Article Text


Original research
Venous collateral drainage patterns predict clinical worsening in dural venous sinus thrombosis
  1. Sunil A Sheth1,
  2. Harry Trieu1,
  3. David S Liebeskind2,
  4. Jeffrey L Saver2,
  5. Viktor Szeder1,
  6. Reza Jahan1,
  7. Satoshi Tateshima1,
  8. Gary Duckwiler1
  1. 1Division of NeuroInterventional Radiology, Department of Radiology, University of California Los Angeles, Los Angeles, California, USA
  2. 2Department of Neurology, University of California Los Angeles, Los Angeles, California, USA
  1. Correspondence to Dr Sunil A Sheth, Department of Neurology, UT Health McGovern School of Medicine, 6431 Fannin Street, MSB7.005E Houston, TX 77030, USA; ssheth{at}


Background Dural venous sinus thrombosis (DVST) is an increasingly recognized cause of a wide array of neurological symptoms, with outcomes that range from complete recovery to death. The condition of approximately 23% of patients with DVST will worsen after initial presentation, as a result of restricted venous outflow and venous hypertension, but early identification of this subset is challenging. A venous collateral scale (VCS) that grades alternative drainage routes may improve prediction of clinical deterioration.

Objective To examine the ability of the VCS system to accurately identify patients with DVST who will experience clinical worsening, based on their imaging at presentation.

Methods From our institutional database, we identified patients with DVST on dedicated venous imaging between January 2010 and July 2016. A VCS was created and calculated from venous imaging at presentation by two reviewers blinded to subsequent data.

Results The 27 patients who met the inclusion criteria for this study had a median age of 42 years and 14 (52%) were female. Initial symptoms included headache without hemorrhage in 30% and focal deficit in 30%. Transverse sinus occlusion was present in 70% and superior sagittal sinus occlusion in 41%. VCS was 0 in 11%, 1 in 37%, and 2 in 52%. A lower VCS was significantly associated with clinical worsening both from time of initial symptom onset (77% vs 29%, VCS 0–1 vs 2, p<0.05) and during hospitalization (62% vs 0%, VCS 0–1 vs 2, p<0.01). In multivariate analysis, VCS but no other presenting features was significantly associated with in-hospital worsening (OR=2, p<0.01).

Conclusions The type and quality of venous collaterals influence outcome in DVST. VCS helps to identify patients whose condition is likely to deteriorate and may need additional aggressive interventions.

  • Hemorrhage
  • Angiography
  • Blood Flow
  • Vein
View Full Text

Statistics from


Dural venous sinus thrombosis (DVST) is an increasingly recognized cause of a wide array of neurological symptoms, with outcomes that range from complete recovery to death.1 ,2 DVST affects an estimated 15.7/million a year, with an incidence that increases with time as awareness has improved.3 Importantly, the condition of approximately 23% of patients with DVST will worsen after initial presentation.4 Early identification of this subset, however, remains challenging.

The etiology of this clinical worsening is thought to be secondary to restricted venous outflow. In the absence of adequate drainage routes, pressure builds within the veins. This venous pressure then causes increased pressure on the parenchyma, which results in cerebral edema, hemorrhage, and ischemia.5 Treatments to mitigate this progression involve systemic anticoagulation, to assist the intrinsic thrombolytic pathways in restoring venous drainage routes and prevent progression of the existing thrombus. Current guidelines recommend that endovascular thrombectomy should be considered in cases of clinical worsening or a persistently poor neurological examination.6 However, an imaging biomarker that could predict clinical worsening would significantly improve our ability to manage these patients. Early identification could ensure that these high-risk patients receive closer neurological monitoring and early aggressive intervention to avert neurological injury. Although prior attempts have been made to characterize venous collaterals, they have failed to identify a system that accurately predicts which patients will be resilient to the effects of the occlusion, and which will eventually deteriorate.7 ,8

To deal with this problem, we developed a venous collateral scale (VCS) that scores the venous drainage from the territory of brain affected by the sinus thrombosis. Unlike previous attempts, this scale is based on the type and quality of venous collateral routes to alternate sinuses that might help to prevent malignant venous hypertension. This study examines the ability of the VCS system to accurately identify patients with DVST who will experience clinical worsening, based on their imaging at presentation.


From our institutional database, we identified patients with documented DVST on dedicated venous imaging (MR, CT, or catheter angiography) between January 2010 and July 2016. Both in-hospital patients and outpatients were included. Demographic, clinical, imaging, and angiographic data were retrospectively collected. Patients were included in this study if they presented with symptoms attributable to a DVST and underwent an investigation of these symptoms including dedicated cerebral venous imaging at the time of presentation. Patients with thrombosis related to arteriovenous fistulae were excluded.

