Article Text
PDF
Abstract
Objective To describe different patterns of associated brain lesions in preterm and full-term infants with cerebral sinovenous thrombosis (CSVT) and to assess whether these different patterns are related to gestational age at onset.
Design Magnetic resonance scans of all neonates (six preterm, 24 full term) with suspected CSVT, collected over a 7-year period in two neonatal intensive care units, were evaluated to assess patterns of associated brain lesions. Comparisons between the two gestational age groups were made.
Results CSVT was confirmed on magnetic resonance venography in 26 of 30 neonates (six preterm, 20≥36 weeks' gestational age). The straight (85%) and superior sagittal (65%) sinus were most often affected. Several sinuses were involved in 81% of infants. White matter damage affecting the entire periventricular white matter was seen in five of six preterm infants. Intraventricular haemorrhage (IVH) was common in both groups (4/6 preterm, 16/20 full term). Frontal punctate white matter lesions with restricted diffusion (15/20) and thalamic haemorrhage associated with IVH (11/20) were the most frequent lesions in full-term infants. Focal arterial infarction was present in four of 20 full-term infants. Six infants died in the neonatal period (four preterm, two full term). Follow-up MRIs at 3 months in all survivors showed evolution of the lesions with frontal atrophy in 13 of 20 (12 full term) and delayed myelination in seven of 20 (six full term).
Conclusions Preterm and full-term neonates show different patterns of associated brain lesions. Extensive white matter damage is the predominant pattern of injury in the preterm infant, while an IVH associated with a thalamic haemorrhage and punctate white matter lesions are more common in the full-term infant.
Statistics from Altmetric.com
Introduction
Cerebral sinovenous thrombosis (CSVT) in neonates is a rare diagnosis that is increasingly being recognised with better neuroimaging techniques. A recent study showed an incidence of 2.6/100 000 neonates in the Netherlands.1 CSVT is a serious condition with a mortality rate of 2–12% and an adverse outcome in approximately 50%.2 3 Survivors often suffer from motor and cognitive impairments as well as epilepsy.3,–,5 Since presentation is non-specific, imaging techniques are needed to confirm the diagnosis, with MRI or magnetic resonance venography (MRV) being the gold standard.6 7 Over the last decade, there has been great improvement in MRI techniques. Because MRI is more often routinely used in the evaluation of neonates with neurological symptoms, visualisation of the spectrum of associated brain lesions in CSVT has improved.
What is already known on this topic
▶ Cerebral sinovenous thrombosis (CSVT) in neonates is a serious disease with a high risk of an adverse outcome.
▶ Data about CSVT in prematurely born infants are limited.
What this study adds
▶ The pattern of associated lesions in newborn infants with CSVT depends on gestational age. Full-term infants show a pattern of intraventricular haemorrhage with a thalamic haemorrhage and punctate white matter lesions, while preterms show extensive white matter injury.
▶ In a preterm infant presenting with late onset bilateral white matter injury and late onset intraventricular haemorrhage, CSVT should be considered.
Associated brain lesions detected with imaging have been reported in several studies of CSVT in neonates.8,–,11 Wu and colleagues11 confirmed CSVT in 34% of full-term infants with intraventricular haemorrhage (IVH) and suspected CSVT in an additional 19%. Infarction, and mainly haemorrhagic infarction, has been associated with a worse outcome in several studies.2,–,5 Although some studies have focused on imaging in CSVT,7,–,9 there are no studies reporting associated brain lesions in neonatal CSVT in relation to gestational age.
The aim of this study is to report associated brain lesions in newborn infants with CSVT and describe possible correlations with gestational age in a population of neonates from two tertiary neonatal intensive care units (NICUs).
