 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
a Division of Neonatal-Perinatal Medicine, Department of Pediatrics
b Center for Clinical Research and Evidence-Based Medicine
c Department of Neuroradiology, University of Texas Medical School at Houston, Houston, Texas
d Department of Neuroradiology, Memorial Hermann Hospital, Houston, Texas
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
1000 g) to their total and regional brain volumes, as measured by volumetric MRI performed at term-equivalent age. METHODS. Among 53 extremely low birth weight infants discharged between June 1 and December 31, 2003, 41 had high-quality MRI studies; 30 of those infants had not received postnatal steroid treatment and 11 had received dexamethasone, all after postnatal age of 28 days, for a mean duration of 6.8 days and a mean cumulative dose of 2.8 mg/kg. Anatomic brain MRI scans obtained at 39.5 weeks (mean) postmenstrual age were segmented by using semiautomated and manual, pretested, scoring algorithms to generate three-dimensional cerebral component volumes. Volumes were adjusted according to postmenstrual age at MRI.
RESULTS. After controlling for postmenstrual age at MRI, we observed a 10.2% smaller total cerebral tissue volume in the dexamethasone-treated group, compared with the untreated group. Cortical tissue volume was 8.7% smaller in the treated infants, compared with untreated infants. Regional volume analysis revealed a 20.6% smaller cerebellum and a 19.9% reduction in subcortical gray matter in the dexamethasone-treated infants, compared with untreated infants. In a series of regression analyses, the reductions in total cerebral tissue, subcortical gray matter, and cerebellar volumes associated with dexamethasone administration remained significant after controlling not only for postmenstrual age but also for bronchopulmonary dysplasia and birth weight.
CONCLUSIONS. We identified smaller total and regional cerebral tissue volumes in extremely low birth weight infants treated with relatively conservative regimens of dexamethasone. These volume deficits may be the structural antecedents of neuromotor and cognitive abnormalities reported after postnatal dexamethasone treatment.
Key Words: extremely premature infants • steroids • bronchopulmonary dysplasia • brain volumes • brain imaging
Abbreviations: BW—birth weight • BPD—bronchopulmonary dysplasia • NICHD—National Institute of Child Health and Human Development • CI—confidence interval • ELBW—extremely low birth weight • CSF—cerebrospinal fluid • GA—gestational age • PMA—postmenstrual age
Neurosensory disabilities occur in up to 50% of extremely immature or extremely low birth weight (ELBW) (birth weight [BW] of 1000 g) survivors.1,2 Complications such as bronchopulmonary dysplasia (BPD) are often associated with such neurosensory abnormalities.3 Unfortunately, reduction in BPD rates with postnatal administration of corticosteroids increases rather than reduces the risk of neurosensory disabilities, limiting corticosteroid utility except perhaps for infants with the most severe lung disease.4–7 Corticosteroid use has not been eliminated from the nursery; the Vermont Oxford Network reported that 23% of 14 321 ELBW infants received postnatal corticosteroid treatment in 2002.8 More recently, the Neonatal Research Network database of the National Institute of Child Health and Human Development (NICHD) indicated that, in 2005, 13% of those who survived for >12 hours received postnatal corticosteroid treatment (NICHD Neonatal Research Network, unpublished data, 2006). Use of dexamethasone, the corticosteroid that has been most often administered to prevent or to treat BPD, has also been associated with poor growth and cognitive impairments at school age.5 The structural antecedents for these functional deficits have not been well characterized.
