MR Evaluation of Brain Perfusion after Radiosurgery of Cerebral Arteriovenous Malformations: A Surgeon’s View

In this issue of the AJNR, Guo et al use MR imaging independent component analysis to show the effects of radiosurgery on the perfusion of cerebral arteriovenous malformations (AVMs). Their analysis shows varied perfusion disturbances involving AVMs and surrounding brain tissue and the gradual changes toward normal perfusion after radiosurgery.

Although the thrust of this study was to demonstrate changes in vascular steal phenomena, the prime objective of stereotactic radiosurgery of AVMs is to prevent hemorrhage (1). Nevertheless, such MR techniques may be of clinical utility. The biggest question raised by this study is whether employing this technique routinely would add significantly to the simple visual inspection of routine MR imaging sections.

The effects of radiosurgery on AVMs occur over a prolonged period as the radiated vessels develop endothelial changes, smooth muscle cell proliferation, and hyalinization and eventually become occluded (2). In general, 1–2 years are required for obliteration of the entire AVM. On occasion, significant change in the lesion may be seen as early as 6 months after treatment, and obliteration may occur as long as 3 years after treatment. Routine follow-up MR imaging of AVMs that respond to radiosurgery shows reduction in nidus size and associated abnormal vasculature, areas of elevated T2 signal intensity surrounding the lesion, and variable degrees of contrast enhancement in and around the lesion. MR imaging may eventually strongly suggest obliteration of the AVM, but usually conventional angiography is used to confirm obliteration. Very small residual lesions may occasionally be seen on follow-up angiograms, particularly signaled by persistent early draining veins that are not well delineated on MR images.

The additional characterization of perfusion of the AVM and surrounding brain might be helpful in early prediction of ultimate obliteration. Collection of additional patient data with long-term follow-up and angiographic correlation might show that this technique can supply data that would help one predict whether an AVM would eventually show a good response to radiosurgery. This might allow a significant degree of reassurance to the patient who is waiting for that response to provide protection from hemorrhage. In terms of risk assessment, perhaps changes in the perfusion data, such as the rate of change of perfusion, might predict the risk of development of significant edema during the follow-up period.

Contrary to what might be inferred from advertising sometimes directed at the lay public, radiosurgery of AVMs is not a uniformly successful enterprise (3, 4). Radiographic follow-up is crucial, and at times radiosurgery of an AVM is repeated if the lesion is not totally obliterated after an appropriate duration of follow-up (5). Conceivably, an AVM might appear obliterated on follow-up MR images, but a residual abnormality on the perfusion study might predict that a small remaining defect may well be present, thus drawing closer attention to detailed follow-up conventional angiography. A lack of change of perfusion data at some relatively early point in follow-up might help one predict eventual inadequate response and raise the possibility of earlier radiosurgical retreatment.

Most neurosurgeons accept the dictum that the patient is not protected from hemorrhage unless the AVM is totally obliterated. Some hold the view that partial obliteration may be somewhat protective and better than no treatment, and there has been some suggestion that radiosurgery affords some reduction in hemorrhage rate even if the lesion is not seen to be obliterated. On the other hand, others hold the view that reducing the size of the AVM without completely obliterating it may increase the risk of hemorrhage because of changes in hemodynamics, with the smaller lesion still subjected to a similar pressure gradient. In cases of incomplete obliteration, follow-up MR perfusion data might separate subgroups with some protective effect of partial obliteration from groups with no protective effect or increased risks of hemorrhage.

Like all studies of stereotactic radiosurgery of nonmalignant disease, determination of the utility of this technique will require extended follow-up of many patients. The patients in this study have been followed up for a relatively limited period. None of them have reached the point at which postradiosurgery angiography is needed. The longest follow-up was 25 months, with only seven of the 19 patients followed for at least 18 months. Only two AVMs were shown to be obliterated on the basis of MR imaging criteria. The AVMs in this study are relatively larger targets; certainly none of them can be called small. The smallest was 10 mL in volume. Application of this perfusion MR imaging technique to small AVMs would also be of interest.

It would be of great interest to correlate the changes visualized by use of MR perfusion imaging with seizure frequency in patients with epilepsy caused by their AVM. Perhaps changes in perfusion short of complete obliteration might affect seizure frequency. Nine of the patients presented with seizures alone as symptomatology; no discussion of response of their seizures to treatment exists.

There is considerable discussion in the article regarding development of radiation-induced edema surrounding the lesion. Some of the signal intensity change around the lesion may reflect gliosis rather than edema. Long-term follow-up will be necessary to differentiate the two, because edema should eventually resolve, and residual long-term T2 signal intensity changes should reflect mainly gliosis. There is no discussion as to whether the edema was symptomatic. The authors state that there was reduced perfusion caused by the radiation-induced edema. An alternative explanation for reduced perfusion in those areas would be actual occlusion of small blood vessels in the brain due to the irradiation.

The authors describe increased perfusion of the ipsilateral hemisphere relative to the normal hemisphere and implicate a steal phenomenon to explain this ratio. Although there is a relative difference, there is not clear proof that the elevated blood flow in the nidus hemisphere is coming at the expense of the contralateral hemisphere. It may be extra flow delivered only at the expense of extra work by the heart.

The authors have presented a technique for additional MR imaging characterization of the effects of stereotactic radiosurgery on brain AVMs and the surrounding parenchyma, and they appropriately speculate about the possible applications of this MR imaging technique for the assessment of endovascular and surgical treatment of AVMs. Further long-term follow-up of more patients with this perfusion technique would be of great interest and would be most useful if accompanied by clinical correlation with the neurologic manifestations of radiation-induced edema, seizure control, and ultimate angiographic and neurologic outcome.

• Copyright © American Society of Neuroradiology