OBJECTIVE. Our objective was to identify clinical predictors of progressive white matter injury.
METHODS. We evaluated 133 infants of <34 weeks of gestation at birth from 2 university hospitals. Infants underwent MRI twice, initially when in stable condition for transport and again at term-equivalent age or before transfer or discharge. Two neuroradiologists who were blinded to the clinical course graded MRI white matter injury severity by using a validated scale. Potential risk factors were extracted from medical charts.
RESULTS. Twelve neonates (9.0%) had progressive white matter injury. In the unadjusted analysis of 10 newborns without Candida meningoencephalitis, recurrent culture-positive postnatal infection and chronic lung disease were associated with progressive white matter injury. Exposure to multiple episodes of culture-positive infection significantly increased the risk of progressive white matter injury. Of the 11 neonates with >1 infection, 36.4% (4 infants) had progressive injury, compared with 5.0% (6 infants) of those with ≤1 infection. Of the 35 infants with chronic lung disease, 17.1% (6 infants) had progressive injury, compared with 4.3% (4 infants) of those without chronic lung disease. After adjustment for gestational age at birth, the association between infection and white matter injury persisted, whereas chronic lung disease was no longer a statistically significant risk factor.
CONCLUSIONS. Recurrent postnatal infection is an important risk factor for progressive white matter injury in premature infants. This is consistent with emerging evidence that white matter injury is attributable to oligodendrocyte precursor susceptibility to inflammation, hypoxia, and ischemia.
Noncystic white matter injury, which includes focal abnormalities, white matter volume loss, thinning of the corpus callosum, and diffuse, excessive, high signal intensity, is emerging as the leading central nervous system lesion detected with MRI among infants born prematurely.1–3 Studies demonstrated an association between this injury and adverse motor and cognitive neurodevelopmental outcomes.4,5 Human and animal research suggests that white matter injury may be attributable to oligodendrocyte precursor susceptibility to oxidative stress (reviewed in ref 3). Previous studies identified perinatal clinical risk factors, including prematurity, hypoxia, ischemia, and fetal-maternal infection.3,6
When evaluated with MRI in the early neonatal period, white matter injury is best detected as focal, noncystic, hyperintense areas on T1-weighted MRI scans, whereas volume loss and signal changes typically are later findings seen among near-term infants.4,7 In studies of correlations of MRI findings and pathologic findings for premature newborns, focal hyperintense lesions on T1-weighted MRI scans most closely corresponded to areas of gliosis.8 Additional studies showed that these lesions were associated with adverse neurologic outcomes.5 These hyperintense lesions on T1-weighted MRI scans normally stabilize or improve on follow-up scans obtained at term-equivalent age, which suggests that the injury is acquired early.5,7,9 In a minority of cases, however, the lesions progress, that is, they are larger or more numerous on repeat MRI scans. The pathophysiologic features of this progressive white matter injury are not known.
We analyzed data for infants enrolled in a cohort study examining MRI predictors of neurodevelopmental outcomes, to identify clinical risk factors for progressive white matter injury in premature infants. We hypothesized that postnatal exposures to inflammation or hypoxia might be risk factors. Identifying the cause of progressive white matter injury is important for advancing understanding of the pathophysiologic features of white matter injury and may lead to future therapies for intervention or prevention.
The study subjects were infants at <34 weeks of gestation at birth who were admitted to the ICUs at the University of California, San Francisco, Medical Center and the Children's and Women's Health Centre of British Columbia and were enrolled in a cohort study examining MRI predictors of neurodevelopmental outcomes. Exclusion criteria included clinical evidence of a congenital malformation, congenital infection (toxoplasmosis, rubella, cytomegalovirus, or herpes simplex infection) or syndromic diagnosis.
