Magnetic Resonance Imaging Regional T1 Abnormalities at Term Accurately Predict Motor Outcome in Preterm Infants
OBJECTIVE. The aim of this study was to assess whether periventricular leukomalacia findings are sufficiently sensitive for predicting the severity of motor prognosis by conventional MRI in the near term.
METHODS. Preterm infants with T1 hyperintensity or cysts in the periventricular regions on term MRI were selected, and their gross motor functions were evaluated at the age of 3 to 5 years. Sixty-two infants had findings of T1 hyperintensity or cysts, and except for infants with these findings, none were diagnosed later as periventricular leukomalacia.
RESULTS. All 37 patients with cerebral palsy had periventricular lesions with T1 hyperintensity or cysts in the corona radiata above the posterior limb of the internal capsule on coronal sections. Small T1 hyperintensity lesions were seen on coronal slices and were often difficult to detect on axial slices. All of the 17 infants with T1 hyperintensity findings sparing the corona radiata above the posterior limb of the internal capsule showed normal motor development, irrespective of findings of ventriculomegaly. There was a tendency for the presence of widespread lesions in corona radiata above the posterior limb of the internal capsule to be correlated with the severity of motor handicap.
CONCLUSIONS. Lesions in the corona radiata above the posterior limb of the internal capsule on a coronal view by term MRI were useful for predicting motor prognosis in preterm infants with periventricular leukomalacia.
Periventricular leukomalacia (PVL) is a major type of brain injury in preterm infants. Ultrasonography and MRI are the standard methods that are used for diagnosing PVL. Ultrasonography clearly detects cystic PVL at the bedside, and MRI is superior to ultrasonography in detecting noncystic white matter lesions.1–5
A number of studies have demonstrated correlations between motor prognosis and PVL findings by cranial MRI.6 These MRI scans were conducted serially over periods of months or several years (late MRI). PVL on late MRI was diagnosed when there were (1) ventriculomegaly with an irregular outline of the body and trigone of the lateral ventricles, (2) a reduced quantity of periventricular white matter, and (3) abnormal signal intensity in the periventricular white matter.7,8
Several studies have reported that neonatal conventional MRI, at around term (term MRI), could be used to predict cerebral palsy (CP) with a high degree of sensitivity and specificity.5,9–12 The PVL findings in these studies were cystic lesions, diffuse T2 hyperintensity, and T1 hyperintensity. In a previous report by us, as well as others, T1 hyperintensity lesions in the periventricular white matter were shown to be associated with PVL on late MRI.7,8,12–15 Some infants with these lesions have developed normally without CP,12,16 and a correlation of PVL findings on term MRI with the severity of motor problems has not been examined. The accurate identification of PVL before discharge is clinically important for the early prediction of motor sequelae and for targeting high-risk infants to appropriate rehabilitation services. The aim of this study was to assess whether the findings of T1 hyperintensity in periventricular white matter by conventional MRI around term are correlated with PVL findings as diagnosed by late MRI, which lesions result in later motor defects, and whether these findings can be used to predict accurately the severity of motor problems in preterm infants.
A total of 2342 patients were admitted to the NICU of Kanagawa Children's Medical Center between January 1993 and July 2000. Of 1119 preterm infants who were born at a gestational age of 24 to 34 weeks, MRI scans were performed before discharge from the hospital for 460 infants who had a birth weight of <1500 g or abnormal ultrasonography findings. A total of 430 MRI scans were conducted at a corrected age of 36 to 43 weeks, and 30 scans were conducted at >44 weeks. Among 430 infants, 13 died after discharge and 26 were not followed. We excluded infants with intracranial hemorrhages, including germinal matrix hemorrhage (61 cases), hydrocephalus (23 cases), brain malformations (3 cases), chromosomal abnormalities (2 cases), anomaly syndrome (7 cases), congenital myotonic dystrophy (3 cases), and intrauterine viral infection (3 cases; Fig 1).
For the remaining 289 infants, we examined the findings of T1 hyperintensity or cystic lesions in the periventricular white matter by term MRI. Sixty-two infants had findings of T1 hyperintensity or cystic lesions in the periventricular white matter and were followed clinically for 3 to 5 years. A total of 227 infants without findings in the periventricular white matter were followed until they could walk, and when their neurologic examinations and development were abnormal or their birth weight was <1500 g, they were followed for >3 years. Of 227 infants with normal term-MRI findings, 1 patient could not walk at the age of 3 years. His neurologic examination revealed hypotonia without signs of spasticity, and his late MRI did not show PVL. Except for 62 infants with the findings of T1 hyperintensity or cysts by term MRI, none was diagnosed later as PVL.
