Article Text

Download PDFPDF

Research paper
Decompressive craniectomy in cerebral venous thrombosis: a single centre experience
  1. Sanjith Aaron1,
  2. Mathew Alexander1,
  3. Ranjith K Moorthy1,
  4. Sunithi Mani2,
  5. Vivek Mathew1,
  6. Anil Kumar B Patil1,
  7. Ajith Sivadasan1,
  8. Shalini Nair3,
  9. Mathew Joseph3,
  10. Maya Thomas1,
  11. Krishna Prabhu4,
  12. Baylis Vivek Joseph4,
  13. Vedantam Rajshekhar4,
  14. Ari George Chacko4
  1. 1Neurology Unit, Department of Neurological Sciences, Christian Medical College & Hospital, Vellore, Tamil Nadu, India
  2. 2Department of Radiology, Christian Medical College & Hospital, Vellore, Tamil Nadu, India
  3. 3Neurocritical Care Unit, Department of Neurological Sciences, Christian Medical College & Hospital, Vellore, Tamil Nadu, India
  4. 4Neurosurgery Unit, Department of Neurological Sciences, Christian Medical College & Hospital, Vellore, Tamil Nadu, India
  1. Correspondence to Dr Mathew Alexander, Neurology Unit, Department of Neurological Sciences, Christian Medical College, Vellore, Tamil Nadu 632004, India; mathewalex{at}


Background Cerebral venous thrombosis (CVT) is an important cause for stroke in the young where the role for decompressive craniectomy is not well established.

Objective To analyse the outcome of CVT patients treated with decompressive craniectomy.

Methods Clinical and imaging features, preoperative findings and long-term outcome of patients with CVT who underwent decompressive craniectomy were analysed.

Results Over 10 years (2002–2011), 44/587 (7.4%) patients with CVT underwent decompressive craniectomy. Diagnosis of CVT was based on magnetic resonance venography (MRV)/inferior vena cava (IVC). Decision for surgery was taken at admission in 19/44 (43%), within 12 h in 5/44 (11%), within first 48 h in 15/44 (34%) and beyond 48 h in 10/44 (22%). Presence of midline shift of ≥10 mm (p<0.0009) and large infarct volume (mean 146.63 ml; SD 52.459, p<0.001) on the baseline scan influenced the decision for immediate surgery. Hemicraniectomy was done in 38/44 (86%) and bifrontal craniectomy in 6/44 (13.6%). Mortality was 9/44 (20%). On multivariate analysis (5% level of significance) age <40 years and surgery within 12 h significantly increased survival. Mean follow-up was 25.5 months (range 3–66 months), 26/35 (74%) had 1 year follow-up. Modified Rankin Scale (mRs) continued to improve even after 6 months with 27/35 (77%) of survivors achieving mRs of ≤2.

Conclusions This is the largest series on decompressive craniectomy for CVT in literature to date. Decompressive craniotomy should be considered as a treatment option in large venous infarcts. Very good outcomes can be expected especially if done early and in those below 40 years.


Statistics from


Cerebral venous thrombosis (CVT) is an important cause for stroke in the young.1 ,2 The mortality rates following CVT range from 6% to 15%.3–7 The International Study on Cerebral Vein and Dural Sinus Thrombosis (ISCVT)8 reported that majority of deaths occur within the first 30 days. The leading cause of death is haemorrhagic conversion of large venous infarcts resulting in herniation. However, unlike in arterial strokes,9 decompressive craniectomy has not gained widespread acceptance in patients with venous strokes. Hence, we retrospectively analysed patients with CVT treated with decompressive craniectomy at our centre.

