INTRODUCTION

Posterior fossa (PF) and cerebellar abnormalities in infants are difficult to visualize using conventional cranial ultrasound (CUS) via the anterior fontanelle because of the presence of the tentorium and the distance from the ultrasound probe.¹ Improved views of the PF can be obtained using the posterolateral fontanelle,² but magnetic resonance (MR) imaging gives the most detailed diagnostic views of the PF.

PF abnormalities are not uncommon. Cerebellar hemorrhage has been reported in 10 to 25% of low-birth-weight infants at post mortem,^3,4,5 but in only 13 out of 525 infants examined by ultrasound in the neonatal period.¹ Cerebellar hemorrhage was seen on MR in 3% of a cohort of infants <30 weeks gestational age.⁶ Cerebellar infarction and atrophy was recognized in 10 out of 73 preterm infants examined using MR and the majority of these infants also had abnormalities elsewhere in the brain.⁷ Historically, cerebellar hemorrhage was associated with the use of tight-fitting bands used to secure masks during continuous positive airway pressure (CPAP) ventilation.^5,8 PF hemorrhage associated with vacuum extraction,⁹ extra corporeal membrane oxygenation (ECMO)¹⁰ and traumatic delivery¹¹ has been described.

MR diagnosis of structural abnormalities of the cerebellum including cerebellar hypoplasia,¹² vermian agenesis¹³ and atrophy associated with metabolic disorders such as the carbohydrate-deficient glycoprotein syndrome¹⁴ have been well described. A classification of cerebellar malformations found on MR imaging has recently been proposed.¹⁵

Over 500 newborn infants have undergone MR brain imaging at the Leeds General Infirmary over the last 5 years. The aim of this study was to use this cohort of infants to describe the nature and frequency of PF abnormalities seen in preterm and term infants, to describe the outcome of infants with PF abnormalities and to compare the detection rate using conventional CUS and MR imaging.

METHODS

All infants who had undergone MR imaging of the brain as part of an ongoing MR research program from January 1996 to February 2001 were included. These infants included all those admitted to the neonatal intensive care unit (NICU) with abnormalities on CUS or abnormal neurological examination that required further neuroimaging (n=414) as well as some normal term and preterm infants (n=144) who were scanned for research purposes. Informed consent was obtained for all MR scans. A retrospective search of a detailed research database was made to identify those infants with abnormalities in the PF on MR imaging. The clinical case notes, MR images and CUS findings of these infants were reviewed in detail. Cranial ultrasound images were obtained via the anterior fontanelle by trained ultrasonographers using a 7.5 MHz sector transducer (Toshiba) in coronal and sagittal planes. At the time of the study, it was not our routine practice to obtain images via the posterolateral fontanelle. All MR images were obtained using a 1.5 T GYROSCAN ACS-NT scanner (Phillips Medical Systems, The Netherlands) with a receive-only quadrature head coil. T1-weighted images were acquired in the sagittal and axial planes and T2-weighted images in the axial plane. A number of infants also had T2-weighted coronal examinations. The acquisition parameters have been described previously.¹⁶ Briefly, for T1-weighted spin-echo images the parameters were repetition time (TR) 800 milliseconds, echo time (TE) 13 milliseconds, field of view (FOV) 180 mm, slice thickness 4 mm, with a 0.4 mm gap, acquisition matrix 256 256, scan time 3 minutes 52 seconds. For T2-weighted fast spin-echo images; TR 6000 milliseconds, TE 200 milliseconds, echo train length of 13, FOV 180 mm, slice thickness 3 mm with a 0.3 mm gap, acquisition matrix 256 256, scan time 5 minutes 12 seconds. Scans were performed after feeding or after chloral hydrate sedation. Infants were monitored with pulse oximetry and an electrocardiogram throughout the examination. Limited follow-up data were obtained where possible by contacting the pediatrician looking after the child.

