Natural History of Brain Lesions in Extremely Preterm Infants Studied With Serial Magnetic Resonance Imaging From Birth and Neurodevelopmental Assessment

Leigh E. Dyet; Nigel Kennea; Serena J. Counsell; Elia F. Maalouf; Morenike Ajayi-Obe; Philip J. Duggan; Michael Harrison; Joanna M. Allsop; Joseph Hajnal; Amy H. Herlihy; Bridget Edwards; Sabrina Laroche; Frances M. Cowan; Mary A. Rutherford; A. David Edwards

doi:10.1542/peds.2005-1866

Abstract

OBJECTIVES. The aim was to survey the range of cerebral injury and abnormalities of cerebral development in infants born between 23 and 30 weeks’ gestation using serial MRI scans of the brain from birth, and to correlate those findings with neurodevelopmental outcome after 18 months corrected age.

METHODS. Between January 1997 and November 2000, consecutive infants born at <30 weeks’ gestational age underwent serial MRI brain scans from birth until term-equivalent age. Infants were monitored after 18 months of age, corrected for prematurity, with the Griffiths Mental Development Scales and neurologic assessment.

RESULTS. A total of 327 MRI scans were obtained from 119 surviving infants born at 23 to 30 weeks of gestation. Four infants had major destructive brain lesions, and tissue loss was seen at term for the 2 survivors. Fifty-one infants had early hemorrhage; 50% of infants with term scans after intraventricular hemorrhage had ventricular dilation. Twenty-six infants had punctate white matter lesions on early scans; these persisted for 33% of infants assessed at term. Early scans showed cerebellar hemorrhagic lesions for 8 infants and basal ganglia abnormalities for 17. At term, 53% of infants without previous hemorrhage had ventricular dilation and 80% of infants had diffuse excessive high signal intensity within the white matter on T2-weighted scans. Complete follow-up data were available for 66% of infants. Adverse outcomes were associated with major destructive lesions, diffuse excessive high signal intensity within the white matter, cerebellar hemorrhage, and ventricular dilation after intraventricular hemorrhage but not with punctate white matter lesions, hemorrhage, or ventricular dilation without intraventricular hemorrhage.

CONCLUSIONS. Diffuse white matter abnormalities and post–hemorrhagic ventricular dilation are common at term and seem to correlate with reduced developmental quotients. Early lesions, except for cerebellar hemorrhage and major destructive lesions, do not show clear relationships with outcomes.

Premature birth is associated with significant neurologic, cognitive, and behavioral problems.^1–4 Outcomes after major destructive lesions such as hemorrhagic parenchymal infarction (HPI) or cystic periventricular leukomalacia are well documented.^5,6 However, most infants with neurocognitive deficits do not exhibit these patterns of injury, which suggests that more subtle changes occur, perhaps altering cerebral maturation or causing more diffuse injury.

MRI can indicate the normal development of the preterm brain and can identify diffuse damage.^7–12 Most studies evaluated preterm infants at term-equivalent ages or acquired single images at earlier gestational ages (GAs).^10,12–15 Outcome studies concentrated on specific lesions and did not consider the full range or natural history of abnormalities.^13,15

Previously, we reported serial imaging findings for a small number of consecutively recruited infants born at GA of <30 weeks.¹⁶ The recruitment of this cohort continued, and we now report our findings from birth to term for a larger consecutive population of infants recruited over 3 years, the majority of whom have undergone neurodevelopmental follow-up monitoring after 18 months of age, corrected for prematurity. This allowed us to perform the first comprehensive MRI survey of the natural history of cerebral abnormalities in an unselected cohort of extremely preterm infants, together with preliminary correlation between imaging and developmental outcomes.

METHODS

Population

Infants born at GA of <30 weeks at the Hammersmith Hospital (London, United Kingdom), between January 1997 and November 2000, were eligible for entry into an unselected, consecutive, cohort study. GA was calculated from the date of the last menstrual period and was confirmed with early prenatal ultrasound scanning. The research ethics committee of Hammersmith Hospital approved the study, and informed parental consent was obtained as soon as possible after birth.

Birth weight and mode of delivery were recorded, and fetal membranes were examined for chorioamnionitis.¹⁷ Intrauterine growth restriction (IUGR), defined as weight of <10th percentile for GA and gender, calculated with Child Growth Foundation software,¹⁸ was recorded. Postnatal steroid use was documented.