Term MRI was usually performed before discharge from the hospital using a 1.5T scanner and consisted of the following: coronal and axial spin-echo (SE) T1-weighted images (400/15/2 [echo time/repetition time/excitations]) and coronal and/or axial SE T2-weighted images (300/81–120/1) with 5-mm slices. All MRIs were evaluated independently by 3 of the authors (Drs Nanba, Matsui, and Aida) without knowledge of the clinical outcome, and in case of discrepancy, the findings were established after the discussion by 3. We selected infants with MRI findings of T1 shortening or cysts in the periventricular white matter, which had been reported as findings of PVL at term MRI.7,8 The following MRI findings were studied: (1) distribution of T1 hyperintensity, (2) the presence or distribution of cystic lesions, (3) degree of ventriculomegaly, and (4) degree of irregularity of the ventricular outline. Ventriculomegaly was divided into 5 groups: none, slight, mild when the ventricular/brain ratio (V/B) at the level of the midbody of the lateral ventricles was <0.34 but the occipital horn was largely dilated, moderate when the V/B exceeded 0.35, and severe when little white matter was seen because of the dilation of ventricles.17 The irregularity of the ventricular outline was classified into 4 degrees: none, slight, apparent, and severe.
Intraventricular hemorrhage (IVH) showed a T1 hyperintensity in the periventricular region on term MRI. IVHs were observed to be of homogeneous extreme hyperintensity in the T1 sequence with T2 hypointensity along the ventricular margin. IVH often coexisted with PVL, but to clarify the prognosis for the infants with the lesion in the periventricular white matter, we excluded patients with IVH, including germinal matrix hemorrhage,
Myelination of the corona radiata (CR) also showed T1 hyperintensity in the periventricular region. The normal myelination of CR has been reported to be evident on axial slices after the corrected age of 36 weeks.8,18,19 In terms of the normal myelination of corticospinal tracts in the periventricular region projecting from the motor cortex down to the posterior limb of the internal capsule (PLIC), which was easily observed in coronal images, T1 hyperintensity appeared after the corrected age of 44 weeks (Fig 2). Therefore, it was difficult to distinguish PVL lesions in periventricular CR related to the corticospinal tract (CR-CSp) from normal myelination, and MRIs that were taken between postmenstrual weeks 36 and 43 were included in this study.
Of the 62 infants studied, follow-up MRIs were performed 1 to 3 years later for 17 infants for clinical indications. Late MRI included axial SE T1-weighted images (360–500/15/2) and T2-weighted images (3000/80–110/1).
All MRIs were performed with the infants in stable condition, and the infants were sedated with pentobarbital (2.5–10 mg/kg body weight). Heart rate and transcutaneous oxygen saturation were monitored during and after the examination, and all examinations were done in a safe manner. Informed consent was obtained from all parents, and the study was based on the ethical guidelines approved by the ethics committee of the hospital.
Gross Motor Function
Infants with the lesions in the periventricular white matter and abnormal neurologic examination were followed up for at least 3 years by pediatricians who had been trained in these procedures. The severity of CP was classified into levels I to V on the basis of the Gross Motor Function Classification System (GMFCS)20 (Table 1). No abnormality in gross motor development was classified into level 0.
Of 289 eligible infants, 62 infants had findings of T1 hyperintensity or cystic lesions in the periventricular white matter and were followed for 3 to 5 years. Of 227 infants with normal term MRI findings, none later received a diagnosis as PVL. It was difficult to identify any particular antecedents from the NICU course, which might be responsible for each MRI finding.
Characteristics of T1 Hyperintensity Lesions
Distributed T1 hyperintensity lesions were observed on both coronal and axial images on term MRI in 9 patients. These were sometimes linear heterogeneous punctate lesions in the periventricular white matter in axial images. Where spotty T1 hyperintensity lesions were seen, the T2 image showed mild hypointensity or isointensity (Figs 3 and 4). Small T1 hyperintensity lesions were seen as spotty on coronal slices in 35 patients. These were often difficult to find on axial slices because of the T1 hyperintensity of normal myelination (Fig 3). These small punctate T1 hyperintensity lesions showed an isointensity on T2 images. As a result, it was essential to investigate coronal T1 images, to detect all of the lesions in the periventricular white matter. Twenty-four of 62 patients had a cystic PVL on term MRI. In the cases with cystic PVL, T1 hyperintensity lesions were often in proximity to regions of cystic changes (Fig 4).
