Cerebral sinus venous thrombosis in Swiss children

What this paper adds

Cerebral sinus venous thrombosis (CSVT) in children is rare. Based on a Canadian registry, its incidence is 0.67 cases per 100 000 children per year; neonates were the most commonly affected.¹ The risk factors in neonates include maternal conditions, delivery complications, and neonatal comorbidities.^1–4 Older children often use prothrombotic agents such as oral contraceptives or suffer from infections of the head and neck or a chronic systemic illness.^5–8 Prothrombotic states are increased in patients who suffer from CSVT.^1,9 The superficial venous system is most commonly involved.^1,6,10 Parenchymal injuries, such as ischaemic or haemorrhagic infarcts, are often observed in conjunction with CSVT.^1–3,11,12 One-third of term neonates with intraventricular haemorrhage suffer from CSVT.¹³

Two to 10 per cent of the neonates^2,4 and 11 to 17% of the older children died after CSVT^2,6. A recurrence of the thrombosis occurred in 6%.¹⁰ Thirty-eight per cent of the patients exhibited neurological deficits, such as motor impairment, cognitive problems, or epilepsy.¹ The long-term outcome was especially poor in neonates.^1,2,4 In older children who suffered from CSVT, cognitive assessments with standardized and age-normalized tests showed cognitive performance within the normal range,^14,15 but individual impairments were seen in up to 20% of the patients.¹⁴

In the present study, we aimed to describe the incidence, clinical manifestation, neuroimaging findings, risk factors, and treatment of children suffering from CSVT in Switzerland. In addition, we systematically assessed outcomes after CSVT and identified predictors of poor outcome.

Method

Results

Sixty-five children with CSVT were registered in the database (21 neonates [32%]: 14 males, seven females; 44 children [68%]: 30 males, 14 females). In the neonates, the mean age at diagnosis was 9 days (SD 8d, range 0d–27d), mean gestational age was 37 weeks (SD 4wks, range 25–41wks), and the mean birthweight was 2720g (SD 889g, range 770–3960g). Four neonates were born preterm. One neonate had Down syndrome. Another neonate was suspected as having a genetic syndrome; he presented with dysmorphic features, congenital heart disease, and hypoplasia of the vermis cerebelli.

The mean age of the children at the time of diagnosis was 8 years 7 months (SD 4y 5mo; range 9mo to 15y 7mo).

The most common symptoms at presentation in the neonates were seizures (n=12) irritability and/or apathy (n=7) apnoea and/or respiratory distress (n=5) poor feeding (n=5) and tone abnormalities (n=5). In the children, the most common symptoms were headaches (n=29) decreased level of consciousness (n=13) nausea and/or vomiting (n=14) visual impairment (n=9) seizures (n=9) hemiplegia (n=4) and ataxia (n=3). The overall incidence of paediatric CSVT in Switzerland was 0.558 per 100 000 per year in inhabitants aged under 16 years, and the incidence of neonatal CSVT was 0.034 per 1000 live births per year.

The risk factors for neonates and children are summarized in Table I. In the neonates, prenatal risk factors were present in 11, perinatal risk factors were present in 12, and neonatal comorbidities were present in 13. In the children, the most common risk factors were infections of the head and neck, and chronic medical conditions. Four patients were treated with polyethylene glycol (PEG)-asparaginase. The use of oral contraceptives was not documented. Laboratory investigations for prothrombotic risk factors were available for 14 neonates and 40 children. One or more prothrombotic risk factors were found in 6 of the 14 neonates and in 17 of the 40 children.