RESULTS

From January 1996 until February 2001, n=558 newborn infants were examined with MR imaging. Of these 558 infants, 422 (76%) had abnormal MR brain scans. In all, 20 had abnormalities in the PF. The overall frequency of PF lesions on MR was therefore 20/558 (3.6%). These 20 infants formed the study group whose images and case notes were reviewed in detail.

Of the 20 infants, 10 were born at term (median gestation 40 weeks, range 47 to 42 weeks) and 10 were preterm (median gestation 28 weeks, range 24 to 36 weeks). Scans were performed at a median postnatal age of 6.5 days in the term group and 65 days (37.5 weeks corrected gestational age) in the preterm group. The preterm group was scanned later when clinically stable and no longer receiving intensive care.

Term Infants

Clinical details (Table 1).

Four infants showed clinical evidence of hypoxic–ischemic encephalopathy (HIE)¹⁷ (Subjects 5,7,9,10). Four (Subjects 1,3,4 and 6) were found to have a congenital or metabolic disorder (a complex brain malformation in one, L-2-hydroxy-glutaric-aciduria in one, Joubert's syndrome in one and Leigh's encephalopathy in the other). One term infant had complex congenital heart disease (Subject 2 – hypoplastic left heart and double outlet right ventricle with interrupted aortic arch) and the other intrauterine growth retardation with hypoglycemia (Subject 8).

Table 1 - Clinical Details and Outcome of Term Subjects.

Full table

MR findings (Table 2).

Six infants had MR evidence of extensive PF hemorrhage. In five subjects, the hemorrhage was predominantly subarachnoid, while in one it was subdural. Four of these infants showed a clinical pattern suggestive of HIE (Subjects 5,7,9,10) and in Subject 7 there was an associated coagulopathy (see Figure 1). Subject 2 showed absence of the corpus callosum, optic nerve atrophy and cerebellar hypoplasia in addition to an extensive subarachnoid hemorrhage. Subject 8 (who had suffered severe hypoglycemia (blood sugar 0.4 mmol/l)) had evidence of subarachnoid bleeding around the cerebellar hemispheres with additional punctate hemorrhages¹⁸ in the periventricular cerebral white matter.

Figure 1.

T2-weighted axial (left) and T1-weighted axial (right) MR images of a term infant with severe hopoxic–ischemic encephalopathy and a coagulopathy (Subject 7). The arrow marks the extensive posterior fossa hemorrhage disrupting the right cerebellar hemisphere. Hemorrhage is shown as low signal intensity on the T2 image.

Full figure and legend (112K)

Table 2 - MR Findings in Term Infants.

Full table

The other four term infants all showed MR evidence of structural CNS abnormality. Subject 1 demonstrated gross hypoplasia of the cerebellar hemispheres and vermis, absent corpus callosum and brain-stem hypoplasia. There was also evidence of delayed myelination. Subject 3 had cerebellar atrophy and delayed myelination (see Figure 2). A diagnosis of L-2-hydroxy-glutaric-aciduria was confirmed by urine analysis. Subject 4 showed absence of the cerebellar vermis with splaying of the cerebellar hemispheres, consistent with Joubert's syndrome. The parents were consanguineous and there was a family history of Joubert's syndrome. Subject 6 demonstrated extensive abnormal signal in the cerebellar vermis and occipitoparietal area of the cerebral white matter. The parents were consanguineous and Leigh's encephalopathy was diagnosed in this infant.

Figure 2.

T1 sagittal (left) and T2 coronal MR images (right) from a term infant (Subject 3) who collapsed at 12 hours of age with L-2-hydroxy-glutaric-aciduria. Note the cerebellar atrophy, with a large extra-axial space around the cerebellum in the posterior fossa. Cerebrospinal fluid is shown as low signal intensity on T1 imaging and high intensity on T2 imaging.

Full figure and legend (118K)

Outcome (Table 1).