A total of 204 of 1186 eligible infants were enrolled between August 2000 and December 2006. One hundred thirty-three infants who underwent MRI twice according to the study protocol (initially when in stable condition for transport and again at term-equivalent age or before discharge from the hospital) were evaluated for progressive white matter injury. Reasons for exclusion from the present analysis were as follows: repeat MRI was not possible because of physician and/or scanner unavailability or early hospital discharge (N = 46), MRI scans were degraded by motion artifact or did not contain adequate sequences for interpretation of white mater injury (N = 13), the infant was withdrawn from the study (N = 5), the infant died before the second scan (N = 3), the maximal grade of white matter injury was present at the time of the first scan (N = 3), and a brain malformation was identified on MRI scans (N = 1). Medical care of the infants, including clinical investigations (eg, sepsis evaluation and lumbar puncture) and treatment (eg, antibiotics and ventilation), was at the discretion of the treating physician.
Clinical Data Collection
Clinical data were extracted prospectively from maternal and infant medical charts by a team of trained neonatal research nurses. Perinatal variables included gestational age at birth, birth weight, maternal characteristics, and type of gestation (single versus multiple). Gestational age was calculated on the basis of the last menstrual period or early ultrasound scans (<24 weeks). In cases in which the difference between the 2 methods exceeded 7 days, the ultrasound date was used. Postnatal variables included steroid exposure, patent ductus arteriosis, necrotizing enterocolitis, chronic lung disease, and days of positive pressure ventilation (intubation with ventilation and/or continuous positive airway pressure treatment). Chronic lung disease was defined if any 1 of the following 3 criteria were present: oxygen requirement (>21%) (1) for 28 days, (2) at postmenstrual age of 36 weeks, or (3) at discharge.
Infectious disease data were collected from microbiology reports and chart review. Postnatal infection was diagnosed when infants were treated for clinical symptoms and had positive culture results from blood, endotracheal tube, urine, skin lesion, or cerebrospinal fluid specimens. Central nervous system infection was diagnosed when cerebrospinal fluid findings suggested infection and the infant was treated for meningitis or encephalitis. Children were considered to have had >1 episode of infection when they were treated for a second organism or were treated for the same organism but after findings of negative culture results and discontinuation of the first course of antibiotics.
Serial MRI scans were obtained according to protocol, initially when the infant was in stable condition for transport (mean postmenstrual age: 31.9 weeks) and again at term-equivalent age or before hospital transfer or discharge (mean postmenstrual age: 37.2 weeks). MRI scans were acquired by using a 1.5-T scanner (General Electric Signa; GE Medical Systems, Milwaukee, WI or Siemens Avanto; Siemans Medical Solutions USA Inc, Malvern, PA) and a specialized, high-sensitivity, neonatal head coil built into the MRI-compatible incubator (General Electric; GE Medical Systems, Milwaukee, WI or Lammers Medical Technologies; LMT Lammers Medical Technology, Luebeck, Germany). MRI scans included axial, spin-echo, T2-weighted MRI scans and coronal, volumetric, 3-dimensional, spoiled gradient echo, T1-weighted MRI scans, acquired as reported previously.5 If necessary, infants were sedated according to institutional guidelines.
Two neuroradiologists who were blinded to the subjects' clinical condition evaluated the MRI scans to identify progressive white matter injury. Each scan was evaluated for focal areas of T1-weighted MRI hyperintensity, which were graded by using a validated 4-point scale with high interrater reliability (κ = 0.84).10 White matter was considered normal if there were no white matter abnormalities. The scan was graded as showing minimal white matter injury if there were ≤3 areas of T1-weighted MRI signal abnormality, each <2 mm. Injury was considered moderate if there were >3 areas of T1-weighted MRI signal abnormality or if those areas measured >2 mm but <5% of the hemisphere was involved; injury was considered severe if >5% of the hemisphere was affected. Infants with severe injury on the initial scan were not at risk for progressive disease according to our scoring system and therefore were excluded from the analysis. White matter injury was considered progressive if the second MRI scan showed a higher grade of injury (or any injury for infants whose first scans demonstrated normal findings). Each study also was evaluated for intraventricular and parenchymal hemorrhage, ventriculomegaly, and qualitative changes in the white matter (diffuse, excessive, high signal intensity).7 Newborns were diagnosed as having mild ventriculomegaly if the largest ventricular diameter (at the level of the glomus of the choroid plexus) measured 8 to 10 mm and moderate/severe ventriculomegaly if the diameter measured >10 mm.
