Neonatal Brain Magnetic Resonance Imaging Before Discharge Is Better Than Serial Cranial Ultrasound in Predicting Cerebral Palsy in Very Low Birth Weight Preterm Infants
Objective. To compare the value of serial cranial ultrasound (US) with a single magnetic resonance imaging (MRI) before discharge in very low birth weight preterm infants to predict cerebral palsy (CP).
Methods. Infants who weighed <1250 g at birth and were <30 weeks' gestational age underwent conventional brain MRI at near term (36–40 weeks' postmenstrual age) using 1.5 Tesla MRI scanner. Sagittal and axial T1 and T2 fluid attenuated inversion recovery and gradient recalled echo images were obtained. Cranial US was also obtained at least twice during the first 2 weeks of life. MRI and US images were interpreted by 2 independent radiologists, who were masked to clinical outcome, and scored as follows: category 1, no abnormality; category 2, subependymal hemorrhage or mineralization; category 3, moderate to severe ventriculomegaly; category 4, focal parenchymal abnormality with or without ventriculomegaly. For the purpose of this study, 1 and 2 were categorized as “normal,” and 3 and 4 were categorized as “abnormal.” The infants were assessed at a mean age of 20 and 31 months using the Amiel-Tison standardized neurodevelopmental examination.
Results. The sensitivity and specificity of MRI for predicting CP were 71% and 91% at 20 month and 86% and 89% at 31 months, respectively. The sensitivity and specificity of US for predicting CP were 29% and 86% at 20 months and 43% and 82% at 31 months.
Conclusions. As a predictor of outcome for CP, MRI at near-term in very low birth weight preterm neonates is superior to US. However, both US and MRI demonstrate high specificity.
In the United States, ∼60 000 very low birth weight (VLBW) preterm infants are born each year.1 Because of the therapeutic advances in perinatal and neonatal intensive care medicine, the majority of these infants survive to discharge.2 However, 10% to 15% of those surviving infants develop cerebral palsy (CP) and major psychomotor and cognitive deficits.3–8 CP has the highest lifetime costs per new case and is among the most common birth defects (Centers for Disease Control and Prevention, National Center of Birth Defects and Developmental Disabilities). Although VLBW preterm infants represent only 1.4% of live births per year, they constitute 25% of all children with CP.9 Early prediction of CP in preterm children will help to reduce the personal and economic lifetime costs associated with CP and is important in directing appropriate therapy and allocating services before discharge.
Perinatal brain injury underlies the development of CP.10–12 Brain injury in preterm infants is commonly diffuse white matter injury. In many cases, it is manifested as a silent chronic injury that is very different from acute hypoxic-ischemic encephalopathy, commonly seen in term infants. Preterm brain injury may be attributable to a variety of chronic insults and may present as periventricular leukomalacia (PVL), intraparenchymal hemorrhage, ischemia, and moderate to severe ventriculomegaly with white matter loss. PVL is composed of focal and diffuse components.13 The focal component is characterized by local necrosis that affects all cells types with subsequent cyst formation in the cerebral white matter. The diffuse type, which is more common, is a less severe injury and involves specific cell types, primarily oligodendrocytes. Although the focal type of PVL may be detected by cranial ultrasound (US), the diffuse type is usually not evident on routine imaging. More than one half of VLBW preterm infants with CP do not have intracranial abnormalities detectable by US.8 Moreover, although the brain injury underlying CP in VLBW preterm infants is likely to occur during the perinatal or early neonatal period in the newborn intensive care unit, the neurologic manifestations of CP may not be apparent until 2 to 3 years of age. This delay in diagnosis of CP limits our understanding of the pathophysiology of this lifetime disorder and leads to late intervention. Magnetic resonance imaging (MRI) is better than US in detecting white matter injury.11,14–17 As stated in the recent “Practice Parameter” for neuroimaging in the neonate from the American Academy of Neurology and Child Neurology Society, there is no prospective study that systematically examines the value of both neonatal US and MRI with respect to long-term neurologic outcomes of VLBW preterm infants.18
The aim of the present prospective study was to investigate the potential advantage of brain MRI before discharge to predict CP at 18 to 22 and 30 to 32 months of age in VLBW preterm infants. We hypothesized that predischarge neonatal brain MRI would be more sensitive for predicting CP than early serial cranial US.
