Diagnosis and Management of Cerebral Venous Thrombosis: A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association

Predisposing causes of CVT are multiple. The risk factors for venous thrombosis in general are linked classically to the Virchow triad of stasis of the blood, changes in the vessel wall, and changes in the composition of the blood. Risk factors are usually divided into acquired risks (eg, surgery, trauma, pregnancy, puerperium, antiphospholipid syndrome, cancer, exogenous hormones) and genetic risks (inherited thrombophilia).

Table 3 summarizes the evidence for a cause-and-effect relationship^10,11 between prothrombotic factors and CVT.^12–55 Evidence for the strength and consistency of association, biological plausibility, and temporality is summarized. These criteria are most closely met for deficiency of antithrombin III, protein C, and protein S; factor V Leiden positivity; use of oral contraceptives; and hyperhomocysteinemia, among others.

The most widely studied risk factors for CVT include prothrombotic conditions. The largest study, the ISCVT, is a multinational, multicenter, prospective observational study with 624 patients. Thirty-four percent of these patients had an inherited or acquired prothrombotic condition.¹⁰ The prevalence of different prothrombotic conditions is summarized in Table 3. Recently, another group in the United States reported that 21% of 182 CVT case subjects in 10 hospitals had a prothrombotic condition.¹¹

Two studies have analyzed the role of natural anticoagulant protein deficiencies (antithrombin III, protein C, and protein S) as risk factors for CVT. One study compared 121 patients with a first CVT with 242 healthy control subjects.³⁶ The other study compared 51 patients with CVT with 120 healthy control subjects.¹² Only 1 patient (2%) had antithrombin III deficiency. The combined odds ratio (OR) of CVT when these 2 studies were combined was 11.1 for protein C deficiency (95% confidence interval [CI] 1.87 to 66.05; P=0.009) and 12.5 for protein S deficiency (95% CI 1.45 to 107.29; P=0.03).

The first study mentioned above found a higher prevalence of antiphospholipid antibodies in patients with CVT (9 of 121) than in control subjects (0 of 242).³⁶ In another study from India with 31 CVT patients, anticardiolipin antibodies were detected in 22.6% of CVT patients compared with 3.2% of normal control subjects.¹² Similar findings (5.9%) were observed in the ISCVT study.¹⁰

Resistance to activated protein C is mainly caused by the presence of the factor V Leiden gene mutation, which is a common inherited thrombophilic disorder. A recent meta-analysis of 13 studies, including 469 CVT cases and 3023 control subjects,²⁸ reported a pooled OR of CVT of 3.38 (95% CI 2.27 to 5.05) for factor V Leiden, which is similar to its association with venous thromboembolism (VTE) in general.²⁸

The prothrombin G20210A mutation is present in ≈2% of whites and causes a slight elevation of prothrombin level.^55,56A meta-analysis of 9 studies,³⁸ including 360 CVT patients and 2688 control subjects, reported a pooled OR of CVT of 9.27 (95% CI 5.85 to 14.67) for this mutation,²⁸ which is stronger than its association with VTE in general.

Hyperhomocysteinemia is a risk factor for deep vein thrombosis (DVT) and stroke but has not been clearly associated with an increased risk of CVT. Five case-control studies evaluated hyperhomocysteinemia in patients with CVT.^{13,16,17,29,30} Researchers from Milan¹³ reported on 121 patients with a first CVT and 242 control subjects, finding hyperhomocysteinemia in 33 patients (27%) and 20 control subjects (8%; OR 4.2, 95% CI 2.3 to 7.6). Low levels of serum folate and the 677TT methylenetetrahydrofolate reductase genotype were not associated with CVT risk, independent of homocysteine level.¹³

A study of 45 patients with CVT and 90 control subjects in Mexico reported an adjusted OR of CVT of 4.6 (95% CI 1.6 to 12.8) associated with high fasting homocysteine and an OR of 3.5 (95% CI 1.2 to 10.0) associated with low folate.²⁹ A small Italian study of 26 consecutive patients with CVT and 100 healthy control subjects reported that 38.5% of case subjects and 13% of control subjects had hyperhomocysteinemia (OR 4.2, 95% CI 1.6 to 11.2).¹⁶ No significant differences were found in the prevalence of prothrombin or methylenetetrahydrofolate reductase mutation. No factor V Leiden mutation was found. Another Italian group¹⁷ found a strong and significant association of the prothrombin G20210A mutation (30% versus 2.5% in patients versus control subjects, respectively, P=0.001; OR 16.2, P=0.002) and hyperhomocysteinemia (43.3% versus 10%, P=0.002; OR 6.9, P=0.002).

