How safe is arteriovenous malformation surgery? A prospective, observational study of surgery as first-line treatment for brain arteriovenous malformations

Andrew S. Davidson, Michael K. Morgan

    Research output: Contribution to journalArticlepeer-review

    86 Citations (Scopus)

    Abstract

    Objectives: Existing studies reporting the risk of surgery for brain arteriovenous malformations (AVMs) are often biased by the exclusion of patients not offered surgery. In this study, we examine the risk of surgery, including cases excluded from surgery because of the high surgical risk. Methods: Data were collected on 640 consecutively enrolled AVMs in a database that included all patients not considered for surgery. Results: Patients with Spetzler-Martin grade 1 to 2 AVMs (n = 296) were treated with a surgical risk of 0.7% (95% confidence interval [CI], 0%-3%); patients with Spetzler-Martin grade 3 to 4 AVMs in noneloquent cortex (n = 65) were treated with a surgical risk of 17% (95% CI, 10%-28%). Patients with Spetzler-Martin grade 3 to 5 AVMs in eloquent cortex (n = 168) were treated with a surgical risk of 21% (95% CI, 15%-28%). However, because 14% of patients in this series with similar AVMs were refused surgery because of perceived surgical risk, these results are not generalizable to the population of patients with similar AVMs. Conclusion: The results of this series suggest that it is reasonable to offer surgery as a preferred treatment option for Spetzler-Martin grade 1 to 2 AVMs. This study also reinforces the predictive value of the Spetzler-Martin grading system, with some caveats.

    Original languageEnglish
    Pages (from-to)498-504
    Number of pages7
    JournalNeurosurgery
    Volume66
    Issue number3
    DOIs
    Publication statusPublished - Mar 2010

    Fingerprint Dive into the research topics of 'How safe is arteriovenous malformation surgery? A prospective, observational study of surgery as first-line treatment for brain arteriovenous malformations'. Together they form a unique fingerprint.

    Cite this