Parents of infants with a birth weight <1250 g and a gestational age (GA) <30 weeks, living within 100 miles of Lucile Salter Packard Children's Hospital at Stanford, and born between July 1996 and August 1999 were asked for permission for their infant to participate in this investigation. Estimation of GA was based on the expected date of confinement, which was calculated by the obstetrician, determined by mothers' last menstrual period, and confirmed by prenatal US. For mothers who did not have prenatal care and were not sure of their dates of last menstrual period, the GA was determined at birth using the revised Ballard GA assessment.19 Exclusion criteria included central nervous system malformations, congenital neuromuscular disease, multiple congenital anomalies, and chromosomal abnormalities. The study was approved by the Stanford Administrative Panel on Human Subjects in Medical Research, and informed written consent was obtained from all parents.
MRI was performed before discharge at near term age (36–40 weeks' postmenstrual age), “neonatal,” using a 1.5 T Signa system (General Electric Medical Systems, Milwaukee, WI). Sagittal and axial T1 conventional spin echo images were obtained with 4-mm slices and 1-mm gaps, repetition times (TR) of 500 ms, echo times (TE) of 18 ms, number of excitations (NEX) 2, a 256 × 192 matrix, and a field of view (FOV) of 20 cm. Axial proton density and T2 fast spin echo images were obtained using a TR/TE/echo train length 2000/28/4 and 3600/108/8, respectively, with the same slice/gap, matrix, FOV, and NEX as described above. Axial fluid attenuated inversion recovery images were also done using TR/TE/inversion time 10002/133/2200, NEX 1, and the same slice/gap, matrix, and FOV. Last, gradient recalled echo (GRE) images were obtained using a 15-degree flip angle, TR/TE 516/30, and 1 NEX plus similar slice/gap, matrix, and FOV.
All MRI examinations were reviewed by an experienced pediatric neuroradiologist (P.D.B.) who was masked to the patient history, neurologic assessment, and previous clinical and radiologic reports, including cranial US. The MRIs were surveyed for abnormal intensities, structural alterations, and ventricular size and configuration. Intensity abnormalities sought included T1 hypointensities and T2 hyperintensities (eg, undermyelination, demyelination, neuronal or axonal loss, gliosis). T1 hyperintensities, T2 hypointensities, and GRE hypointensities were also recorded (eg, old hemorrhage, mineralization [ie, calcification or iron deposition], gliosis). Myelination was not specifically assessed because the causes for absence of normal myelination could not be distinguished (eg, delayed myelination vs hypomyelination vs demyelination). Furthermore, the issue of myelination in this patient population is addressed in another publication by the same authors.20 Structural alterations included white matter signal intensity changes, cysts, cavitations, and atrophy. Ventricular size was assessed using ventriculocephalic ratios for the lateral ventricles (ie, bifrontal and biparietal ratios) and assigned as normal or as minimal, mild, moderate, or severe enlargement. Ventricular size and configuration assessment also included asymmetric or disproportionate enlargement with regard to the fourth, third, and lateral ventricles (eg, temporal horns, frontal horns, occipital horns, atria), and the contour of the ventricular chambers (ie, irregular or smooth). The MRI findings were categorized for the degree of brain injury as follows: category l (C1), no abnormality; C2, minimal subependymal hemorrhage or mineralization (ie, iron or calcium deposition; eg, grade 1 and 2 germinal matrix/intraventricular hemorrhage [IVH]); C3, moderate or severe ventriculomegaly (ie, >25%–50% ventriculocephalic ratio; eg, grade 2 IVH, posthemorrhagic hydrocephalus, or atrophy; Fig 1); C4, parenchymal abnormality, including evidence of injury (eg, T1, T2, or GRE intensity abnormalities or structural alterations) as a result of hemorrhage or ischemia (eg, grade 4 IVH or PVL; Fig 1). For the purposes of this study, C1 and C2 were categorized as “normal,” and C3 and C4 were categorized as “abnormal.”