Preliminary investigations suggested that lesions detected with advanced quantitative MRI technologies, such as volumetric MRI, before nursery discharge correlated well with neurodevelopmental deficits assessed at 1 to 2 years of age.9–11 Volumetric MRI was also used to characterize the neuroanatomic effects of prolonged steroid therapy at term-corrected age in one reported study.12 In that report, Murphy et al12 noted dramatic volume reductions in cerebral cortical gray matter (35%) and total brain tissue volume (30%) in 7 dexamethasone-treated preterm infants, compared with 11 untreated infants. A difficulty in conducting such studies is that the results may be inadvertently biased with the use of convenience samples, unmasked evaluations, and incomplete adjustment for potential confounders, including BPD and timing of brain MRI studies. Moreover, the relevance of the findings of Murphy et al12 is limited by shifts in current practice prompted by recommended guidelines to restrict postnatal corticosteroid use.13
An assessment of current ELBW infants is needed to identify whether relatively conservative dexamethasone regimens affect brain development in high-risk ELBW infants and, if so, to identify the most vulnerable regions that may serve as the neural substrates of permanent neurosensory disabilities. We hypothesized that postnatal systemic dexamethasone therapy would be associated with reduced cortical tissue and total brain tissue volumes, after adjustment for timing of MRI, presence of BPD, and other potential confounders. We also hypothesized that reductions in total and regional volumes, if any, would be considerably less than previously described by Murphy et al.12
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Among the 53 ELBW infants discharged from our NICU between June 1, 2003, and December 31, 2003, 12 were excluded because of motion artifacts/poor-quality MRI scans (9 infants) or incomplete coronal sequences (3 infants). The remaining 41 infants formed our study cohort (none had a congenital anomaly of the central nervous system). Of these 41 infants, 11 (27%) had been treated with systemic dexamethasone therapy (8 because of evolving BPD and 3 because of presumed airway edema); 30 (73%) had not received postnatal corticosteroid treatment. Relevant prenatal, perinatal, and postnatal data were collected prospectively by a research nurse. The best obstetric estimate of gestational age (GA) was used to indicate pregnancy length, on the basis of last menstrual period date or first-trimester ultrasonographic results. The pediatric estimate of GA16 was used for 5 infants for whom obstetric estimates were unavailable. The physiologic definition of BPD17 was assigned at postmenstrual age (PMA) of 36 ± 1 weeks and included 2 subpopulations of BPD infants, (1) neonates receiving positive pressure support or >30% supplemental oxygen, who were assigned the outcome of BPD and were not tested further; and (2) infants receiving 30% oxygen or >30% effective oxygen with saturations of >96% who failed a room-air challenge (saturation of <90% during weaning off oxygen). Abnormal cranial ultrasound results were defined as echodense intraparenchymal lesions, periventricular leukomalacia, porencephalic cysts, or ventriculomegaly, with or without intraventricular hemorrhage, in the 10- to 14-day cranial ultrasound studies.18 This cluster of lesions includes periventricular and intraventricular hemorrhage of grades 3 and 4.19
Training and Experience of Investigators
An extensive training period preceded the study for 2 of the authors (Drs Parikh and Lasky), to develop software expertise, to identify and to define neuroanatomic boundaries for smaller cerebral structures, to standardize image quality, and to pretest a robust procedure of manual and semiautomated scoring for all of the preselected structures. One of the authors (Dr Lasky) was already highly experienced in performing volumetric MRI studies with nonhuman primates.20 Primary anatomic references included the neuroanatomy atlas by Haines21 and various Internet-accessible human atlases.22,23
MRI Acquisition
All images were obtained by using GE-LX (General Electric, Milwaukee, WI) or GE-Horizon (General Electric) 1.5-T scanners. Coronal T2-weighted MRI brain scans were acquired by using the following sequence parameters: repetition time, 7500 ms; echo time, 175 ms; matrix size, 512 x 512; number of excitations, 2; field of view, 18 cm; slice thickness, 4 mm; gap, 0.5 mm. All MRI scans were completed at Memorial Hermann Children's Hospital and were digitally transferred to a workstation for analysis.
MRI Analysis
Coronal T2-weighted MRI scans were imported into Analyze 4.0 software (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN) for manual and semiautomated whole-brain segmentation and volume rendering (Fig 1A). Tissue segmentation was performed by manually segmenting high-intensity cerebrospinal fluid (CSF) ventricular spaces (right and left lateral ventricles, third ventricle, and fourth ventricle) on the basis of pixel intensity and known spatial neuroanatomic boundaries, after equating image intensity with predefined algorithms. Smaller cortical and subcortical structures were next segmented manually in a similar manner. Scored structures included the corpus callosum, midbrain, pons, medulla, and left and right analogs of the caudate, hippocampus, amygdala, and subcortical gray matter (Fig 1B). Subcortical gray matter was defined as structures with lower signal intensity (consistent with gray matter tissue) medial to the external capsule, lateral to the midline, inferior to the anterior horn of the lateral ventricle, and superior to the third ventricle. These included the thalamus, hypothalamus, globus pallidus, putamen, and claustrum. The MRI scans did not readily differentiate the internal capsule from surrounding gray matter tissue; therefore, subcortical gray matter volumes also included this white matter structure. Larger structures, including cerebellum, cortical gray matter, cortical white matter, and extraaxial CSF, were identified and labeled last, on the basis of signal intensity and spatial location. The total brain tissue volume (cerebrum plus cerebellum) was defined as the total brain volume minus all CSF spaces. For a given structure/region, the two-dimensional segmented area was multiplied by the thickness of the slice, and the resulting three-dimensional volume was summed for each slice containing the segmented structure, to yield the absolute volume (in cubic millimeters). All images were scored by one of the authors (Dr Parikh), who was masked with respect to clinical history, including dexamethasone therapy.
|
Statistical Analyses
Data were screened for accuracy and examined for normality to conform to the assumptions of the parametric statistics used. Covariates known or suspected to affect the primary outcome were compared between the dexamethasone-treated and untreated infant groups. Fisher's exact test was used to determine differences in categorical baseline variables, and a 2-tailed t test was used to determine differences in continuous variables. A P value of <.05 was considered to be statistically significant.
