Encephalopathy of the Preterm—Clinical Aspects

Relevant Neuropathology

The hallmark of the encephalopathy of prematurity is cerebral WMI (see Chapter 18 ). To understand the value and the challenges of neuroimaging of the encephalopathy in the neonatal period, we should review briefly the neuropathology of WMI. WMI consists of two distinct components: focal necrosis , with loss of all cellular elements, dorsal and lateral to the lateral ventricles, and a more cell-specific diffuse injury involving pre-OLs in cerebral white matter and marked by astrogliosis and microgliosis. A spectrum of severity is recognized and in living infants is based on neuroimaging (principally MRI). Thus WMI can be categorized into three subtypes ( Table 20.5 ). The most severe form involves focal necroses that are macroscopic in size (i.e., relatively large, more than several millimeters) and evolve to tissue dissolution and cystic change over a period of weeks. This focal necrotic/cystic form is often simply termed cystic PVL or cystic WMI (see Fig. 20.1 ) and is readily detected by cranial ultrasonography or MRI (see later, Fig. 20.11 ). Cystic PVL/WMI is now uncommon and has an overall incidence in living VLBW infants of 3% or less. Recent data from the Vermont Oxford network on >40,000 preterm infants with a birthweight of <1500 g showed that the incidence of cystic WMI was static from 2014 to 2021 at 2.7% to 2.8%, with the highest rates of occurrence in the preterm infants weighing <750 g at 4% to 5% (see Fig. 20.6 ).

In the largest neuropathological series of preterm infants ( n = 41), approximately 40% exhibited PVL with focal necroses, but macroscopic necroses were observed in only 18% of these PVL cases (only 7% of the total of autopsied preterm infants). A moderate form of WMI involves focal necrotic lesions that are 1 to 2 mm in size and, upon tissue dissolution, evolve not to cysts but rather to focal glial scars, sometimes visible as punctate areas of increased signal intensity on T ₁ -weighted MRI ( Fig. 20.7 ). This form, often termed noncystic or punctate WMI , occurs in approximately 25% of living infants. Finally, the least severe form of PVL involves a focal necrotic component <1 mm in size; that is, microscopic and so small as to be invisible to neuroimaging. This mild form, like all forms of WMI, exhibits the diffuse gliosis neuropathologically as described in Chapter 18 . The degree of the diffuse abnormality generally correlates with the severity of WMI (see Table 20.5 ). The MRI correlate of the diffuse gliosis likely is the diffuse signal change described later. The mild form of PVL is likely present in a substantial minority of premature infants (see later). In the neuropathological series just noted, fully 82% of neuropathologically defined PVL cases (34% of the total series of autopsied premature infants) had focal necrotic lesions that were <1 mm and therefore likely below the detection of conventional clinical MRI scanners. Such cases may fail to be classified as WMI by MRI. In addition, the timing of the MRI may be crucial in recognition of WMI (see later), in that term equivalent MRI may fail to detect as much as 30% of all WMI in the preterm infant. All three subtypes of WMI include the diffuse component, characterized by astrogliosis/microgliosis and, after the initial pre-OL cell death, an excess of oligodendroglial progenitors. These oligodendroglial progenitors, however, fail to differentiate into myelin-producing cells (see Chapters 18 and 19 ). The severity of the diffuse component appears to parallel the severity of the PVL.

A form of cerebral WMI with no necroses and only the diffuse white matter gliosis must also be considered . It is noteworthy that in the group of 41 autopsied premature infants by Pierson and coworkers, approximately 40% had only diffuse gliosis, with no focal necroses. In strict terminology, these cases without any necroses do not lie within the neuropathological spectrum of “PVL.” These diffuse gliotic lesions only may be the least severe form of cerebral WMI, but because its MRI appearance is likely identical to the least severe form of PVL/WMI—that is, with microscopic necroses—distinction in vivo is not currently possible . Whether these infants with the diffuse gliotic lesions only consistently exhibit diffuse signal abnormality on MRI is unknown. Moreover, whether infants with the diffuse gliotic component without any necrotic component subsequently exhibit the neuronal/axonal dysmaturational effects of the encephalopathy of prematurity is not known. The particular potential importance of diffuse white matter gliosis without focal necroses relates not only to its high frequency in neuropathological studies of premature infants but also to the observations of Buser et al, who not only confirmed the high frequency but who also showed in cerebral white matter the characteristic excess of pre-OLs and their maturational arrest , as observed neuropathologically in the diffuse component of PVL. These findings suggest that such infants in vivo could develop the impairment of myelination and perhaps also the secondary dysmaturational effects on neuronal/axonal structures consequent to disrupted myelination (see Chapters 18 and 19 ). The important point in this context is that accurate identification of the diffuse white matter cellular abnormalities in vivo in the neonatal period currently is not clearly possible . Experimental data based on imaging at 11.7 T suggest potential for identifying and quantifying white matter gliosis (as well as microscopic necrosis). However, the imaging was carried out ex vivo , and the safety of imaging human preterm brain at 11.7 T has not been established. Finally, a recent series of studies in a fetal lamb model of WMI demonstrated the sensitivity of low-field MRI (0.35 T) on day 5 of life. They demonstrated distinct signal abnormality with noncystic WMI and also minor signal abnormalities with gliosis ( Fig. 20.8 ). They also noted that this may not be apparent after 2 weeks if MRI occurs later in the neonatal course.