US was obtained with an 8-MHz transducer (Acuson Sequoia, San Jose, CA) at least twice during the first 2 weeks of life and thereafter as clinically indicated at the discretion of the attending physician. Cranial US studies performed after 2 weeks of age were limited primarily to cases in which abnormal early findings required additional follow-up. Five to 6 coronal plane images and 4 to 5 sagittal plane images were obtained via the anterior fontanelle and supplemented with posterior fontanelle or mastoid images. These images were scored by an experienced pediatric radiologist (K.K.) who was masked to the MRI findings and the clinical history. The US examinations were surveyed for abnormal echoarchitecture (abnormally increased or decreased echoes), structural alterations, and ventricular size and configuration. The US findings were categorized using the same method as for the MRI review and based on the presence or absence of hemorrhage or mineralization, ventriculomegaly (>1 cm measured at the midbody of the lateral ventricle on sagittal scan), or parenchymal injury.
The Amiel-Tison standardized neurologic examination21 was performed by certified masked developmental specialists who had been trained to reliability in the examination by the National Institute of Child Health and Human Development Neonatal Research Network.6 The infants' examinations were scored as normal when no abnormalities were observed in the neurologic examinations and functional gross motors skills were age appropriate. CP was characterized by abnormal muscle tone/movement in at least 1 extremity and abnormal control of movement and posture. The severity of CP was classified further as mild, moderate, and severe on the basis of Gross Motor Function classification for CP.22 Infants were examined twice: once at a mean age of 20 months' corrected age and the second time at a mean age of 31 months. The Bayley Scales of Infant Development,23 including Mental Developmental Index and Psychomotor Developmental Index, were performed by masked examiners who were trained by 1 of the 4 National Institute of Child Health and Human Development study network “gold standard” examiners.6
We calculated the sensitivity, specificity, positive predictive value, and negative predictive value of the MRI and US findings in relation to the neurologic outcomes (CP and non-CP) at 20 months' corrected age and at 31 months.
A total of 238 infants were eligible during the recruiting period. Twenty-seven parents declined, 21 infants died, 31 infants transferred to other hospitals, 11 infants had congenital anomalies, and for 32 we were not able to obtain consent. Of 116 infants enrolled, 6 withdrew from the study and 11 later died. Of the 99 infants who participated, we were able to obtain an MRI in 88 cases before discharge home. Twenty-six examinations were excluded because of movement artifacts or congenital brain anomalies. Sixty-two infants had MRI at 36 to 40 weeks' postmenstrual age. One infant later received a diagnosis of muscular dystrophy in follow-up and was excluded from additional analysis. The final study population consisted of 31 male and 30 female infants with birth weight ranging between 502 and 1240 g and a GA ranging between 23 and 29 weeks. Among infants recruited, there were 30 of multiple gestation. Demographics of the study population are shown in Table 1. There were no significant differences in birth weight or other demographic data between CP and non-CP infants. Only 2 of the 7 CP infants weighed >1000 g at birth. All infants except 3 had at least 3 USs. These cases were excluded from additional analysis of US. None of these had CP at 20 months, and they did not return for the 31-month follow-up.
Table 2 shows MRI findings and neurologic outcome at 20 months of age. Among 7 infants who had a diagnosis of CP at this age, there were 5 true positive and 2 false negative MRIs. The MRI was truly negative in 50 but falsely positive in 4 non-CP infants. Table 3 summarizes the neurologic and Bayley results of 43 infants (including all CP cases) who returned for follow-up at a mean age of 31 months. One subject (case k), considered normal at 20 months, received a diagnosis of CP at 31 months of age. This increased the number of true-positive MRI results to 6 and decreased the number of false-positive MRI results from 4 to 3. Moreover, another infant (case g) who was considered to have CP at 20 months was categorized as non-CP in follow-up at 31 months of age, decreasing the number of false-negative MRI results to 1 (Table 3). The Bayley results were >2 standard deviations below the mean (ie, <70) in all of the infants who had abnormal MRI findings except in case i (Table 3). Furthermore, only 1 infant (case f) with abnormal neurologic outcome had Bayley scores >70. Infants with visual and auditory deficits are identified in Table 3 to clarify some of the low Mental Developmental Index/Psychomotor Developmental Index scoring as a result of the sensory impairment.
