Venous Thrombotic Recurrence After Cerebral Venous Thrombosis
A Long-Term Follow-Up Study
Background and Purpose—After cerebral venous thrombosis (CVT), the risk of venous thrombotic events was estimated at 2% to 3% for a new CVT and 3% to 8% for extracranial events. However, because of the paucity of prospective studies, the clinical course of CVT is still largely unknown. We aimed to prospectively evaluate the rate of thrombosis recurrence in a cohort of CVT patients with a long-term follow-up and to detect predisposing factors for recurrence.
Methods—Consecutive CVT patients with complete clinical, radiological, biological, and genetic data were systematically followed up. New venous thrombotic events were detected after hospital readmission and imaging confirmation.
Results—One-hundred eighty-seven patients (mean age 45±18 years, 67% women) with angiographically confirmed CVT were included. Cause was found in 73% of patients. Coagulation abnormality and JAK2 gene mutation were detected in 20% and 9%, respectively. Median follow-up length was 73 months (range 1–247 months). Mean duration of the oral anticoagulant treatment was 14 months. Mortality rate was 2.5% per year, with 2% in-hospital mortality. During follow-up, CVT reoccurred in 6 patients, whereas 19 subjects had a symptomatic extracranial venous thrombotic event, with cumulative venous thrombotic recurrence rates of 3% at 1 year, 8% at 2 years, 12% at 5 years, and 18% at 10 years. A previous venous thrombotic event (hazard ratio, 2.8; P=0.018), presence of cancer or malignant hemopathies (hazard ratio, 3.2; P=0.039), and unknown CVT causes (hazard ratio, 2.81; P=0.024) were independently associated with recurrence.
Conclusions—In our cohort of CVT patients followed on average for >6 years, subjects with a previous venous thrombotic event, cancer/malignant hemopathies, and unknown CVT causes were found to be at higher risk of recurrence.
Cerebral venous thrombosis (CVT) is an uncommon disease that mainly affects subjects younger than 40 years of age, patients with thrombophilia, and women who are pregnant or receiving hormonal contraceptives. The annual incidence is estimated to be 3 to 4 cases per million,1 which increases to 12 cases per 100 000 deliveries during the peripartum period.2
Mortality and recurrence rates are lower when compared with arterial stroke, with an estimated annual risk of new venous thrombotic events after a first CVT at 2% to 3% for a new CVT and 3% to 8% for extracranial events.3–5
However, because of a paucity of prospective studies with long-term follow-up, recurrence of venous thrombosis (VT) after CVT is still largely unknown, as are predisposing factors for recurrence. In fact, mixed results exist on the factors associated with recurrence risk. In a multicenter observational study of 624 adult patients with CVT, male sex and polycythemia/thrombocythemia were found to be the only independent predictors of venous thromboembolism after CVT.6 These results were not confirmed by other authors who observed an association between recurrence and a history of previous extracerebral venous thrombotic events,7 whereas in another study no variable was found to be significantly associated with recurrent VT.4 No study to date has estimated risk of VT recurrence with different CVT causes.
We, therefore, aimed to prospectively evaluate the incidence of recurrent cerebral and extracerebral venous thrombotic events in a cohort of CVT patients and to detect clinical, biological, and radiological factors associated with VT recurrence.
A database was established to prospectively follow up a cohort of consecutive patients with a first episode of angiographically proven CVT admitted to primary (n=3) and comprehensive (n=1) stroke centers in the French Poitou-Charentes region (1.8 million inhabitants, mean age 42 years, 48% <40-year old, 51% women) from January 2005 to December 2014. The study was approved by the local ethics committee, and all enrolled subjects signed informed consent forms. Clinical diagnosis of CVT was confirmed by brain computed tomography (CT) with CT venography, magnetic resonance imaging combined with magnetic resonance venography, conventional angiography, or multimodal imaging. Complete clinical, radiological, biological, and genetic data were collected, including demographic data, symptoms and signs from onset to diagnosis, Glasgow Coma Scale and National Institutes of Health Stroke Scale scores in the acute phase, thrombus localization, and presence of parenchymal lesions. Thrombophilic screening was assessed in all the study subjects, including factor V Leiden, G20210A mutations, lupus anticoagulant, and anticardiolipin antibodies. However, proteins C and S and antithrombin III were not systematically assessed before anticoagulation was started and not controlled after treatment interruption; therefore, they were not analyzed in this study. Patients underwent lumbar puncture unless contraindicated. The JAK2 V617F mutation analysis was systematically assessed in all subjects included from 2008 using a real-time polymerase chain reaction method.
