Cerebral sinovenous thrombosis (CSVT) occurrence is rare in children, with an incidence of about 0.67 cases per 100 000 children per year. This is probably underestimated as diagnosis is often delayed or missed.^{1, 2} It is a potentially severe condition, as children with CSVT are at risk for complications in the acute phase, such as intracranial hypertension, seizures, ischemic and/or haemorrhagic venous stroke, and for long-term sequelae, such as visual impairment, and motor or cognitive disabilities.¹ Despite recent advances, diagnostic and therapeutic strategies for CSVT in children still suffer from lack of evidence, and consequent heterogeneity.^1-3 Paediatric management recommendations are mostly adapted from adult literature.^{3, 4} Treatment in adults relies on anticoagulation therapy, initially with unfractionated heparin or low-molecular-weight heparin, followed by vitamin K antagonists or direct oral anticoagulants, regardless of the presence of intracranial haemorrhage if secondary to the thrombosis.² Therapeutic studies are lacking in paediatric CSVT, especially randomized controlled trials. Despite concerns about age-related differences in the coagulation and neurological systems, a consensus has formed in favour of anticoagulation therapy in older children (i.e. excluding neonates) with CSVT.^{3, 5} A 3- to 6-month treatment has been proposed as the standard duration in most studies,³ and may be adapted to each situation and risk factor. While CSVT symptoms are quite similar in children and adults,⁶ the clinical context in which they appear may be different. Infection-related CSVT, mainly in the head and neck (ear, sinus, mouth, face, and neck), explains only 8.2% of adult cases, while it is much more common in paediatric cases (23%–49%).^{1, 2, 7-9} Although data from a few studies suggest a ‘less severe’ condition, with a lower risk of spontaneous recurrence,^{3, 10, 11} paediatric CSVT occurring in an infectious setting has not been compared to CSVT occurring in other settings. The present study focused on a comparison between paediatric patients with CSVT with and without head/neck infection to identify potential adaptations of management between groups.

METHOD

RESULTS

A total of 84 consecutive patients (36 patients in centre A, 48 in centre B) verified the inclusion criteria. There were 46 males and 38 females (sex ratio 1.2), with a median age of 4 years 6 months (range 3 months–17 years 5 months). Most patients (n = 80, 95.2%) had an identified aetiology of their CSVT episode. An associated head/neck infection was found in 65.4% of all cases (INF+: n = 55), a similar proportion in both centres (72.2% vs 60.4%, non-significant), and represented the main identified CSVT aetiology. Infections were mainly otogenic (oto-mastoiditis n = 36, isolated otitis media n = 1). Other infections were meningitis (n = 10), sinusitis (n = 5), cervical adenophlegmon or parapharyngeal abscess (n = 4), and facial cellulitis (n = 1). Only 53% of patients with mastoiditis had bacterial identification, the most frequent germ being fusobacterium. Potential risk factors for CSVT were identified in almost every patient without head/neck infection (n = 27/29): 44.8% had a chronic or subacute prothrombotic condition, 27.5% had loco-regional non-infectious conditions, and 20.6% had a transient or acquired prothrombotic risk factor. Chronic or subacute prothrombotic condition included nephrotic syndrome (n = 4), Behcet disease (n = 4), systemic vasculitis (n = 1), sickle cell disease (n = 2), lupus associated with antiphospholipid syndrome and nephrotic syndrome (n = 1), and lymphoblastic acute leukaemia with L-asparaginase treatment (n = 1). Loco-regional non-infectious conditions included central nervous system tumour (n = 2), head trauma (n = 4), cerebrovascular malformation (n = 1), and acute disseminated encephalomyelitis (n = 1). Transient or acquired prothrombotic risk factors were anaemia (n = 2; iron deficiency and inflammatory anaemia), dehydration (n = 2), disseminated intravascular coagulation related to sepsis (n = 1), and recently introduced combined oestrogen/progestin oral contraception (n = 1). Of note, one patient with mastoiditis who also had sickle cell anaemia and one patient with meningitis occurring after a traumatic skull fracture were classified in the INF+ group as the infection was considered as the final trigger for CSVT.

