Objectives. To investigate the effect of gestational age at birth on the frequency of ultrasound-detected brain lesions in infants born at <33 weeks of gestation and to investigate whether the relationship between neonatal cranial ultrasound diagnosis and neurodevelopmental outcome at 8 years of age was independent of gestational age.
Methods. Eight hundred forty-seven infants born at <33 weeks of gestation, admitted to a single tertiary referral center between 1983 and 1988, underwent serial neonatal cranial ultrasound. At 8 years of age neurodevelopmental outcome was assessed by structured neurologic examination, psychometric tests (Wechsler Intelligence Scale for Children), tests of visuomotor integration (Beery) and motor impairment (Henderson-Stott). Infants were subdivided into a group born at <28 weeks and a group born at between 28 and 32 weeks. Neurodevelopmental outcome was analyzed for each ultrasound diagnosis.
Results. Hemorrhagic lesions such as germinal matrix/intraventricular hemorrhage and hemorrhagic parenchymal infarction were more frequent in infants born at <28 weeks. There was no difference in the frequency of cystic periventricular leucomalacia between the 2 groups. When neurodevelopmental outcome for each ultrasound diagnosis was compared, no significant difference was found between the 2 gestational age groups.
Conclusion. In the gestational age range studied, adverse neurodevelopmental outcome depends primarily on the type of the intracranial lesion rather than on gestational age.
- very preterm infant
- neonatal cranial ultrasound
- neurodevelopmental outcome
- brain lesions
- gestational age
As a result of advances in perinatal management, survival for infants born before 32 weeks of gestation is continuing to increase. However, rates of neurodevelopmental impairments have remained substantially unchanged,1–3 and the long-term outcome of extremely immature infants remains a major concern. For example, school performance of infants weighing <1000 g is suboptimal when compared both with their term peers and with infants weighing between 1000 and 1500 g.4–7 Most neurodevelopmental impairments are likely to be the consequence of brain damage of perinatal origin, and the majority of these lesions can be identified in the neonatal period with the use of cranial ultrasound.8,9 As a result, ultrasound detection of brain lesions has proved very useful for providing early prognostic information about outcome in childhood.10–12 Several studies have reported the impact of individual lesions on neurodevelopmental outcome.11,13–16 At the same time, a number of studies have shown that gestational age and neurodevelopmental outcome are closely related, with the worst outcome in infants born at the lowest gestational ages (for review, see ref 3). Although implicit in other studies, no previous study has specifically addressed this issue by looking at neurodevelopmental outcome of preterm infants with the same lesion at different gestational ages.
The aims of this study were, first, to establish the frequency of ultrasound-detected brain lesions at different gestational ages in infants born before 33 weeks of gestation and, second, to investigate whether the relationship between these lesions and neurodevelopmental outcome at 8 years of age was independent of gestational age.
All infants <33 weeks of gestation born between 1983 and 1988 who were admitted within 1 week of birth to the Neonatal Intensive Care Unit of University College Hospital, London were enrolled into the study.
Neonatal Cranial Ultrasound
Cranial ultrasound investigations were performed daily for the first 5 days of life and then weekly until discharge. A mechanical sector scanner (Diasonics DS1, Siemens, Erlangen, Germany) was used between 1983 and 1986, with a 5-, 6-, or 7.5-MHz probe. After 1987, an Ultramark 4 Scanner (Siemens, Erlangen, Germany) was used with a 7.5-MHz probe for routine use and a 10-MHz probe to clarify suspicious findings. The images were stored on videotape. These tapes were reviewed by at least 2 experienced observers, and the diagnosis was agreed by consensus. The definitions for the ultrasound diagnoses are given in Appendix 1.