At our institution, standard of care for treatment includes systemic anticoagulation if not contraindicated. In a subset of patients who continue to worsen despite system anticoagulation, additional treatment with endovascular thrombectomy is performed. The decision to perform endovascular therapy involves a discussion about the risks/benefits of the procedure in relation to the overall clinical situation of the patient. Routine clinical care included follow-up imaging in all patients.

DVST location was determined by blinded review of initial presentation imaging, including dedicated venous imaging. If multiple modalities were available (ie, CT venogram, MR venogram, catheter angiogram), all were reviewed. In addition, from these initial studies, venous collaterals were graded according to the VCS as shown in table 1. Briefly, VCS 0 indicates that no meaningful venous drainage from the territory of brain affected by the sinus occlusion could be visualized. VCS 1 indicates that while a draining vein from the affected region is seen, the vein does not connect to a patent sinus. VCS 2 indicates that a draining vein from the affected territory is visualized and this vein then connects to a patent sinus, and as such functions as a suitable venous collateral. Examples of VCS 0–2 are shown in figure 1. Patients were considered to have multiple sinus occlusions if more than one of the dural sinuses (straight, superior sagittal, right and left transverse) were involved. For pairwise comparisons, the VCS was collapsed into a binary scale of VCS 0–1 vs VCS 2.

Table 1

Venous collateral scale

Figure 1

Representative angiographic images of the patients with dural venous sinus thromboses demonstrating the three venous collateral scale (VCS) grades. (A and B) Anteroposterior (AP) and lateral views of a patient with left transverse sinus occlusion, with no evidence of collateral vein (white circle), representing VCS 0. (C and D) AP and lateral views of a patient with right transverse sinus occlusion. A cortical vein (vein of Labbé) is visualized; however, there is no connection to a patent sinus (white arrow), corresponding with VCS 1. (E and F) AP and lateral views of a patient with a superior sagittal sinus and proximal right transverse sinus occlusion. There is a large superior anastomotic vein that connects to the Sylvian vein (white arrow), and then into the patent portion of the right transverse sinus through the inferior anastomotic vein of Labbé, corresponding with VCS 2.

We then examined the patients in our study for evidence of clinical worsening. Clinical worsening was determined by blinded review of neuroimaging studies subsequent to the initial study within the same hospitalization, as well as review of the medical records. For inpatients, worsening was defined as any new hemorrhage within the territory of the DVST associated with worsening in neurological examination or expansion of edema within the territory of the DVST if also associated with worsening in neurological examination. Thus, both radiographic and clinical progressions were required. Clinical worsening in the prehospital setting was defined as progression from mild neurologic symptoms before presentation to the hospital, to new hemorrhage or edema with associated neurological worsening upon presentation to the hospital.

Univariate comparisons between categorical variables were made using Fisher's exact test and the χ2 test, and between continuous variables using the Mann–Whitney U test. All statistical analyses performed using commercially available software (Prism 7, GraphPad Software, La Jolla, California, USA and SAS 9.4, SAS Institute Inc, Cary, North Carolina, USA). Multivariate regression was performed using Matlab (Mathworks Inc, Natick, Massachusetts, USA). p<0.05 was considered statistically significant and p<0.12 was considered a trend. Variables were included in the multivariate analysis if they demonstrated significance of a trend towards difference in univariate analysis. This study was approved by the institutional review board of the local institution and was conducted in compliance with the Health Information Portability and Accountability Act.


The 27 patients who met the inclusion criteria for this study had a median age of 42 (IQR 24–58) and 52% (14/27) were female. As shown in table 2, 11% (3/27) of patients used oral contraceptives, 19% (5/27) had a history of smoking, and 33% (9/27) showed evidence of a hypercoagulable state. The total percentage of patients with either oral contraceptive use or evidence of hypercoagulability was 44%. A history of smoking was associated with a higher VCS score. There was a trend towards a higher VCS score for patients using oral contraceptives. The most common presenting symptoms were focal neurological deficit without hemorrhage and headache without neurological deficit or hemorrhage. In-hospital mortality was 19%. Location of occlusion, treatment modality, and in-hospital mortality were not significantly associated with VCS grade. Transverse sinus involvement was more common than superior sagittal sinus or straight sinus. Almost all patients were treated with anticoagulation. There was a strong trend towards increased duration of time from symptom onset to initiation of anticoagulation in patients in the VCS 2 group.

Table 2

Patient demographics

VCS score was calculated from CT venograms in 5/27 (19%), MR venogram in 18/27 (67%) and digital subtraction angiography is 12/27 (44%). A total of eight (30%) patients underwent multiple imaging modalities in close proximity to each other and at the time of hospital arrival. In this group of eight patients, all had two qualifying studies, of which 6/8 pairs (75%) resulted in consistent dichotomized VCS (0–1 vs 2).