Cranial Ultrasonography

Cranial ultrasonography is of great value in diagnosis of WMI ( Tables 20.6 to 20.8 ). The cysts of cystic PVL are readily visible by cranial ultrasonography (see Table 20.6 , Figs. 20.9 to 20.10 ).

They are located primarily just lateral to the atria of the lateral ventricles and extend anteriorly, and to some extent posteriorly, with increasing severity (see Fig. 20.7 ). They are sometimes only visible for a few weeks and may have fully resolved by term equivalent, but their presence is predictive of subsequent CP. Their “resolution” is likely due to development of a gliotic scar with collapse of the cyst (see Chapter 18 ). ( Fig. 20.12 ). As a result, their manifestation at term equivalent may be subtle, consisting of mild to moderate ventricular dilatation and an irregularly contoured ventricular wall.

The ultrasonographic imaging correlates of noncystic WMI are generally not obvious ( Tables 20.7 and 20.8 ). Some data suggest that cerebral white matter echodensity at least as echogenic as the choroid plexus and persisting for at least 7 days is significant. Further, the presence of white matter inhomogeneity or a “patchy” appearance on ultrasonography should also alert the clinician to potential white matter abnormalities ( Fig. 20.13 ). This patchy appearance can sometimes be hemorrhagic in origin, and a combination of diffusion- and susceptibility-weighted MRI can help identify a hemorrhagic component. Although PVL is principally a nonhemorrhagic lesion, occasionally secondary hemorrhage can be dramatic ( Fig. 20.14 ).

An extensive evaluative ultrasonography protocol was developed by the eurUS.brain group for documentation of abnormalities in the white matter of preterm infants ( Table 20.9 ), including levels of severity of WMI.

Magnetic Resonance Imaging

Compared with cranial ultrasonography, MRI is more sensitive for delineating the presence and severity of WMI in the preterm infant. WMI is classified on MRI with four key patterns —cystic-PVL/WMI, punctate white matter lesions, diffuse white matter abnormality, and diffuse excessive high signal intensity. The focal necrotic/cystic component of PVL , observed well on cranial ultrasound, is readily demonstrated by MRI as well (see Table 20.5 ). However, the much more common moderate (“noncystic”) forms of PVL often also are detected by MRI and generally not by cranial ultrasonography (see Table 20.7 ). Indeed, the high frequencies of noncystic cerebral white matter abnormality, manifested either as focal punctate white matter lesions or, even more commonly, as the mildest form of PVL (with microscopic necroses), as diffuse MRI white matter signal abnormality and subsequent volume loss, were unexpected until the application of MRI (see Table 20.5 ).

Accurate evaluation of the preterm brain requires optimal MRI acquisition (see Table 20.10 ) and expert neuroradiological interpretation.

Several structured evaluative tools have been developed to better delineate the nature of WMI and its relation to cerebral dysmaturation to enhance prediction of adverse neurodevelopmental outcome. These systems are outlined in Table 20.11 .