Statistical analyses were performed by using Stata 9.2 software (Stata Corp, College Station, TX). Differences between clinical predictors were assessed by using Student's 2-tailed t test for continuous variables and the χ2 test or Fisher's exact test for categorical variables. Logistic regression analysis was used to evaluate the association between clinical risk factors and progressive white matter injury. The likelihood ratio test P value was used to determine statistical significance in the logistic regression models.
Clinical characteristics of the infants enrolled from the University of California, San Francisco (N = 111), and the University of British Columbia (N = 22) were similar (Table 1). The mean ± SD gestational age at birth was 28.4 ± 2.5 weeks, and 57% of the children were male.
Progressive White Matter Injury
For 12 infants (9.0%), the second (near-term) MRI scan demonstrated a higher grade of white matter injury, with an increased number of T1-weighted MRI hyperintense lesions, compared with the early MRI scan (Fig 1). There was no significant difference in the timing of or interval between the scans for infants with and without progressive white matter injury.
None of the infants in this cohort had diffuse, excessive, high signal intensity or cystic white matter injury. The neonatal courses and neurodevelopmental outcomes of the affected infants were heterogeneous (Table 2). Newborns with progressive white matter injury had high incidence rates of infection, blood pressure instability, and chronic lung disease. Two infants had Candida meningoencephalitis. Both children had initially absent or mild injury that was graded as severe on follow-up MRI scans. In both cases, the T1-weighted MRI hyperintense lesions were gadolinium-enhancing and were thought to represent Candida microabscesses. Because of the distinct pathophysiologic features, these 2 infants were excluded from further analysis. Of the remaining infants, 1 had moderate injury, which progressed to severe injury after placement of a ventriculoperitoneal shunt. Another had mild injury that progressed to severe injury after multiple infections and meningitis. A third had multiple infections and chronic lung disease; this child had moderate injury after normal findings on the first MRI scan. Finally, the 7 remaining infants were free of white matter injury on the first MRI scans but had mild injury on repeat MRI scans.
We used logistic regression analyses to evaluate the association between potential perinatal and postnatal clinical risk factors for progressive white matter injury (Table 3). Increasing gestational age at birth had a protective effect (odds ratio [OR]: 0.8; 95% confidence interval [CI]: 0.6–1.0). There were no statistically significant differences in infant gender or birth weight between the groups.
Chronic lung disease was associated with progressive white matter injury. Of infants with chronic lung disease, 17.1% (6 infants) had progressive injury, compared with 4.3% of those without chronic lung disease (4 infants; P = .02). Similarly, the infants with progressive white matter injury had a longer duration of positive pressure ventilation treatment (mean: 32 ± 8.9 days vs 15.3 ± 1.8 days; P = .03).
There was no statistically significant association between maternal chorioamnionitis, steroid exposure, patent ductus arteriosis, intraventricular hemorrhage, or ventriculomegaly and progressive white matter injury. No child with progressive white matter injury had definite necrotizing enterocolitis.
Recurrent (>1 episode), culture-positive infection was associated with increased risk of progressive white matter injury. Increased risk also was observed with all other categories of infection (including any infection, central nervous system infection, and infection between the MRI scans), although these associations were not statistically significant. Of infants with any culture-positive infection, 13.0% (6 infants) had progressive injury, whereas only 4.7% (4 infants) without infection were affected (OR: 3.2; 95% CI: 0.8–11.8). Exposure to multiple episodes of culture-positive infection significantly increased the risk of progressive white matter injury. Of infants with recurrent infection, 36.4% (4 infants) had progressive injury, compared with 5.0% (6 infants) of those with ≤1 infection (OR: 10.9; 95% CI: 2.5–47.6). Of the 10 newborns without Candida meningoencephalitis, 5 of the infants with progressive white matter injury underwent cerebrospinal fluid analysis. The association between central nervous system infection and progressive white matter injury did not reach statistical significance (OR: 6.6; 95% CI: 0.5–80.1). Of the affected infants, 2 had infections diagnosed in the interval between the scans. One of those infants had no white matter injury initially and minimal injury at the follow-up evaluation. The other infant had minimal injury initially and moderate injury at the follow-up evaluation. The association between infection diagnosed specifically in the interval between the 2 scans and progressive white matter injury was not statistically significant (OR: 2.1; 95% CI: 0.4–10.8).