Pregnancy and the puerperium are common causes of transient prothrombotic states.⁵⁷ Approximately 2% of pregnancy-associated strokes are attributable to CVT.³¹ The frequency of CVT in the puerperium is estimated at 12 cases per 100 000 deliveries, only slightly lower than puerperal arterial stroke.⁵⁸

In a study from Mexico, ≈50% of CVT occurred during pregnancy or puerperium.³² Most pregnancy-related CVT occurs in the third trimester or puerperium. Seven of 8 CVTs among 50 700 admissions for delivery in Canada occurred postpartum.³³ During pregnancy and for 6 to 8 weeks after birth, women are at increased risk of venous thromboembolic events.³⁴ Pregnancy induces several prothrombotic changes in the coagulation system that persist at least during early puerperium. Hypercoagulability worsens after delivery as a result of volume depletion and trauma. During the puerperium, additional risk factors include infection and instrumental delivery or cesarean section. One study reported that the risk of peripartum CVT increased with increasing maternal age, increasing hospital size, and cesarean delivery, as well as in the presence of hypertension, infections, and excessive vomiting in pregnancy.³⁵ Recently, it was reported that in pregnant women, hyperhomocysteinemia was associated with increased risk of puerperal CVT (OR 10.8, 95% CI 4.0 to 29.4) in a study of 60 case subjects and 64 control subjects.³⁰

A 1998 study compared the prevalence of several risk factors, including use of oral contraceptives, among 40 female patients with CVT, 80 female patients with DVT of the lower extremities, and 120 female control subjects.³⁶ Nearly all CVT case subjects were using oral contraceptives (96%), which conferred 22.1-fold increased odds of CVT (95% CI 5.9 to 84.2). The OR for women with the prothrombin G20210A mutation who used oral contraceptives was 149.3 (95% CI 31.0 to 711.0) compared with those with neither characteristic. Stratification for the presence of factor V Leiden or prothrombin mutation and the use of oral contraceptives showed similar point estimates for the coagulation abnormalities alone and the use of oral contraceptives alone, whereas the presence of both risk factors gave an OR of 30.0 (95% CI 3.4 to 263.0) for factor V Leiden and 79.3 (95% CI 10.0 to 629.4) for the prothrombin mutation. A study in the Netherlands³⁷ found that of 40 female CVT patients, 85% used oral contraceptives, with an adjusted OR of 13 (95% CI 5 to 37). The combination of oral contraceptives with a prothrombotic condition also dramatically increased the risk of CVT. A study from Brazil showed similar results.¹⁸ In a meta-analysis that included 16 studies, the authors reported an increased risk of CVT in oral contraceptive users (relative risk 15.9, 95% CI 6.98 to 36.2).⁵⁹ In another meta-analysis of 17 studies,²⁸ an increased risk of CVT was found in patients who used oral contraceptives (OR 5.59, 95% CI 3.95 to 7.91; P<0.001). It is clear that the use of oral contraceptives is associated with an increased risk of CVT, that the great majority of younger nonpregnant women with CVT are oral contraceptive users, and that the risk of CVT with oral contraceptive use in women is greater among those with a hereditary prothrombotic factor.

In the ISCVT,¹⁰ 7.4% of cases of CVT were associated with cancer. It has been speculated that CVT could be more frequent in cancer patients, particularly in patients with hematologic malignancies; however, there are no studies with a control group. Potential mechanisms for an association of cancer with CVT include direct tumor compression, tumor invasion of cerebral sinuses,^39–41 or the hypercoagulable state associated with cancer.⁶⁰ Chemotherapeutic and hormonal agents used for cancer treatment may also play a role.