For the diagnostic evaluation, MRI was performed in 20 neonates and 32 children. Magnetic resonance venography was performed in 7 neonates and 17 children. Four neonates and 31 children underwent CT. All 21 neonates had a power Doppler ultrasound, which detected 48% of the instances of CSVT. The localization of the CSVT differed between neonates and children (Table II). In the neonates, the deep venous system (straight sinus, vein of Galen, internal veins, jugular veins) was more commonly affected (neonates: n=15 children: n=15 p=0.005). Multiple sinus involvement was more common in the neonates (neonates: n=16; children: n=20 p=0.018). In the children, the lateral sinuses were predominantly involved (neonates: n=9 children: n=34 p=0.007). Parenchymal lesions were more common in the neonates (neonates: n=18 children: n=8 p<0.001). The parenchymal injuries in the neonates were haemorrhagic infarctions of the cerebral white matter (n=10) intraventricular haemorrhage (n=8) haemorrhagic lesions of the basal ganglia/the thalamus (n=7) arterial ischaemic stroke (n=2) and subarachnoid haemorrhage (n=1). One child had cerebellar hypoplasia in the context of an underlying dysmorphic syndrome. The parenchymal injuries in the children were haemorrhagic lesions of the cerebral white matter (n=5) haemorrhagic lesions of the basal ganglia/the thalamus (n=2) intraventricular haemorrhage (n=1) and chronic subarachnoid bleeding/hygroma (n=1).

Seven neonates and 38 children received an anticoagulant medication. The therapeutic treatment consisted of standard heparin (20–30U/kg/h) in 2 neonates and in 24 children. Low-molecular-mass heparin was used in 5 neonates and in 12 children (enoxaparine 1.5mg/kg/dose or nadroparine 85 anti-FXa U/kg/dose twice daily). Prophylactic treatment consisted of low-molecular-mass heparin in 7 neonates and in 26 children (enoxaparine 0.75mg/kg/dose twice daily or nadroparine 60 anti-FXa U/kg/dose once daily) as well as oral anticoagulation (phenprocoumone, acenocoumarolum, and danaparoid sodium) in 11 children. The duration was based on clinical outcome and follow-up investigations, and ranged from 1 week to 9 months. One patient with chronic inflammatory bowel disease suffered from significant gastrointestinal bleeding during oral anticoagulation. Otherwise, no serious side effects were reported. In two children no anticoagulant treatment was initiated, but prophylactic treatment with aspirin was started. Follow-up imaging studies were available for 41 patients. Eighteen patients showed complete recanalization, 19 showed partial recanalization, and 4 showed no recanalization.

Two patients died. One neonate who had dysmorphic features and presented with respiratory distress and seizures died a few months after presentation. His MRI revealed multiple CSVTs of the deep and superficial venous system and hypoplasia of the vermis cerebelli. The details of his death are unknown but are probably not related to CSVT. One male infant died during the acute event. He was born preterm and had had intraventricular haemorrhage with parenchymal involvement in the perinatal period; he had epilepsy and presented with sudden loss of consciousness at the age of 3 years 8 months. The MRI showed multiple CSVTs, with involvement of the deep and superficial venous system. In the laboratory investigations, a very low level of protein S was found, and congenital protein S deficiency was suspected.

Fifty-two patients with a follow-up duration of more than 18 months were considered for follow-up assessment. Five patients or parents refused participation and 10 patients could not be contacted. A standardized follow-up assessment was performed in 13 neonates (three were born preterm, one was born at <32wks gestational age; the neonate who had Down syndrome was not included) and in 24 children with a mean follow-up duration of 26 months. In summary, the outcome was considerably worse in the neonates. Five of the 13 neonates were diagnosed with epilepsy whereas epilepsy was not reported in any of the children. For the neurological examination, moderate or marked motor impairment was found in 7 neonates (spastic hemiplegia in four and spastic tetraplegia in three). One child had mild spastic hemiplegia, which was not functionally limiting. Two children developed idiopathic intracranial hypertension.

A developmental/neuropsychological test battery was performed in 32 patients (11 neonates, 21 children; BSID-II in 10, K-ABC in 6, WISC-III in 14, and WAIS-III in 2). The results are shown in Table III. In summary, intellectual disability or developmental delay (IQ or developmental quotient <70) was observed in 8 neonates, but no child exhibited intellectual disability. In the neonates, the mean BSID-II mental index and motor index scores were 1.6 standard deviations below the typical population. In one neonate, no BSID-II was performed, but a K-ABC at 8 years showed intellectual disability (IQ 47). In the children, the mean values for the total scores were within the normal range for all test batteries. There were no significant differences compared with the healthy population in any of the subtests of the WISC-III (data for subtests not shown). More children than expected showed individual cognitive weakness on the WISC-III in one or more subtests. Individual cognitive weakness in one subtest was found in four children. Eight children had individual weaknesses in two subtests, one child had individual weaknesses in three subtests, and one child had individual weaknesses in four subtests.

