Spontaneous Postembolization Hemorrhage

Spontaneous postembolization hemorrhage is among the most vexing and devastating complications of AVM embolization. There are multiple theories of spontaneous AVM hemorrhage before treatment, and it is believed that these factors contribute to hemorrhage after embolization as well, usually due to changes in transmural vessel pressures or blood flow patterns. The most important concept to be discussed in this regard is that of normal perfusion pressure breakthrough.

The normal perfusion pressure breakthrough hypothesis was posited in 1978 by Spetzler as a mechanism to explain postoperative hemorrhage seen in the brain after complete excision of a high-flow AVM. In normal brain parenchyma, perfusion pressures are maintained within a narrow window by vessel caliber regulation through means that are both extrinsic and intrinsic to the vessels themselves. Cerebral capillary perfusion pressures can be regulated over a wide range of arterial blood pressures as a result. Although extrinsic sympathetic regulation of vascular tone does play a role in the central nervous system, it is primarily through intrinsic smooth muscle autoregulation that capillary perfusion is maintained. Importantly, arterioles appear to have a greater ability to dilate in response to hypotension than they do to constrict in response to hypertension.

In the brain of a patient harboring an AVM, there is an area of low vascular resistance due to the loss of the capillary bed and enlargement of feeding arterial and draining venous calibers. Parallel vascular beds are therefore subjected to relative hypotension with respect to perfusion pressures. The AVM essentially presents a short circuit through which blood will preferentially flow. As a result, vessels irrigating normal brain near the AVM will become chronically dilated in a constant effort to maintain flow to these relatively ischemic regions of the brain. These vessels may either lose their autoregulation abilities, or their autoregulatory set points may be dramatically shifted to remain dilated even in the face of relatively high capillary perfusion pressures. Chronically low tissue oxygen tension may also result in effusive neovascularization with frail, “leaky” capillaries.

The result of these phenomena is a vascular system in which feeding arteries are chronically dilated with reduced capacity to respond appropriately to changes in perfusion pressure heads, and with downstream capillary beds that are prone to rupture. Embolization of an AVM results in decreased flow across the fistula and a compensatory increase in flow through these fragile systems. It would be expected, therefore, that postembolization hemorrhage would be common—and it is.

Several recent studies have evaluated hemorrhage risk after endovascular embolization. Asadi et al. reported post- or intraprocedural hemorrhages in 57 of 199 patients (29%) treated with endovascular embolization, 47% of which occurred spontaneously at a site remote from the embolization after uncomplicated completion of the embolization procedure. This hemorrhage rate is relatively high compared with other series, probably because the majority of patients in this series presented with hemorrhage from the AVM; in addition, almost all patients underwent multiple embolizations (i.e., the risk of complications from an individual embolization procedure is a fraction of these numbers), and an aggressive strategy of endovascular therapy was pursued, with 45% of patients achieving cure of the AVM without supplemental radiosurgery or open microsurgical resection.

Baharvahdat et al. reported post- or intraprocedural hemorrhages after endovascular embolization—again with the goal of complete obliteration of the lesion—in 92 of 408 patients (23% of patients, 11% of embolization procedures). Fifty-two percent of these complications were felt to be related to spontaneous hemorrhage unrelated to inadvertent vessel rupture or perforation during the procedure. Eighty-one percent of these occurred just hours or days after the procedure was complete, with a mean of 34 hours postprocedure. Once again, this study is remarkable for the fact that endovascular embolization was utilized as the sole therapeutic modality in 91% of cases. Spontaneous hemorrhages resulted in 37 patients suffering an acquired neurologic deficit, 58% of which resulted in permanent disability and 10% in death. Types of hemorrhages encountered included intracerebral, intraventricular, and subarachnoid hemorrhage; 71% of spontaneous hemorrhages were intraparenchymal. Importantly, the authors of this study identified premature embolization of the draining vein as a contributing factor in delayed hemorrhage, with 17% of hemorrhages occurring in this context, although other authors have reported successful transvenous embolization strategies in small patient samples.

In neurosurgical series using embolization as an adjunctive therapeutic modality followed by either microsurgical resection or stereotactic radiosurgery, hemorrhagic complications appear to be less common. Crowley et al. reported permanent neurologic morbidity from all causes in 9.6% of patients with only 1 death out of 327 patients treated. Six of 153 patients (4%) suffered unexpected postprocedural neurologic deficits secondary to hemorrhagic complications in an earlier, smaller series published by the same group. Only 13% and 2.3% of procedures in the smaller and larger series, respectively, were curative, whereas 70% and 79%, respectively, were performed before microsurgical resection of the lesion, thus highlighting differences in treatment strategies employed around the world and by different subspecialists. One should bear this in mind when comparing studies, because the surgical complications after craniotomy may not be included in the reported data. In an older surgical series, Taylor et al. reported permanent neurologic deficits after embolization in 18 of 201 patients (9%) and death in 4 patients (2%). Again, in this series embolization was used primarily as an adjunct to surgical resection.

In summary, periprocedural intracerebral hemorrhage after endovascular embolization of is most common in patients in whom: (1) complete embolization of the lesion is attempted in one session, (2) when embolic material enters the draining veins during subtotal embolization, and (3) when high-risk features (e.g., intranidal aneurysms) are left untreated. Hemorrhage in these situations is caused either by worsening of venous or intraaneurysmal hypertension or by normal perfusion pressure breakthrough. Avoidance of hemorrhage may therefore be accomplished by staging attempts at embolization, targeting high-risk features of the lesion early in treatment, and maintaining strict blood pressure control after embolization. The risks of hemorrhage from attempted endovascular cure of the AVM should be weighed against the risks of microsurgical resection, which are patient- specific but may be accurately predicted by the SM grading scheme.

At our institution, all cerebral AVMs are treated by neurosurgeons with training in both open surgical and endovascular treatment modalities, and with access to state-of-the-art radiosurgical therapy. Embolizations are generally staged with no more than 30% to 50% of the lesion embolized in a single session, except in very small lesions fed by a single artery, when cure may be achieved in a single sitting and partial embolization is not possible. Postoperatively, systolic blood pressures are controlled to <120 mm Hg, frequently requiring intravenous infusions of nicardipine. Labetalol is a second-line agent that is also used as a continuous infusion or, more commonly, on an as-needed basis for breakthrough hypertension. Postembolization, all patients are monitored in the neurosurgical intensive care unit at least overnight. Neurologic deterioration from delayed postembolization hemorrhage generally prompts immediate evacuation of the hematoma and resection of the AVM.