A total of five patients were treated with endovascular thrombectomy—four in the VCS 0–1 group and one in the VCS 2 group. In all five patients, endovascular thrombectomy was performed owing to clinical worsening. Two patients in the VCS 0–1 group showed worsening from the initial symptoms, and the other two worsened in-hospital. The one patient in the VCS 2 group who underwent thrombectomy experienced worsening from the initial symptoms. The rate of embolectomy between the VCS 0–1 and VCS 2 groups was not statistically significant.

As shown in figure 2A, 3 patients (11%) were found to have VCS 0, 10 (37%) VCS 1, and 14 (52%) VCS 2. There was a correlation between VCS and multiple sinus occlusions (figure 2B). Compared with patients with VCS 0, patients with VCS 2 were less likely to have multiple sinus occlusions (100% vs 14%, VCS 0 vs VCS 2, Fisher's exact test, p<0.05).

Figure 2

Venous collateral scale (VCS) characteristics within the studied cohort. (A) Distribution of patients by VCS grade demonstrates increased frequency of higher grades. (B) Distribution of patients with multiple sinus involvement by VCS grade. *Represents p<0.05, Fisher's exact test.

A total of 14 patients (52%) experienced clinical worsening related to their DVST. All these patients had a new or expanded hematoma, with the exception of one patient who had a new ischemic event. As shown in table 3, the likelihood of worsening was far greater in patients with lower VCS. All clinical worsening (including both prehospital and in-hospital instances of clinical worsening) was nearly 2.7-fold higher in patients with VCS 0–1. Similarly, in-hospital worsening was also predicted by the VCS, with significantly higher rates of worsening in patients with VCS 0–1 compared with VCS 2.

Table 3

Likelihood of clinical worsening

We then performed a multivariate analysis to evaluate the independent effect of VCS on likelihood of in-hospital worsening, relative to the demographic features that were associated with the score. As shown in table 4 VCS alone, but not sex, oral contraceptive pill use, smoking, or time from symptom onset to initiation of heparin was significantly associated with in-hospital worsening. Compared with those with VCS 2, patients with VCS 0–1 had an OR of two towards worsening while in hospital.

Table 4

Multivariate predictors of in-hospital worsening


In this analysis of 27 patients with DVST at a single tertiary care center over a 6-year-period, VCS measured at the time of presentation demonstrated excellent discrimination in identifying patients likely to have clinical worsening. The rate of clinical worsening, in combined prehospital and in-hospital settings, was 52%, and patients with a lower VCS score were more likely to have multiple sinus involvement. In both the prehospital and in-hospital settings, patients with lower VCS were far more likely to show clinical deterioration subsequently. In multivariate analysis, VCS was significantly associated with in-hospital worsening.

The VCS we evaluate in this study differs from a prior attempt to classify venous collaterals. This previous venous collateral classification system by Qureshi also included three grades.7 In that system, grades are assigned based on the type of sinus that a collateral vein drains into: grade I for connecting to the same sinus, grade II for connecting to a different sinus, and grade III for connecting to a different circulation (ie, deep and superficial). Our scale, on the other hand, does not take into account which sinus the collateral drains into, but considers all patent sinuses that can receive flow and reduce pressure from tissue that would have otherwise drained to the occluded sinus as equivalent. With this idea in mind, the VCS was designed so that increasing values of the scale represent graded improvements in the capacity of the venous system to distribute and manage accumulating venous pressure. The Qureshi venous collateral system was evaluated in an independent cohort of patients with DVST.8 In that study, the frequency of grade I collaterals was 3% and grade II and III 27% and 58%, respectively, and the collateral grade was not associated with presenting features or clinical outcome.

From an angioarchitectural point of view, there are a finite number of collateral routes that the venous system can develop to bypass occlusion of a major sinus. In our study as well as many others on DVST, the most common sites of occlusion are the transverse/sigmoid sinus and superior sagittal sinus.1 ,3 ,9 The most common routes to alternate sinuses from the territories drained by these sinuses would be through the Sylvian vein into the ipsilateral superior anastomotic vein (in the case of transverse sinus occlusion) or ipsilateral inferior anastomotic vein (in the case of superior sagittal sinus occlusion). Less common, but also possible, pathways to patent sinuses would include connections from the Sylvian or uncal veins to the ipsilateral cavernous sinus, or potentially, connections to the deep venous system. In our study we found that patients with multiple sinus occlusions were more likely to have VCS 0, a finding consistent with the fact that the number of available alternative sinuses is limited and which emphasizes the physiological limitations of the type of venous collateral that can be developed. In addition, previous studies have identified the presence of deep venous system thrombosis as a risk factor for poor outcome.1 ,10 This finding can also be easily rationalized within the framework of the VCS, as the deep system is far less likely to find collateral routes to alternative sinuses, given the relative infrequency of robust connections to the superficial veins.