Comparison between the INF+ and INF– groups

The INF+ group were significantly younger (median age 3 years 8 months vs 10 years 7 months old, p < 0.001), and more frequently had fever at presentation (96.4% vs 10.3%, p < 0.001) than the INF– group (Table 1). Patients in the INF+ group displayed a milder clinical presentation (no CSVT-related neurological signs in 30.9% vs 6.9%, p = 0.014) than the INF– group. They also had less frequent symptoms of intracranial hypertension (21.8% vs 64.3%, p < 0.001), or associated intracranial haemorrhage (1.8% vs 24.1%, p = 0.002). Diagnostic CT venography was performed in 82% of patients, magnetic resonance venography in 40%, and both modalities in 27%. CSVT associated with head/neck infection was less frequently extensive (27.3% vs 65.5%, p = 0.001) (Table 2, Figure 1). No patient required endovascular intervention. Anticoagulation treatment was comparable in the acute phase, with 81 out of 84 patients treated by unfractionated heparin or low-molecular-weight heparin. Of note, 76% of the INF+ patients (data available only for centre B) had surgical intervention; anticoagulation treatment was started shortly after surgery in these cases. Bridging to vitamin K antagonists was less frequent in the INF+ group (32.7% vs 58.6%, p = 0.034). Median anticoagulation treatment duration tended to be shorter in the INF+ group (3 vs 4 months, p = 0.08, non-significant). Most patients had a thrombophilia biological screening at the acute phase (78% and 96% in the INF+ and INF– groups respectively, non-significant). This workup was less frequently positive in the INF+ group (32% vs 44%, non-significant). In seven patients from the INF+ group with a positive antiphospholipid antibody result, a control test was performed several months later, which showed normalization of this parameter in all patients.

Table 1. Comparison between children with cerebral sinovenous thrombosis (CSVT) and head/neck infection (INF+) and no head/neck infection (INF–)

	INF+	INF–	p
n	55	29
Median age, years:months (IQR)	3:8 (1:11–6:7)	10:7 (6:4–14:2)	<0.001
Males	32 (58.2)	14 (48.3)	0.490
Acute phase
Fever	53 (96.4)	3 (10.3)	<0.001
Intracranial hypertension	12 (21.8)	18 (64.3)	<0.001
CSVT-related neurological symptoms	38 (69.1)	27 (93.1)	0.014
Associated intracranial haemorrhage	1 (1.8)	7 (24.1)	0.002
Associated ischaemic lesion	1 (1.8)	2 (6.9)	0.273
Outcomes
Median follow-up, months (IQR)	7 (6–24)	30 (12–51.5)	0.002
Median time to complete recanalization, days (IQR)	89 (35–101)	112 (83–177)	0.005
Complete recanalization at 3 months	29 (65.9)	8 (36.4)	0.035
Complete recanalization at end of follow-up	44 (86.3)	21 (77.8)	0.355
Anticoagulant treatment	52 (96.3)	29 (100.0)	0.540
Median anticoagulant treatment duration, months (IQR)	3 (3–4)	4 (3–6)	0.080
mRS = 0 in survivors	45/52 (86.5)	15/26 (57.7)	0.009

• Data are n (%) unless otherwise stated. Bold type indicates statistical significance. Abbreviations: IQR, interquartile range; mRS: modified Rankin Score.

Table 2. Cerebral sinovenous thrombosis (CSVT) location: comparison between children with CSVT and head/neck infection (INF+) and without regional infection (INF–)

	INF+	INF–	p
n	55	29
Affected sinus
Superficial venous system CSVT
Transverse sinus	37 (67.3)	21 (75.0)	0.614
Sigmoid sinus	8 (14.5)	1 (3.6)	0.261
Internal jugular vein	17 (30.9)	9 (32.1)	1.000
Superior sagittal sinus	7 (12.7)	13 (46.4)	0.001
Deep venous system CSVT
Inferior sagittal sinus	0 (0)	0 (0)	-
Straight sinus	0 (0)	0 (0)	-
Cavernous sinus	3 (5.5)	0 (0)	0.547
Petrosal sinus	0 (0)	0 (0)	-
Extensive CSVT
CSVT affecting several venous segments	15 (27.3)	19 (65.5)	0.001

• Data are n (%) unless otherwise stated. Bold type indicates statistical significance.