Follow-up visits were performed at 6 months, including neuroimaging follow-up, at 12 months, and yearly thereafter at outpatient evaluation that included neurological examination. If in-person evaluation was impossible, alternative methods included a telephone interview of the patient or an interview with a relative or general practitioner. Subjects known to be hospitalized for any reason during follow-up underwent direct inpatient neurological evaluation.
Follow-up data recorded included: disability (modified Rankin Scale), death, recurrent symptomatic CVT (new symptoms with new thrombosis on repeated neuroimaging), and other symptomatic venous thrombotic events.
Primary outcome was cerebral or extracranial VT recurrence. All the new venous thrombotic events were detected after readmission to regional hospitals and received imaging confirmation. In case of multiple recurrences, only the first one was included in the time-to-event analysis.
Clinically suspected deep vein thrombosis (DVT) was confirmed by duplex ultrasonography, CT, or magnetic resonance imaging venography. Symptomatic pulmonary embolism (PE) was confirmed by contrast-enhanced CT or ventilation–perfusion lung scan interpreted as high probability for PE. Symptomatic portal vein thrombosis was confirmed by contrast-enhanced CT or duplex ultrasonography. A recurrent CVT was distinguished from the original thrombosis by comparing serial imaging. To be classified as a recurrent CVT, new or recurrent symptoms had to develop with new filling defects in the follow-up study not present in the first images or in an interval study showing thrombus resolution.
The Kaplan–Meier product-limit method was used to calculate cumulative recurrence rates and SE. Cox proportional hazards models were used for univariate and multivariate prognostic analyses. All variables with a P value <0.2 at univariate analysis were entered in the multivariate analysis. The final model was obtained after applying a backward elimination procedure; all variables with a P value <0.05 were retained. All analyses were performed with SAS release 9.3 software.
One-hundred ninety-four patients with confirmed CVT were enrolled in our prospective observational study. Seven (3.6%) patients were lost to follow-up after discharge and excluded from the study. Of the remaining 187 patients (mean age 45±18 years, 67% women, 97% white), 98% had an admission Glasgow Coma Scale score between 9 and 15, and 2% were comatose (Glasgow Coma Scale score of <9). Sixty-two patients (33%) presented with isolated intracranial hypertension. Lumbar puncture was performed in 99 patients: 20 patients had >5 cells and 47 had >45 mg/dL protein. Demographic, clinical, and imaging features are shown in Table 1.
CVT cause was found in 73% of subjects (Table 2). Genetic or acquired thrombophilia was detected in 20% of subjects, mainly represented by factor V or II mutation. JAK2 gene analysis was assessed in 120 patients, and a mutation was detected in 9% (n=11) of them. A pharmacological cause was observed in 47% of patients, mainly (91%) represented by hormonal contraceptives or replacement therapy and less frequently by chemotherapy.
In the acute phase, 185 (99%) patients were anticoagulated with intravenous heparin (94%) or subcutaneous low-molecular–weight heparin (6%) in therapeutic dosages followed by oral vitamin K antagonists with a target international normalized ratio of 2:3.8 One patient was treated with local endovascular thrombolysis, and 1 underwent mechanical thrombectomy. Jugular vein stenting was performed in 2 patients, and 1 had optic nerve fenestration. Additional treatment included antiepileptic medication (59 patients [32%]), mannitol (13 [7%]), and acetazolamide (22 [12%]).
Mean and median follow-up lengths were 80 (SD 60) and 73 months (range 1–247 months), respectively. Patients received oral anticoagulant treatment for 14 months on average. Mortality rate was 2.5% per year with 2% in-hospital mortality and 17% overall mortality, mainly related to cancer or hematologic malignancies. Discharge modified Rankin Scale score was 0 to 1 in 74% of patients, whereas 8% of them had a slight residual disability (modified Rankin Scale score of 2) and 16% moderate to severe disability (modified Rankin Scale score of 3–5).
During follow-up, 30 new venous thrombotic events occurred in 25 CVT patients. CVT reoccurred in 6 patients, whereas 19 subjects had new extracranial venous thrombotic events, represented by DVT (12 patients), PE (5 patients), and portal vein thrombosis (2 patients). A single new event occurred in 23 patients, whereas 1 patient experienced CVT followed by PE, and 1 had 5 episodes of DVT. The first recurrent event was included in the analysis.
The cumulative venous thrombotic recurrence rate was 3±1% at 1 year, 8±2% at 2 years, 12±3% at 5 years, and 18±4% at 10 years. Recurrence rate is represented in Figure. In patients with a new CVT, mean recurrence delay was 28.5 months (SD 19.7).