Patients and methods
Patients
Patients with suspected CSVT who were admitted during the neonatal period to the NICU of the Wilhelmina Children's Hospital between March 2002 and June 2010 and to the NICU of the Isala Clinics between January 2006 and September 2009 were studied retrospectively. Both full-term and preterm infants were included. A total of 32 neonates were identified. Two neonates were excluded: one neonate had CSVT with multiple cerebral abnormalities probably due to non-accidental injury, and the other died before MRI could be performed. The remaining 30 were included in this study and 10 of these infants have been reported previously.12
Informed parental consent for MRI and permission for the study were obtained from the authors' medical ethics committee.
Clinical data and follow-up
Clinical data were retrieved from the medical records. All surviving patients were seen at 3, 9 and 18 months in the authors' follow-up clinic and two have now been assessed at 5.5 and 7 years of age. Normal outcome was defined as the absence of cerebral palsy and epilepsy and a normal motor and cognitive development (Griffiths' developmental quotient > 85). The patients were tested during the first year with standardised items from Amiel-Tison and Grenier13 and Touwen14 and the Alberta Infant Motor Scale.15 From 12 months onwards, the Griffiths' developmental scale16 was used to assess neurodevelopmental outcome.
Imaging
All neonates had at least one MRI or MRV within the first week after admission to confirm the diagnosis of CSVT. A definite diagnosis was made when MRV showed clear lack of flow in a sinus at a threshold of 300 and 150 mm/s in combination with signal abnormalities suggesting thrombosis on T1-weighted and T2-weighted images.17 The MR investigations were performed on a 1.5 T ACS-NT system or a 3.0 T whole-body Achieva system (Philips Medical Systems, Best, the Netherlands). At 3 months of age, a follow-up MRI was performed to assess recanalisation.
MRI included T1-weighted sagittal images, T2-weighted and T1-weighted or inversion recovery axial images and diffusion-weighted images (DWIs). Since 2003, two-dimensional phase contrast (2D PC) sagittal images had been added to the protocol, but were used routinely starting in 2006. This 2D PC image was performed for rapid visualisation of the superior sagittal sinus and the straight sinus (repetition time (TR) 11 ms; echo time (TE) 6.6 ms; velocity sensitivity 150–300 mm/s; scan duration 46 s). From 2008 onwards, three-dimensional (3D) MRV became available (TR 18 ms; TE 6.6 ms; voxel size 0.8 mm; PC 150 mm/s; scan time 234 s). From 3D MRV, maximum intensity projections were reconstructed using the system software. Susceptibility-weighted imaging, to improve visualisation of haemorrhagic lesions, became available in 2009. A paediatric MR protocol was used for the infants who had a repeat MRI over 1 year of age, including a fluid-attenuated inversion recovery sequence.
All MRIs were re-evaluated independently by an experienced paediatric radiologist and two experienced neonatologists. No blinding was applied. All associated lesions were recorded and severity was scored. Associated lesions seen on conventional MRI and DWI were defined as haemorrhagic or non-haemorrhagic lesions. A comparison was made between the MRI findings at the time of diagnosis and at 3 months.
Results
Six preterm and 24 full-term neonates were included. Mean gestational ages were 32+6 weeks (range 30+0 to 35+1 weeks) for the preterm and 39+4 weeks (range 36+2 to 42+2 weeks) for the near-term/full-term group. Mean postnatal ages at presentation were 9 days (range 1–28) and 5 days (range 1–19), respectively. CSVT was confirmed on MRI or MRV in 26 neonates. MRV did not give a conclusive diagnosis in four full-term infants with clinical and/or imaging suspicion of CSVT. These infants were excluded from the analysis. MR scans used for diagnosis were carried out at an average age of 4 days after presentation (range 0–9 days). Clinical characteristics are presented in online table S1.
MRV findings
A summary of the sinuses involved is given in table 1. The straight sinus (85%) and the superior sagittal sinus (65%) were most often affected and the majority of the infants had involvement of several sinuses (81%). Involvement of the transverse sinus was often seen in combination with that of other sinuses. One infant showed isolated involvement of the transverse sinus, with absence of flow on MRV and intensity changes of the sinus on T2-weighted imaging. Involvement of the straight sinus and the deep venous system was seen in nine patients. In 13 patients, the superior sagittal sinus and the straight sinus were affected. Four patients had involvement of all major sinuses.