Outcomes

Two patients in the INF+ group died because of a bacterial meningitis complicated by massive intracranial hypertension. Median follow-up duration for survivors was 12 months. Median time to complete recanalization was significantly shorter in the INF+ group (89 days vs 112.5 days, p = 0.005). Taking into account censored patients and different follow-up durations, the difference in the time to complete recanalization remained significant (p = 0.031, Figure 2). The rate of complete recanalization was significantly higher in the INF+ group at 3 months (65.9% vs 36.4%, p = 0.035) but not significantly different at the end of follow-up (86.3% vs 77.8%, p = 0.35). No heparin-induced thrombocytopenia or any symptomatic haemorrhage during anticoagulant treatment was reported. Three patients in the INF– group experienced a CSVT relapse during follow-up; two had Behcet disease and one had no identified condition. Clinical outcome in survivors was significantly better in the INF+ group, with an excellent outcome (modified Rankin Scale = 0) at last visit in 86.5% versus 57.7% in the INF– group (p = 0.009).

Patients with CSVT associated with otogenic infection and no neurological sign

Among the INF+ group, a subgroup of patients with otogenic infection (mastoiditis with or without otitis) without any neurological sign at the acute phase was specified. It represented 13 out of 36 otogenic infections. Children in this subgroup displayed the above-mentioned specific features in an even more pronounced way: they were significantly younger than the other patients with CSVT (median age 2 years 5 months vs 6 years 4 months, p = 0.003) and had an even shorter time to complete recanalization compared to the INF– group and the rest of the INF+ group (median 64 days vs 90 days, p = 0.026) (Figure 3).

DISCUSSION

This study compared children with CSVT with and without concurrent head/neck infection. The children in the INF– group resembled adults in terms of aetiology, clinical presentation, and evolution under anticoagulant treatment. Thus, current paediatric recommendations for CSVT, which are largely adapted from adult guidelines, seem appropriate for this subgroup. On the contrary, our study delineated a specific subpopulation of children with regional infection (INF+) with clinical, radiological, and outcomes specificities, for whom slightly different management strategies could be considered.

This study included one of the largest samples of patients with paediatric CSVT (n = 84) over a limited period in only two centres. Patients with concurrent head/neck infection outnumbered previous reports, which allowed a comprehensive study of this subpopulation and its comparison with paediatric CSVT from other causes. In our sample, most CSVTs (95.2%) were related to an underlying condition (previously diagnosed or revealed by the CSVT) or a precipitating event, as previously reported (95% in Dlamini et al.).¹³ Associated head/neck infection was the most frequent aetiology (65.5%), with a higher rate than reported in the years 2001 to 2005 (23%–49%).^{1, 6} Increased diagnosis of asymptomatic CSVT since the implementation of systematic CT imaging in mastoiditis workup^{14, 15} may account for this difference.

Children with CSVT associated with head/neck infection were younger and had milder clinical and radiological presentations than children with other causes. They also had better outcomes. In the literature, reported outcome in this setting is very heterogeneous, with variable measurement tools. A prevalence of 25% to 74% of long-term sequelae is reported,¹⁶ usually associated with a severe initial presentation and/or a persisting risk factor for recurrence.¹ In our study, the prevalence of sequelae (modified Rankin Scale ≥1) was 23.1% in the whole sample, and significantly lower in the INF+ group, although two patients from this group had early fatal outcome related to severe bacterial meningitis. These results were more significant in the subgroup of children with CSVT associated with otogenic infection without neurological symptoms, in which all patients had a modified Rankin Scale score of 0 at the end of their follow-up.

