Background Transvenous embolization (TVE) is an emerging technique for the endovascular treatment of cerebral arteriovenous malformations (AVMs). The aim of this study was to investigate two in vivo AVM models and to assess TVE techniques using these models.

Methods Blood flow in the porcine rete mirabile (RM) was modified by either creating a carotid–jugular fistula or by placing a balloon guide catheter in the carotid artery. The RM was embolized with precipitating hydrophobic injectable liquid (PHIL) 25% via transarterial embolization (TAE; control group) and compared with TVE applying the transvenous retrograde pressure cooker technique and TVE using a Woven EndoBridge (WEB) device for flow control (n=6, respectively). The embolization extent (penetration of the RM), the number of events of reflux or embolization distal to the RM and the procedure times were assessed.

Results The modified RM could be successfully used for embolization in all cases. There were no significant differences regarding the outcome parameters between the two AVM models (fistula or balloon). TVE using the pressure cooker technique led to a higher extent of embolization (median 98.8% vs 63.5%; p=0.008), a lower number of reflux or distal embolization events (p<0.001) and a shorter procedure time (p<0.001) compared with conventional TAE. TVE using a WEB device for flow control was technically feasible and achieved a moderate extent of embolization (median 83.8%).

Conclusion After surgical or endovascular modification, the porcine RM is a feasible in vivo AVM model for the investigation of TVE techniques. TVE using the pressure cooker technique is superior to conventional TAE in this experimental model.