Volume determination and predictive models in the management of cerebral arteriovenous malformations

Abstract

Cerebral arteriovenous malformations (AVMs) entail a high cumulative risk of severe neurological symptoms and are usually treated with surgery, radiosurgery, embolisation or combinations thereof. There are well supported models to predict the outcome of radiosurgery, but similar models do not exist for surgery or embolisation. The aim of this work was to improve the models and implement them in the general management of AVM patients, notably those treated by embolisation.

Aims of the present investigation: 1) To improve predictive models for radiosurgery and adapt them to use without dose planning. 2) To create a method for uncomplicated measurement of AVM volume from angiography. 3) To devise a method to correct for the geometric distortion in digital subtraction angiography. 4) To make a hypothetical comparison of two major management strategies, using the methods and models as reference standard.

Materials and methods: 1) Roughly 1500 patients treated with Gamma Knife radiosurgery for AVM during the period 1970-1993 were retrospectively studied. 2) The "intersecting cone model" for AVM measurement was created and validated. 3) Qualitative and quantitative aspects of the geometric distortion were investigated and corrected for with a calibration-correction scheme. 4) A prospective study of 88 consecutive patients was done during 1997-1999. AVM volumes and clinical information were recorded before the first and after the last embolisation.

Results and conclusions: 1) The obliteration rate depended only on the minimum dose to the periphery of the AM The complication rate depended also on AVM volume, location and patient history. The risk for haemorrhage in the latency period after radiosurgery depended on the minimum dose to the AM patient age and AVM Volume. 2) The "intersecting cone model" call be used instead of volume data from the dose planning system. 3) Geometric distortion may affect high precision stereotaxy but its impact on volume measurement is limited. It can be corrected. 4) For AVM> 1oml the outcome models are not very well substantiated. However, embolisation had comparative advantages. For AVM=1oml the hypothetical outcome of combined treatment with embolisation and radiosurgery was equal to that of radiosurgery alone.

Outcome models from radiosurgery can be used in clinical practise as a reference standard in the management of AM However, further studies are necessary to identify specific patient and AVM subgroups amenable to each treatment.