All brain volumes were adjusted for PMA at the time of brain MRI. Because infants are growing rapidly around the time of discharge and age differences at the time of MRI can affect brain component volumes dramatically,24 adjustment for PMA at MRI was critical, so as not to obscure associations between dexamethasone treatment and brain volumes. Multivariate linear regression techniques were used to evaluate the relationship of postnatal dexamethasone administration to component and total brain volumes (all continuous outcomes) and to adjust for potential confounders. Multivariate linear regression analysis was also used to assess the relationship between dexamethasone doses and brain volumes.
To avoid bias in selecting the regression models, we used the following predefined criteria to decide which covariates could be included: group differences at baseline at P < .10, an association with the primary outcome at P < .25, a >20% change in the postnatal steroid regression coefficient when the variable was used in the model, or, for some variables (eg, BW, BPD, and abnormal cranial ultrasound results), a priori evidence25 that the variable might affect cerebral tissue volumes. Because of the limited number of infants studied, the individual regression equations included no more than 4 predictor variables, 2 of which were PMA at MRI and dexamethasone treatment. The third was a potential confounder tested individually with PMA at MRI and dexamethasone in the regression equation. The last variable (BPD) was added on clinical grounds.
On the basis of the predefined criteria listed above, the following covariates were considered for inclusion in the model: GA, BW, presence of BPD, presence of cranial ultrasound abnormalities, gender, and prenatal steroid use. With our limited sample size, none of these variables was associated significantly with brain volumes. BW was associated marginally with cortical tissue volume (P = .10). No association was observed between cortical tissue volume and GA (P = .66), prenatal steroid use (P = .29), BPD (P = .55), or abnormal cranial ultrasound scans (P = .58) when these variables were included in the model one at a time with dexamethasone and PMA at MRI. The same regression equations with total brain tissue volumes as the dependent variable resulted in similar P values (results not shown). When included with PMA and dexamethasone in the model, BW was associated marginally with the 2 main dependent variables, cortical tissue volume (P = .10) and total brain tissue volume (P = .12). On the basis of this association and on clinical grounds, BW and BPD were included in the final model (in addition to PMA at MRI and dexamethasone treatment), to control for potential confounding.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Figure 2A depicts dexamethasone-associated cerebral volume effects, adjusted only for PMA at MRI, in 5 affected regions. Total brain tissue volume was 10.2% (95% CI: 1.0%–19.3%) smaller in the dexamethasone-treated infants, compared with untreated infants (P = .03). Cortical tissue volume was 8.7% (95% CI: –0.5% to 18.0%; P = .06) smaller. Cerebellum and subcortical gray matter volumes were significantly lower, by 20.6% (95% CI: 6.5%–34.8%) and 19.9% (95% CI: 8.2%–31.6%), respectively. Although cortical gray matter (not shown) was 11.1% (95% CI: –0.5% to 22.8%) smaller in dexamethasone-treated infants, this did not reach statistical significance (P = .06); total brain gray matter volume was reduced significantly, by 11.7% (95% CI: 0.7%–22.6%; P = .04).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The gray matter volume reductions associated with steroid use were severalfold greater in the study by Murphy et al12 than in our study. This difference might be attributable to subject characteristics, interventions, and study designs. In particular, we studied the effects of lower cumulative doses of dexamethasone prescribed later in life to higher-risk ELBW infants. In addition, to minimize bias, we enrolled an unselected cohort of all ELBW infants, performed masked evaluations, and assessed a larger sample size, which allowed more-complete adjustment for potential confounders. We tested the association of global and regional cerebral volume abnormalities with multiple clinical variables, individually and in combination with other variables, including dexamethasone administration. The strongest association, not surprisingly, was with PMA at MRI, which is why all analyses were conducted only after adjustment for this factor. The period beyond 29 weeks of gestation involves normal rapid increases in total brain tissue volumes.24 Therefore, differences in age of even 1 to 2 weeks would result in large differences in global and regional volumes. BW also showed a strong relationship with brain volumes. Large differences in cortical gray matter volume (35%) in the infants studied by Murphy et al12 might have been more reflective of incompletely adjusted group differences in BW and other important covariates and/or exposure to comparatively higher cumulative doses of dexamethasone in lower-risk infants. In univariate analysis, we also observed significantly smaller cortical gray matter volumes in association with steroid administration; however, this difference in multivariate analysis was not as dramatic (7.5%) and was not statistically significant (P = .10). Our findings are disturbing because we observed differences even after adjustment for BPD and other risk factors and with relatively conservative use of dexamethasone. This could be a result of residual confounding or a real effect of dexamethasone on brain volumes, despite use limited to high-risk infants.