Relation Between T1 Hyperintensity or Cystic Lesions and GMFCS
Of 62 infants with the T1 hyperintensity, 25 had normal motor development and 37 had CP. We prepared a list of these term-MRI findings that were classified by GMFCS level (Tables 2 and 3). All of the 17 infants with T1 hyperintensity findings that spared the CR-CSp showed normal motor development without any signs of spasticity, irrespective of findings of ventriculomegaly or an irregularity in the ventricular outline (Table 2, Fig 5). In 4 infants with spotty T1 hyperintensity findings that spared the CR-CSp on term MRI, the late MRIs after the age of 1 showed widespread T2 hyperintensity in the periventricular white matter with a mild irregularity in the ventricular outline with or without a reduced quantity of periventricular white matter.
Among 45 children who showed T1 hyperintensity or cysts in the CR-CSp, 8 had normal development without any signs of spasticity and 37 had spastic motor defects as a result of PVL (Table 3). All 8 infants with normal motor development had small spotty T1 hyperintensity lesions, and none had widespread lesions or cysts in the CR-CSp; on their MRIs, a slight ventriculomegaly was sometimes seen, although irregularity of the ventricular walls was absent. Infants with findings in the CR-CSp, ventriculomegaly, and irregularity of the walls developed CP.
Sensitivity and Specificity of PVL Findings on Term MRI in Predicting the Severity of Motor Problems Among 289 Eligible Infants
The sensitivity and the specificity of cystic lesions in the periventricular white matter for detecting CP (GMFCS level I or higher) among the 289 infants were 62% (23 of 37) and 87% (251 of 289), respectively. The sensitivity and the specificity of lesions in the CR-CSp for detecting CP (GMFCS level of I or higher) among the 289 infants were 100% (37 of 37) and 97% (244 of 252), respectively (Table 4).
The widespread findings in the CR-CSp tended to correlate with a worse ventriculomegaly or irregularity in the ventricular wall. Of 13 infants with T1 hyperintensity in the CR-CSp and without any irregularity in the ventricular outline, 8 had no abnormality in gross motor development (GMFCS 0) and 10 were able to walk without assistive devices (GMFCS 0–I). The lesions in the CR-CSp with ventriculomegaly (V/B ≥0.35) were correlated with GMFCS V. These sensitivity and specificity for detecting GMFCS V were 100% (11 of 11) and 100% (278 of 278), respectively. Ventriculomegaly (V/B ≥0.35) was always accompanied by an apparent irregularity in the ventricular outline. Lesions in the CR-CSp with an apparent irregularity in the ventricular wall were correlated with GMFCS IV or V. For these lesions, the sensitivity and the specificity for detecting GMFCS IV to V were 90% (18 of 20) and 100% (268 of 269), respectively.
There was a tendency for the severity of the MRI findings in CR-CSp to be correlated with the degree of motor disability. For detecting GMFCS levels I to III (they were able to walk with or without assistive devices), findings of T1 spotty hyperintensity without cysts in the CR-CSp had a sensitivity of 78% (14 of 18) and a specificity of 96% (260 of 271). For detecting GMFCS level IV (they were able to sit, but independent mobility was very limited), findings of T1 band-like hyperintensity in the CR-CSp had a sensitivity of 75% (6 of 8) and a specificity of 99% (278 of 281). The findings of cystic formation in the CR-CSp for detecting GMFCS V (they were not able to maintain antigravity head and trunk postures in prone and sitting positions; they were almost bedridden) had a sensitivity of 73% (8 of 11) and a specificity of 99% (275 of 278).
The major sequela of PVL is spastic diplegia, because most PVL occurs in the region of the white matter that is traversed by descending fibers from the motor cortex corresponding to the legs.6 We demonstrated that the presence of T1 hyperintensity lesions or cysts and their distribution at CR-CSp on a coronal view were important for detecting PVL related to motor defects. Coronal sections were superior to axial sections for detecting findings of T1 hyperintensity. On axial slices, we were unable to distinguish small spotty T1 hyperintensity lesions from myelination in the CR; as a result, punctate lesions on coronal slices were not often detected on axial views. On coronal slices, myelination in the CR above the PLIC that we noticed was not remarkable until a corrected age of 43 weeks, and we could easily define the lesions using the location of the PLIC as a hallmark. In previous studies, T1 hyperintensity was observed in preterm infants on MRI that was performed between the neonate and term-equivalent period.1–3,7,8,12–16,21 However, some reported that T1 hyperintensity did not affect the prognosis.16,21 These reports were based on axial or sagittal sections, and it might be difficult to determine the precise position of the findings with relation to the corticospinal tracts. We conclude that coronal MRIs between 36 and 43 weeks' corrected age would be the most useful for diagnosis of PVL and the prediction of the severity of gross motor function. In this study, we evaluated MRIs with 5-mm slices; 2- to 3-mm slices would improve the sensitivity.