Surviving infants were enrolled in a prospective follow-up program. Neurodevelopmental outcome was assessed at 8 years of age by investigators who were blind to the ultrasound diagnoses. Details of these assessments have been published elsewhere12,13 and are summarized in Appendix 2. Based on the findings at these assessments, the children were assigned to 1 of 3 outcome categories. These outcome categories were 1) no neurodevelopmental impairment; 2) impairment but without disability (neuromotor impairment without functional consequences, high-tone hearing loss without requirement for aiding, or IQ 70–79); 3) impairment with disability (neuromotor impairment with functional consequences, sensory neural hearing loss requiring aiding, blind or registered partially sighted, IQ >2 standard deviations (SDs) below the test mean or requirement for formal educational provision). If a child’s disabilities were sufficiently severe to prevent completion of the tests, then the arbitrary test result was assigned at 3 SDs below the mean for the Wechsler Intelligence Scale for Children-Revised (WISC-R)/Wechsler Intelligence Scale for Children-III (WISC-III) and below the 5th centile for the Beery test, and a maximum error score was assigned for the Henderson-Stott Test of Motor Impairment, as discussed previously.13
To investigate the effect of gestation, the cohort was divided into two groups: group A comprised all infants born before 28 weeks of gestation and group B comprised all infants born between 28 and 32 weeks of gestation.
The data were analyzed with SPSS for Windows (version 7.5). Group differences were analyzed by using the χ2 test, Fisher Exact Test, and Student t test.
This study was part of a long-term follow-up program for very preterm infants for which approval was given by the Joint University College London/University College London Hospitals Committees on the Ethics of Human Research.
Eligible for prospective enrollment in this study were 852 infants born between 1983 and 1988. Figure 1 shows the flow of patients from enrollment to follow-up at 8 years of age. Five infants, who died on the first day of life before ultrasound scanning was performed, were excluded. The analysis looking at the relationship between gestational age and lesions detected by ultrasound was performed on the remaining 847 subjects. The number of long-term survivors in group A was 151 and 484 in group B. Table 1 gives perinatal details of the infants by gestational age groups. The infants who died had a significantly (P < .0001) lower birth weight (mean: 635 g, SD: 213 g) compared with the survivors (mean: 941 g, SD: 186 g) in group A. Similarily, in group B the infants who died had a significantly lower (P < .0001) birth weight (mean: 1201 g, SD: 353 g) when compared with the survivors (mean: 1442 g, SD: 319 g). Gestational age in both groups was significantly lower in the infants who died (group A: P < .001, mean gestational age of the nonsurvivors, 25 weeks, SD: 1.4 weeks; survivors, 26 weeks, SD: 0.9 weeks; group B: P < .0001, mean gestational age of the nonsurvivors, 29 weeks, SD: 1.3 weeks; survivors, 30.2 weeks, SD: 1.3 weeks). Apgar and first pH were significantly lower (P < .001) and base excess was significantly worse (P = .001) in the nonsurvivors in both gestation groups. In group A there was a significant difference between surviving and nonsurviving infants for mode of delivery and small for gestational age. There was no significant difference for these variables in group B.
At 8 years of age, the surviving infants (n = 635) were recalled for follow-up. Of these 582 (91%) children were assessed; 137 of the 151 survivors born before 28 weeks of gestation and 445 of the 484 survivors born between 28 and 32 weeks of gestation were available for follow-up.
The ultrasound diagnoses of those children who could not be assessed at 8 years of age (“lost to follow-up”) are given in Table 2.
Table 3 shows the neonatal brain lesions in all infants and in surviving infants in group A and group B, respectively. Comparing the individual ultrasound diagnoses for all infants in group A with those in group B, a significantly higher proportion (P < .0001) of infants had normal ultrasound in group B. In group A, a significantly higher proportion of children had germinal matrix/intraventricular hemorrhage (GMH/IVH; P < .0001) and hemorrhagic parenchymal infarction (HPI; P < .0001), ventricular dilatation (P = .02), and GMH/IVH combined with ventricular dilatation and periventricular flare (P = .02). There was no significant difference between the groups for hydrocephalus or flare. Notably, there was no difference in the proportion of infants with cystic periventricular leukomalacia (PVL) between group A and group B.
When lesions of surviving infants were analyzed, a normal scan was significantly more frequent in the more mature group (P < .001). HPI was significantly (P < .0001) more frequent in group A than in group B. This difference was also shown for posthemorrhagic hydrocephalus (P = .02) and flare (P < .02). There was no difference in the proportion of cystic PVL between the 2 gestational age groups. The less severe lesions such as GMH/IVH and a combination of GMH/IVH, ventricular dilatation, and flare were not different in their distribution between the 2 gestational age groups, although there was weak evidence that these lesions were more frequent in the more immature group (P = .09 for GMH/IVH, P = .06 for the combination of GMH/IVH with flare, and P = .06 for ventricular dilatation only).