MRI

A dedicated, neonatal, 1.0-T, MRI scanner (Phillips Medical Systems, Best, Netherlands) installed within the NICU was used for the majority of imaging; Phillips 1.5-T Eclipse and 1.0-T HPQ scanners were used for 9 and 3 infants at term, respectively. Infants were sedated with orally administered chloral hydrate (20–50 mg/kg), and ear protection was applied. A neonatologist trained in MRI supervised all imaging, with monitoring described previously.¹⁹ The MRI scans acquired included T1-weighted conventional spin echo (repetition time: 600 milliseconds; echo time: 20 milliseconds) and T2-weighted fast spin echo (repetition time: 3500 milliseconds; echo time: 208 milliseconds) scans in transverse and coronal planes.

The first MRI scan was obtained as soon as possible after birth. Infants underwent serial MRI scanning during admission to the NICU. Scan timing was dependent on clinical stability, but images were not acquired or used for clinical purposes. An image obtained at or soon after postmenstrual age of 36 weeks was defined as a term MRI scan.

Image Analysis

Qualitative image analysis was performed by 2 experienced observers; both observers assessed all scans, and any discrepancies were resolved through consensus. Scans were categorized as shown in Table 1 (Figs 1–3). ^16,20,21 Images were reported as normal if they had age-appropriate white matter details and myelination, normal-sized ventricles, and no diffuse or focal pathologic findings.

FIGURE 1

T1-weighted MRI scan in the transverse plane for an infant born at GA of 29 weeks 6 days and scanned at 3 days of age. There are high signal intensity punctate lesions bilaterally in the white matter of the centrum semiovale.

FIGURE 2

T2-weighted MRI scan in the transverse plane for an infant born at GA of 23 weeks 3 days and scanned at GA of 25 weeks. There is a cerebellar hemorrhagic lesion involving the whole right hemisphere.

FIGURE 3

Classification of DEHSI. T2-weighted MRI scans in the transverse plane at term-equivalent age show no DEHSI present within the white matter (A), DEHSI present but mainly periventricular (B), or severe DEHSI present, with increased signal intensity extending into subcortical white matter (C).

View this table:

TABLE 1

Classification of Developmental Features and Cerebral Pathologic Lesions on Brain MRI Scans

Follow-up Evaluation

Follow-up assessments were performed between 18 and 36 months of corrected age, by experienced developmental pediatricians who were blinded to the MRI findings. The Griffiths Mental Development Scales^22,23 provided an overall developmental quotient (DQ), with subscales assessing skill areas (locomotor, personal-social, hearing and speech, eye/hand coordination, and performance, as well as practical reasoning for children >2 years of age). The neurologic examination was based on our published protocol,²⁴ although, because children were examined beyond the age range for which this examination has been standardized, the scoring system was not used. Abnormalities of tone, posture, and movement consistent with cerebral palsy (CP) were recorded. Severity of CP was defined with the gross motor function classification.²⁵ Weight and head circumference were measured, and the SD scores (SDSs) with respect to the age-adjusted means were calculated with Child Growth Foundation software.¹⁸

Statistical Analyses

Data were analyzed with StatsDirect statistical software (StatsDirect, Sale, United Kingdom). Statistical results are reported for groups that were large enough to analyze, and the results were significant at a level of P ≤ .05. Exploratory analyses of outcomes were undertaken, and the quoted statistics are univariate. Where appropriate, infants with significant focal lesions were excluded from analysis, so that the effect of an abnormality on outcome could be examined in isolation. Parametric and nonparametric tests were used as appropriate.

RESULTS

Study Group

A total of 162 infants were born at GA of <30 weeks during the study period. Consent was obtained for 140 infants. There were no significant differences in birth weight or GA between recruited infants and those without consent; however, there were fewer male infants in the group without consent (P = .046). Twenty-one infants died after consent was obtained but before a scan could be performed. These infants were significantly younger (P < .001), with a lower birth weight (P = .01), than the remaining infants and were more likely to be from a multiple birth (P = .01) (Table 2).