Materials and methods

This study was conducted in the neurological sciences department of a quaternary level teaching hospital. Patients with CVT who had undergone decompressive craniectomy over a period of 10 years (2002–2011) were identified from the computerised discharge summaries. Patient data were collected from the inpatient records and the Stroke clinic where the patients were being followed up postoperatively. Radiological images were reviewed from Picture Archiving and Communication System (PACS, GE). The diagnosis of CVT was based on MRI/MRV. The venous infarct size was measured on MRI images and the maximum dimension of the lesion on fluid attenuated inversion recovery (FLAIR) images was used. Volume was calculated by measuring the area on contiguous images and multiplying by the slice thickness taking into account the slice gap. Haemorrhage was not separately assessed. Venous infarct is the lesion (haemorrhage and oedema together). Lateral shift of pineal gland and septum pellucidum were used to measure the midline shift. Haemorrhage was established either on plain CT scan or on T2W MRI.

Surgical procedure and postoperative management

Thirty-eight patients with unilateral hemispheric involvement with midline shift to the contralateral side underwent ipsilateral frontotemporoparietal decompressive craniectomy. The dura was opened and a lax duraplasty performed with pericranial graft. The bone flap was not replaced. No attempt was made to remove infarcted brain, except in four patients who had significantly large haematomas. Six patients who had bilateral frontal lobe involvement underwent bifrontal decompressive craniectomy, the posterior limit of the craniectomy being tailored to the posterior extent of the infarcted brain, with lax duraplasty. Postoperatively, the patients were paralysed and ventilated with normocarbia in the postoperative period for a period of 48 h during which time mannitol was continued. Anticoagulation was restarted/started 48 h after the surgical procedure. Of the 25/44 (56.8%) patients in whom decision for decompression was deferred at admission, anticoagulation (intravenous unfractionated heparin) was started in 23/44 (52%) to keep the target PTT>1.5 times the control. (This was reversed with intravenous (IV) protamine before the surgery.) Anticonvulsants were started in those patients who presented with seizures and also were given prophylatically to others with large infarcts involving the cerebral cortex. Attempts to reduce the raised intracranial pressure were made by using acetazolamide, IV dexamethasone and hypertonic (3%) saline. Repeat imaging was done as per the discretion of the treating team for 18/25 (72%) patients.

Patients who were willing for the same underwent cranioplasty at 6 months or more after discharge from hospital.

Modified Rankin Scale (mRs)10 was used to assess the outcome at discharge and at 3, 6 and 12 months follow-up after surgery.

Statistical analysis

Variables were compared using univariate analysis. All factors reaching significance were analysed using a logistic regression model to perform a multivariate analysis. Two-tailed Fisher's exact test was used to look for significant associations. Statistical analysis was done using a statistical package STATA (StataCorp).


Clinical features

Over a 10-year period (2002–2011), 587 patients with CVT were managed of whom, 44 patients (7.4%) underwent decompressive craniectomy.

There were 27 females (61.3%) and 17 males (38.6%). Eleven of the 27 females (40.7%) presented in the postpartum period. The mean age at presentation was 35.6±11.7 years (range 19–60 years). The mean duration of symptoms at the time of presentation was 6.47±SD days (range 1–18 days). Headache was the most common symptom followed by seizures and motor deficits. Glasgow Coma Score of <9/15 at admission was seen in 14/44 (30%) (table 1).

Table 1

Clinical features at presentation of the 44 patients with CVT who underwent decompressive craniectomy

Radiological features

The venous infarcts were involving the right side in 18/44 (41%) patients, left side in 16/44 (36%) patients and 10/44 (23%) patients had bilateral involvement. An acute on chronic presentation of CVT was seen in 2/44 (4.5%) patients (table 2). The most common dural venous sinus involved was the superior sagittal sinus in 36/44 (82%) patients. In 36/44 (82%) patients there was associated cortical vein involvement. In 9/44 (20.4%) patients the vein of Labbe and in 4/44 (9%) patients the vein of Trollard were the prominent veins involved. Associated deep venous system involvement was noted in 3/44 (6.8%) patients.