Interestingly, children with CSVT associated with a head/neck infection displayed a significantly higher rate of venous recanalization at 3 months follow-up and a global quicker recanalization. In guidelines, assessing recanalization to support the decision of anticoagulant treatment weaning is often recommended at 3 months post-CSVT, and at 6 months if recanalization was not complete at the first control.³ In our study, CSVT management focused on these recommendations as most patients had control imaging at these predetermined delays. Actual recanalization delays may be shorter: in half of the patients with otogenic infection and no neurological symptom recanalization was complete 65 days after CSVT onset (Figure 2), that is, nearly 1 month before first recommended imaging control. One could suggest reimaging earlier in this setting, for example at 6 to 8 weeks from onset, and stopping anticoagulant treatment if recanalization was obtained and the initial infection cured. To our knowledge, some centres have already implemented this practical management, but no formal report or comparative study is available so far to estimate its efficiency (early patency diagnosis allowing shorter anticoagulation treatment vs no patency leading to supplemental reimaging). The role of anticoagulation treatment in the recanalization rate has been debated.^{3, 17} In this study most children in both groups had anticoagulation treatment – as recommended in Europe³ – suggesting that the observed difference in recanalization delay instead relates to the cause of CSVT. In our patients, anticoagulant therapy was well tolerated and no major treatment-related complication was reported, as suggested in previous paediatric reports.³

Concerning thrombophilia workup, we observed large variations in assessed parameters according to centres and over time. Globally, it was positive in 37% of patients, with a lower prevalence in the INF+ group. Previous studies reported positive thrombophilia screening in childhood CSVT in up to 60% of cases.¹⁶ It is currently unknown if these abnormalities are either causal, causal factors in association with other risk factors, or if they may be epiphenomena.^{16, 18} In mastoiditis complicated by CSVT, Coudert et al.¹⁹ reported a significant predominance of Fusobacterium necrophorum, found in 53.8% in this setting. Similarly, in our study, Fusobacterium necrophorum was the most prevalent germ, found in 57% of patients with an identified germ. Indeed, Fusobacterium necrophorum is known to particularly induce an inflammatory state, which becomes a prothrombotic state.^{20, 21} In our study, a late thrombophilia workup in a subgroup of patients with mastoiditis and positive initial workup showed the normalization of antiphospholipid antibody dosage after the infection episode, pointing to a risk of false-positive results at the acute phase of an infection. These findings suggest that the need for prolonged treatment may be infrequent in CSVT associated with head/neck infection and that thrombophilia workup could be delayed to 1 to 3 months after CSVT onset to avoid false-positives related to the acute inflammatory phase.

As mentioned above, specificities of children with CSVT and concurrent head/neck infection were even more pronounced in the subgroup of children with otogenic infection without neurological symptoms. As the number of patients in this subgroup was quite small (n = 13), we could not carry out further statistical analyses, but these results were consistent with a previous reported study on otogenic paediatric CSVT.¹⁰ These findings support the suggestion to adapt current recommendations concerning CSVT management in children, with earlier control imaging and delayed thrombophilia workup.

The major limitation of our study is its retrospective nature, with associated intrinsic biases. Participating centres were academic tertiary centres with highly specialized paediatric departments, including paediatric ear nose and throat and paediatric neurosurgery departments, which may have modified the repartition of CSVT causes. To evaluate outcomes, limited available data precluded the use of the dedicated Pediatric Stroke Outcome Measure scale, and we used the modified Rankin Scale adjusted for children, although it is less precise in this setting. Another limitation is related to anticoagulant treatment. In this sample, nearly all patients were treated. The above-mentioned suggestions for adapted management of CVST with regional infection are thus supported by data from treated patients only. This study does not address the issue of the anticoagulant treatment indication in paediatric otogenic CSVT, which is debated in the literature. Guidelines usually recommend anticoagulant treatment for non-neonatal paediatric CSVT,^{3, 16, 22} with a suggested individualized approach for some circumstances, including otogenic infection.¹⁶ Some authors suggest that adequate infection treatment (associating antibiotics and local surgery) may be sufficient,^{17, 23} whereas others suggest that anticoagulant treatment may prevent CSVT extension during the inflammatory prothrombotic acute phase. The tolerance of anticoagulant treatments in children is satisfactory in this setting, including for recent drugs.^{22, 24} But compliance may be difficult to obtain in case of repeated injections (low-molecular-weight heparin) or lab tests (vitamin K antagonists). Thus, alleviation of anticoagulant treatment burden, either by modifying galenic forms or by reducing treatment duration, remains an important goal to improve the patients' quality of life. Additional studies are encouraged to verify the relevance of the suggested measures.