The outcome among the term group was poor. Four infants died within the first year of life (Subjects 1, 2, 3 and 6). Of those who survived, two have cerebral palsy (subjects 5 and 10) and Subject 4 has severe developmental delay, hypotonia and a shunt for hydrocephalus. The remaining three infants (Subjects 7,8,9) have only mild deficits or were lost to follow-up (Subject 7 has microcephaly and a left-sided squint, Subject 8 has no motor deficit but a higher-order language disorder).

Preterm Infants

Clinical details (Table 3).

The gestational age of these infants ranged from 24 to 36 weeks (median 28 weeks). Eight of the infants were being treated for complications of prematurity. Three of these had intestinal perforation, one spontaneously (Subject 13) and two in association with necrcotizing enterocolitis (NEC) (Subjects 12 and 17). Two of the more mature babies were being treated for congenital heart disease. Subject 18 had DiGeorge's syndrome with an interrupted aortic arch and a ventricular septal defect, while the other (Subject 15) had Goldenhar's syndrome with transposition of the great arteries and aqueduct stenosis (detected antenatally by the presence of severe hydrocephalus). Antenatal risk factors in addition to prematurity, included maternal sepsis, antepartum hemorrhage and maternal lupus requiring plasmaphoresis. Postnatal risk factors included hypotension requiring inotropic support (five infants), confirmed sepsis (four infants), NEC (two infants), seizures (two infants), patent ductus arteriosus requiring indomethacin (one infant) and cardiac arrest (one infant).

Table 3 - Clinical Details and Outcome of Preterm Infants.

Full table

MR findings (Table 4).

Five infants showed evidence of focal intraparenchymal cerebellar hemorrhage (Subjects 13,16,17,19,20) (see Figures 3 and 4). Three infants had much more extensive cerebellar hemorrhage with destruction or secondary atrophy of most of one cerebellar hemisphere (Subjects 11,12,14) (see Figure 5). The remaining two infants were those with the associated cardiac abnormalities. Subject 15 showed evidence of aqueduct stenosis with massive dilatation of the lateral and third ventricles, and Subject 18 showed generalized cerebellar atrophy with reduced white matter throughout the cerebral hemispheres. None of the preterm group showed evidence of extra-axial hemorrhage.

Figure 3.

T2-weighted axial MR image from a 35-week infant (Subject 17). Arrow indicates the focal hemorrhagic lesion in the right cerebellar hemisphere.

Full figure and legend (116K)

Figure 4.

T2-weighted axial MR image from a 24-week gestation infant performed at 12 weeks of age (Subject 16). Note the multiple punctate hemorrhagic lesions in the right cerebellar hemisphere (arrows).

Full figure and legend (86K)

Figure 5.

T2-weighted coronal and axial MR images from a 29-week gestation infant (Subject 14). Arrows indicate the extensive destruction of the left cerebellar hemisphere.

Full figure and legend (118K)

Table 4 - MR Findings in Preterm Infants.

Full table

Outcome (Table 3).

The outcome data for the preterm group are not yet complete as many were scanned recently. All 10 infants survived. Subject 11 has a flaccid paralysis of both legs but normal cognitive function. Subject 12 is blind following retinal detachment. Subject 14 has mild motor delay with some hypotonia. Subject 15 has a shunt for hydrocephalus. Subject 17 has microcephaly but no focal motor deficit at 20 months. Subject 19 has no motor deficit but has speech and language delay secondary to deafness.

Cranial ultrasound.

In our cohort, CUS had been performed in 19 of the 20 infants. Although CUS did detect abnormalities in the brain in 15/19, only seven of these were within the PF. In the remaining 12 infants, the lesions seen on MR had not been detected on a contemporaneous CUS.

Associated pathology.

MR imaging demonstrated additional pathology above the tentorium (i.e. outside the PF) in 17/20 cases. This included dilatation of the lateral ventricles in 9/20 (four term, five preterm) and intraventricular hemorrhage (IVH) in 6/20 (two term, four preterm). In one preterm infant, there was evidence of IVH on early CUS that had resolved by the time of the MR scan. Cerebral white matter abnormality was seen on MR in 10/20, the majority of lesions being punctate lesions in the periventricular white matter (high T1, low T2).¹⁸ Subject 14 had ischemic change in the basal ganglia (high T1, low T2) and Subject 20 showed cortical 'highlighting' (high T2). Two infants (Subjects 17 and 20) had extensive extra-axial bleeding above the tentorium as well as in the PF.