Sinuses affected
Associated imaging findings
Only two patients, who had an isolated thrombosis of the straight sinus, did not show any associated lesions. All other patients had at least one associated lesion (table 2).
Associated lesions (MRI at diagnosis)
Associated lesions in the preterm infant
In five of the six preterm infants, severe white matter lesions affecting the entire periventricular white matter were seen (figure 1). This pattern was seen in only one full-term neonate (p<0.001). Associated IVH was present in four of these five infants on MRI. The onset of this IVH, as assessed by regular cranial ultrasound, was not before the seventh day of life in three infants. Two infants had a thalamic haemorrhage and parietal-occipital infarction was seen in one infant (see online figure S1).
Example of the preterm lesion pattern. The first MRI, performed 6 days after onset of symptoms, shows extensive involvement of the white matter, best seen on diffusion-weighted imaging compared with the T2-weighted image (A,B). On a repeat MRI 8 days later, the T2-weighted spin echo sequence now shows haemorrhagic conversion and cystic evolution (C). There was propagation of the thrombosis in the superior sagittal sinus and straight sinus, already clearly visible on the mid-sagittal T1-weighted image (D). At term-equivalent age, extensive cysts are seen throughout the white matter (E). At 6 months (4 months' corrected age), the cystic lesions have resolved, resulting in an irregular, ex vacuo ventricular dilatation, with sulci abutting the ventricular wall (fluid attenuated inversion recovery (FLAIR) sequence) (F).
Associated lesions in the full-term infant
The most frequently associated finding was an IVH, seen in 16 of the 20 full-term infants. A thalamic haemorrhage was the second most common associated lesion, occurring in 11 of these 16 infants (figure 2). All infants with a thalamic haemorrhage had an associated IVH and the straight sinus was occluded in 10. One infant had three neonatal MRIs showing propagation of the thrombus with progression from unilateral to bilateral thalamic haemorrhages and subsequent development of posthaemorrhagic ventricular dilatation.
Sequential MRI in a full-term infant with thrombosis of the straight sinus, showing a large left-sided thalamic haemorrhage (A) in the neonatal period (day 5) and resolution of the haemorrhage at 3 and 24 months (B,C) with cavitation. White matter loss and delayed myelination of the affected hemisphere becomes more obvious over time. Gliosis is clearly visible on the fluid attenuated inversion recovery (FLAIR) image at 24 months (D). Outcome is moderately abnormal with a global developmental delay and epilepsy at 24 months.
White matter lesions in the full-term infants were mostly punctate white matter lesions (PWMLs) with restricted diffusion at the level of the veins draining the white matter, seen in 15 full-term infants (75%). This injury was predominantly frontal and considerably less extensive than the white matter lesions seen in the preterm infants (figure 3).
Full-term infant with thrombosis of the straight sinus, showing left-sided anterior punctate white matter lesions (T2-weighted spin echo) (A,B), with a dilated left internal cerebral vein. Dilatation of the left anterior caudate vein is seen on the susceptibility weighted imaging (C). High-signal intensity on diffusion-weighted imaging is most marked in the corpus callosum and frontal white matter (D). Loss of flow is seen in the straight sinus on the three-dimensional magnetic resonance venography (E). The T2-weighted spin echo at 3 months of age shows resolution of the lesions (F).
Occipital infarction was present in four neonates and parieto-occipital infarction in one. In two of these infants occipital infarction was bilateral, associated with hypoglycaemia (<2.5 mmol/l) in one. The caudate nucleus was affected in two full-term infants and restricted diffusion included the corpus callosum in three.
Follow-up MRI
Follow-up MRIs at 3 months were available in 19 out of 20 survivors and at 7 months in the remaining one. Later imaging was performed to see whether flow was restored in the sinuses, so low molecular weight heparin (LMWH) could be discontinued and also to assess the residual damage.
