[Radiosurgery of cerebral arteriovenous malformations: a prescription algorithm].

PURPOSE: To study prognostic factors of obliteration and risk factors of brain radiation necrosis in order to propose an algorithm for radiosurgery prescription for cerebral arteriovenous malformations (cAVM).

MATERIAL AND METHODS: One hundred and seventy-nine patients were analysed. Radiosurgery delivered 6 or 10 MV X-rays by arc therapy in 84% of cases, or by fixed field in 16% of cases using two different micro-multileaf collimators (micro-MLC). Follow-up consisted of screening radiation necrosis by MRI every 6 months, and assessing local control by arteriography every 2 years. Obliteration was defined as at least 95% reduction of cAVM volume. Cox proportional hazard model was used to evaluate the local control and the appearance of radiation necrosis over time.

RESULTS: Local control rate was 82.7% with the mean follow-up of 3.1 years (0.5-11). Significant prognostic factors were: simple nidus (RR=2.8, p<0.0001), number of embolizations before radiosurgery below 4 (RR=2.9, p<0.0001), prescribed dose to the periphery of at least 18 Gy (RR=2, p=0.0002), nidus volume below8cm(3) (RR=1.9, p=0.0002), and number of table positions below six (RR=1.4, p=0.05). Radiation necrosis rate was 11.2% with a mean time to onset of 18 months. Significant predictive factors were: fixed field versus arc therapy (according to MLC RR=9.1, p<0.0001, and RR=15.1, p=0.01), age below 30 years (RR=2.5, p=0.04), depth of cAVM greater than or equal to 7 cm (RR=7.6, p=0.008), and volume of brain tissue covered by the 12 Gy isodose (V12 Gy) of at least 11 cm(3) (RR=7.8, p=0.05).

CONCLUSION: A radiosurgery prescription algorithm taking into account the prescribed dose to the periphery (> or = 18 Gy) and reduction of V12 Gy was elaborated from these data.