Occlusive hyperemia: a new way to think about an old problem

Abstract

A L-RODHAN and his coworkers at the Mayo Clinic' have introduced the new term to explain on a hemodynamic basis the edema, either with or without hemorrhage, that may complicate excision of arteriovenous malformations (AVM's), usually larger AVM's with high flows. An identical occlusive hyperemia may follow preoperative embolization and, in our institution for one, postembolization occlusive hyperemia has occurred more often than postexcision occlusive hyperemia. In fact, we have encountered a handful of cases in which occlusive hyperemia (cerebral edema with associated hemorrhage) caused abrupt neurological deterioration after spontaneous thrombosis of a major draining vein, a condition that we have termed overload. Venous overload is a common accompaniment of both symptomatic and asymptomatic AVM's, with the pathognomonic findings of venous outflow restriction and aberrant venous drainage patterns shown by angiography and findings of cerebral edema demonstrated on computerized tomography and magnetic resonance (MR) imaging. In our opinion, the evolution of a nidus aneurysm or angiographically obvious venous outflow restriction (or both) precedes rupture of many, if not most, AVM's that have spontaneous hemorrhage as their mode of presentation. Finally, precisely the same manifestations of occlusive hyperemia accompany dural arteriovenous fistulas (AVF's) when venous outflow becomes restricted and eventually occluded. Apart from ruptured aneurysms, interference with venous drainage represents the single factor responsible for almost all clinically significant complications of AVM's and AVF's preceding, during, and following treatment. Restriction of venous outflow becomes the most critical determinant of what happens within and surrounding an AVM nidus. What are the exceptions? First, rupture of an AVM nidus (which, incidentally, is believed to occur nearer the venous than the arterial end of the nidus) can undoubtedly occur in the absence of any venous occlusive process. Second, pure postoperative hemorrhage from a retained portion of an AVM represents rebleeding, whether it is a matter of blowing out a clotted plug or of arterial overload as blood is diverted to the retained nidus. However, in the instance of a retained nidus coupled with venous restriction, the venous overload (occlusive hyperemia) becomes malignant! To eliminate or reduce the possibility of a retained nidus, we rely on intraoperative postresection angiography and strongly recommend it. AI-Rodhan, et aL,~ do not speculate on the wide range of intervals, from less than 3 hours to 11 days, between the time the patient leaves the operating room and the onset of neurological deterioration. We suspect that early deterioration, possibly during the 1st day or so, reflects differing degrees of venous outflow restriction, and that later deterioration, certainly after several days, can be attributed to outright delayed thrombosis of one or more critical draining veins or even a dural venous sinus. Clearly, the matter of early versus late delay is not a black-and-white issue, but at least the possible mechanisms fit more or less into such a pattern. Stagnation of arterial flow may be difficult to characterize in a background of postoperative edema, intraoperative occlusion of feeding arteries, and coexisting venous obstruction. We have the impression that stagnation of arterial flow is common in the immediate postoperative period and more likely to occur after resection of larger AVM's. The person directing postoperative care may unwittingly promote venous thrombosis by keeping the patient dry as a well-accepted means of reducing postoperative reactive edema. Better postoperative management would be to assure an adequate blood volume, assiduously avoid hemoconcentration or even a hematocrit over 35%, and immediately after the operation commence a course of aspirin and low-dose heparin therapy. The known and unknown factors that cause spontaneous extracranial venous thrombosis after craniotomy must apply intracranially as well. A1-Rhodan, et al., ~ present two cases as examples, and both cases deserve comment. In Case 1 the respon-