View this table:

TABLE 2

Demographic Data for the Total Cohort

Therefore, 119 infants were entered into the study, with a median GA of 27 weeks 4 days (range: 23 weeks to 29 weeks 6 days) and median birth weight of 880 g (range: 370–1606 g). Thirty-two infants (26.9%) exhibited IUGR. Sixty-seven infants (56%) were male, and 33 (28%) were delivered vaginally. Placental histologic results were unavailable for 10 infants, but 51 infants (47%) with histologic findings had evidence of chorioamnionitis in the placental membranes. All except 2 infants received prenatal steroid therapy, and 17 infants (14%) received postnatal steroid therapy. A total of 106 (76%) of the infants with consent survived to discharge.

The median time after delivery for the first brain MRI scan was 2 days (interquartile range: 1–5 days). Infants underwent a median of 2 scans (range: 1–8 scans), and 24 had >3 scans. A total of 327 scans were performed. Eighty-seven surviving infants (82%) had a term MRI scan, at a median postmenstrual age of 40 weeks 6 days (range: 36 weeks to 53 weeks 2 days).

MRI Analysis

Fifty-one infants (44%) had normal initial MRI scans, with the exclusion of 2 infants who first underwent scanning near term-equivalent age (69 and 89 days). Infants with abnormal initial imaging results had lower GAs (P = .02). The abnormalities found on initial scans are summarized in Fig 4. Seven (8%) of the term MRI scans were reported as normal, and 5 of those infants had normal imaging results throughout the neonatal period. The remaining 2 infants both had widened extracerebral spaces (ECSs), with unilateral germinal layer hemorrhage (GLH) or unilateral intraventricular hemorrhage (IVH), on their initial scans. These abnormalities were no longer seen at term. There were no significant differences in GA or birth weight between infants with and without normal imaging results at term. None of the infants with normal term imaging results received postnatal steroid treatment. The abnormalities found on the term scans are shown in Fig 5. The majority of infants in the cohort had several abnormalities on each of their scans, with a median of 2 abnormalities per scan (range: 0–6 abnormalities).

FIGURE 4

Findings on the first MRI brain scans after birth (n = 119).

FIGURE 5

Findings on the term MRI brain scans (n = 87).

Major Destructive Parenchymal Lesions

Two extremely preterm infants, born at GAs of 24 weeks 6 days and 25 weeks, had cystic periventricular leukomalacia and experienced episodes of cardiovascular collapse during the neonatal period. Cysts were noted on a MRI scan obtained at GA of 31 weeks 2 days in the first case but were absent on a previous MRI scan obtained at 30 weeks 3 days. Cysts were first identified at term in the second case; however, the only previous scan was acquired at birth, which prevented accurate timing with MRI. Two infants experienced HPI. One infant, born at GA of 23 weeks, had bilateral HPI and died. The other infant, born at GA of 27 weeks 4 days, had unilateral HPI. Both infants also had evidence of IVH. The term scan of the unilateral HPI showed a small porencephalic cyst on the left and symmetrical myelination within the posterior limb of the internal capsule (Fig 6).

FIGURE 6

Imaging of an infant with unilateral left HPI born at GA of 27 weeks 4 days and scanned at term. A indicates a T2-weighted MRI scan in the transverse plane. There is a small left porencephalic cyst with a low-signal intensity rim, consistent with previous hemorrhage, in the centrum semiovale; B, inversion recovery MRI scan in the transverse plane. There is symmetrical high signal intensity within the posterior limb of the internal capsule.

Hemorrhage

GLH

GLH was present for 47 infants (39%) and was bilateral for 20. Hemorrhage occurred in the matrix over the caudate head for 33 infants (70%), in the matrix at the roof of the temporal horn for 5 infants (11%), and in both areas for the rest. There was no difference in GA at birth between infants with GLH and infants with no hemorrhage. GLH was present at term for 9 infants.

IVH

IVH occurred for 29 infants (24%) and was bilateral for 24. Twenty-four infants also had GLH. Infants with IVH had significantly lower GAs at birth, compared with infants with GLH or no hemorrhage (P = .0035). IVH was more common after vaginal delivery (P = .006), with a risk ratio of 2.54 (95% confidence interval: 1.37–4.57). Twenty-two infants with IVH underwent term imaging, and the hemorrhage had resolved in all cases. Eleven infants (50%) developed ventricular dilation.