Table 2

Radiological features: dural sinus involvement at presentation of the 44 patients with CVT who underwent decompressive craniectomy

The mean infarct volume at admission was 105.9±59.8 ml (range 15–240 ml). Midline shift was seen in 35/44 (79.5%) patients. (table 3) The mean midline shift was 7.08±4.6 mm (range 0–17). A midline shift of 10 mm or more was seen in 15/44 (34%) patient at admission. Infarct volume correlated positively with degree of middle shift (0.667).

Table 3

Radiological features (at admission) of the 44 patients with CVT who underwent decompressive craniectomy

The admission imaging showed features of subfalcine and transtentorial herniation in 33/44 (75%) patients while only subfalcine herniation was seen in five patients (11%). Seven patients (15.9%) had features of transtentorial herniation alone, of which four had vein of Labbe involvement along with transverse sigmoid sinus thrombosis.

Time of deciding on surgery after admission

The decision for decompressive craniectomy was taken at admission itself in 19/44 (43%) patients. The factors that influenced decision for taking patients for immediate decompression on univariate analysis were the degree of midline shift (table 4) and the infarct volume.

Table 4

Time of decision for surgery and survival of the 44 patients with CVT who underwent decompressive craniectomy

In the group who underwent immediate decompression, 12/19 (63%) had midline shift of ≥10 mm on the baseline imaging compared to only 3/25 (12%) in the delayed surgery group (p=0.0009). Also, the mean infarct volume in those who underwent immediate decompression was 146.63±52.5 ml (range 65–240 ml) versus 73.5±44.2 ml (range 15–186 ml) in delayed decompression group (p<0.001).

Clinical parameters at admission like low Glasgow Coma Scale (GCS) (p=0.5724), pupillary asymmetry (p=0.810), papilloedema (p=0.278) and fever (p=0.406) were not found to have any influence on the decision regarding immediate decompression.

Of the 25/44 (56.8%) patients in whom conservative management was attempted, 22/25 (88%) had a drop in GCS before undergoing surgery. Of these 6/25 (24%) developed pupillary asymmetry before undergoing surgery (two with fixed and dilated pupils). Repeat imaging was done in 18/25 (72%). New infarcts were noted in 8/25 (32%), and 17/25 (68%) had increase in the size of the old infarct. In 11/25 (44%) there were new areas of haemorrhage.

There were 10 patients with bifrontal infarcts due to anterior superior sagittal sinus thrombosis. Of these, 5/10 (50%) did not have any midline shift. The infarct volume in this group was < 30 ml except in one. Only this patient who had a large infarct volume of 117 ml got operated within the first 12 h and survived; in all the remaining 4/5 patients, surgery was done after 12 h and all 4 died.

There were five patients with acute vein of Labbe thrombosis. All of them had haemorrhagic venous infarcts with associated transverse—sigmoid sinus thrombosis. Without superior sagittal sinus involvement, only one patient in this group had a midline shift of >10 mm.

In all the 9/44 patients who had no midline shift on the initial imaging (five with anterior superior sagittal sinus involvement and three with vein of Labbe infarcts) repeat imaging showed extension of the earlier infarction with midline shift and herniation. Also new infarcts (5/9, 55%) and new areas of haemorrhage were seen (8/9, 88.8%).

Postoperative complications

Significant immediate postoperative complications were seen in nine patients (20%). Two patients had cerebrospinal fluid (CSF) leak, one each being associated with flap necrosis and pyogenic meningitis. Another patient had a brain abscess at the site of the previous infarct. Three patients had postoperative septicaemia and one patient had coagulopathy. One patient who had intraventricular extension of the haemorrhage developed obstructive hydrocephalus postoperatively and needed external ventricular drain insertion. No permanent CSF diversion procedure was performed. One patient developed arterial infarct in the posterior cerebral artery territory and worsened. Two patients developed bilateral extensive lower limb deep venous thrombosis (IVC filter was placed in one) and one patient had a fatal pulmonary embolism.