The results for the predictors of poor outcome are listed in Table IV. The univariate analysis showed a significant association between poor outcome and neonatal age (OR 43.21, 95% CI 1.89–986.80, p=0.007), the presence of parenchymal injuries (OR 20.01, 95% CI 0.82–487.87, p=0.040), and the absence of anticoagulant treatment (OR 12.57, 95% CI 1.16–135.95, p=0.030). In the multivariate analysis neonatal age (OR 17.78, 95% CI 0.85–373.35) and the absence of anticoagulant treatment (OR 7.54, 95% CI 0.58–97.08) remained the strongest predictors of poor outcome. The analysis was performed separately for patients without neurological comorbidities and chronic disease; the results were similar.

Discussion

The aim of the present study was to describe the incidence, manifestation, neuroimaging findings, and risk factors for CSVT in Swiss children younger than 16 years. Furthermore, we aimed to evaluate differences between neonates and older children, to assess the outcome after CSVT, and to identify predictors of outcome. All cases were prospectively collected by the SNSR, a nationwide collaborative project including all Swiss neuropaediatricians. The yearly incidence of CSVT was 0.558 per 100 000 inhabitants under 16 years old. This is somewhat lower than the findings from the Canadian registry, which included patients younger than 18 years.¹

Compared with the Canadian study, we report fewer neonates.¹ In Switzerland, transcranial Doppler ultrasonography is the standard neuroradiological investigation in neonates. CT or MRI is often not routinely performed in sick neonates. Compared with CT/MRI, ultrasonography detected only 48% of the instances of CSVT. Although ultrasonography is a powerful tool for diagnosing neonatal CSVT in the superior sagittal sinus,¹⁷ instances with involvement of the deep venous system might have been missed. We assume that CSVT is underdiagnosed, especially in neonates, and therefore plan to develop evidence-based guidelines for the diagnostic evaluation of CSVT. In contrast to the Canadian study¹, we found significant differences between neonates and children in the location of the CSVT. The deep venous system, particularly the straight sinus, was more commonly involved in neonates than previously reported.^1,4,5,11 We do not have a satisfactory explanation for this finding. The risk factors for CSVT in neonates and children were similar to previous findings.^1–8

Previous studies have assessed the outcome after CSVT with varying time intervals and included children with a very short follow-up. In the present study, only patients with a follow-up duration of 18 months or more were included for the outcome assessment. The outcome was considerably worse in neonates. Moreover, neonatal age at presentation was the strongest predictor of poor outcome. In contrast, children who experienced a CSVT after the neonatal period showed intellectual abilities within the normal range, but a considerable number of the children showed a heterogeneous cognitive profile with partial disturbances of cognitive performance. This might indicate that, despite normal general intellectual functioning, children suffering from CSVT may have attention and perception problems that only become evident under increasing academic demands.

The choice of medication and the length of treatment vary between centres; this was also the case in the present study. Guidelines for the treatment of CSVT have been published only recently.^18,19 In the children who were registered in the SNPSR, anticoagulation therapy was given to 33% of the neonates and 86% of the children. It was not the primary aim to evaluate the efficacy and safety of anticoagulation therapy, although for the study population anticoagulation therapy was associated with a better outcome and few side effects. Randomized controlled studies are needed to analyse the value of anticoagulation therapy in the treatment of paediatric CSVT.

Conclusion

CSVT is rare in the paediatric population. However, its incidence may be underestimated. Clinical features and risk factors differ between neonates and children. The outcome after CSVT is considerably worse in neonates. Despite the relatively good neurological outcome and intellectual functioning in children, some patients may suffer from individual neuropsychological impairments.