The observed smaller subcortical (deep nuclear) and total gray matter volumes in dexamethasone-treated ELBW infants, without similar differences in white matter, suggest that neurons may be particularly vulnerable to the toxic effects of dexamethasone. Very preterm infants are exposed frequently to hypoxia-ischemia or other insults. Dexamethasone may impair protective mechanisms against such insults; dexamethasone pretreatment in adult rats before right cerebral artery occlusion to induce an infarction resulted in a 10-fold greater infarction volume.26 Alternatively, such gray matter volume deficits may reflect the greater degree of illness in the dexamethasone-treated infants, compared with untreated infants. These differences might have persisted despite our efforts to control for confounding. In children and adolescents who were born prematurely, volumetric MRI studies observed brain volume deficits in cortical and deep nuclear gray matter consistently.27–30 Subplate neurons and axonal development are crucial for cortical and thalamic neuronal development. Volpe31 hypothesized that sublethal injury to subplate neurons and/or disrupted axonal development resulting from reactive oxygen species generated through ischemia and inflammation could lead to profound neuronal abnormalities in cortical and deep nuclear gray matter. These neuronal abnormalities are identifiable as early as term-equivalent age and are greatest among the most-immature infants and those with white matter injury.9,10
The detrimental effect on the cerebellum we described has not been reported for human subjects after corticosteroid administration. In neonatal mice, corticosterone administration impaired brain DNA, RNA, and protein synthesis, with resulting permanent reductions in size and weight of the cerebellum.32 Structural and biochemical brain alterations were also observed in the cerebellum of rats after commonly used doses of dexamethasone.33 Conversely, direct injury to developing cortical gray or white matter might have resulted in cerebellar volume deficits through trophic transsynaptic negative effects on cerebellar growth.34,35 Therefore, direct injury to the cerebellum is not a prerequisite for cerebellar volume disturbances. Despite these interesting observations, however, the mechanisms of the neurologic effects of dexamethasone, particularly among vulnerable ELBW infants, are likely still more complex and remain to be clarified.
Despite our use of a relatively more-robust design, we recognize certain limitations with assessing brain volumes in vivo with advanced volumetric MRI. The use of 4-mm brain slices rather than thinner three-dimensional contiguous slices might have resulted in less-reliable measurements of smaller structures such as the corpus callosum and hippocampus. This limitation might have masked true differences in volumes for vulnerable structures such as the hippocampus. The dichotomous physiologic definition of BPD, without quantification of disease severity, might have hampered our efforts to assess and to control for group differences in lung disease adequately. Also, the functional implications of the described volume deficits can be estimated at best without long-term, neurologic, follow-up assessment. Finally, as in all observational studies, the results might be distorted by residual confounding attributable to unknown or incompletely addressed confounders. These study limitations would be best addressed within a well-designed randomized trial with 2-year neurosensory follow-up assessment, as we are undertaking currently.36
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We are grateful to our neonatal transport team for transporting our ELBW infants skillfully and safely for MRI and to Christine Domonoske, PharmD, for assistance in verifying all steroid data.
|
FOOTNOTES |
|---|
Address correspondence to Nehal A. Parikh, DO, University of Texas Medical School at Houston, 6431 Fannin, MSB 3.236B, Houston, TX 77030. E-mail: nehal.a.parikh{at}uth.tmc.edu
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
brains/humanatlas. Accessed January 6, 2005This article has been cited by other articles:
 |
|
 |
|
  H. Needelman, M. Evans, H. Roberts, M. Sweney, and J.B. Bodensteiner Effects of Postnatal Dexamethasone Exposure on the Developmental Outcome of Premature Infants J Child Neurol, April 1, 2008; 23(4): 421 - 424. [Abstract] [PDF]
|
|
|
|
 |
|
 K J Rademaker, L S de Vries, C S P M Uiterwaal, F Groenendaal, D E Grobbee, and F van Bel Postnatal hydrocortisone treatment for chronic lung disease in the preterm newborn and long-term neurodevelopmental follow-up Arch. Dis. Child. Fetal Neonatal Ed., January 1, 2008; 93(1): F58 - F63. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. T. Vasileiadis Neonatal Brain Volumetric Studies: Regression Analysis and Interpretation Pediatrics, June 1, 2007; 119(6): 1251 - 1252. [Full Text] [PDF]
|
|
|
|
|
|
N. A. Parikh, R. E. Lasky, K. A. Kennedy, and J. E. Tyson Neonatal Brain Volumetric Studies: Regression Analysis and Interpretation: In Reply Pediatrics, June 1, 2007; 119(6): 1252 - 1253. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2007 American Academy of Pediatrics. All rights reserved. |
||
|
||