New neuroimaging procedures have increased the ability to detect CVT in recent years and have also helped to identify other potential causes, including infections, mainly in parameningeal locations (ear, sinus, mouth, face, and neck). These causes only explained 8.2% of all cases in the ISCVT series.² In contrast, CVT caused by infection is more common in children. In a recent series of 70 children with CVT in the United States, 40% had infection-related CVT.¹⁶ Conversely, a French study of 62 adults with isolated lateral sinus thrombosis found that only 3 cases were related to parameningeal infections.⁴²

Other conditions have been associated with CVT in case reports or small series, including paroxysmal nocturnal hemoglobinuria,⁴⁸ iron deficiency anemia,⁴⁹ thrombocythemia,⁵⁰ heparin-induced thrombocytopenia,⁶¹ thrombotic thrombocytopenic purpura,¹⁴ nephrotic syndrome,⁵¹ inflammatory bowel disease,^10,62 systemic lupus erythematosus,⁵² Behçcet disease,⁵³ mechanical precipitants, epidural blood patch,⁴⁵ spontaneous intracranial hypotension,⁴⁶ and lumbar puncture.⁴⁷

The diagnosis of CVT is typically based on clinical suspicion and imaging confirmation. Clinical findings in CVT usually fall into 2 major categories, depending on the mechanism of neurological dysfunction: (1) Those that are related to increased intracranial pressure attributable to impaired venous drainage and (2) those related to focal brain injury from venous ischemia/infarction or hemorrhage. In practice, many patients have clinical findings due to both mechanisms, either at presentation or with progression of the underlying disease. Headache, generally indicative of an increase in intracranial pressure, is the most common symptom in CVT and was present in nearly 90% of patients in the ISCVT.¹⁰ Similar headache frequency has been reported in other populations studied.⁶³ The headache of CVT is typically described as diffuse and often progresses in severity over days to weeks. A minority of patients may present with thunderclap headache, suggestive of subarachnoid hemorrhage, and a migrainous type of headache has been described.⁶⁴ Isolated headache without focal neurological findings or papilledema occurs in up to 25% of patients with CVT and presents a significant diagnostic challenge.⁶⁵ CVT is an important diagnostic consideration in patients with headache and papilledema or diplopia (caused by sixth nerve palsy) even without other neurological focal signs suggestive of idiopathic intracranial hypertension. When focal brain injury occurs because of venous ischemia or hemorrhage, neurological signs and symptoms referable to the affected region are often present; most common are hemiparesis and aphasia, but other cortical signs and sensory symptoms may occur. Psychosis, in conjunction with focal neurological signs, has also been reported.⁶⁶

Clinical manifestations of CVT may also depend on the location of the thrombosis (Figure 2). The superior sagittal sinus is most commonly involved, which may lead to headache, increased intracranial pressure, and papilledema.⁶⁷ A motor deficit, sometimes with seizures, can also occur. Scalp edema and dilated scalp veins may be seen on examination.⁶⁸ For lateral sinus thromboses, symptoms related to an underlying condition (middle ear infection) may be noted, including constitutional symptoms, fever, and ear discharge. Pain in the ear or mastoid region and headache are typical. On examination, increased intracranial pressure and distention of the scalp veins may be noted. Hemianopia, contralateral weakness, and aphasia may sometimes be seen owing to cortical involvement.⁶⁹ Approximately 16% of patients with CVT have thrombosis of the deep cerebral venous system (internal cerebral vein, vein of Galen, and straight sinus), which can lead to thalamic or basal ganglial infarction. Most patients present with rapid neurological deterioration. CVT may be confused with other medical conditions.^70–75 Cortical vein thrombosis is also uncommon, and specific clinical syndromes related to the larger cortical veins are rarely seen (eg, temporal lobe hemorrhage associated with vein of Labbé thrombosis).⁷⁶

Several important clinical features distinguish CVT from other mechanisms of cerebrovascular disease. First, focal or generalized seizures are frequent, occurring in ≈40% of patients. Second, an important clinical correlate to the anatomy of cerebral venous drainage is that bilateral brain involvement is not infrequent. This is particularly notable in cases that involve the deep venous drainage system, when bilateral thalamic involvement may occur, causing alterations in level of consciousness without focal neurological findings. Bilateral motor signs, including paraparesis, may also be present due to sagittal sinus thrombosis and bihemispheric injury. Finally, patients with CVT often present with slowly progressive symptoms. Delays in diagnosis of CVT are common and significant. In the ISCVT, symptom onset was acute (<48 hours) in 37% of patients, subacute (>48 hours to 30 days) in 56% of patients, and chronic (>30 days) in 7% of patients. The median delay from onset of symptoms to hospital admission was 4 days, and from symptom onset to diagnosis, it was 7 days.¹⁰

There are several clinical scenarios in which misdiagnosis, or delay in diagnosis, of CVT frequently occurs.

