Radiosurgery is a term applied to high precision localised irradiation given in one session. One technique uses cobalt sources arranged in a hemisphere and focused on to a central target (described as a gamma knife). A gamma knife unit has been in operation in Sheffield since 1986. Identical high precision radiosurgery can be delivered by appropriately adjusted linear accelerators and has been available in Britain since 1989. Currently, at least six radiosurgery facilities are available to NHS patients. The limited usefulness of the technique for treating brain tumours suggests that the existing NHS facilities are sufficient for the expected workload.

The aim of radiosurgery is to deliver a sphere of high dose irradiation more localised than would be achieved with conventional radiotherapy. However, this is possible only for small lesions less than 3.5-4cm in diameter. Radiosurgery was used initially for treating inoperable arteriovenous malformations and subsequently for treating acoustic neuromas, solitary brain metastases, and a mixture of other tumours. Despite many years of experience, there is no single randomised trial or robust case-control study testing the efficacy and safety of radiosurgery in comparison with other established treatments. Most reports claiming benefit are from retrospective studies of enthusiastic application of radiosurgery to patients with small brain tumours.

It is generally agreed that single fraction radiosurgery obliterates 80-90% of small arteriovenous malformations. The aim is to reduce the risk of subsequent haemorrhage from an annual untreated rate of rebleeding of 2-4%. In the first two years after radiosurgery the reported annual rebleeding rate is 4-8%,^2,3 and long term data on the incidence of rebleeding at 5-10 years are poor. No information exists on the survival of treated compared with untreated patients. The treatment is not without toxicity: the risk of radiation induced damage seen on magnetic resonance imaging is 20-30% for 2 cm and 40-50% for 3 cm diameter lesions, and these are often symptomatic when in eloquent regions of the brain.⁴

The tumour control of acoustic neuroma after radiosurgery is 91% at five years with a 17% risk of VIIth and a 45% risk of VIIIth neuropathy at five years.⁵ Radiosurgery has been advocated for patients with other benign tumours. Early results suggest a recurrence rate of small benign meningiomas of >10% at five years with a 6% risk of neurological toxicity.⁶ Long term tumour control of pituitary adenoma after radiosurgery is not known. However, damage to the optic apparatus, causing visual impairment, has been reported in 3.4%⁷ and 24%⁸ of patients after radiosurgery at two years, and 30% patients developed temporal lobe damage.⁹ Optic neuropathy is related to the proximity of the tumour to the optic apparatus and limits the technique to small sellar lesions. In contrast, the risk of radiation induced visual impairment after conventional fractionated radiotherapy is 1-2%, with a tumour control rate of about 90% at 20 years for adenomas of all sizes.¹⁰ Radiosurgery of benign pituitary adenomas and meningiomas is associated with mortality ranging from 1.6%⁷ to 24%.¹¹ The deaths are a direct consequence of damage from high dose single fraction radiation and have not been reported after fractionated treatment.

Few effective treatments exist for brain metastases and malignant gliomas, and patients will accept any offer of hope. Radiosurgery is a non-invasive alternative to surgery for solitary brain metastases. The median survival is 6-12 months and is related to performance and the state of systemic malignancy.¹² Radiosurgery has no advantage over whole brain radiotherapy for patients with multiple brain metastases.¹³

Radiosurgery alone is not the appropriate primary therapy for malignant gliomas. However, a boost after conventional surgery and radiotherapy of malignant glioma is claimed to be associated with marginal prolongation of survival, but this may be explained by patient selection.¹⁴ Two multicentre randomised studies in the United States and Europe are currently examining this issue. Patients with malignant brain tumours should be encouraged to take part in trials to define the role of radiosurgery in treatment.

Radiosurgery as delivered by the gamma knife has major limitations. Each “shot” consists of a small radiation sphere 4-18 mm in diameter, and these need to be multiple for the treatment of larger lesions. In contrast, linear accelerator techniques offer the possibility of treating larger and moreirregular lesions with a technique described as conformal stereotactic radiotherapy. Single fraction radiation as delivered by the gamma knife in doses greater than 8 Gy to critical structures is associated with a high risk of injury. Giving treatment in multiple small doses (the principle of fractionation) allows for higher radiation doses to the tumour without increased risk of damaging the central nervous system. Conformal stereotactic radiotherapy coupled with fractionation is increasingly being explored as a potentially safer method of delivering high precision localised irradiation.