The sensitivity and specificity of MRI for detecting CP at 20 months' follow-up were 71% and 91%, respectively. The sensitivity of MRI at 31 months of age increased to 86% with the specificity of 89% (Fig 2). Although the specificity of US was comparable to MRI, the sensitivity of US was 29% at 20 months and 43% at 31 months. The positive predictive value of US for CP was much lower compared with MRI (Fig 2).
US is currently considered the standard of care for neuroimaging of the preterm newborn because it is faster than MRI, is more readily available, and can be performed at bedside. However, results from this prospective study indicate that cranial US is not a highly sensitive diagnostic modality. Abnormalities on near-term (“neonatal”) MRI had significantly greater sensitivity and positive predictive value for the outcome of CP among VLBW preterm infants than results from early serial cranial US. We have previously shown, among other investigators, low sensitivity (56%) and low positive predictive value (42%) for US in predicting CP in a population of 100 preterm infants who were <30 weeks' GA and followed prospectively until 2 years of age.24 In that study, weekly US was performed until discharge in all infants. One limitation of the present study is that we did not have simultaneous MRI and US examinations, and MRI was always the last examination before discharge. This may partly be responsible for relative lower sensitivity and positive predictive value of US in the present study. However, a number of investigators have previously shown low sensitivity for US even when applied at term-equivalent age.8,15,24–29 In a recent study in which the value of serial US during the first 4 to 6 weeks of life in a group of 96 VLBW preterm infants was compared with a single MRI at term-equivalent age, the authors found low sensitivity (26%) for US in detecting noncystic white matter injury found on MRI.30 Although no follow-up studies were conducted to demonstrate the clinical significance of these findings, these results support our findings and suggest the relative deficiency of US in detecting subtle-diffuse brain injury and later neurologic impairment. Another limitation of our study is that we had a 1-mm gap between the successive MRI sections. Although this procedure eliminates cross-talk among sections when no gap is used, it may be responsible for missing small lesions. Newly developed methods of 3-dimensional MRI with thin (1–3 mm) serial sections with no gap will be superior to the method used in this study. The last but not the least limitation of this study is the small number of subjects. Although this is the first study of its kind to have neonatal MRI and US as well as long-term follow-up to 31 month of age, the total number of 43 infants followed by this age limits the generalizability of our findings.
In 1 study in which paired MRI and US were done in a group of preterm infants who were <30 weeks' GA,16 US and MRI were comparable for detecting germinal matrix hemorrhage and IVH. However, more diffuse, noncystic white matter injury was better detected by MRI. The majority of brain injuries in preterm infants occurs in the white matter, and most of the lesions are diffuse rather than focal or cystic.13 This explains, in part, the discrepancy between US and MRI along with the low predictive value of US for long-term neurodevelopmental disability.7,8,11,15,28 Furthermore, Maalouf et al31 did serial MRI in VLBW (<30 weeks' GA) preterm infants starting within 1 week of life and demonstrated that diffuse white matter injuries were usually not detected until term or near-term corrected age. There were no findings of cystic PVL among the 41 cases that they studied. The authors concluded that although brain injury may occur antenatally, it often becomes clinically detectable only later in or beyond the neonatal period. This, too, explains the higher positive predictive value of MRI relative to US. Moreover, another study in which a single MRI was performed at 2 to 3 weeks of age and compared with serial US findings in the same infants also showed the low predictive value of US.32 The latter finding shows the lower sensitivity of US as a diagnostic modality rather than the timing of US.