Anecdotal evidence using positron emission tomography showed a reduction of the cerebral blood flow after ligation of the superior sagittal sinus with a concomitant venous infarction.¹⁵⁶ An increased regional cerebral blood volume was also observed in a young adult with sagittal sinus thrombosis.¹⁵⁷ A prolonged mean transit time and increased cerebral blood volume have been suggested as venous congestion, contrary to the pattern observed in patients with an ischemic arterial stroke (prolonged mean transit time with reduction in cerebral blood volume).^111,124

The positive findings of intraluminal thrombus are the key to a confident diagnosis of CVT by CT or MRI. Unfortunately, these findings are not always evident, and the diagnosis rests on nonfilling of a venous sinus or cortical vein (Figure 15). Given the variation in venous anatomy, it is sometimes impossible to exclude CVT on noninvasive imaging studies. Anatomic variants of normal venous anatomy may mimic sinus thrombosis, including sinus atresia/hypoplasia, asymmetrical sinus drainage, and normal sinus filling defects related to prominent arachnoid granulations or intrasinus septa.^{2,71,72,95,97,106,108,109,125,142–150,158} Angiographic examination of 100 patients with no venous pathology¹⁵⁹ showed a high prevalence of asymmetrical lateral (transverse) sinuses (49%) and partial or complete absence of 1 lateral sinus (20%).

Flow gaps are commonly seen on TOF MRV images, which sometimes affects their interpretation. The hypoplastic dural sinus may have a more tapering appearance than an abrupt defect in contrast-enhanced images of the sinus. The lack of identification of a thrombus within the venous sinus on MRI or contrast-enhanced MRV or CTV is helpful to clarify the diagnosis.¹⁶⁰

As mentioned, sinus signal-intensity variations may also affect the interpretation of imaging in the diagnosis of CVT.⁷⁰ Direct cerebral venography may be difficult to interpret owing to retrograde flow of contrast from the point of injection, and the venous pressure may not be accurate because of relative compartmentalization within the system.⁷⁰

To address treatment of CVT in adults, we reviewed systematic reviews and guideline statements of the Cochrane Collaboration,¹⁶¹ the American College of Chest Physicians,^162,163 and the European Federation of Neurological Sciences,¹⁶⁴ in addition to performing a literature review using search terms in PubMed: (“cerebral vein thrombosis” OR “cerebral venous thrombosis” OR “sinus thrombosis”) AND randomized trial; (“cerebral vein thrombosis” OR “cerebral venous thrombosis” OR “sinus thrombosis”) AND treatment guideline. Secondary sources of data included reference lists of articles reviewed and cohort studies that related treatment to outcomes. A summary algorithm for the diagnosis and management of patients with CVT is provided (Figure 4).

Organized care has been defined as collaborative, high-quality, standardized, effective and cost-effective care given by an interdisciplinary team using protocols based on best practices.¹⁶⁵ According to the Stroke Unit Trialists' Collaboration, the most important components of organized stroke care are assessment by a stroke neurologist, admission to a stroke unit with stroke-directed nursing care, physiotherapy, and occupational therapy.^166–169 Organized care is one of the most effective interventions to reduce mortality and morbidity after acute stroke.^166,167 For example, stroke unit care was associated with a 14% reduction in the odds of death at 1 year (OR 0.86, 95% CI 0.76 to 0.98; P=0.02), death or institutionalization (OR 0.82, 95% CI 0.73 to 0.92; P<0.001), and death or dependency (OR 0.82, 95% CI 0.73 to 0.92; P=0.001). These benefits were independent of age, sex, stroke severity, and stroke subtype.^167,169,170

CVT is an uncommon but potentially serious and life-threatening cause of stroke. On the basis of findings for stroke unit care in general, management of CVT in a stroke unit is reasonable for the initial management of CVT to optimize care and minimize complications. Additional specialist input as needed to provide therapeutic anticoagulation is appropriate.

There are several rationales for anticoagulation therapy in CVT: To prevent thrombus growth, to facilitate recanalization, and to prevent DVT or PE. Controversy has ensued because cerebral infarction with hemorrhagic transformation or ICH is commonly present at the time of diagnosis of CVT, and it may also complicate treatment. A summary table is provided with data from observational studies and randomized clinical trials^{10,84,136,171–181} (Table 5) of CVT.