We further analyzed the data to see if the timing of death had a significant impact on the distribution of lesions with gestation. Infants who died before 2 weeks of age were compared with those who died after 2 weeks of age. The purpose of this analysis was to take into account the fact that cystic PVL takes longer to develop. Table 4 shows the results of this analysis. In group A only, GMH/IVH (P = .004), flare (P = .05), ventricular dilatation (P = .04), hydrocephalus (P = .001), and the combination of GMH/IVH with flare and ventricular dilatation (P = .04) were significantly more common in infants who died before 2 weeks of age. Cystic PVL was significantly more frequent (P = .025) in those who died after 2 weeks of age. It has to be kept in mind, however, that there were only 2 infants with cystic PVL who died in either gestational age group. Therefore, this result must treated with caution. There was no significant difference for normal ultrasound scans or HPI. In contrast, in group B, flare was the only ultrasound lesion that was significantly different (P = .003) between early and late death.
Neonatal Brain Lesions and Neurodevelopmental Outcome of Long-Term Survivors at 8 Years of Age
Table 5 shows neurodevelopmental status of the long-term survivors for both gestational age groups A and B at 8 years of age. The WISC-R version was used for children born between 1983 and 1986 and WISC-III was used for children born in 1987 and 1988. When means of full-scale IQ before and after the change in the WISC version were compared, there was no significant difference in the mean values. Outcome measures have been expressed in terms of total impairment (which includes infants with nondisabling impairments and disabling impairments) and disabling impairments alone. Outcome at the age of 8 years was significantly worse for children born at <28 weeks of gestation (P = .008 for all impairments and P = .001 for disabling impairments).
Table 6 shows the neurodevelopmental outcome at the age of 8 years in the surviving infants by individual neonatal ultrasound diagnosis. When outcome was compared between the 2 gestational age groups within each ultrasound diagnosis, there was no statistically significant difference for total impairment or disabling impairment between the 2 groups.
In addition, in a separate analysis for the Beery and WISC tests there was no significant difference between group A and B within one ultrasound category.
We have demonstrated that the increased risk of morbidity in infants <28 weeks of gestation was associated with an excess of certain intracranial lesions, especially hemorrhagic lesions. However, when each lesion was considered separately, the relationship with overall neurodevelopmental outcome was independent of gestational age.
Intracranial Lesions and Gestational Age
Our data suggest that, in this cohort of very preterm infants, hemorrhagic lesions were particularly associated with birth at very low gestational age, whereas cystic periventricular leucomalacia was evenly distributed between 24 and 32 weeks of gestation. We suspected that this finding might be the result of the time taken for the lesions to become visible on cranial ultrasound scanning, and that the true prevalence of cystic PVL could, as a result, have been underestimated in those who died early. We found more cystic PVL in the very immature infants who died after 2 weeks, but this difference was not apparent in the more mature age group. However, because there were only 2 infants with cystic PVL who died in either gestation group, these results have to be interpreted with caution. Nevertheless, it seems likely that the time of death was not the only explanation for the difference in the frequency of hemorrhagic and ischemic lesions. In surviving infants, serial ultrasound examinations were continued at least up to term, so that it is unlikely that we failed to identify lesions, such as localized cystic PVL, which can develop beyond the first month of life.16,17 Ultrasound equipment has changed over time and, with the use of mechanical-sector scanning, detection of PVL has been become more reliable. Roth et al18 demonstrated that the change of equipment, however, has not had a significant effect on the accuracy of prediction of neurodevelopmental outcome. Furthermore, a change in the ability of detecting lesions would have affected both groups in the same way.
Other studies have also found that gestational age is an independent risk factor for the development of both hemorrhagic and ischemic lesions.19,20 DeVries et al20 found an increased incidence of extensive cystic PVL in the more mature infants, whereas pure hemorrhagic lesions were more common in the most immature infants. The changing frequency of hemorrhagic lesions with gestational age is most likely caused by physiologic changes and maturation of the vascular system, especially changes in the subependymal matrix.21 The equal distribution of ischemic lesions across the gestational age groups in our study is probably indicative of different etiologic processes in the development of the two pathologies. The mechanisms underlying the vulnerability of very preterm infants to periventricular white matter injury are a matter of continuing debate.21–23
Ultrasound Diagnosis, Gestational Age, and Neurodevelopmental Outcome at 8 Years
As in many previously published studies (for review, see Hack and Fanaroff3), we found that children in the lower gestational age group were more likely to develop neurodevelopmental impairments. In both groups the proportion of children with impairments seems high, but it must be borne in mind that, in the “all-impairments” group, children with any impairment were included (nondisabling and disabling) and that a great number of children in this group had only minor impairments.