Although the total number of subjects who were fully studied was small, the sensitivity of 86% for neonatal MRI with many true-negative and a few false-positive findings in the present study makes neonatal MRI a clinically useful tool for predicting long-term neurologic outcome of VLBW preterm infants. Although the prediction of CP was the primary objective in this study, it is interesting that most of the infants with an abnormal MRI (true positive) had abnormal Bayley score at 31 months. In a recent prospective study, neonatal MRI had a sensitivity of 82% and a specificity of 97% for predicting CP.25
We are currently using diffusion tensor MRI (DTI) using 4-mm sections with no gap to quantify the microstructural changes in the white matter.20,33 In the latter study, we investigated the advantages of DTI in a group of 63 preterm infants with normal term conventional MRI. Thirteen of these infants developed neurologic abnormalities, including CP. DTI results of these infants showed reduced myelination in the internal capsule in these infants compared with control subjects. Our results and those recently published in the literature support the additional advantages of DTI for investigating white matter integrity, which may enhance the predictive value of conventional MRI in preterm infants.17,34–37 A repeat MRI, perhaps including functional MRI at a later age, may also prove to be useful in understanding the underlying brain pathology and developmental differences between CP and controls as well as between preterm and term infants.38–43
Ment et al18 developed an evidence-based practice parameter on neuroimaging of the neonate after extensive review of the literature. On the basis of the strength of the available evidence, the following recommendations were made. “Routine screening cranial US should be performed on all infants of <30 weeks GA once between 7 and 14 days of age and should be optimally repeated between 36 and 40 weeks' postmenstrual age. This strategy detects lesions such as IVH, which influences clinical care, and those such as PVL and low-pressure ventriculomegaly, which provide information about long-term neurodevelopmental outcome. There is insufficient evidence for routine MRI of all VLBW preterm infants with abnormal results of cranial US.”18 At the time of that review, the literature was also considered inadequate to make a recommendation regarding the use of MRI in VLBW preterm infants with “normal” US results. We agree that VLBW preterm infants with severely abnormal US findings probably do not require additional MRI for the specific purpose of counseling and prediction of later CP. However, on the basis of the findings of the present prospective study, as well as the other reports from other investigators,15,16,25,29–32,44–48 we recommend using brain MRI before discharge as a routine level of care for at-risk VLBW preterm infants with normal early US or US abnormalities that may not be predictive of CP. MRI should not be considered as a replacement for US but indeed as an additional clinical test that can improve brain assessment and detection of anatomic abnormalities.
In conclusion, we offer the following revision to the neuroimaging practice parameter for screening of the “at-risk” VLBW preterm infant: routine screening cranial US should be done once between 7 and 14 days of age with repeat US and MRI before discharge between 36 and 40 weeks' postmenstrual age. Larger, multicenter studies comparing neonatal neuroimaging and their power to predict later neuromotor and cognitive deficits in VLBW preterm infants are needed.
This work was supported in part by a grant from the Packard Foundation, a grant from the Child Health Research Fund, National Institutes of Health Grant M01 RR00070 awarded to the General Clinical Research Center, National Institute of Neurological Disorders and Stroke Grant R21 NS40374, the Mary L. Johnson Research Fund, Innovation in Patient Care, and Lucile Packard Foundation for Children's Health, and a Dana Foundation Clinical Hypotheses Program in Imaging grant.
We are grateful for the continuous support of Roger B. Baldwin, Bethany M. Ball, Anne De Battista, and the staff of the Infant Development Clinic at LPCH for contributions in our follow-up program.
- ↵MacDorman MF, Minino AM, Strobino DM, Guyer B. Annual summary of vital statistics—2001. Pediatrics.2002;110 :1037– 1052
- ↵Lemons JA, Bauer CR, Oh W, et al. Very low birth weight outcomes of the National Institute of Child health and Human Development Neonatal Research Network, January 1995 through December 1996. NICHD Neonatal Research Network. Pediatrics.2001;107 (1). Available at: www.pediatrics.org/cgi/content/full/107/1/e1