In all infants there was a decrease in the extent of the lesions and a restoration of flow in the previously affected sinuses. The most frequent imaging finding was frontal atrophy, in 13 infants (12 full term, one preterm). A delay in myelination was seen in seven (six full-term, one preterm) and early gliotic changes were present in five full-term infants. Atrophy or cavitation at the site of a previous lesion was seen in 10 full-term infants (figure 2). One of the preterm infants, without signs of postnatal infection, showed extensive cystic lesions throughout the white matter. The two infants without any associated lesions on initial imaging also did not have any lesions on follow-up MRI.
Five children had an MRI at an older age (range 1.5–7 years). Gliosis was present at the site of the lesions, which was not yet visible on the MRI performed at 3 months of age (figure 2). Thalamic cavitation was seen in four.
Neurodevelopmental outcome
Six infants died during the neonatal period. Two full-term infants died, one due to severe encephalopathy (see online figure S2) and the other due to late complications of severe mesenteric thrombosis. Four preterm infants died during the neonatal period after withdrawal of intensive care treatment due to severe intracranial lesions.
Early outcome was normal in 14 of the 18 survivors born at term, 12 of whom were treated with LMWH for 3 months according to the authors' previously described protocol.12 Of the other four full-term infants, also treated with LMWH, three showed a global developmental delay, including the child with bilateral thalamic haemorrhages and posthaemorrhagic ventricular dilatation. He required insertion of a ventriculo-peritoneal shunt and developed cerebral visual impairment. Two of these four infants developed epilepsy and required antiepileptic medication.
Both preterm infants who survived have severe psychomotor retardation. One died at the age of 3 years due to superior vena cava syndrome caused by massive thrombosis of a central line used for parenteral nutrition for severe feeding problems associated with several chromosomal abnormalities. The other child developed cerebral palsy and therapy-resistant epilepsy.
Discussion
A different spectrum of associated brain lesions could be recognised in a large group of neonates with CSVT, depending on their gestational age at the time of developing CSVT. Extensive white matter lesions throughout the periventricular white matter were the predominant pattern of injury associated with CSVT in preterm infants, while an IVH associated with a predominantly unilateral thalamic haemorrhage and PWMLs were most often seen in the full-term infant. Almost all infants had associated lesions.
While the association of CSVT with IVH and unilateral thalamic haemorrhage or venous infarction has been reported, data on CSVT and associated white matter injury in the preterm infant are scarce.18 The present data do suggest that CSVT should be considered in preterm infants with late onset IVH followed by cystic periventricular leukomalacia (c-PVL). Without an MRI with MRV studies shortly after the development of late onset IVH, a subsequent diagnosis of c-PVL may be made without considering earlier CSVT as the underlying aetiology.
An explanation for the spectrum of associated brain lesions, with differences in gestational age, may be the difference in vulnerability at different stages of maturation. Volpe recently postulated that the rapid development of the brain during the preterm period makes it especially vulnerable to exogenous and endogenous insults. As a result of the combination of destructive and developmental disturbances, c-PVL may occur.19
In full-term infants, there was a correlation between involvement of the straight sinus and a mostly unilateral thalamic haemorrhage, which can be understood given the anatomic location of the straight sinus. As described previously, there tends to be a good correlation between the occluded sinuses and the location of the lesions in neonates with CSVT.9
An IVH was present in the majority of both preterm and full-term infants. This is a common finding in CSVT.11 An IVH was present in all neonates with a thalamic haemorrhage. It is likely that pressure builds up proximal to the occluded vein and then first gives rise to an IVH. When the reduction in pressure is not enough, the thin walls of the thalamic veins will also rupture. An IVH with or without a thalamic haemorrhage was often associated with PWMLs, predominantly involving the frontal white matter. The PWMLs are seen as areas of increased signal intensity on DWI and are most likely due to impaired venous drainage of the medullary veins following occlusion of the draining sinus.18 A repeat MRI at 3 months no longer showed these PWMLs, but atrophy of the frontal white matter was often seen as a late sequel.