There are 2 available randomized controlled trials comparing anticoagulant therapy with placebo or open control in patients with CVT confirmed by contrast imaging. Taken together, these trials included only 79 patients. One trial of 20 patients assessed intravenous unfractionated heparin (UFH) using dose adjustment to achieve an activated partial thromboplastin time twice the pretreatment value compared with placebo.¹⁷¹ This study used a heparin bolus of 3000 U followed by continuous intravenous infusion. The primary outcome was a CVT severity scale at 3 months, which evaluated headache, focal signs, seizures, and level of consciousness. The secondary outcome was ICH. The trial was stopped early after 20 of the planned 60 patients were enrolled because there was a benefit of treatment. Among 10 patients in the heparin group, 8 recovered completely and 2 had mild deficits at 3 months. Among 10 patients in the placebo group, 1 recovered completely, 6 had minor deficits, and 3 died by 3 months. Two patients treated with placebo and none treated with heparin developed ICH. One patient in the placebo group had unconfirmed pulmonary embolus.

The other trial of 59 patients compared subcutaneous nadroparin dosed on the basis of body weight (180 anti-factor Xa units per kilogram daily in 2 divided doses) with placebo for 3 weeks followed by 3 months of oral anticoagulation (without placebo control) in those randomized to nadroparin.¹⁷² The study was blinded during the first 3 weeks and open label thereafter. Primary outcomes were scores for activities of daily living, the Oxford Stroke Handicap Scale, and death. Secondary end points were symptomatic ICH and other major bleeding. At 3 months, 13% of patients in the nadroparin group had a poor outcome compared with 21% given placebo (treatment difference in favor of nadroparin −7%; 95% CI −26% to 12%). There was no symptomatic ICH in either group (1 nonfatal hemorrhage with nadroparin and 1 fatal unconfirmed pulmonary embolus with placebo). Six patients on active treatment (12%) and 8 control subjects (28%) had full recovery over 3 months.

Meta-analysis of these 2 trials¹⁶¹ revealed a nonstatistically significant relative risk of death or dependency with anticoagulation (relative risk 0.46, 95% CI 0.16 to 1.31), with a risk difference in favor of anticoagulation of −13% (95% CI −30% to 3%). The relative risk of death was 0.33 (95% CI 0.08 to 1.21), with a risk difference of −13% (95% CI −27% to 1%).

A third trial randomized 57 women with puerperal CVT confirmed only by CT imaging and excluded those with hemorrhage on CT.¹⁸² Treatment was with subcutaneous heparin 5000 IU every 6 hours, dose adjusted to an activated partial thromboplastin time 1.5 times baseline for at least 30 days after delivery. Outcome assessment was not blinded. Three patients in the control group either died or had residual paresis compared with none in the heparin group.

In the special situation of CVT with cerebral hemorrhage on presentation, even in the absence of anticoagulation, hemorrhage is associated with adverse outcomes. Highlighting this, in 1 trial of nadroparin, all 6 deaths in the trial overall occurred in the group of 29 patients with hemorrhage on their pretreatment CT scan. None of the deaths were attributed to new or enlarged hemorrhage. These 29 patients were equally divided between treatment groups. Thus, cerebral hemorrhage was strongly associated with mortality but not with cerebral bleeding on treatment. Other studies^171,175 suggested low rates of cerebral hemorrhage after anticoagulation for CVT.

In the special situation of a patient with a major contraindication for anticoagulation (such as recent major hemorrhage), the clinician must balance the risks and benefits of anticoagulation, depending on the clinical situation. In these settings, as for venous thrombosis in general, consultation with an expert in anticoagulation management may be appropriate, and low-intensity anticoagulation may be considered if possible in favor of no anticoagulation until such time as it might be safe to use full-intensity anticoagulation.

A number of observational studies, both prospective and retrospective, are available, primarily from single centers.^{10,136,175–178} Not all studies reported specifically on outcomes of anticoagulation treatment, because the majority of patients in most studies were treated with intravenous UFH or low-molecular-weight heparin (LMWH) at the time of diagnosis, with eventual use of vitamin K antagonists. Data are summarized in Table 5. Mortality rates were low, typically <10%, often due to the underlying disease (eg, cancer) rather than CVT and rarely due to ICH. The majority of patients fully recovered neurological function, and few became disabled.