The association between recurrent infection and progressive white matter injury persisted after adjustment for gestational age at birth (OR: 8.3; 95% CI: 1.5–45.3; P = .016). After adjustment for gestational age at birth, however, the association between chronic lung disease and progressive white matter injury was no longer statistically significant (OR: 3.7; 95% CI: 0.6–21.4; P = .13). The effect of recurrent infection was independent of chronic lung disease; with adjustment for both gestational age and presence of chronic lung disease, the effect of recurrent infection remained statistically significant, with a similar effect size (OR: 6.22; 95% CI: 1.0–37.0; P = .043).
In our cohort of premature infants admitted to 1 of 2 university hospital ICUs, progressive white matter injury was an uncommon problem that occurred most frequently in the setting of severe postnatal systemic illness. Of 133 premature infants who underwent MRI twice in the neonatal period, <10% had more numerous T1-hyperintense lesions on MRI scans obtained at near-term postmenstrual age, compared with scans obtained, on average, 5 to 6 weeks earlier. After adjustment for gestational age at birth, recurrent, culture-positive infection was associated with a more than eightfold increase in the odds of progressive white matter injury. Three of the affected infants had Gram-negative infections. Although infants with progressive white matter injury also had higher rates of chronic lung disease, this association was not statistically significant after adjustment for gestational age at birth.
Our data add to a growing body of evidence that suggests a link between infection and white matter injury in premature infants. Leviton and Gilles11 first made this association >30 years ago. Two meta-analyses by Wu and Colford12,13 showed that maternal chorioamnionitis is a risk factor for both cystic periventricular leukomalacia and cerebral palsy. Stoll et al14 showed that extremely low birth weight infants with culture-positive episodes of infection had impaired growth and increased rates of adverse neurodevelopmental outcomes, including cerebral palsy and lower Bayley Scales of Infant Development II Mental and Psychomotor Developmental Index scores, as well as vision and hearing impairments. Our study is the first to use MRI to show a direct in vivo relationship between recurrent infection and progressive postnatal white matter injury in human subjects.
The pathogenesis of progressive white matter injury in the setting of recurrent infection is unknown. For some infants, the association between recurrent infection and white matter injury in the postnatal period may be related to direct invasion by microorganisms, as presumed for the infants with Candida meningoencephalitis. In infants without central nervous system infection, however, the white matter injury may be attributable to damage to the preoligodendroglial cells, which are sensitive to free radicals, reactive oxygen species, and inflammatory cytokines that are generated during periods of ischemia and reperfusion (reviewed in ref 3). This hypothesis is supported by experimental animal models that suggest a relationship between cytokine responses, hypotension, and white matter injury15–17 and studies in premature human infants that show a relationship between inflammatory cytokine levels and white matter injury.18,19 Pressure-passive cerebral circulation in the setting of cardiorespiratory compromise also is important.20 Its role in the pathophysiologic processes of progressive white matter injury requires further elucidation. Because accurate blood pressure data were not collected for our cohort, we were not able to assess the impact of hypotension on progressive white matter injury. Recurrent postnatal infection was neither necessary nor sufficient to cause progressive white matter injury in this cohort, which suggests that there are unmeasured causes or mediating variables. With exclusion of the children with Candida meningoencephalitis, 6 (60%) of the 10 infants with progressive white matter injury did not have recurrent infection and 7 (63%) of the 11 infants with recurrent infection did not have progressive white matter injury. Finally, the association between infection and progressive white matter injury could be related to another, unknown and unmeasured, confounding variable.
