Statistics from Altmetric.com
Multiple concurrent cerebral haemorrhages in the absence of trauma or a bleeding diathesis suggest venous sinus thrombosis, multiple haemorrhagic infarcts, and haemorrhagic metastases. Iatrogenic venous obstruction is another possible cause. Patient 1 was a 55 year old obese woman who underwent a left posterior fossa craniotomy (Jannetta procedure) for trigeminal neuralgia. Examination and a CT of the head were normal. At surgery she was positioned on her right side with her head held in slight lateral flexion to place the left occipital area uppermost. During the operation the anaesthetist reported that on two occasions the pulse rate slowed only to return to normal when the retractor was immediately removed. The patient’s head was repositioned to provide increased lateral flexion and because of her short neck and broad shoulders, slight Trendelenberg tilt was applied. A large artery was found indenting the left fifth nerve. On repositioning the retractor, swelling of the left hemisphere was noted which bulged though the craniectomy. A small cerebellar resection was performed and the head repositioned with upward tilt. The operation was then abandoned.
Postoperatively she failed to regain consciousness. Brain CT showed moderate swelling of the left cerebellar hemisphere and both occipital lobes appeared hypodense. There were also high density haemorrhagic lesions in the left temporal, right frontal and occipital lobes. She died the next day.
At postmortem the brain was swollen and markedly congested with patchy areas of subarachnoid haemorrhage scattered over the cerebral convexities. The venous sinuses and cortical veins were patent. The vessels of the circle of Willis were patent and free of atheroma. Multiple haemorrhagic venous infarcts were present in the right anterior frontal (40×30×20 mm), right posterior frontal (30×20×20 mm), right parietal (15×10×10 mm), right temporal (20×20×10 mm), right occipital (25×10×20 mm) and left posterior frontal (50×35×40 mm) lobes. The overlying subarachnoid space and sulci in the affected areas were widely distended with blood (sulcal haematomas) and adjacent cerebral cortex and underlying subcortical white matter were disrupted by confluent small haemorrhages (haemorrhagic infarction of venous type). Microscopical assessment of the haemorrhagic lesions disclosed venous and capillary congestion and coalescence of multiple microhaemorrhages typical of venous obstruction. The left cerebellum was lacerated and swollen.
In 1978 Jannetta et al reviewed 825 cases of posterior fossa surgery.1 Five were complicated by supratentorial haemorrhage, and in four the cause remained elusive. Two came to postmortem examination disclosing haemorrhage without evidence of underlying neoplasia or vascular malformation but it was not clear whether they were arterial or venous. It was considered that patient position (modified sitting) may have been in some way implicated and subsequent procedures were done in the lateral decubitus position. In 1988 Hanakita and Kondo2 reviewed 278 patients who underwent microvascular decompression all in the lateral decubitus position. There were two fatal intracerebral haemorrhages but there was no postmortem. Excessive cerebellar traction and a disturbance of venous return with an increase in blood pressure was assumed to be the cause in one, but no explanation was given for the second.
The haemorrhages which occurred in our case were simultaneous and widespread and pathologically were venous yet the venous sinuses were patent macroscopically.
We postulate then that the venous haemorrhages were due to acute intraoperative venous hypertension secondary to neck positioning, possibly contributed to by the short thick neck. It is not the lateral decubitus position versus the sitting position which so important but the neck positioning.
In support of the hypothesis of mechanical venous obstruction, Gooding and Stimac3 have demonstrated in animal and infant cadaver models that turning the head to one side results in torsion and compression of the ipsilateral jugular vein. They postulate that jugular venous occlusion on the side of the dominant venous drainage can result in severely limited cerebral venous drainage. Emerson and Parker4 showed that with obstruction there was a marked reduction in cerebral blood flow with an increase in cerebral venous pressure to 25 mm Hg. Doppler studies have shown that cranial venous drainage is often asymmetric, being more dominant on the right and with unilateral venous occlusion, efferent contralateral shunting of blood is only possible to the dominant side. It was the right side of the neck which was compressed at surgery in our case.
Case 2 was a 40 year old woman with cryptogenic cirrhosis who underwent orthotopic liver transplantation. On admission for transplantation a central venous line was inserted into the right internal jugular vein. A Swan-Ganz catheter was inserted into the left internal jugular vein intraoperarively.
Postoperatively transient thrombocytopenia was noted, the platelet count falling to 68 000 rising to 100 000 by day 2 and 335 000 by day 9. On day 4 the Swan-Ganz catheter was removed and on day 7 the central venous line was taken out. In addition to cirrhosis of uncertain aetiology the liver showed multifocal well differentiated hepatocellular carcinoma.
On the ninth post-operative day she complained of dizziness, blurred vision, and headache and the next day she was transiently dysphasic. Brain CT disclosed multiple superficial and deep haemorrhages in both hemispheres (fig 1). There were no cutaneous or systemic bleeding manifestations at this or any other time. Brain MRI confirmed the haemorrhagic lesions and disclosed patent venous sinuses. A subsequent ultrasound of the jugular veins 4 days later showed circumferential thrombus in both the right and left internal jugular veins with some venous flow. A repeat CT 1 month later was normal (fig 2). The patient remains well.
Neurological complications after liver transplantation include cerebral haemorrhage but we could find no case report of multiple perioperative intracerebral haemorrhages without a significant bleeding diathesis.
Radiologically there were multiple haemorrhagic lesions without venous sinus thrombosis. Clinically there was no coagulopathy, no vasculitis, or evidence of haemorrhagic metastases. The complete resolution of the haemorrhages without evidence of encephalomalacia and the multifocal haemorrhages not conforming to any particular arterial territory supported the hypothesis that they were venous.
We therefore postulate that jugular venous catheterisation induced jugular venous thrombosis and venous hypertension with secondary multiple haemorrhages. There is one case5 reported of an infant aged 2 months who died of multiple cerebral haemorrhages thought to be produced by bilateral internal jugular vein thromboses caused by jugular vein catheters.
The concept of venous hypertension causing cerebral haemorrhage is not generally recognised. Neck positioning or traction and jugular vein catheterisation causing extracranial venous obstruction are two possible mechanisms with important management implications.
We thank the Department of Radiology and Department of Media and Illustration, Flinders Medical Centre, for their assistance in the preparation of radiology and pathology illustrations. We also thank Vicky Frangoulis for manuscript preparation.
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.