- ↵Vohr BR, Wright LL, Dusick AM, et al. Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National Institute of Child Health and Human Development Neonatal Research Network, 1993–1994. Pediatrics.2000;105 :1216– 1226
- ↵Aziz K, Vickar DB, Sauve RS, Etches PC, Robertson CM. Province-based study of neurologic disability of children weighing 500 through 1249 grams at birth in relation to neonatal cerebral ultrasound findings. Pediatrics.1995;95 :837– 844
- ↵Pinto-Martin JA, Riolo S, Onaan A, Holzman C, Susser MW, Paneth N. Cranial ultrasound prediction of disabling and nondisabling cerebral palsy at age 2 in a low-birth-weight population. Pediatrics.1995;95 :249– 254
- ↵Wilson-Costello D, Borawski E, Friedman H, Redline R, Fanaroff AA, Hack M. Perinatal correlates of cerebral palsy and other neurological impairment among very low birth weight children. Pediatrics.1998;102 :315– 322
- ↵Dammann O, Leviton A. Brain damage in preterm newborns: might enhancement of developmentally regulated endogenous protection open a door for prevention? Pediatrics.1999;104 :541– 550
- ↵Volpe JJ. Neurology of the Newborn. 4th ed. Philadelphia, PA: WB Saunders; 2001
- ↵Maalouf E, Duggan P, Counsell S, et al. Comparison of findings on cranial ultrasound and magnetic resonance imaging in preterm infants. Pediatrics.2001;107 :719– 727
- ↵Ment LR, Bada HS, Barnes P, et al. Practice parameter: neuroimaging of the neonate: report of the Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society. Neurology.2002;58 :1726– 1738
- ↵Arzoumanian Y, Mirmiran M, Barnes PD, et al. Diffusion tensor brain imaging findings at term-equivalent age may predict neurologic abnormalities in low birth weight preterm infants. AJNR Am J Neuroradiol.2003;24 :1646– 1653
- ↵Amiel-Tison C. Neuromotor status. In: Taeusch H, Yogman M, eds. Follow-up Management of the High-Risk Infant. Boston, MA: Little, Brown and Company; 1987:115–126
- ↵Bayley N. Bayley Scales of Infant Development. 2nd ed. San Antonio, TX: The Psychological Corporation; 1993
- O'Shea TM, Klinepeter KI, Dillard RG. Prenatal events and the risk of cerebral palsy in very low birth weight infants. Am J Epidemiol.1998;147 :362– 369
- ↵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 :805– 809
- ↵Mirmiran M, Arzoumanian Y, Barnes P, et al. Quantitative brain magnetic resonance imaging in preterm infants may predict cerebral palsy. Pediatr Res2002;51 :439A
- ↵Huppi PS, Murphy B, Maier SE, et al. Microstructural brain development after perinatal cerebral white matter injury assessed by diffusion tensor magnetic resonance imaging. Pediatrics.2001;107 :455– 460
- McKinstry RC, Miller JH, Snyder AZ, et al. A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns. Neurology.2002;59 :824– 833
- Counsell SJ, Allsop JM, Harrison MC, et al. Diffusion-weighted imaging of the brain in preterm infants with focal and diffuse white matter abnormality. Pediatrics.2003;112(suppl) :1– 7
- ↵Hoon AH Jr, Lawrie WT Jr, Melhem ER, et al. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology.2002;59 :752– 756
- Peterson BS, Vohr B, Kane MJ, et al. A functional magnetic resonance imaging study of language processing and its cognitive correlates in prematurely born children. Pediatrics.2002;110 :1153– 1162
- Peterson B, Anderson A, Ehrenkranz R, et al. Regional brain volumes and their later neurodevelopmental correlates in term and preterm infants. Pediatrics.2003;111 :939– 948
- Cooke RW, Abernethy LJ. Cranial magnetic resonance imaging and school performance in very low birth weight infants in adolescence. Arch Dis Child Fetal Neonatal Ed.1999;81 :F116– F121
- ↵Abernethy LJ, Palaniappan M, Cooke RW. Quantitative magnetic resonance imaging of the brain in survivors of very low birth weight. Arch Dis Child.2002;87 :279– 283
- Miller SP, Cozzio CC, Goldstein RB, et al. Comparing the diagnosis of white matter injury in premature newborns with serial MR imaging and transfontanel ultrasonography findings. AJNR Am J Neuroradiol.2003;24 :1661– 1669
- Childs AM, Ramenghi LA, Cornette L, et al. Cerebral maturation in premature infants: quantitative assessment using MR imaging. AJNR Am J Neuroradiol.2001;22 :1577– 1582
- Debillon T, N′Guyen S, Muet A, Quere MP, Moussaly F, Roze JC. Limitations of ultrasonography for diagnosing white matter damage in preterm infants. Arch Dis Child Fetal Neonatal Ed.2003;88 :F275– F279
- Copyright © 2004 by the American Academy of Pediatrics