DISCUSSION

Frequency of Cerebellar and Other PF Abnormalities

The overall frequency of PF lesions seen on MR was lower than the incidence reported in other studies using MR¹ and post-mortem examination^3,4,5 but comparable to the findings of Merrill et al.,¹ who found PF lesions in 13/525 neonates using ultrasound via the posterolateral fontanelle. All these studies show some selection bias, since the study populations were sick preterm infants receiving intensive care, often in tertiary referral units. It is therefore difficult to estimate an overall incidence in the preterm and term population. In our study, the infants were highly selected in that they were being treated in a regional NICU that receives tertiary referrals, and many of the infants had abnormal CUS imaging or clinical evidence of an intracranial abnormality (e.g. seizures or abnormal neurological signs). The fact that the incidence of PF abnormalities is low (3.6%) despite this selection bias towards sicker infants is reassuring.

The preterm infants were scanned at a postmenstrual age of 37.5 weeks and it is possible that in this group minor degrees of PF hemorrhage may have resolved spontaneously, but it is most unlikely that larger or more significant lesions would not have still been evident. The optimal time to scan infants using MR is contentious. Most would recommend an early scan 10 to 14 days after the insult has occurred. However, these lesions are often clinically silent and practical considerations of scanning sick, often ventilated infants mean that in some imaging may have to be delayed. Late scanning at 36 weeks or later may be better at predicting subsequent neurodevelopment.¹⁹

CUS versus MR for Imaging the PF

The fact that routine CUS detected the PF abnormality in only 7/19 cases confirms the view that many PF lesions may be missed by conventional CUS via the anterior fontanelle.¹ There is evidence that a posterolateral fontanelle approach improves the sensitivity of ultrasound at detecting PF lesions.^1,2 This approach was not routinely used for CUS during the study period at our institution. Comparative studies of MR and post-mortem histological findings show that even high-resolution MR may not detect some cerebellar injury.²⁰ Amongst our cohort of 20 infants, seven had congenital abnormalities, metabolic or neurodevelopmental disorders associated with brain malformations. The high frequency of these associated abnormalities is not unexpected since cerebellar lesions are often associated with more widespread CNS malformation syndromes and metabolic disorders.^15,21 In this group of infants, MR is invaluable at accurately demonstrating neuroanatomical abnormalities, including characteristic cerebellar appearances associated with several metabolic disorders^14,21 and neuronal migration disorders.

PF Lesions in Term Infants

In the term group, all the acquired PF lesions (n=6) were extra-axial hemorrhages and 4/6 of these infants showed evidence of HIE. In one subject, the extra-axial hemorrhage was associated with other CNS abnormalities and congenital heart disease. The other term infants had PF abnormalities associated with their underlying metabolic or neurodevelopmental disorders. Only one infant had intracerebellar hemorrhage with a rim of blood at the periphery of each hemisphere. The poor outcome among the term infants reflects the frequency of serious congenital abnormality and the high incidence of asphyxia.

PF Lesions in Preterm Infants

In the preterm group, the majority of the PF abnormalities were acquired and were intracerebellar hemorrhages. In the majority, these hemorrhages were focal and unilateral, often towards the periphery (dorsal area) of the cerebellar hemisphere. Similar hemorrhages have been described in six very-low-birth-weight (VLBW) infants, and were not associated with instrumental or traumatic delivery.¹ None of our infants had mask CPAP requiring an occipital band, which has been reported to cause intra-cerebellar hemorrhage.⁵ These peripheral cerebellar hemisphere hemorrhages may represent germinal matrix bleeding within the subpial external granule cell layer, which is particularly prominent from 24 to 30 weeks gestation.¹ Three preterm infants showed extensive destruction of one or both cerebellar hemispheres with secondary atrophy. These changes probably represent extensive old hemorrhage, although infarctions of cerebellar tissue in preterm infants in the distribution of the posterior cerebellar arteries have been reported.⁷ Cerebellar infarction seen on MRI has been associated with later cerebral palsy and visual problems in a cohort of extremely low-birth-weight infants.²² The exact etiology of these hemorrhagic cerebellar infarctions remains unclear.^7,22