Concerning the fourth key pattern of WMI, diffuse excessive high signal intensity (DEHSI), the neuropathological correlates of this white matter signal abnormality are unclear. A reasonable speculation is that the abnormality reflects the diffuse gliotic (astrogliosis and microgliosis) component of WMI. It is seen best on T ₂ -weighted images at term equivalent age and is a common finding (see earlier). Though these white matter signal intensity changes are associated with increased water diffusion coefficient values on diffusion imaging, the qualitative identification of signal changes on T ₂ -weighted images is subjective. Further, recent studies have failed to identify any relationship between these signal changes and outcome at either 18 or 30 months of age. However, the absence of a later clinical correlate to DEHSI does not necessarily mean that a degree of WMI is not present. Indeed, in the neuropathological series referred to earlier, the focal necrotic component was <1 mm in most cases and likely below resolution of clinical MRI scanners. Moreover, as noted earlier, in 42% of the entire series of autopsied premature infants, diffuse gliosis without focal necrosis was present. Whether the subsequent neuronal/axonal abnormalities observed in the encephalopathy of prematurity and likely important for long-term outcome were present subsequently in infants with DEHSI is unknown. Moreover, the lack of later clinical abnormalities also could reflect the beneficial effects of plasticity.

DWI as a mode for the detection of acute WMI was first shown in 1999 to identify the ischemic portion of WMI before more confluent injury is identified on cranial ultrasound or conventional imaging. Since that observation, a small number of similar cases have been described. All of the infants in these studies with early DWI changes subsequently developed moderate-severe WMI, emphasizing

the potential utility of this technique in the early detection of WMI.

Using diffusion-based MRI, a series of studies in premature infants without severe WMI measured fractional anisotropy (FA) and used measures in the cerebral white matter as a means to detect mild WMI and their associated subsequent dysmaturational consequences (see Fig. 20.11 ). One large study examined 491 preterm infants (52% male) without focal necrotic WMI using diffusion-based MRI and identified the key marker of abnormality in the white matter with lower FA in the cerebral white matter, related to high radial diffusion rather than axial diffusion. The high radial diffusion is consistent with an impairment of pre-OL ensheathment. The findings suggest that impaired pre-OL maturation is the critical underlying pathology in mild cerebral WMI. The lower FA values were independently associated with increased number of days on ventilation, perhaps consistent with chronic hypoxia or related insults and with fetal growth restriction. The latter has been shown to be associated with a degree of hypoxia and in experimental studies to lead to delayed OL maturation. The white matter findings were also related to prolonged parenteral nutrition and suggest that impaired nutrition may lead to impaired pre-OL development (see later). Importantly, the abnormal FA values in the large study of Barnett et al were associated with impaired neurodevelopmental performance at age 20 months.

Other potential MRI correlates of diffuse cerebral WMI in the neonatal period include abnormalities of ¹ H spectroscopy, with both involving areas of white matter that may appear normal on conventional T ₁ - and T ₂ -weighted imaging.

Spastic Diplegia

Premature infants may exhibit a variety of major motor deficits related to PVL, severe IVH (see Chapter 28 ), and stroke (see Chapter 25 ), among other pathologies. In the context of this chapter on the encephalopathy of prematurity, the principal major motor deficit is spastic diplegia . This motor disturbance has as its central feature spastic paresis of the extremities with greater affection of lower more than upper limbs. The incidence of this motor deficit has declined markedly in the past 2 decades and is generally approximately 2% to 3%, although in the extremely preterm infant the incidence can be as high as 10%.

Two major lines of evidence indicate that the focal necroses of PVL result in spastic diplegia and its variants. First, the topography of the focal lesions includes the region of cerebral white matter traversed by descending fibers from motor cortex, and those subserving function of lower extremities are more likely to be affected by the periventricular locus of the necrosis (see Figs. 20.15 and 20.16 ). More severe lesions, with lateral extension into the centrum semiovale and corona radiata, would be expected to affect upper extremities and intellectual functions as well. Indeed, patients with spastic diplegia with significant involvement of upper extremities exhibit other manifestations of more severe cerebral disturbance, including intellectual deficits ( Table 20.13 ).

A second line of evidence linking spastic diplegia and focal PVL, related to the first, is that the ultrasonographic and MRI correlates of the focal necrotic component of PVL, especially cystic PVL, are associated with spastic diplegia. The severity and extent of injury, as manifested by cystic change or ventricular dilation or both, are correlated with more severe involvement of lower limbs, prominent involvement of upper limbs, and impairment of cognitive function. Further, in a study in which the location of cysts and/or signal abnormality on MRI was compared with motor outcome, cysts located more anteriorly, near the corticospinal tracts, were most strongly associated with motor deficits ( Fig. 20.17 ). The motor deficits in premature infants with mild spastic diplegia may disappear in the first several years of life, especially if the white matter lesions are noncystic by ultrasonography.

