and cerebral palsy was evaluated in a meta-analysis and chorioamnionitis was found to be a risk factor for both cerebral palsy (RR 1.9; 95% CI 1.5-2.5) and cystic PVL (RR 2.6; 95% CI 1.7-3.9) (Wu and Colford 2000). It was recently shown that chorioamnionitis is not only associated with raised IL-1 beta and IL-6 but also with a decrease in mean and diastolic blood pressure after birth (Yanowitz et al. 2002). It is possible that these inflammation-induced haemodynamic disturbances make the newborn more susceptible to perinatal brain injury.Prevalence and clinical significance of PVL
The incidence of preterm WMI (white matter injury) varies among reports, partly due to definitions based on different imaging techniques (CUS or MRI) and their particular timelines and diagnostic roles. A systematic review concluded that the prevalence of preterm WMI, including both cystic and non-cystic, is 14.7% based on ultrasound diagnosis and 32.8% based on MRI (Romero-Guzman et al. 2017). The prevalence is 39.6% in infants born below 28 weeks GA; 27.4% below 32 weeks and 7.3% below 37 weeks. In very low birth weight infants (VLBW < 1500 g) there is a reported 20-50% overall incidence of some degree of white matter injury. There is a peak incidence at 27-28 weeks. Different groups used different classifications in different cohorts. Often only infants with extensive cystic lesions are included, and sometimes a unilateral porencephalic cyst, usually due haemorrhagic venous infarction, is lumped with cases of bilateral cystic PVL and then referred to as "echolucencies". Even in the interpretation of echolucencies interobserver agreement has been remarkably low. Most studies in the eighties reported an incidence of 3 to 10% for bilateral cystic leukomalacia (de Vries et al. 1988a, Fawer et al. 1985, Trounce, Rutter, & Levene 1986, Lemons et al. 2001). When non-cystic PVL ("flare") was included an incidence of 26% has been found (Trounce, Rutter, & Levene 1986). A survey in Japan, where many units have access to MRI, revealed an incidence of 5% (in babies less than 33 weeks gestation) using ultrasound, and 8% when MRI was used to make the diagnosis (Fujimoto et al. 1998). Leukomalacia, in its classic form, evolving from coagulation to later cystic necrosis has become less common (Hamrick 2004). In the unit in Utrecht the incidence in VLBWs has dropped from 4% in the early nineties to about 1% over the last couple of years. It has become clear that other forms of brain injury account for the majority of neurodevelopmental outcomes in VLBW infants. In a centre with a high prevalence of cPVL (17 % below 33 w of gestation) pre- and perinatal factors like indomethacin tocolysis and low Apgar scores have been found as risk factors (Murata et al. 2005). Hypocarbia may cause PVL (Okumura et al. 2001), but spontaneous hyperventilation with hypocarbia may also be one of the only clinical signs in preterms with extensive white matter injury, together with irritability (a clinical feature difficult to quantify). Although, in VLBW survivors, severe neurologic disability mainly defined by cerebral palsy (CP) and intellectual disability, is reported at a lower rate than previously, such outcomes are still observed in 5%–15% of VLBW survivors (Hamrick et al. 2004, Robertson et al. 2007, van Haastert et al. 2011, Moore et al. 2012, Cheong et al. 2017, Pierrat et al. 2017). Milder cognitive disabilities, learning difficulties and behavioral problems detected at (pre)school age are observed in 25%–50% of VLBW and represent a significant public health burden (Marlow et al. 2005, Larroque et al. 2008, Cheong et al. 2017). The focal necrotic lesions observed in cPVL classically precede a clinical picture of bilateral spastic CP (BSCP) and visual dysfunction, as well as moderate-to-severe cognitive impairment and learning disability (van Haastert et al. 2011). Epilepsy may affect around 20 % of spastic diplegic children. Some pronounced non-cystic forms may also lead to CP and cognitive dysfunction (Ment et al. 1999, Pappas et al. 2018). The relatively recent shift to less severe WMI explains a shift to milder forms of impairment predominantly in cognition and behaviour, often without motor involvement or CP (Larroque et al. 2008, Hutchinson et al. 2013, Schneider and Miller 2019). Specific deficits affecting former preterms with normal IQ influence executive functions (Johnson et al. 2016), motor coordination and visuomotor processing, attention and hyperactivity (Anderson et al. 2011), language and learning (Grunau et al. 2002, Johnson et al. 2016, Joseph et al. 2016), and mental health (autism spectrum disorder and internalizing symptoms) (Grunau et al. 2004). The growing brain adapts to the lesions with a myriad of options. There are plenty of indications that CUS and MRI do not pick up all changes in the brain of preterms with PVL. The focal lesions are destructive, but also interrupt developmental pathways, affecting areas proximal or distal, but distant to the focal lesions:- periventricular leukomalacia complex (Grunnet 1982, Paneth 1990 and 1994)
- late neuronal migration disorders / cortical dysplasia (Marin-Padilla 1997, 1999 and 2000)
- cytokines in cortical neurons in acute stage (Kadhim et al. 2003)
- APP in pyramidal neurons (Deguchi et al. 1999)
- reduced cortical grey matter volume in ex-VLBW on MR (Inder et al. 1999 and 2005)
- reduced glucose metabolism in rolandic area