Overall neurodevelopmental outcome at the age of 8 years was not different when both groups were compared within each ultrasound diagnosis.
Our results suggest that the increased risk of impairment in the group born before 28 weeks of gestation was primarily attributable to an increased incidence of hemorrhagic lesions in the most immature infants, and that gestational age in itself does not have an independent effect on neurodevelopmental outcome for the child with a specific lesion.
It could be speculated that injury to the very immature brain might lead to more potential for recovery and, hence, intracranial lesions in the very young infants might be associated with less severe long-term outcome. Our data, however, fail to support this. Neuropathologic studies have shown that injuries to the transient fetal subplate zone, which contains “waiting” axons and neurons, result in disruption of processes that are essential for normal brain organization.24,25 The subplate zone has its developmental peak between 22 and 34 weeks of gestation.21 Hence, intracranial lesions that affect the subplate zone in the 2 gestational age groups we studied may have similar adverse effects on brain organization and subsequent outcome.
In summary, the results of this study suggest that, in very preterm infants born between 24 and 32 weeks of gestation, the risk of adverse overall neurodevelopmental outcome depends primarily on the presence and type of intracranial lesion. Although there is some controversy,26 our data suggest that if further methods to reduce the incidence of hemorrhagic lesions in the most immature infants can be found, this may be expected to lead to improvement in their long-term outcome.
APPENDIX 1: DEFINITION OF INDIVIDUAL CRANIAL ULTRASOUND LESIONS27
Uncomplicated GMH/IVH. Hemorrhage into the germinal layer or lateral ventricle, including subependymal pseudocyst but not associated with periventricular flare, ventricular dilatation with cerebrospinal fluid, HPI, or loss of brain tissue.
Subependymal pseudocyst (SEPC). Cystic degeneration within a germinal-layer hemorrhage without cystic changes in the surrounding brain parenchyma.
Hemorrhagic parenchymal infarction (HPI). Markedly increased echodensities within the brain parenchyma, wedge shaped, and extending from the ventricular margin, in association with an ipsilateral GMH/IVH.
Periventricular flare (PVF). Abnormally increased echodensities in the periventricular white matter without an ipsilateral GMH/IVH.
Cystic PVL. Cystic lesion within the periventricular white matter, not preceded by HPI at the same site.
Ventricular dilatation. Dilatation of the lateral ventricle with cerebrospinal fluid such that the depth of the frontal horn immediately anterior to the thalamocaudate notch is >3 mm (97th centile).28
Posthemorrhagic hydrocephalus. Marked pressure-driven dilatation of a lateral ventricle with cerebrospinal fluid such that its width is 5 mm or more above the 97th centile for this.28
Loss of brain tissue. Loss of brain tissue from any cause, including cystic PVL, porencephalic cyst following HPI, irregular enlargement of the lateral ventricles, and generalized cerebral atrophy. Does not include isolated subependymal pseudocyst.
APPENDIX 2: FOLLOW-UP PROTOCOL AT 8 YEARS OF AGE
Structured neurologic examination29
Vision test (Snellen`s charts)30
Henderson-Stott Test of Motor Impairment (TOMI)31
Beery test of visual-motor integration32
(WISC-R was used for children born between 1983 and 1986 and WISC-III, for children born in 1987 and 1988.)
NON–SYMPTOM DRIVEN IMAGING
“The history of imaging since the discovery of x-rays has been one of an exponential rise in the volume and accuracy of information, acquired against a background of firstly increasing and then reducing invasiveness—and rising costs. This has allowed such investigations to move tentatively from being purely symptom driven to being non-symptom driven. It is small wonder that the flood of information from these investigations and our knowledge of how to deal with it may be several years out of step.