In 1 of the 6 patients with a recurrent CVT and 1 of the 19 patients with a new extracranial venous thrombotic event (PE), mortality was directly attributable to the recurrence.
In univariate analysis, recurrence was associated with older age, a previous venous thrombotic event, absence of hormonal agents, presence of cancer or malignant hemopathy, and unknown causes (Table 3). In addition, patients who discontinued estrogens after CVT displayed decreased recurrence rates compared with subjects who had not taken hormonal treatment (P=0.0036).
No association was found between the duration of anticoagulation treatment and the recurrence risk in the overall study population (P=0.70) or the subgroup of patients with CVT of undetermined cause (P=0.30).
When all potential risk variables with P value <0.2 at univariate analysis were entered into multivariate model, a previous venous thrombotic event (hazard ratio [HR], 2.8; P=0.018), presence of cancer or malignant hemopathy (HR, 3.2; P=0.039), and unknown CVT cause (HR, 2.81; P=0.024) were independently associated with any first VT recurrence (Table 3).
Our study showed that cumulative risk of recurrent VT continues to increase over time. We also detected factors associated with VT recurrence in subjects who were diagnostically evaluated and treated for the initial CVT.
Early recurrence rates in our study are comparable with the results of other studies.3–5,7,9–11 Details on the studies that evaluated venous thrombotic events recurrence in CVT patients are displayed in Table 4.
In the ISCVT study (International Study on Cerebral Vein and Dural Sinus Thrombosis), a large prospective multinational observational study conducted on 624 adult cases of CVT, a 2.2% overall CVT recurrence rate and 3% extracranial venous thrombotic recurrence rate were observed during a median follow-up period of 16 months (mean=18.6, SD=11.1).3 Ruling out subjects lost to follow-up immediately after discharge (n=80), cerebral and extracerebral venous thrombotic recurrence rates were 2.6% and 3.4%, respectively.
Similarly, Gosk-Bierska et al4 found a 2.2% annual sinus thrombosis recurrence rate and a 2.8% annual extracerebral thrombosis (ie, DVT or PE) recurrence rate during a mean follow-up period of 36±47 months (range 0–262 months) after a CVT, with a median time to recurrence of 10 months.
An annual CVT recurrence rate of 1.8% was observed by Preter et al9 in 77 patients with CVT during a median follow-up period of 63 months, mainly (89%) occurred within the first 12 months, whereas in the VENOPORT study (Cerebral Venous Thrombosis Portuguese Collaborative Study Group), a 1.6% overall CVT recurrent thrombosis rate was reported in 126 CVT patients with a mean follow-up length of 1.8 years.10
Similarly, Martinelli et al5 found an event rate of 2.03 per 100 patient-years in a population of 145 patients with a first episode of CVT followed up for 6 years after discontinuation of anticoagulant therapy.
In a large, retrospective cohort study, an overall recurrence incidence of 23.6 events per 1000 patient-years (95% confidence interval [CI], 17.8–28.7) was observed during a median follow-up period of 40 months (range 6–297 months).7
Recently, in a retrospective single-center study conducted on 161 CVT patients with a median follow-up period of 39 months (interquartile range 14–95 months), Hiltunen et al11 observed a 1.1% annual venous thromboembolic recurrence rate (5 DVT, 2 PE, and 3 superficial vein thrombosis) with a 0.1% annual CVT recurrence rate. Authors stated that low incidence of recurrence could be explained by the frequent use of antithrombotic agents in this study population because 39% of patients received permanent anticoagulation and 25% received aspirin.
Similarly to other studies, in our population, new venous thrombotic events mainly (52%) occurred in the first 24 months.
In our cohort of CVT patients with long-term follow-up, subjects with a previous venous thrombotic event, presence of cancer or malignant hemopathies, and unknown CVT causes were found to be at higher risk of recurrence in the multivariate analysis.
The increased risk of both cerebral and extracerebral venous thrombotic recurrence in CVT patients with a history of previous extracerebral venous thrombotic events (HR, 2.70; 95% CI, 1.25–5.83; P<0.011) has previously been observed by Dentali et al,7 while the association with cancer was found only in the univariate analysis.