In a retrospective study of 102 patients with CVT, 43 had an ICH. Among 27 (63%) who were treated with dose-adjusted intravenous heparin after the ICH, 4 died (15%), and 14 patients (52%) recovered completely. Of the 13 patients who did not receive heparin, mortality was higher (69%) with lower improvement in functional outcomes (only 3 patients completely recovered).¹⁷¹

The largest study by far was the ISCVT, which included 624 patients at 89 centers in 21 countries. Nearly all patients were treated with anticoagulation initially, and mortality was 8.3% over 16 months; 79% had complete recovery (modified Rankin scale [mRS] score of 0 to 1), 10.4% had mild to moderate disability (mRS score 2 to 3), and 2.2% remained severely disabled (mRS score 4 to 5).¹⁰ Few studies had sufficient numbers of patients not treated with anticoagulation to adequately address the role of anticoagulation in relation to outcome. Data from observational studies suggest a range of risks for ICH after anticoagulation for CVT from zero to 5.4%.^{136,171,181,183}

In conclusion, limited data from randomized controlled clinical trials in combination with observational data on outcomes and bleeding complications of anticoagulation support a role for anticoagulation in treatment of CVT, regardless of the presence of pretreatment ICH. On the basis of the available data, it is unlikely that researchers will have equipoise on this question, so a new randomized trial may not be feasible. Anticoagulation appears safe and effective. There was consensus in the writing group to support anticoagulation therapy in the management of patients with CVT. If anticoagulation is given, there are no data supporting differences in outcome with the use of UFH in adjusted doses or LMWH in CVT patients. However, in the setting of DVT or PE, a recent systematic review and meta-analysis of 22 studies showed a lower risk of major hemorrhage (1.2% versus 2.1%), thrombotic complications (3.6% versus 5.4%), and death (4.5% versus 6.0%) with LMWH.¹⁸⁴

Pregnancy induces changes in the coagulation system that persist into the puerperium and result in a hypercoagulable state, which increases the risk of CVT. Incidence estimates for CVT during pregnancy and the puerperium range from 1 in 2500 deliveries to 1 in 10 000 deliveries in Western countries, and ORs range from 1.3 to 13.^238–240 The greatest risk periods for CVT include the third trimester and the first 4 postpartum weeks.²⁴⁰ Up to 73% of CVT in women occurs during the puerperium.²⁴¹ Cesarean delivery appears to be associated with a higher risk of CVT after adjustment for age, vascular risk factors, presence of infections, hospital type, and location (OR 3.10, 95% CI 2.26 to 4.24).³⁵

Vitamin K antagonists, including warfarin, are associated with fetal embryopathy and bleeding in the fetus and neonate and thus are generally believed to be contraindicated in pregnancy. Therefore, anticoagulation for CVT during pregnancy and early in the puerperium consists of LMWH in the majority of women.²²⁰

In contrast to UFH, LMWH is not associated with teratogenicity or increased risk of fetal bleeding. The American College of Chest Physicians guidelines for antithrombosis address prevention and treatment of DVT and pulmonary embolus in pregnancy and the puerperium, recommending LMWH over UFH (recommendation 4.2.1).^241a They recommend that treatment be continued throughout pregnancy and for at least 6 weeks postpartum (for a total minimum duration of treatment of 6 months). Although these recommendations are directed to systemic venous thrombosis, it is logical to apply them to CVT for several reasons. First, safety in terms of teratogenicity and fetal/newborn/maternal bleeding complications should be similar, and second, the recommendations are concordant with treatment of non–pregnancy-associated CVT. In a retrospective cohort study of 37 high-risk pregnancies, once-daily tinzaparin was studied for the prevention of initial or recurrent cerebral thrombosis. During treatment, no systemic venous thrombosis occurred; however, 1 parietal infarct and 1 postpartum CVT were documented.²⁴² As in nonpregnant women, fibrinolytic therapy is reserved for patients with deterioration despite systemic anticoagulation, and its use has been reported during pregnancy.²⁴³

Patients with previous VTE are at increased risk of further venous thrombotic events compared with healthy individuals.^244,245 Similarly, women with a history of VTE appear to have an increased risk of thrombotic events (ie, DVT, PE) in future pregnancies.⁵⁷ Pregnancy, and in particular puerperium, are known risk factors for CVT. Six studies investigated the outcome and complications of pregnancy in women who had CVT,^{10,117,175,246–248} with a total of 855 women under observation, of whom 83 became pregnant (101 pregnancies) after their CVT.