In addition to the evidence cited above indicating that spastic diplegia is caused by periventricular injury to descending corticospinal tracts, injury to thalamocortical sensory afferents may play a role in the motor deficits associated with PVL. Several studies employing diffusion tensor imaging with diffusion anisotropy measures (see Chapter 13 ) to evaluate white matter integrity found correlations between the degree of white matter abnormality in thalamocortical tracts and motor deficits in older children who had been born prematurely. Thus it has been hypothesized that disruption of sensorimotor loops involved in motor control plays an important role in the motor deficits associated with PVL. Nevertheless, in one report in which both corticospinal and thalamocortical tracts were assessed, correlation of motor dysfunction in spastic diplegia was greater for impairment of the former than the latter.

Other Motor Disturbances

At least one-half of preterm infants exhibit motor disturbances despite the absence of CP ( Table 20.14 ). These include difficulties in gross and/or fine motor skills, especially involving balance, visual motor integration, hand-eye coordination, manual dexterity, and related motor impairments. These disturbances are manifest in infancy as delayed motor milestones; at school age with impairment in activities ranging from running, ball skills, and drawing; and in adolescence and adulthood with clumsiness and dyscoordination. Although these deficits are minor in comparison to CP, they are much more common and may affect quality of life. These disorders are now codified as developmental coordination disorder (DCD). The American Psychiatric Association defined DCD as impairment in coordinated motor skills that significantly interferes with performance in everyday activities. Abilities assessed include manual dexterity, aiming, catching, and balance. Scores above the 15th percentile are considered normal, scores in the 6th to 15th percentiles are at risk, and scores in less than or equal to the 5th percentile are consistent with significant motor difficulty. Although motor delays are evident in early childhood, the diagnosis of DCD is often not made until school age. In a recent study, it was reported that all motor impairments for children born VP were worsening over time from 13% in 1991 to 1992, to 15% in 1997, and to 26% in 2005 using the same evaluative measures. Because the rates of CP were relatively stable over these time periods, as described previously, the deterioration in motor function resulted from an increase in non-CP motor impairments, such as DCD, in survivors born VP.

The clinicopathological correlates of DCD are likely multiple. Consistent with a relation to the WMI of the encephalopathy of prematurity, long-term follow-up studies of preterm infants with white matter signal abnormalities detected on conventional (T ₁ - and T ₂ -weighted) MRI at term equivalent were strongly predictive of motor impairment at ages 2 and 5 years. Further, diffusion abnormalities of white matter, including the internal capsule and corticospinal tracts, have been associated with similarly abnormal motor outcomes. That gray matter structures involved in the encephalopathy of prematurity (e.g., basal ganglia, thalamus) are also involved seems likely but has not been studied specifically. Recent work suggests importance also for cerebellar impairment. More data are needed on this important issue.

Cognitive Deficits

Cognitive performance involves a complex series of functions that can be parsed in many ways. In this section, we review overall IQ measures (often from younger populations), followed by outcome measures involving more specific cognitive functions, including language and executive function.

Concerning overall cognitive performance , fully 30% to 50% of VP survivors exhibit cognitive deficits, and as a group, preterm infants have lower developmental quotients than their term-born counterparts. It has been consistently reported that children born VP (<32 weeks gestation) perform approximately 12 points (0.8–0.9 SD) below term-born controls, and when restricted to children born <28 weeks, an average weighted 14-point deficit was reported. In infants born <27 weeks gestation, a greater IQ difference was observed relative to children born extremely preterm and at term, with a reduction in mean IQ score of 17 points in the extremely preterm cohort. For children born <26 weeks gestation, the mean IQ was reported to be 25 points below term-born counterparts (with greater within- and between-individual variability than controls). Taken together, this difference of approximately 1 SD in IQ between children born VP from term-born children has tangible consequences for academic and functional outcomes. For example, a 10-point reduction in IQ has been equated to a 29% increased risk of school dropout and a 10% increased risk of poverty in high-income countries like the United States.

