Angioarchitecture associated with haemorrhage in cerebral arteriovenous malformations: a prognostic statistical model

Neuroradiology. 1997 Jan;39(1):52-8. doi: 10.1007/s002340050367.


The overall haemorrhagic risk of a cerebral arteriovenous malformation (cAVM) is 2-4% per year. However, the individual risk of haemorrhage has never been determined. This study was undertaken to assess the haemorrhage risk of an individual cAVM. Neuroangiographic findings of 160 cAVM were analysed retrospectively, looking at 30 angiographic features. A statistical model was established by logistic regression to evaluate the risk of an individual cAVM. We statistically correlated 15 parameters with the haemorrhage risk. The statistical model includes five independent parameters. Four are unfavourable: exclusively deep drainage, venous stenoses, venous reflux and the radio of afferent to efferent systems; one is favourable: venous recruitment. This model quantifies the individual risk of haemorrhage. When this model is applied to the population studied, the error rate is 5%. This model can contribute to therapeutic strategy, and to a better understanding of the natural history of cAVM.

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Cerebral Angiography
  • Cerebral Arteries / diagnostic imaging*
  • Cerebral Hemorrhage / etiology*
  • Cerebral Veins / diagnostic imaging*
  • Child
  • Female
  • Humans
  • Intracranial Arteriovenous Malformations / complications*
  • Intracranial Arteriovenous Malformations / diagnostic imaging*
  • Logistic Models
  • Male
  • Middle Aged
  • Prognosis
  • Retrospective Studies
  • Risk Factors