The straight sinus was the most common site of thrombosis in the neonates. This is in contrast to other studies, reporting an occlusion of the deep venous system in only about one third of their patients.2 4 18 There is no clear explanation for this difference. Perinatal factors, postnatal age at diagnosis and underlying prothrombotic factors in the population are similar to those reported by others, but diagnosis in local hospitals and referral patterns may differ.
Occipital focal infarction was found in about 20% of infants (one preterm, four full term), with cystic evolution or focal atrophy on follow-up MRI at 3 months. Occipital infarction is also often reported to be associated with symptomatic neonatal hypoglycaemia.20 At least one infant with bilateral occipital infarction experienced hypoglycaemia (<2.5 mmol/l) as well as CSVT. It is unlikely that other perinatal factors, such as asphyxia or assisted vaginal delivery, contributed to the associated brain lesions in the infants.
Previous studies on CSVT reported associated lesions in 19–70% of cases.2 4 9 11 The authors found associated lesions in all but two patients (92%). This difference may be explained by the diversity in study objectives and the lack of reporting associated lesions. Another explanation may be that only more severely affected infants with CSVT, presenting with encephalopathy and/or neonatal seizures, were referred to the authors' level three units. Finally it is possible that newer MR techniques, such as DWI and susceptibility-weighted imaging, provide more detailed information about associated brain lesions. The type of lesions among the studies however was comparable.4 8 10 11
All initial imaging was done within 9 days after first presentation, with MRV as part of the protocol. It is important to perform appropriate imaging as soon as possible in infants suspected of having CSVT because they may benefit from anticoagulation, reducing the risk of propagation of the thrombosis.21
A limitation of this study was the late introduction of 3D MRV. The authors therefore cannot exclude that infants may have been missed prior to the introduction of 3D MRV, but most likely only infants with an occlusion of the internal cerebral veins, because those with a superior sagittal or straight sinus thrombosis would have been identified with a 2D PC image. It is also possible that the authors missed the diagnosis in a few preterm infants with c-PVL born during the first years of the study period, before 2D PCA became part of the imaging protocol.
All surviving neonates had a repeat MRI at 3 months showing recanalisation of the affected sinus, as was recently also shown by Moharir et al.22 While the authors and others recommend a repeat MRI or MRV to monitor recanalisation, the exact timeframe differs; others have suggested 6 weeks instead of 3 months.23 Evidence of gliotic changes was only clearly seen in the six patients who had an MRI beyond the first year of life.
In four full-term infants, CSVT could not be confirmed on imaging. Their associated brain lesions, however, did follow a pattern similar to that of the neonates with confirmed CSVT. Two neonates had a thalamic haemorrhage; one presented with extensive bilateral white matter lesions and one with haemorrhagic PVL. It is possible that these neonates had rapid recanalisation and that their thrombosis was missed.
In conclusion, different patterns of associated lesions can be recognised in newborn infants with CSVT, depending on their gestational age at the time of developing the CSVT. The predominant preterm pattern associated with CSVT is extensive white matter injury, while in the full-term infant an IVH associated with a thalamic haemorrhage and PWML is most often seen. CSVT should be considered in the preterm infant, presenting with late onset bilateral white matter injury, associated with an IVH.
Acknowledgments
The authors would like to thank the MRI technicians for their support, and Dr C Koopman-Esseboom and Drs I C van Haastert for performing follow-up assessments.
References
Supplementary materials
Web Only Data adc.2010.201129
Files in this Data Supplement:
Footnotes
-
Competing interests None.
-
Ethics approval Informed parental consent for MRI and permission for the study were obtained from the authors' medical ethics committee.
-
Provenance and peer review Not commissioned; externally peer reviewed.