There are several limitations to this study. First, the small size of the cohort and the rare occurrence of the outcome limited the statistical modeling. Many of the variables we examined had an estimated association with progressive white matter injury but did not reach statistical significance because of wide variability in the data. Although the sample size was small, the University of California, San Francisco/University of British Columbia cohort is the largest group of premature infants with data from serial MRI, which is technically difficult and requires specialized equipment and personnel. Second, we were unable to assess meaningfully the association between progressive white matter injury and neurodevelopmental outcomes. Only 7 affected infants were old enough for cognitive testing at the latest follow-up evaluation, and most of the children in the group with progressive white matter injury were only mildly affected and had normal neurodevelopmental outcomes. For these reasons, the clinical implications of progressive white matter injury remain unclear. However, previous studies suggesting an association between white matter injury (and especially severe injury) and adverse neurodevelopmental outcomes make further evaluation of the association between progressive white matter injury and outcomes worthwhile. Third, the presence of central nervous system infection (meningitis or encephalitis) was measured imprecisely, because the decision to perform a lumbar puncture was left to the discretion of the treating physician. This potential measurement error may explain why central nervous system infection was not significantly associated with outcomes. Fourth, there may be unmeasured mediating variables (such as hypoxia or hypotension) or confounding variables (such as medications administered) that would better explain the association between infection and white matter injury. Finally, our study design, in which infants underwent MRI initially when in stable condition for transport and again before hospital discharge, adds variability to the outcome measures. Because little is known about the timing of progression or regression of white matter injury in these children, it is possible that we missed some infants with progressive disease. Such a measurement error would have the effect of reducing our ability to detect any differences between the groups.
This is the first study using MRI in premature infants to show that recurrent postnatal infection is an important risk factor for progressive white matter injury. Larger studies that include more infants with moderate or severe white matter injury are needed to assess its impact on neurodevelopmental outcomes. If these results are confirmed in larger studies, then infants with recurrent infection may be an ideal group for investigation of novel neuroprotective agents, such as those targeting inflammatory pathways.
This publication was made possible by grant UL RR024131-01 from the National Center for Research Resources, a component of the National Institutes of Health and National Institutes of Health Roadmap for Medical Research. This research also was supported by the National Institutes of Health (grant NS40117) and the Canadian Institutes of Health Research (operating grant 151135 CHI). Dr Miller is a Canadian Institutes of Health Research Clinician Scientist and Michael Smith Foundation for Health Research Scholar. Dr Glass is supported by a National Institute of Neurological Disorders and Stroke Neurological Sciences Academic Development Award (grant NS01692). Dr Chau is supported by Bourse McLaughlin de l'Université Laval and the Fondation pour la Recherche sur les Maladies Infantiles.
We thank the children and their families, Kerry Moore for administrative support, and the nurses of the Neonatal Clinical Research Center for infant screening and transport.
- Accepted November 20, 2007.
- Address correspondence to Steven P. Miller, MDCM, FRCPC, Division of Neurology, BC Children's Hospital, University of British Columbia, K3-180, 4480 Oak St, Vancouver, BC, V6H 3V4 Canada. E-mail:
The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health.
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Noncystic white matter injury is the leading central nervous system lesion seen in infants born prematurely. Focal lesions observed early in the postnatal course typically stabilize or resolve by term-equivalent age. Hypoxia, ischemia, and inflammation are risk factors for injury.
What This Study Adds
Progressive white matter injury is uncommon and is seen in the setting of critical illness. Recurrent postnatal infection is a risk factor for progressive white matter injury.
- Inder TE, Anderson NJ, Spencer C, Wells S, Volpe JJ. White matter injury in the premature infant: a comparison between serial cranial sonographic and MR findings at term. AJNR Am J Neuroradiol. 2003;24 (5):805– 809
- ↵Dyet LE, Kennea N, Counsell SJ, et al. Natural history of brain lesions in extremely preterm infants studied with serial magnetic resonance imaging from birth and neurodevelopmental assessment. Pediatrics. 2006;118 (2):536– 548
- Dalitz P, Harding R, Rees SM, Cock ML. Prolonged reductions in placental blood flow and cerebral oxygen delivery in preterm fetal sheep exposed to endotoxin: possible factors in white matter injury after acute infection. J Soc Gynecol Invest. 2003;10 (5):283– 290
- ↵Duncan JR, Cock ML, Suzuki K, Scheerlinck JP, Harding R, Rees SM. Chronic endotoxin exposure causes brain injury in the ovine fetus in the absence of hypoxemia. J Soc Gynecol Invest. 2006;13 (2):87– 96
- Copyright © 2008 by the American Academy of Pediatrics