These studies found that the risk of complications during future pregnancies was low. In fact, 88% of the pregnancies ended in a normal birth, the remainder being terminated prematurely by voluntary or spontaneous abortion. There was only 1 case of recurrent CVT and 2 cases of DVT; however, a high proportion of spontaneous abortion was noted.

On the basis of the available evidence, CVT is not a contraindication for future pregnancies. Considering the additional risk that pregnancy confers to women with a history of CVT, prophylaxis with LMWH during future pregnancies and the postpartum period can be beneficial.

The incidence of pediatric CVT is 0.67 per 100 000 children per year.⁹¹ When neonates are excluded, the reported incidence is 0.34 per 100 000 children per year.²⁴⁹ Neonates present with seizures or lethargy, whereas older infants and children (similar to adults) usually present with seizures, altered levels of consciousness, increasing headache with papilledema, isolated intracranial hypertension, or focal neurological deficits.

Consideration of endovascular treatment for neonates and children with CVT is driven by the high rates of adverse outcomes. No randomized clinical trials have been conducted in pediatric CVT. Therefore, treatment practices have been extrapolated primarily from adult studies.

In children, and increasingly in neonates, the mainstay of CVT treatment is anticoagulation, including LMWH, UFH, and warfarin. Individual and regional practices vary widely in pediatric CVT and particularly in neonatal CVT. Seizures were observed in >50% of the pediatric population with CVT.¹⁰² Given the higher frequency of epileptic seizures in children, continuous electroencephalography monitoring may be considered for unconscious or mechanically ventilated children.

Neurological worsening may occur in 23% of patients, even several days after diagnosis. Neurological worsening can feature depressed consciousness, mental status disturbance, new seizure, worsening of or a new focal deficit, increase in headache intensity, or visual loss. Approximately one third of patients with neurological deterioration will have new parenchymal lesions when neuroimaging is repeated. Patients with depressed consciousness on admission are more likely to deteriorate.^1,278

Approximately 3% to 15% of patients die in the acute phase of the disorder.²⁸ Most early deaths are a consequence of CVT. In the ISCVT,¹⁰ 21 (3.4%) of 624 patients died within 30 days from symptom onset; however, in a recent retrospective/prospective multicenter study¹⁶ from the United States, higher mortality (13%) was reported. Case series from developing countries also have higher figures for early deaths, with 6% reported in a large Pakistan-Middle East registry⁶³ and 15% in a single-center case series from Iran.²⁶¹

In the largest study, the ISCVT, risk factors for 30-day mortality were depressed consciousness, altered mental status, and thrombosis of the deep venous system, right hemisphere hemorrhage, and posterior fossa lesions. The main cause of acute death with CVT is transtentorial herniation secondary to a large hemorrhagic lesion,⁵ followed by herniation due to multiple lesions or to diffuse brain edema. Status epilepticus, medical complications, and PE are among other causes of early death.^136,279

Deaths after the acute phase are predominantly related to the underlying conditions, in particular malignancies.^10,14

In the ISCVT study,⁵⁵ complete recovery at last follow-up (median 16 months) was observed in 79% of the patients; however, there was an 8.3% overall death rate and a 5.1% dependency rate (mRS score ≥3) at the end of follow-up (12.6% if we consider patients who survived with an mRS score ≥2). In a systematic review that included both retrospective and prospective studies, overall mortality was 9.4%, and the proportion of dependency (mRS score ≥3 or Glasgow Outcome Scale score ≥3) was 9.7%.²⁸ Two retrospective/prospective studies were reported after this review. In the Pakistan-Middle East registry,⁶³ the dependency rate (mRS score ≥3) was higher (11%), whereas in the US multicenter registry,¹⁶ 28% of patients were dependent at 12 months. Of note, some studies include patients transferred to tertiary care centers, whose strokes are usually more severe, with the potential for a referral bias. Among the 7 cohort studies (including the prospective part of retrospective/prospective studies in which information can be analyzed separately), the overall death and dependency rate was 15% (95% CI 13% to 18%).¹⁰

There is little information on the long-term neuropsychological and neuropsychiatric outcome in CVT survivors.^260,272 Despite the apparent general good recovery in most patients with CVT, approximately one half of survivors feel depressed or anxious, and minor cognitive or language deficits may preclude them from resuming their previous jobs.^260,272

Abulia, executive deficits, and amnesia may result from thrombosis of the deep venous system, with bilateral panthalamic infarcts. Memory deficits, behavioral problems, or executive deficits may persist.^263,280

Aphasia, in general of the fluent type, results from left lateral sinus thrombosis with temporal infarct or hemorrhage. Recovery is usually favorable, but minor troubles in spontaneous speech and naming might persist.