Several studies of PVL investigated correlations between the severity of clinical features and findings in the corticospinal tract on late MRI but not on term MRI.22–25 In our experience, the distribution of T1 hyperintensity on term MRI was more limited than that of T2 or fluid-attenuated inversion-recovery hyperintensity on late MRI, and term MRI might more clearly demonstrate the focal lesions that are responsible for symptoms, rather than late MRI. Recent diffusion-weighted MRI studies that were performed in the neonatal period revealed defect in the corticospinal tract with PVL.26–30 They may be used for detecting the focus of white matter abnormalities in the future, but this technique is difficult to use as a screening method.
The findings of T1 hyperintensity on term MRI pathologically corresponded to cellular reactions of glial cells and macrophages, as well as formation of microcalcifications.15,31,32 T1 hyperintensity lesions were often adjacent to regions of cystic change, consistent with necrotic changes without cyst formation. Because of the shrinkage of the necrotic lesions,6 the broad findings of T1 hyperintensity lesions as well as cystic lesions led to ventriculomegaly and an irregularity. The severe findings in CR-CSp associated with severe ventriculomegaly and irregularity resulted in the worst outcome.
The possibility that infants with T1 hyperintensity sparing CR-CSp have cognitive or other defects by reason of a reduction in white matter cannot be excluded.33 In this study, cognitive or visual dysfunctions were not evaluated, and additional study in this area is needed. We examined the findings of T1 hyperintensity in preterm infants. Similar findings are also observed in term infants, and it needs to be elucidated whether those findings in term infants have the same pathology as white matter injury in prematurity.
PVL lesions in term MRI cannot be correctly evaluated by axial slices, which are misleading. We conclude that coronal T1 sequences in term MRI are useful as a screening method for the diagnosis of PVL and the prediction of motor outcome in preterm infants.
We thank Dr Yoshiaki Saito and Kousaku Ohno, Division of Child Neurology, Institute of Neurologic Sciences, Tottori University School of Medicine, for help in the preparation of this manuscript.
- Accepted December 12, 2006.
- Address correspondence to Yukiko Nanba, MD, 2-10-1 Okura Setagaya-ku Tokyo, 157-8535 Japan. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵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
- 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
- ↵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
- Maalouf EF, Duggan PJ, Counsell SJ, et al. Comparison of findings on cranial ultrasound and magnetic resonance imaging in preterm infants. Pediatrics.2001;107 :719– 727
- ↵Volpe JJ. Neurology of the Newborn. 4th ed. Philadelphia, PA: WB Saunders; 2001
- ↵Barkovich AJ. Pediatric Neuroimaging. 4th ed. New York, NY: Lippincott-Williams & Wilkins; 2005:209– 220
- ↵Rutherford MA. MRI of the Neonatal Brain. Philadelphia, PA: WB Saunders; 2002:45– 47, 155–164
- ↵Mirmiran M, Barnes PD, Keller K, et al. Neonatal brain magnetic resonance imaging before discharge is better than serial cranial ultrasound in predicting cerebral palsy in very low birth weight preterm infants. Pediatrics.2004;114 :992– 998
- ↵Aida N, Nishimura G, Hachiya Y, Matsui K, Takeuchi M, Itani Y. MR imaging of perinatal brain damage: comparison of clinical outcome with initial and follow-up MR findings. AJNR Am J Neuroradiol.1998;19 :1909– 1921
- Keeney SE, Adcock EW, McArdle CB. Prospective observations of 100 high-risk neonates by high-field (1.5Tesla) magnetic resonance imaging of the central nervous system. II. Lesions associated with hypoxic-ischemic encephalopathy. Pediatrics.1991;87 :431– 438
- ↵Cornette LG, Tanner SF, Ramenghi LA, et al. Magnetic resonance imaging of the infant brain: anatomical characteristics and clinical significance of punctate lesions. Arch Dis Child Fetal Neonatal Ed.2002;86 :F171– F177
- ↵Counsell SJ, Maalouf EF, Fletcher AM, et al. MR imaging assessment of myelination in the very preterm brain. AJNR Am J Neuroradiol.2002;23 :872– 881
- Hoon AH, Lawrie WT, Melhem ER, et al. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology.2002;59 :752– 756
- 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
- ↵Hüppi 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
- ↵Garel C, Delezoide AL, Elmaleh-Berges M, et al. Contribution of fetal MR imaging in the evaluation of cerebral ischemic lesions. AJNR Am J Neuroradiol.2004;25 :1563– 1568
- ↵Inder TE, Hüppi PS, Warfield S, et al. Periventricular white matter injury in the premature infant is followed by reduced cerebral cortical gray matter volume at term. Ann Neurol.1999;26 :755– 760
- Copyright © 2007 by the American Academy of Pediatrics