On the contrary, in the ISCVT, male sex (HR, 2.6; 95% CI, 1.4–5.1; P=0.004) and polycythemia/thrombocythemia (HR, 4.4; 95% CI, 1.6–12.7; P=0.005) were the only independent predictors of venous thromboembolism after CVT.6
Similarly, in the study conducted by Martinelli et al,5 male CVT patients had a 7- to 8-fold increased risk of recurrence compared with female patients, while severe thrombophilia was associated with an increased risk of DVT or PE (HR, 4.71; 95% CI, 1.34–16.5), but not with CVT recurrence. However, these findings may not be applicable to the general CVT population because patients with malignancies or in need of permanent anticoagulation were not included in this study. Differences in the inclusion criteria might be the reason of the discrepancy between our results and those of the Italian group because we did not find any correlation between sex and recurrence risk.
To our knowledge, this is the first time that cryptogenic CVT was found to be associated with the risk of recurrence. The prospective design and long-term follow-up of our study might explain this difference. Our cryptogenic CVT rate is similar to that reported in other studies; however, a wide range of idiopathic thrombosis rates, from 9% to 44%, has been reported.3–5,7,11 In our opinion, unknown causes may limit secondary prevention measures in this subgroup of patients, leading to an increase in the risk of recurrence. This finding underscores the need for further research to better identify both congenital and acquired VT risk factors to develop more specific treatment strategies.
Another finding of our study is that women receiving hormonal agents such as oral contraceptives or replacement therapy at the time of their first CVT event had lower recurrence risk compared with women who had not taken estrogens. These data suggest that subjects whose CVT is favored by hormonal agents are at lower risk of new venous thrombotic events after estrogen discontinuation compared with CVT patients with other risk factors or whose CVT cause was not detected. Similar data were obtained in a study conducted on patients with a first event of DVT and PE; a trend toward a lower risk of recurrence after discontinuation of oral contraceptive was observed compared with subjects who had not taken estrogens, although this result was not confirmed for postmenopausal hormones.12
One limitation of our study is that we included patients with incomplete thrombophilia evaluation. In fact, because of a relatively high percentage of subjects whose thrombophilic screening was not repeated after anticoagulation interruption, a decision was made to only consider clear prothrombotic abnormalities. As a consequence, idiopathic CVT percentage may have been slightly overestimated. However, because of the rigorous selection process and considering the robust statistical results, we think that our findings were not significantly affected. Another limitation is that, because of the low number of female subjects under replacement therapy before CVT, we cannot exclude the possibility that the significant association between hormonal agents and recurrence risk was mainly driven by oral contraceptives more than by postmenopausal hormones. A third limitation is that our study is a French single-region study, which almost exclusively included white patients; we cannot rule out the possibility that our results were influenced by genetic and environmental factors and, therefore, should be verified in different populations before being applied to other contexts. However, this design assures better homogeneity in selection and follow-up procedures. A forth limitation is represented by the higher cancer and hematology malignancy prevalence compared with previous follow-up study populations; this probably explains the relatively high mortality rate during follow-up. However, a decision was made to include all consecutive subjects with a first CVT event and at least 1 follow-up evaluation, including those with malignancies, to better evaluate recurrence rate in the global CVT population.
Finally, it is worth specifying that all new qualifying events, including PE, occurred after CVT resolution at follow-up imaging and should thus be considered as new events and not as a possible direct consequence of CVT. In fact, PE has been described in CVT patients even in the absence of DVT, probably as a consequence of a detached thrombus from the lateral sinus in subjects with a global prothrombotic state in patients without thrombosis resolution.13,14
Our study provides data that could contribute to the detection of subjects at higher risk of venous thrombotic recurrence, who may benefit from prolonged anticoagulation.
We are grateful to Pr Andrei V. Alexandrov for his precious comments and suggestions.
- Received September 2, 2016.
- Revision received November 3, 2016.
- Accepted November 10, 2016.
- © 2016 American Heart Association, Inc.
- Lanska DJ,
- Kryscio RJ
- Ferro JM,
- Canhão P,
- Stam J,
- Bousser MG,
- Barinagarrementeria F
- Martinelli I,
- Bucciarelli P,
- Passamonti SM,
- Battaglioli T,
- Previtali E,
- Mannucci PM
- Miranda B,
- Ferro JM,
- Canhão P,
- Stam J,
- Bousser MG,
- Barinagarrementeria F,
- et al
- Dentali F,
- Poli D,
- Scoditti U,
- Di Minno MN,
- De Stefano V,
- Stefano VD,
- et al
- Saposnik G,
- Barinagarrementeria F,
- Brown RD Jr,
- Bushnell CD,
- Cucchiara B,
- Cushman M,
- et al
- Preter M,
- Tzourio C,
- Ameri A,
- Bousser MG
- Cakmak S,
- Nighoghossian N,
- Desestret V,
- Hermier M,
- Cartalat-Carel S,
- Derex L,
- et al