Extracerebral Hemorrhage

Six infants (5%) had extracerebral hemorrhage present on their first scans. Hemorrhage occurred only in the posterior fossa for 4 infants, was hemispheric for 1 infant, and was found in both locations for 1 infant. Five of 6 infants also had evidence of GLH or IVH. There was no difference in mode of delivery between infants with extracerebral hemorrhage and the rest of the cohort. One infant developed hemispheric extracerebral hemorrhage later in the neonatal period, after an episode of cardiovascular collapse. Hemorrhage resolved by term in all cases.

Punctate White Matter Abnormalities

Twenty-six infants (22%) had punctate white matter abnormalities, and these were identified on the first MRI scans for 13 infants. The lesions were first noted on the term scans for 3 infants. The number of lesions identified per infant ranged from 1 to 9, with a median of 2. Five infants with punctate lesions on early scans still had lesions noted at term, although the lesions were less obvious and less numerous. There were no significant differences in GA or birth weight for infants with and without punctate white matter abnormalities. There were also no associations with mode of delivery, chorioamnionitis, or IUGR.

Diffuse Cerebral Abnormalities

Diffuse White Matter Abnormalities

Two infants with term MRI scans had severe ventricular dilation and had insufficient white matter volume to assess diffuse excessive high signal intensity (DEHSI) within the white matter on T2-weighted imaging scans. Sixty-eight infants (80%) had DEHSI present within their white matter; this was classified as severe in 10 cases. There was a lower incidence of DEHSI among infants with IUGR (risk ratio: 0.74; 95% confidence interval: 0.5–0.94; P = .018). There were no significant differences in GA, birth weight, mode of delivery, chorioamnionitis, or postnatal steroid exposure between infants with and without DEHSI. There was also no relationship with earlier lesions, such as hemorrhage or punctate white matter abnormalities, or with ventricular dilation or an enlarged ECS. There were no significant differences in the proportions of infants with each abnormality between the whole cohort and the subgroup of infants with DEHSI.

Dilation of the Lateral Ventricles

Thirty-six infants (30%) had ventricular dilation identified on their first scans, and these infants were born at significantly lower GAs (P < .001), with lower birth weights (P = .013). Sixteen of these infants also had IVH identified. The dilation was no longer present at term for 4 infants. Forty-seven infants (39%) had ventricular dilation on their final scans; for 25, it was associated with GLH or IVH. Two infants had marked post–hemorrhagic hydrocephalus and required ventriculoperitoneal shunts (Fig 7). No other infants required treatment for ventricular dilation. There was no difference in the development of ventricular dilation after isolated GLH, compared with IVH.

FIGURE 7

Serial T2-weighted MRI scans acquired in the transverse plane. These images show the development of posthemorrhagic ventricular dilation in an infant born at GA of 27 weeks 5 days. Multiple revisions of a ventriculoperitoneal shunt were required, exacerbating white matter destruction posteriorly on the right (E).

ECS

Fifteen infants (17%) had a widened ECS on their term scans. Another 45 infants had a widened ECS on earlier scans that was no longer present on their final scans. Resolution of a widened ECS was dependant on its location. A posteriorly widened ECS was likely to resolve by term and is usually considered a normal finding until that time.⁹ An anteriorly widened ECS was still present at term more frequently (P = .0016).

Other Lesions

Cerebellar Hemorrhagic Lesions

Cerebellar hemorrhagic lesions were present on the first scans for 8 infants (7%), who were born earlier (P = .007) and with lower birth weights (P = .03) than the rest of the cohort. Infants with cerebellar hemorrhagic lesions were more likely to have been born vaginally (risk ratio: 4.53; 95% confidence interval: 1.25–16.3; P = .037). Six of 8 hemorrhagic lesions were unilateral, and 4 were associated with IVH. Cerebellar atrophy developed for 4 of 6 infants with term scans.

Basal Ganglia and Thalamic Abnormalities

Seventeen infants (14%) had basal ganglia or thalamic abnormalities. These findings are summarized in Table 3. The caudate nucleus was affected most frequently. Hemorrhage within the caudate was usually associated with either GLH or IVH. All hemorrhagic and cystic lesions had resolved by term for infants who underwent subsequent imaging. Five infants had bilateral atrophy of the thalamus identified at term. None of these infants had previous abnormalities within the thalamus, and all of them had associated white matter atrophy, with marked ventricular dilation. There were no infants with basal ganglia or thalamic injuries consistent with an acute hypoxic ischemic injury.