Concerning the clinicopathological correlates of the cognitive deficits in premature infants, consistent with the concept of the encephalopathy of prematurity, WMI appears central. A relation of severe cognitive deficits to white matter involvement is apparent in infants with cystic WMI and spastic involvement of both upper and lower extremities, as described earlier (see Table 20.13 ). Notably, most infants with noncystic WMI exhibit only minor motor deficits but prominent cognitive disturbance. Involvement of cerebral white matter fibers subserving visual, auditory, somesthetic, and associative functions may be crucial in this context. Indeed, the peritrigonal region, a site of predilection of PVL, is a region containing a high concentration of interhemispheric callosal commissural fibers, intrahemispheric associative fibers, and ascending (thalamocortical) and descending (cortical to deep nuclear structures and to brainstem/cord) projection fibers. In an assessment of the relationship between the amount of white matter involved with PVL and cognitive outcome, the involvement of frontal white matter particularly predicted adverse cognitive and motor outcomes (see Fig. 20.6 ). However, this regional distinction is not dramatic and has not been emphasized by others.

The importance of injury to the white matter in relation to cognitive outcome was suggested by a series of studies that reported an association of impaired neurodevelopment in premature infants studied in later infancy, childhood, and adolescence in whom neuroimaging during the neonatal period showed white matter signal abnormality, ventricular dilation, and qualitative measures of diminished white matter volume. The largest well-characterized study to date ( n = 167) of VP infants (<30 weeks gestation) showed clearly the relationship between the frequency and severity of these neonatal MRI measures (obtained at term equivalent age) and subsequent cognitive motor deficits ( Table 20.15 ). The increasing severity of WMI was accompanied by abnormalities of gray matter such as increased size of the subarachnoid space and impaired gyral maturation. The likely relation of the latter dysmaturation to primary injury to white matter was detailed in Chapter 18 . Since the seminal work just described, there have been a large number of studies principally confirming the earlier observation. White matter abnormality can also be quantified with diffusion imaging, particularly through measures of diffusion anisotropy (see Chapter 13 ). For example, in a study of preterm children with follow-up at 2 years, cognitive scores were correlated with anisotropy values in the corpus callosum.

It is also apparent that the primary cerebral WMI is related to the secondary developmental disturbances of cerebral cortex, thalamus, and basal ganglia in premature infants (see Chapter 18 ). In a demonstration of the imaging disturbance consistent with WMI and its broader related disturbances, a study defined a common neonatal image phenotype that appeared to predict adverse neurodevelopmental outcome in children born preterm. Based on a study of 80 preterm infants (mean gestational age 29 weeks) by MRI at term equivalent age, the common image phenotype was characterized by diffuse WMI and tissue volume reduction in the periventricular white matter, corona radiata, central region of the centrum semiovale, as well as the thalamus and globus pallidum. The abnormal image phenotype was associated with reduced median Developmental Quotient (DQ) (DQ = 92) at 2 years compared with control infants (DQ = 112). In a similar study of adolescent children, Soria-Pastor et al also found that lower white matter volume was associated with IQ and processing speed.

Within other MRI studies it has been demonstrated that widespread alterations in WM bundles are associated with impairments in language, reading, or mathematic skills in VP infants. Among all anatomical sites, the corpus callosum has been established as a major interhemispheric track related to adverse cognitive outcomes, although other minor tracts have also been identified. Intellectual quotient and reading skills have been positively associated with anisotropy in the occipital corpus callosum at age seven. Reduced anisotropy in the corpus callosum has also been linked to impaired attention performance, specifically in orientation scores. Prematurity and WMI both appear to act as independent factors leading to the altered white matter microstructure that is related to adverse outcomes. Both slower maturation—that is, a slower increase of anisotropy measures during the preterm period—and a lower fractional anisotropy at term equivalent age in the posterior limb of the internal capsule and throughout the white matter have also been associated with worse cognitive outcomes at 18 to 24 months. In a different longitudinal cohort study, higher diffusivity in both the axial and radial directions in the cerebellum and the inferior occipital region were related to cognitive impairments. Finally, based on still more sophisticated diffusion measures, preterm teenagers with mathematic difficulties were found to have lower neurite density and lower anisotropy in the corona radiata, internal and external capsule, and left fronto-occipital fasciculus.