We found evidence of greater venous collaterals in patients with a history of smoking, a finding that may at first seem counterintuitive, as one would expect more robust vasculature in patients with fewer risk factors. Indeed, smoking is known to have an effect on the type and quality of arterial cerebral collaterals, and as a result directly influence outcome in acute ischemic stroke.11 On the other hand, the consequences of smoking may be less pronounced on the venous system. Although smoking is often considered a risk factor for the development of DVST, unlike better-characterized thrombophilic disorders, the association is not clear.12 ,13 Further, this relationship also holds true in systemic venous thromboembolic disease, in which neither former nor current smoking contribute a significant relative risk.14 Further, in multivariate analysis, history of smoking did not have any effect on the likelihood of in-hospital worsening independent of VCS.

In addition, we also found a trend towards greater delay in initiation of anticoagulation in our VCS 2 cohort compared with VCS 0–1. This finding may be related to the higher (though not significant) incidence of traumatic or postsurgical patients in this group. Importantly, in multivariate analysis, delay in anticoagulation did not significantly influence the likelihood of in-hospital worsening.

We found a rate of clinical worsening in patients with DVST of 52%, which is higher than that found in other studies, which range from 12% to 23%.1 ,4 This difference is probably due to the design of our study, in which we included patients who worsened in the prehospital setting. This design was chosen to optimally evaluate the hypothesis that the likelihood of clinical worsening in DVST is related to the type and quality of venous collaterals. Patients with poor venous collaterals are more likely to progress rapidly from the time of initial occlusion, and by the time they present to the hospital, may already have had neurological symptoms and hemorrhage. By limiting the study to the in-hospital setting alone, this progression, and its relationship to venous collateral status, would have been missed. However, by factoring in the patient's progress from the time of symptom onset, we can more completely compare the effect of venous collaterals on clinical worsening. Because another important use of venous collateral scales is to identify which patients to treat more aggressively once they have arrived in the hospital, we also performed a second analysis to examine only in-hospital worsening. Our rate of in-hospital worsening of 30% is more consistent with the results of prior studies. Importantly, we were able to show that the VCS is predictive of clinical worsening starting in both prehospital and in-hospital settings. In addition, we found in multivariate analysis that VCS is significantly associated with the likelihood of in-hospital worsening.

In this study we focus on venous collateral imaging and its prediction of in-hospital progression, whereas prior studies have evaluated clinical and demographic features as predictors of poor ultimate outcome. Studies of long-term prognosis have identified features at presentation such as encephalopathy, poor neurological examination, intracranial hemorrhage and deep venous system thrombosis, as mentioned above.1 ,15 ,16 Some of these features, including the presence of intracranial hemorrhage and deep venous system thrombosis, are closely tied to the presence and degree of venous hypertension, which is correlated to the quality of venous collaterals. Further studies may reveal the degree of correlation between venous anatomy and other risk factors for poor final outcome.

Our study has some limitations—particularly the limited sample size and retrospective nature. Thus, subgroup analyses within the cohort were limited. Further validation in an independent cohort will improve the generalizability of our findings. In addition, heterogeneous imaging modalities were used in scoring VCS. Further studies of this score based on a consistent single modality will be important to validate the results.


The diagnosis and effective management of DVST remain challenging. The VCS we present in this manuscript provides a measure of the type and quality of venous collaterals. It represents an anatomic assessment of the network's ability to redirect venous flow and mitigate malignant venous hypertension. We find that this scale accurately identifies patients who are likely to show clinical deterioration and, as such, may warrant earlier aggressive measures, to proactively intervene before the likely subsequent injury.


View Abstract


  • Contributors SAS, HT, and GD: substantial contributions to the conception and design of the work, drafting and critically revising it and its final approval. He agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. DSL, JLS, VS, RJ, and ST: acquisition of data, revising the article and its final approval. The authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This work has not been presented in part or whole at any prior conference or submitted to another journal for publication.

  • Funding This work was supported by the Society for NeuroInterventional Surgery Foundation Pilot Research Award (principal investigator (PI): SAS) and the American Heart Association/American Academy of Neurology/American Brain Foundation Lawrence Brass Stroke Research Award (PI: SAS, award number 15CRP22900006).

  • Competing interests DSL: consulting fees from Stryker, Covidien, and Zoll (modest). ST: consulting fees from Penumbra, Covidien, Stryker, and Reverse Medical. RJ: consulting fees from Covidien and Stryker.

  • Ethics approval Institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.