Cerebellar hemorrhages destroying part or all of a cerebellar hemisphere were reported in 14% of infants less than 32 weeks at post mortem.⁴ A post-mortem study described two distinct patterns of cerebellar hemorrhage: (1) massive hemorrhage that destroyed more than one-third of the cerebellum and was usually associated with intraventricular hemorrhage and (2) small multiple microscopic and macroscopic punctate cerebellar hemorrhages, which were present in up to 21% of neonatal specimens at post mortem.³ Such multiple punctate hemorrhages were seen in at least one of our cases (Subject 16). Similar punctate lesions have been described in the periventricular white matter and do not seem to be associated with significant adverse outcome, if found in isolation.¹⁸ These punctate lesions that give a low signal on T2-weighted images probably represent small areas of hemorrhage or hemorrhagic infarction. The outcome of infants with punctate hemorrhages isolated to the cerebellum has not been systematically studied.¹⁹ Peripheral cerebellar hemorrhages appear to be clinically silent, often found on routine imaging.¹ Interestingly, six of the 10 preterm infants that we studied had significant hypotension requiring inotropic support. Other studies have described a very high proportion of preterm infants with cerebellar hemorrhage needing cardiac massage or resuscitation with epinephrine.¹ It seems likely that major cardiovascular instability in the perinatal period represent a risk factor for cerebellar hemorrhage.

Limited follow-up in previous studies suggest that these lesions have a good outcome with normal motor development at 13 to 37 months but with some cognitive delay.¹ It is increasingly recognized, however, that the cerebellum has important functions other than the control of voluntary movement. There is increasing clinical and experimental evidence that the cerebellum is involved in cognitive functions and other nonmotor behaviors.²³ In our study, the follow-up data are insufficient to draw any firm conclusions, although in general the outcome in the preterm group was better than in the term group.

CONCLUSION

Cerebellar and other PF lesions may have important neurodevelopmental sequelae. Lesions in the PF are well visualized using MR imaging. We would recommend that MR imaging be considered in the investigation of any infant thought to have a complex brain malformation, metabolic disorder or in whom there is antenatal suspicion of PF abnormality. While CUS is still the most convenient imaging modality in the sick preterm infant receiving intensive care, MR imaging of the PF should be considered in any preterm infant found to have parenchymal lesions elsewhere in the brain and in any infant who has major episodes of cardiovascular instability. Infants found to have cerebellar hemorrhage or infarction will require careful neurological follow-up.

In our cohort, there was a higher incidence of intracerebellar hemorrhage in preterm infants compared with term infants. These intracerebellar hemorrhages tended to be focal, unilateral and were often unsuspected clinically. Further long-term follow-up studies are required to determine the neurodevelopmental significance of these focal cerebellar lesions in preterm infants.