Because our current concept of the encephalopathy of prematurity involves primary WMI leading to secondary dysmaturation of gray matter structures (see Chapter 18 ), it would be predicted that imaging studies of preterm infants later in life would show evidence of such developmental disturbances of these neuronal-axonal structures. Gray matter tissue signal abnormalities on conventional MRI in preterm infants are less common in the neonatal period than are white matter signal abnormalities, perhaps consistent with the uncommon occurrence of acute neuronal injury. Gray matter abnormality is more typically manifest near term equivalent age as alterations of cortical folding and/or enlarged extracerebral spaces . As discussed in Chapter 7 , dramatic changes in cortical folding take place during the third trimester of gestation. This process can be disrupted in preterm infants. For reasons that have not been fully elucidated, these cortical folding abnormalities most commonly involve the temporal and inferior frontal areas as well as the cingulate sulcus. In a study of 167 very preterm infants, gray matter abnormalities were found in half of the infants and consisted of abnormal/immature cortical folding patterns and/or enlarged subarachnoid space. These abnormalities were associated with an increased risk of severe cognitive delay, psychomotor delay, and CP at age 2 years, but the association was less robust than that with white-matter abnormalities in the same study. The potential links between cerebral WMI and cortical folding abnormalities were discussed in Chapter 7 but may involve impairment of late migrating GABAergic neurons to superficial layers of cerebral cortex or of tension generated by underlying developing white matter.

Finally, the relation of white matter abnormality to gray matter disturbance was suggested by a large study of neurodevelopmental outcomes in preterm children in which correlations were sought between abnormalities on MRI at term equivalent age and outcome at age 7 years, and T ₁ - and T ₂ -weighted MRI studies were scored for signal abnormality and/or volume loss in white matter, cortex, deep nuclear gray matter, and cerebellum. Higher global, deep gray matter, cerebellar, as well as white matter abnormality scores were related to poorer IQ and motor function ( Fig. 20.18 ). Notably, abnormalities of the area of basal ganglia showed a strong relation to cognitive outcome. Though area reduction could be related to direct injury to deep nuclear gray matter, it is more likely that basal ganglia area reduction reflects a disruption of reciprocal connections between basal ganglia and cortex as a consequence of WMI or cortical dysmaturation or both (see Chapter 18 ).

Other studies of regional development of the cerebral cortex, basal ganglia, and cerebellum have also identified disturbances with VP birth associated with impairment in cognition. In particular, the Griffiths’ developmental quotient was shown to be positively correlated with microstructural developmental within the frontal, parietal, and occipital cortices during the neonatal period.

In addition to findings on conventional MRI, MR volumetry (see Chapter 13 ) has also been applied to evaluate preterm children and cognitive outcomes. A recent meta-analysis of 367 publications documented in the preterm infant that mean total brain volume at term equivalent age was 379 mL (SD 72 mL; based on n = 756), cerebellar volume 21 mL (6 mL; n = 791), cortical gray matter volume 140 mL (47 mL; n = 572), and unmyelinated white matter volume 195 mL (38 mL; n = 499). In comparison to term-born infants, preterm infants consistently demonstrated reductions in total brain volume and deep nuclear gray matter volume and increased cerebrospinal fluid. A number of studies have related reductions in regional brain volumes to neurodevelopmental outcome in the preterm infant. At short-term follow up (<2 years), neurodevelopmental disability was associated with volumetric reductions in cerebral white matter, cerebral cortex, deep nuclear gray matter, hippocampus, total cerebral tissue, and cerebellum.

Executive function is a broad term that refers to coordination of many interrelated processes and involves purposeful, goal-directed behavior that is instrumental in cognitive, behavioral, emotional, and social functions. Though there is not yet consensus on the exact components of executive function, it may be considered to include cognitive flexibility, goal setting, attentional control, and information processing. Executive dysfunction, then, is not a unitary disorder but reflects a range of impairment phenotypes.

One of the functional implications of executive dysfunction is poor academic performance (which may also be caused by low IQ), and a number of studies have shown poorer academic performance for preterm children on standardized testing. Deficits in academic performance may also be manifest as learning disability. In a study of 75 children born <800 g at birth who had a Verbal or Performance IQ ≥85, 65% of preterm children met criteria for learning disability, compared with 13% of the control group. Findings on subtests were consistent with the large body of evidence showing that academic underachievement arises, at least in part, from executive dysfunction.

Concerning clinicopathological correlations, the association between imaging findings and executive/academic dysfunction in preterm children has been evaluated in multiple studies. In the study noted earlier with outcomes at age 7 years (see Fig. 20.18 ), higher cerebral white matter, deep gray matter, and global abnormality scores were related to spelling and math computation. Studies focused on WMI have also noted associated impairments in executive functioning, verbal and visuospatial working memory, and learning. Finally, a strong correlation was shown between reduced FA in the left cingulum, the superior longitudinal fasciculi, and attention, orientation, or alerting performance. At adolescence, the same pattern between WM disruption of those identical bundles remained in relation with poor executive function. Overall the frequent concurrence of cerebral white matter abnormality with disturbances of gray matter structures is consistent with the central concept of the encephalopathy of prematurity.