Risk factors for poor long-term prognosis in the ISCVT cohort were central nervous system infection, any malignancy, thrombosis of the deep venous system, intracranial hemorrhage on admission CT/MRI, Glasgow Coma Scale score <9, mental status disturbance, age >37 years, and male sex.⁵⁵ Brain herniation leading to early death was more frequent in young patients, whereas late deaths due to malignancies and less favorable functional outcome were more frequent in elderly patients.^6,10,89 Table 6 summarizes demographic, imaging, and clinical variables associated with poor prognosis.^281,282 A Glasgow Coma Scale score of 14 to 15 on admission, a complete or partial intracranial hypertension syndrome (including isolated headache) as the only manifestation of CVT, and absence of aphasia were variables associated with a favorable outcome.^117,177

Despite the overall favorable outcome, ≈15% of CVT patients die or become dependent after CVT.^10,283 Risk stratification scores might improve the ability to inform CVT patients of their individual prognosis and to select those who might benefit most from intensive monitoring and invasive treatments. One study created and validated a risk score model to predict a poor outcome. The risk score model range from 0 (lowest risk) to 9 (highest risk), and a cutoff of ≥3 points indicated a higher risk of death or dependency at 6 months. Two points were assigned for the presence of malignancy, coma, or thrombosis of the deep venous system and 1 point for male sex, presence of decreased level of consciousness, or ICH. The predictive ability (c-statistics) in the derivation cohort was 85.4%, 84.4%, and 90.1% in the validation samples. Sensitivity and specificity in the combined samples were 96.1% and 13.6%, respectively.

Another study²⁸⁴ incorporated age >37 years and central nervous system infection into this model and assigned a weighted index to each variable. The study validated the score in 90 CVT patients and obtained an area under the receiver operator characteristic curve of 0.81 to predict mortality. With a cutoff score of ≥14, sensitivity was 88% and specificity was 70%. The predictive value for good outcome, defined as an mRS score <2, was 95%, and for bad outcome, it was 39%.

In a systematic review of 5 small studies,²⁸ recanalization rates of CVT at 3 months and 1 year of follow-up were 84% and 85%, respectively. The highest rates of recanalization are observed in deep cerebral veins and cavernous sinus thrombosis and the lowest rates in lateral sinus thrombosis.¹⁹³ In adults, recanalization of the occluded sinus is not related to outcome after CVT.^41,194

To address the diagnosis and management of CVT, we systematically searched in PubMed on the following terms: “cerebral vein thrombosis” OR “cerebral venous thrombosis” OR “sinus thrombosis.” Then, we refined our search by combining these with “epidemiology,” “management,” “diagnosis,” “imaging,” “MRI, “randomized trial,” “prognosis,” and “outcome.” These terms were searched with regard to adults, pregnant women, children, and neonates. Our last search was undertaken on July 7, 2010. No language restriction was placed on the searches. Because the intention was to guide readers on the management of CVT based on a comprehensive review of the literature, including sometimes specific and/or uncommon clinical situations, no formal restrictions or further quality assessment was undertaken.

For the treatment section, we reviewed systematic reviews and guideline statements of the Cochrane Collaboration,¹⁶¹ the AHA/American Stroke Association,²⁸⁵ the American College of Chest Physicians,^162,163 and the European Federation of Neurological Sciences,¹⁶⁴ in addition to literature reviews and treatment guidelines. For specific therapeutic alternatives, we combined (“cerebral vein thrombosis” OR “cerebral venous thrombosis” OR “sinus thrombosis”) with “hemicraniectomy,” “thrombolysis,” or “endovascular.” Secondary sources of data included reference lists of articles reviewed and cohort studies that related treatment to outcomes.

Authors assigned to each section were responsible for checking for additional references for their specific topic. For the section on “CVT in the Pediatric Population,” we also reviewed the guideline statements of the AHA²⁶⁷ and the “American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition)” on antithrombotic therapy in neonates and children.²⁶⁹ For the section on “CVT During Pregnancy,” we also reviewed the guideline statements from the American College of Chest Physicians.^241a