View this table:

TABLE 3

Basal Ganglia and Thalamic Abnormalities

Congenital Abnormalities

One infant had bilateral posterior cortical clefts that were present initially and became more obvious on subsequent scans. This infant was a monozygotic twin whose sibling died in utero, and this abnormality was likely developmental. Subsequently, this infant developed ventricular dilation and a wide ECS. Another infant had an absent septum pellucidum and developed marked ventricular dilation, DEHSI within the white matter, and thalamic atrophy.

Neurodevelopmental Examination Results

Sixteen infants (13.4%) died before follow-up assessment, having had ≥1 MRI brain scan. These infants were born at lower GAs (P = .0001), had lower birth weights (P = .002), and had a higher incidence of chorioamnionitis (P = .05). Cerebellar hemorrhagic lesions also were more prevalent in these infants (P = .012) (Table 4). Sixty-eight surviving infants (66%) underwent neurodevelopmental follow-up assessment, at a median age of 23.93 months (range: 19.5–34.43 months) (Fig 8).

FIGURE 8

Age at follow-up evaluation (corrected for prematurity).

View this table:

TABLE 4

Differences Between Follow-up Groups

The results of the Griffiths Mental Development Scales assessment are given in Table 5. The mean overall DQ corrected for GA was 97 ± 18 and was correlated significantly with GA (r² = 0.079; P = .021) and birth weight (r² = 0.064; P = .039). There was no relationship between overall DQ and age at follow-up assessment. Four children (6%) had CP, 1 with hemiplegia, 2 with diplegia, and 1 with hemiplegia and diplegia. With the gross motor function classification, 1 infant was classified in level 2, being unable to walk independently at 2 years of age; the other infants were all classified in level 1. The mean SDS for head circumference at follow-up evaluation was −1.34 ± 1.6. This was not related significantly to GA, birth weight, or SDS of the head circumference at birth.

View this table:

TABLE 5

Distribution of Overall DQ and Subscale Quotients for the Griffiths Mental Development Scales Assessment at Between 18 and 36 Months of Corrected Age (Corrected and Not Corrected for GA at Birth)

Infants exposed to postnatal steroid therapy were more likely to undergo follow-up monitoring (P = .01); otherwise, there were no differences between infants with and without follow-up evaluation (Table 5). Thirty-five infants did not undergo follow-up evaluation between 18 and 36 months of corrected age, including 2 families (3 children) who declined consent for additional study, 1 child who was living abroad, and 3 children who did not attend their appointments. The remaining 28 infants were evaluated either before or after the defined time period. The percentage of infants in the whole cohort who underwent follow-up monitoring at any age was therefore 93%. There were 2 additional infants within the total cohort who had significant motor handicaps; 1 had severe nonambulatory CP, microcephaly, and associated deficits, and the other had diplegia. The rest of the infants were independently mobile.

Correlation of MRI and Follow-up Findings

To assess the effects of specific abnormalities on outcomes, the analyses were performed before and after exclusion of infants with other significant focal lesions. Eleven infants with HPI, basal ganglia and thalamic lesions, cerebellar hemorrhage, and congenital abnormalities were excluded. The overall DQ of the cohort after exclusion of these infants was 100 ± 16.2 and was not significantly different from that of the whole cohort.

Thirty-two infants (68%) with normal initial MRI scans and 35 infants (51%) with abnormal initial MRI scans underwent follow-up monitoring. There was no difference in the number of infants who died before follow-up evaluation between the 2 groups. Infants with abnormal initial imaging results had a higher DQ (normal initial scan: 93.5 ± 15.27; abnormal initial scan: 101.8 ± 18; P = .048). There was no relationship between overall DQs and numbers of abnormalities per scan. There were no significant differences in SDSs for weight or head circumference at follow-up evaluation between the 2 groups. Six infants with normal term scans underwent developmental follow-up assessment. These infants had a higher DQ than did infants with abnormal imaging results at term (112 ± 15 vs 96 ± 16.8; P = .034).