Selection of treatment modalities for cerebral arteriovenous malformations: a retrospective analysis of 348 consecutive cases

The objective was to establish the selection criteria for the optimal manag ement modalities for cerebral arteriovenous malformations. We analysed the complications and late outcomes in 348 consecutive cases (132 microsurgical resections, 202 stereotactic radiosurgeries, 8 embolisations only, 6 combi ned treatments) managed at Yonsei University Hospital from 1988 to 1997. Fi les for all patients were analysed. The outcome was classified into good fo r the patients who returned to their previous job with or without neurologi cal deficits, fair for the patients who were unable to return to work but p erformed daily activities independently with minor deficits, and poor for t he patients who were performing dependent daily activities with major defic its. The outcome of microsurgery was considered good in 108 patients (81.8% ), fair in 18 (13.6%), poor in 4 (3.1%), and 2 (1.5%) patients died. Initia l insults and haemodynamic complications were the major cause of an unfavou rable outcome. The cumulative occlusion rate of the nidus after radiosurger y was 10.2% within 12 months, 75.3% within 24 months, and 89.8% within 36 m onths. Perilesional imaging changes with neurological deficits (4 permanent and 6 transient, 4.8%) and haemorrhage (16 patients, 7.7%) during the late nt interval were the major cause of an unfavourable outcome (1 poor, 4 dead after radiosurgery). Postradiosurgery bleeding occurred frequently within 6 months (6 patients), and between 13 and 24 months (8 patients). In conclu sion, selection of treatment modality for cerebral AVMs depends on the preo perative evaluation of the risk/benefit ratio in each case. Microsurgical r emoval, which eliminates the risk of bleeding immediately, is preferred for lesions in non-eloquent areas. Radiosurgery is an effective treatment moda lity for small lesions in eloquent areas, but has a substantial risk of hae morrhage during the latency period. Results of this study suggest that micr osurgical removal should be considered for lesions in Eloquent areas with h igh haemorrhage risk, such as prior haemorrhage, medium to large size lesio n, and single deep venous drainage. (C) 2000 Harcourt Publishers Ltd.