Mortality in the patients with CVT who had to undergo decompressive craniotomy was 9/44 (20.4%). There were two patients whose condition had worsened and were discharged at request; they were also included as mortality. Among the patients who died, three were due to non-neurological complications. One patient had recovered to an mRs of 3, but on the tenth postoperative day, had a fatal pulmonary embolism. One 60-year-old lady died following complications of an acute coronary syndrome and one patient succumbed following Methicillin resistant Staphylococcus pneumonia and sepsis.

The variables that were found to be significant predictors of mortality after surgery (table 5) on the univariate analysis (25% level of significance) were age >40 years, deep venous system involvement, involvement of >2 dural sinuses, acute on chronic presentation, surgery delayed by >12 h and involvement of cortical veins. On multivariate analysis (5% level of significance) only age >40 years and delay in surgery by >12 h had significant influence on the mortality—area under curve 0.867 (95% CI 0.755 to −0.978, p=0.001). Out of two patients who had bilaterally fixed and dilated pupils before surgery, one survived.

Table 5

Predictors of postoperative mortality for decompressive craniectomy for cerebral venous thrombosis

Follow-up outcome

Out of the 35/44 survivors, 3/35 (8.5%) were lost to follow-up. There were 32/35 (91%) patients with a follow-up of 3 months, 30/35 (85.7%) with a follow-up of 6 months and 26/35 (74%) who were followed up for more than a year. Cranioplasty was done in 6/35 (17%). There were 16 patients with a follow-up of more than 2 years. The average follow-up was 25.5 months (range 3–66 months).

Between 6 months and 1 year, four patients (mRs 2=2, mRs 1=1 and mRs 3=3) were lost to follow-up. In the remaining patients, the mRs was continuing to improve even after 6 months (table 6). Overall at 6 months and 1 year 27/35 (77%) had achieved an mRs of ≤2.

Table 6

Modified Rankin Score of 35 survivors who underwent decompressive craniectomy

During the follow-up, 9/35 (25.7%) developed complications requiring hospitalisation/emergency room visits. Seizures 3/33 (9%) was the most common complication. One patient had breakthrough seizures and two had drug withdrawal seizures due to poor compliance. One patient developed severe carbamazepine induced rash, one patient who had a bifrontal decompressive craniectomy with duraplasty developed an extradural abscess which was drained. One patient developed sunken flap syndrome (syndrome of the trephined). One patient developed anticoagulant-induced GI bleed requiring blood transfusion. One patient became pregnant after 18 months and underwent medical termination of the pregnancy. One patient had a recurrence of CVT after 54 months.


The cause of death in CVT is mainly due to increased intracranial pressure (ICP) leading to transtentorial herniation. In CVT, the ICP rises due to various mechanisms: (1) mass effect secondary to infarction—especially if there is associated haemorrhage; (2) venous obstruction; (3) obstruction to CSF flow and absorption, especially if the dural sinuses with the arachnoid granulations are involved and (4) brain swelling secondary to ischaemia. Venous stroke with haemorrhagic infarcts, especially on the right side, has a worse prognosis.11 In patients with CVT, the predictors for death at the time of admission are seizures, low GCS score <9, radiological findings of deep venous system involvement and posterior fossa involvement.8 For the subset of CVT patients with high risk of mortality, many modalities of therapies are available. Selective catheterisation and endovascular urokinase infusion12–15 and mechanical thrombectomy devices16–18 have been described. However, unlike in arterial strokes, the role of decompressive craniectomy is not well established in CVT. In ISCVT2 (624 patients), decompressive craniotomy was performed only in 1.4%. Stam et al19 tried endovascular thrombectomy and thrombolysis for patients with severe CVT. In their series, patient with rapid onset of symptoms and large haemorrhagic infarcts with midline shift had a higher mortality even with endovascular treatment. The same authors subsequently treated such patients with large haemorrhagic venous infarcts successfully with decompressive surgery.20

Bousser's group21 has coined a new term ‘malignant CVT’ for a subset of patients with supratentorial venous infarcts due to superficial venous system thrombosis having decreased consciousness and dilated pupils or radiological signs of transtentorial herniation.