“But there is a more sinister danger. Because of medical setting and on-site medical expertise are not necessary for our new imaging techniques, the ties between the medical indications for a particular test and the motives for carrying it out are inevitably loosened. And thus the opportunity for financial gain moves into conflict with clinical need.”
Hayward R. VOMIT (victims of modern imaging technology): an acronym for our times. BMJ. 2003;326:1273
Submitted by Student
This study was supported by the Medical Research Council, United Kingdom; SPARKS (SPorts Aiding medical Research for KidS); and the Airways Trust.
We acknowledge the vital contributions of the following colleagues to this work: Phil Amess, Anthony Costello, David Edwards, Peter Hope, Pat Hamilton, Vincent Kirkbride, David McCormick, Judith Meeks, Juliet Penrice, Jan Townsend, the staff of the Neonatal Unit of University College Hospital, and the parents and children for their cooperation.
- ↵Tin W, Wariyar U, Hey E, for the Northern National Network. Changing prognosis for babies of less than 28 weeks’ gestation in the north of England between 1983 and 1994. BMJ.1997;314 :107– 111
- Emsley HC, Wardle SP, Sims DG, Chiswick ML, D’Souza SW. Increased survival and deteriorating developmental outcome in 23 to 25 week old gestation infants, 1990–4 compared with 1984–9. Arch Dis Child Fetal Neonatal Ed.1998;78 :F99– F104
- Stewart A, Thorburn R, Hope P, Goldsmith M, Lipscomb A, Reynolds E. Ultrasound appearance of the brain in very preterm infants and neurodevelopmental outcome at 18 months of age. Arch Dis Child.1983;58 :598– 604
- Whitaker A, Feldman J, Van Rossem R, et al. Neonatal cranial ultrasound abnormalities in low birth weight infants: relation to cognitive outcomes at six years of age. Pediatrics.1996;98 :719– 729
- ↵Pierrat V, Duquennoy C, vanHaastert IC, Ernst M, de Vries LS. Ultrasound diagnosis and neurodevelopmental outcome of localised and extensive cystic periventricular leucomalacia. Arch Dis Child Fetal Neonatal Ed.2001;84 :F151– F156
- ↵Hesser U, Katz-Salomon M, Mortensson W, Flodmark O, Forssberg H. Diagnosis of intracranial lesions in very-low-birthweight infants by ultrasound: incidence and association with potential risk factors. Acta Paediatr Scand.1997;419 :16– 26
- ↵Volpe J. Neurology of the Newborn. Philadelphia, PA: WB Saunders Company; 2001
- Kadhim H, Tabarki B, Verellen G, De Prez C, Rona AM, Sebire G. Inflammatory cytokines in the pathogenesis of periventricular leucomalacia. Neurology.2001;56 :1278– 1284
- ↵Back SA, Luo LL, Borenstein NS, Levine JM, Volpe J, Kinney HC. Late oligodendrocyte progenitors coincide with the developmental window of vulnerability for human perinatal white matter injury. J Neurosci.2001;21 ;1302– 1312
- ↵Kostovic I, Lukinovic N, Judas M, et al. Structural basis of the developmental plasticity in the human cerebral cortex: the role of the transient subplate zone. Metab Brain Dis.1989;17– 23
- ↵Levene MI, Starte DR. A longitudinal study of post-haemorrhagic ventricular dilatation in the newborn. Arch Dis Child.1981;56 :905– 910
- ↵Amiel-Tison C, Stewart A. Follow-up studies during the first years of life: a pervasive assessment of neurological function. Arch Dis Child.1989;64 :496– 502
- ↵Sheridan MD. Stycar Vision Test. Windsor: National Foundation for Educational Research; 1968
- ↵Henderson EH, Stott HD, Moyes FA. Test of Motor Impairment. London, United Kingdom: Harcourt Brace Jovanovic; 1987
- ↵Beery KE. Revised Administration, Scoring and Teaching Manual for the Developmental Test of Visual-Motor Integration. Toronto, Canada: Modern Curriculum Press; 1982
- ↵Wechsler D. Wechsler Intelligence Scales for Children. New York, NY: Psychological Corporation; 1974
- ↵Wechsler D. The Wechsler Intelligence Scale for Children-III. New York, NY: Psychological Corporation; 1991
- Copyright © 2003 by the American Academy of Pediatrics