Language delay is common in preterm children. The language and communication problems are often attributed to delayed development with the expectation of later catch-up. However, meta-analysis indicates that language deficits persist in preterm children with a severity comparable to that documented in other cognitive domains. Children born VP have been shown to perform below term-born peers across language domains of phonological awareness, semantics, grammar, discourse, and pragmatics at both 7 and 13 years of age. For semantics, a meta-analysis found an approximate 0.4 to 0.6 SD reduction in performance for children born VP/VLBW compared with controls. It has been suggested that complex language skills may be a more useful index of language functioning in preterm children, given the association of language skills with other higher-order skills and functional domains (e.g., motor development). An impact of language deficits in preterm children on quality of life, including friendship quality, reading skills, and overall academic achievement, has been suggested.

When considering the imaging correlates of language dysfunction, current concepts indicate that the anatomical substrate of language function involves widely distributed areas. Consistent with the central pathology of WMI, reduced corpus callosum volumes have been related to impaired verbal fluency. Functional MRI studies have been particularly useful in the investigation of language networks and alterations in the preterm infants. Predominantly resting state functional MRI studies have shown that language networks are detectable in preterm infants, despite the fact that the infants will not have any fluent speech for more than a year. The imaging correlates of language impairment in preterm children are likely multiple, but white matter disturbance is identified commonly. For example, WMI has been associated with impaired language skills. Similarly, functional resting state studies in preterm infants have revealed reduced covariance in both language and frontoparietal networks, findings suggesting that functional connections in the language networks in individuals born preterm are less complex. As preterm infants mature, the organization of the language networks appears to become further altered with functional connectivity in Wernicke’s region at 8 years of age demonstrating increased right inferior frontal gyrus and bilateral supramarginal gyri connectivity in preterm children compared with term controls. Altered language networks have also been shown in preterm children at 10 years of age, with increased connectivity between the frontal cortex (right and left) and the salience network, default mode network, and the central executive network. Several studies have correlated differences in functional connectivity in language regions to language outcomes. Adolescents who were born preterm exhibited less lateralization in the right hemisphere language homologs than did controls born at term, and this difference inversely correlated with receptive language. Similarly, lateralization of functional connectivity in the superior temporal lobe language areas inversely correlated with verbal comprehension abilities. For specific language pathways in adolescents and adults born preterm, the strengths of connections between the left cerebellum and bilateral inferior frontal gyrus and between Wernicke’s area and right supramarginal gyrus have correlated with receptive vocabulary and verbal comprehension scores, respectively. Cerebellar lesions in the absence of overt cerebral lesions have also been associated with abnormalities of receptive and expressive language. Recall, however, that most cerebral WMI in premature infants may be below the resolution of conventional MRI (see earlier).

In summary , language disturbance is common in the preterm infant, and the underlying imaging phenotype appears to vary in relation to the age of study. In the neonatal period, a relation of WMI and reduced functional connectivity within the language networks is typical. As the preterm infant matures, altered hemispheric connectivity with an imbalance of hemispheric connectivity and related fronto-temporal networks appear to underlie impaired language performance. Further longitudinal investigation of these language networks is required to define the impact of neonatal WMI on later brain development and functioning.

Behavioral Disturbances

Elevated rates of social-emotional deficits and psychiatric disorders have been recognized among children born preterm, with increasing numbers of reports detailing the “preterm behavioral phenotype” comprised of inattention, anxiety, and social communication deficits. These comorbid symptoms and the related disorders of attention deficit hyperactivity disorder (ADHD), anxiety, and autism spectrum disorder (ASD) are two to four times more common among preterm children. As with other neurodevelopmental impairments, the most immature preterm children are at greatest risk for these social-emotional impairments and psychiatric diagnoses. Further, studies examining the trajectory of these symptoms demonstrate their persistence into adolescence. Importantly, rates of these disorders remain elevated even after accounting for the increased frequency of other neurodevelopmental disabilities, including motor and intellectual impairments.