In this study, we have seen that patients with bifrontal infarcts due to anterior superior sagittal sinus thrombosis may not have large infarct volume and midline shift on the baseline imaging. Similarly, patients with vein of Labbe thrombosis can develop uncal herniation without much midline shift. In this subset of patients, surgery was deferred due to the initial imaging findings and had a higher mortality. This implies that there is a subset of patients not fitting into the definition of malignant CVT at presentation but are at a high risk for early deterioration.

We had decided on decompressive craniectomy in 81% of the cases within the first 24 h. Claudia et al22 using perfusion and diffusion MRI have demonstrated a penumbra like state of metabolically compromised but still viable brain tissue in humans with CVT. Diffusion-weighted imaging can also demonstrate early ischaemic changes, and can differentiate conventional T2-weighted MR areas of cytotoxic from vasogenic oedema. In view of this potential for late improvement in cerebral tissues affected by CVT, it is better to avoid resecting brain tissue during surgery.

This series shows that decompressive craniectomy in CVT is not only life-saving but that the survivors can have an excellent long-term outcome. Compared with arterial strokes where an mRs of 2 after 12 months was achieved only in 14%,9 we had 68% of survivors having an mRs of ≤2 by 6 months itself (table 6). It is interesting to note that the mRs improved maximally after the third month. Similar findings were observed by Bousser's group.21

This is the largest series so far, to our knowledge, on decompressive surgery for CVT from a single centre (table 7).

Table 7

Studies on decompressive surgery for CVT from the literature

In a recent systematic review29 of 69 patients from 22 centres who had undergone decompressive surgery (craniectomy or haematoma evacuation) for CVT, unfavourable outcome (mRs score of 5 or death) was seen in only 12/69 (17.4%).

The present study has limitations. We had patients who were advised decompression, but opted against and got discharged against medical advice. We don't have follow-up on these cases; these cases can cloud such a retrospective comparison between the surgical and the non-surgical arms. These are the limitations of this being a retrospective study.

In conclusion, decompressive surgery should be considered in large malignant venous infarcts with midline shift. Survivors can expect excellent outcome. In patients with anterior superior sagittal sinus thrombosis with bifrontal infarcts and vein of Labbe thrombosis, the initial infarct volume and midline shift can be deceptive and such patients will need close monitoring for prompt intervention. A randomised controlled study may be required for early patient selection and identifying the optimal time for surgical intervention.


The authors thank Ms Grace Rebekah J Msc Biostatics, Dept of Biostatics, Christian Medical college, Vellore, India for helping with the data analysis.



  • Contributors Decompressive craniectomy in cerebral venous thrombosis—a single centre experience. SA DM (Neurology): Neurology Unit, Patient follow-up, Data collection, analysis, compilation of the medical, surgical and radiological information and writing the paper. RKM MCh (Neurosurgery) Neurosurgery Unit. Data collection and analysis of the surgical details and writing the paper. SM MD (Radiology): Department of Radiology. Data collection and analysis of the radiological details and writing the paper. VM DM (Neurology): Neurology Unit, Patient follow up, data collection, analysis and writing the paper. AKBP DM (Neurology): Neurology Unit. Patient follow up, data collection and analysis. AS DM (Neurology): Neurology Unit. Data collection and analysis. SN MD (TB & CD) Asst. Professor Neurocritical care unit, Analysis, writing up the intensive care aspects. MJ MCh (Neurosurgery) Neurocritical care unit Analysis, writing up the intensive care aspects. MT DM (Neurology): Neurology Unit, Analysis and writing up the paper. KP MCh (Neurosurgery). Analysis of the surgical details and writing the paper. BVJ MCh (Neurosurgery) Neurosurgery Unit, Analysis of the surgical details and writing the paper. VR MCh (Neurosurgery) Neurosurgery Unit. Analysis of the surgical details, writing the paper and corrections. AGC MCh (Neurosurgery) Neurosurgery. Analysis of the surgical details, writing the paper and corrections.

  • Competing interests None.

  • 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.