References

1. Merrill JD, Piecuch RE, Fell SC, Barkovich AJ, Goldstein RB. A new pattern of cerebellar hemorrhages in preterm infants. Pediatrics 1998;102:E62.
2. Luna JA, Goldstein RB. Sonographic visualization of neonatal posterior fossa abnormalities through the posterolateral fontanelle. AJR 2000;174:561–567.
3. Martin R, Roessmann U, Fanaroff A. Massive intracerebellar hemorrhage in low-birth-weight infants. J Pediatr 1976;89:290–293.
4. Grunnet ML, Shields WD. Cerebellar hemorrhage in the premature infant. J Pediatr 1976;88:605–608.
5. Pape KE, Armstrong DL, Fitzhardinge PM. Central nervous system pathology associated with mask ventilation in the very low birthweight infant: a new etiology for intracerebellar hemorrhages. Pediatrics 1976;58:473–483.
6. Rutherford MA. Haemorrhagic lesions of the newborn brain. In: Rutherford MA (ed). MRI of the Neonatal Brain. London: W.B. Saunders; 2001. p. 171–200.
7. Mercuri E, He J, Curati WL, Dubowitz LM, Cowan FM, Bydder GM. Cerebellar infarction and atrophy in infants and children with a history of premature birth. Pediatr Radiol 1997;27:139–143.
8. Tuck S, Ment LR. A follow-up study of very low-birthweight infants receiving ventilatory support by face mask. Dev Med Child Neurol 1980;22:633–641.
9. Huang LT, Lui CC. Tentorial hemorrhage associated with vacuum extraction in a newborn. Pediatr Radiol. 1995;25(Suppl 1):S230.
10. Bulas DI, Taylor GA, Fitz CR, Revenis ME, Glass P, Ingram JD. Posterior fossa intracranial hemorrhage in infants treated with extracorporeal membrane oxygenation: sonographic findings AJR AM J Roenterol 1991;156(3):571–575.
11. Serfontein GL, Rom S, Stein S. Posterior fossa subdural hemorrhage in the newborn. Pediatrics 1980;65:40–43.
12. deSouza N, Chaudhuri R, Bingham J, Cox T. MRI in cerebellar hypoplasia. Neuroradiology 1994;36:148–151.
13. Adamsbaum C, Moreau V, Bulteau C, Burstyn J, Lair MF, Kalifa G. Vermian agenesis without posterior fossa cyst. Pediatr Radiol 1994;24:543–546.
14. Antoun H, Villeneuve N, Gelot A, Panisset S, Adamsbaum C. Cerebellar atrophy: an important feature of carbohydrate deficient glycoprotein syndrome type 1. Pediatr Radiol 1999;29:194–198.
15. Patel S, Barkovich AJ. Analysis and classification of cerebellar malformations. Am J Neuroradiol 2002;23:1074–1087. | PubMed | ISI |
16. Childs AM, Ramenghi LA, Evans DJ, et al. MR features of developing periventricular white matter in preterm infants: evidence of glial cell migration. Am J Neuroradiol 1998;19:971–976.
17. Sarnat HB, Sarnat MS. Neonatal encephalopathy following fetal distress. A clinical and electroencephalographic study. Arch Neurol 1976;33:696–705. | PubMed | ISI | ChemPort |
18. 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.
19. Roelants-van Rijn AM, Groenendaal F, Beek FJ, Eken P, van Haastert IC, de Vries LS. Parenchymal brain injury in the preterm infant: comparison of cranial ultrasound, MRI and neurodevelopmental outcome. Neuropediatrics 2001;32(2):80–89. | Article | PubMed | ISI | ChemPort |
20. Felderhoff-Mueser U, Rutherford MA, Squier WV, et al. Relationship between MR imaging and histopathologic findings of the brain in extremely sick preterm infants. Am J Neuroradiol 1999;20:1349–1357.
21. Steinlin M, Blaser S, Boltshauser E. Cerebellar involvement in metabolic disorders: a pattern-recognition approach. Neuroradiology 1998;40:347–354. | Article | PubMed | ISI | ChemPort |
22. Johnsen SD, Tarby TJ, Lewis KS, Bird R, Prenger E. Cerebellar infarction: an unrecognized complication of very low birthweight. J Child Neurol 2002;17:320–324.
23. Boltshauser E. Cerebellar imaging — an important signpost in paediatric neurology. Childs Nerv Syst 2001;17:211–216.

Acknowledgments

We are indebted to all the staff of the neonatal unit and the MR department for their assistance. We are also grateful to all the pediatricians who provided us with follow-up data. We particularly thank the parents of the babies for allowing their infants to be scanned.