With specific regard to attention deficit hyperactivity disorder (ADHD) , this behavioral disturbance is relatively common among preterm children, with prevalence rates of 10% to 20%. In a recent meta-analysis, preterm children had a relative risk of ADHD of 2.64 compared with term-born children. This increased prevalence is greater in later childhood, with 11-year-old children born <26 weeks gestation having four times the odds of ADHD compared with term controls and 10 times the odds for the ADHD inattentive subtype. Similarly, inattentive ADHD was the most common type of ADHD diagnosis in an ELBW population at 8 years. The prevalence also varies with degree of prematurity and in one study was 17% for infants born weighing <750 g compared with 6% in infants weighing 750 to 1499 g at birth. ADHD has been associated with neonatal medical factors such as chronic lung disease, sepsis, and necrotizing enterocolitis, as well as intracranial hemorrhage and WMI. Some data suggest that reduction of perinatal systemic inflammation could reduce the incidence of ADHD. Notably, systemic inflammation is an important precipitant of cerebral WMI (see Chapter 17, Chapter 19 ).

Socialization Deficits—Autism Spectrum Disorders

Approximately 20% of preterm toddlers have abnormalities on early screening tests for ASD, such as the Modified Checklist for Autism in Toddlers, a finding suggesting that they are at high risk for ASD. Major motor, cognitive, visual, and hearing impairments may lead to false positive screening tests, and these impairments may account for more than half of the positive Modified Checklist for Autism in Toddlers screens in extremely low gestational age newborns. Nevertheless, even after those with such impairments are eliminated, 10% of children—nearly double the expected rate—screen positive. The perinatal clinical correlates of ASD and other social/behavioral issues in preterm infants are not yet fully delineated. However, the risk of ASD does increase with the degree of prematurity, either with or without intellectual disability. From a neuroimaging standpoint, ASD has been associated with cystic white matter lesions and cerebellar abnormalities during the perinatal period, although in studies with relatively small numbers of subjects.

Symptoms of ADHD, anxiety, and ASD comprising the preterm behavioral phenotype have also been linked to altered neonatal structural and functional connectivity in key brain regions. Although WMI is clearly a major pathogenic factor, additional recent evidence suggests that preterm birth may predispose children to higher rates of emotion dysregulation and social-emotional disorders due to stress experienced during the NICU hospitalization via changes in hypothalamic-pituitary-adrenal axis function and brain connectivity. For instance, alterations in connectivity of the glucocorticoid-rich amygdala, which has a prominent role in emotion processing, have been linked to NICU stress exposure in preterm infants.

Visual Deficits

A number of injuries can lead to visual impairment in preterm children, ranging from retinopathy of prematurity (ROP) through injury to white matter visual pathways, thalamus, and cerebral cortex. In this section, we focus primarily on visual impairment caused by brain injury. Much of the visual impairment found in preterm children can be classified as cortical or cerebral visual impairment (CVI), which is defined as bilateral impairment of visual acuity due to damage to cerebral visual areas. In a study of 105 preterm infants and 67 control infants, 24% of the preterm children met criteria for CVI, compared with 7% of controls (OR = 3.86; 95% CI, 1.40–10.70). Several studies have evaluated the association between brain injury and visual impairment. Perhaps not surprising, WMI has a strong association with visual impairment, because the primary area of injury includes the optic radiations (geniculocalcarine tracts) and visual association fibers. More severe WMI correlates with poorer future vision, and lesions of the peritrigonal white matter and optic radiations, as well as of the occipital cortex, have been associated with abnormality of visual acuity and function.

CVI, though defined on the basis of visual acuity, is also associated with impaired visual perception and visual-motor integration. Geldof et al proposed that these impairments are related to injury involving occipital-parietal-frontal neural circuitries. These circuitries are mediated by cerebral white matter regions particularly likely to be affected in cerebral WMI in premature infants. Consistent with the importance of cerebral WMI in recovery is the finding that only 42% of children with PVL and CVI show improvement in visual function, a proportion considerably lower than the 78% of a heterogeneous group of infants with primarily striate cortical injury.

A number of studies have evaluated the relationship between the quality of MRI diffusion parameters of the geniculocalcarine tract (usually fractional anisotropy; see Chapter 13 ) and visual function. In general, higher anisotropy values corresponded to better visual function. In one study of 142 preterm children and 32 control children who were evaluated at age 7 years, diffusion abnormalities were also found in association with WMI on conventional MRI and with ROP. The association of abnormalities of the geniculocalcarine tract with ROP suggests that cerebral white matter changes not only can cause visual impairment but can be caused by other abnormalities. In the case of ROP, it has been postulated that the abnormalities of the geniculocalcarine tract are a consequence of transsynaptic effects. Other studies concerning structure-function relationships include a volumetric study in which reduced occipital regional volumes were associated with impaired visual function.