The validity of classification for the clinical presentation of intracranial dural arteriovenous fistulas.

A number of classification schemes for intracranial dural arteriovenous fistulas (AVFs) have been published that claim to predict which lesions will present in a benign or aggressive fashion based on radiological anatomy. We have tested the validity of two proposed classification schemes for the first time in a large single-institution study. A series of 102 intracranial dural AVFs in 98 patients assessed at a single institution was analyzed. All patients were classified according to two grading scales: the more descriptive schema of Cognard, et al. (Cognard) and that recently proposed by Borden, et al. (Borden). According to the Borden classification, 55 patients were Type I, 18 Type II, and 29 Type III. Using the Cognard classification, 40 patients were Type I, 15 Type IIA, eight Type IIB, 10 Type IIA+B, 13 Type III, 12 Type IV, and four Type V. Intracranial hemorrhage (ICH) or nonhemorrhagic neurological deficit was considered an aggressive presenting clinical feature. A total of 16 (16%) of 102 intracranial dural AVFs presented with hemorrhage. Eleven of these hemorrhages (69%) occurred in either anterior cranial fossa or tentorial lesions. When analyzed according to the Borden classification, none (0%) of 55 Type I intracranial dural AVFs, two (11%) of 18 Type II, and 14 (48%) of 29 Type III intracranial dural AVFs presented with hemorrhage (p < 0.0001). After exclusion of visual or cranial nerve deficits that were clearly related to cavernous sinus intracranial dural AVFs, nonhemorrhagic neurological deficits were a feature of presentation in one (2%) of 55 Type I, five (28%) of 18 Type II, and nine (31%) of 29 Type III patients (p < 0.0001). When combined, an aggressive clinical presentation (ICH or nonhemorrhagic neurological deficit) was seen most commonly in intracranial dural AVFs located in the tentorium (11 (79%) of 14) and the anterior cranial fossa (three (75%) of four), but this simply reflected the number of higher grade lesions in these locations. Aggressive clinical presentation strongly correlated with Borden types: one (2%) of 55 Type I, seven (39%) of 18 Type II, and 23 (79%) of 29 Type III patients (p < 0.0001). A similar correlation with aggressive presentation was seen with the Cognard classification: none (0%) of 40 Type I, one (7%) of 15 Type IIA, three (38%) of eight Type IIB, four (40%) of 10 Type IIA+B, nine (69%) of 13 Type III, 10 (83%) of 12 Type IV, and four (100%) of four Type V (p < 0.0001). No location is immune from harboring lesions capable of an aggressive presentation. Location itself only raises the index of suspicion for dangerous venous anatomy in some intracranial dural AVFs. The configuration of venous anatomy as reflected by both the Cognard and Borden classifications strongly predicts intracranial dural AVFs that will present with ICH or nonhemorrhagic neurological deficit.