Charles J. Wrobel, Edward H. Oldfield, Giovanni Di Chiro, Edward C. Tarlov, Richard A. Baker and John L. Doppman
✓ Arteriovenous malformations (AVM's) of the spine commonly cause progressive myelopathy. Occasionally, myelography reveals serpentine filling defects characteristic of a spinal AVM, but an AVM or arteriovenous (AV) fistula cannot be demonstrated arteriographically, despite selective catheterization of all vessels known to have the potential of supplying the spinal cord and spinal dura. Often, and particularly in the setting of subacute or acute deterioration, this has been attributed to spontaneous thrombosis of the veins (the Foix-Alajouanine syndrome). Three patients are reported in whom intracranial dural AV fistulas, supplied by branches of the internal and external carotid arteries, drained into spinal veins and produced myelopathy. In one patient, motor and sensory deficits were limited to the lower extremities. In all three patients, disconnection of the fistula from its spinal venous drainage permitted arrest of a rapidly progressive myelopathy and partial recovery. These findings indicate that some patients who appear to have spinal cord AVM's but exhibit negative spinal arteriography are suffering from cranial dural AV fistulas and therefore need carotid as well as spinal arteriography. The considerable distance of these fistulas from the level of neurological expression supports venous hypertension as a pathophysiological mechanism of spinal cord injury. Interruption of a cranial dural fistula draining into spinal veins permits recovery of the myelopathy.
Bruce Rosenblum, Edward H. Oldfield, John L. Doppman and Giovanni Di Chiro
✓ The medical records and arteriograms of 81 patients with spinal arteriovenous malformations (AVM's) were reviewed, and the vascular lesions were classified as dural arteriovenous (AV) fistulas or intradural AVM's. Intradural AVM's were further classified as intramedullary AVM's (juvenile and glomus types) and direct AV fistulas, which were extramedullary or intramedullary in location. Dural AV fistulas were defined as being supplied by a dural artery and draining into spinal veins via an AV shunt in the intervertebral foramen. Intramedullary AVM's were defined as having the AV shunt contained at least partially within the cord or pia and receiving arterial supply by medullary arteries.
Of the 81 patients, 27 (33%) had dural AV fistulas and 54 (67%) had intradural AVM's. Several dissimilarities in clinical and radiographic findings of the two subgroups were evident. The patients with intramedullary AVM's were younger; the age at onset of symptoms averaged 27 years compared to 49 years for dural AV fistulas. The most common initial symptom associated with dural AV fistulas was steadily progressive paresis, whereas hemorrhage was the most common presenting symptom in cases of intramedullary lesions. No patients with dural AV fistulas had subarachnoid hemorrhage. Activity exacerbated symptoms more frequently in patients with dural lesions. Associated vascular anomalies occurred only in cases of intradural AVM's. In 96% of the dural lesions the AV nidus was in the low thoracic or lumbar region; in only 15% did the intercostal or lumbar arteries supplying the AVM also provide a medullary artery which supplied the spinal cord. In contrast, most intradural AVM's (84%) were in the cervical or thoracic segments of the spinal cord and all of them were supplied by medullary arteries. Transit of contrast medium through the intradural AVM's was rapid in 80% of cases, suggesting high-flow lesions. Forty-four percent of the patients with AVM's of the spinal cord had associated saccular arterial or venous spinal aneurysms. No dural AV fistulas displayed these characteristics. A good outcome occurred in 88% of patients with dural AV fistulas after nidus obliteration, while 49% of patients with intramedullary AVM's did well after surgery or embolization.
These findings suggest that dural and intradural AVM's differ in etiology (acquired vs. congenital) and that they have different pathophysiology, radiographic findings, clinical presentation, and response to treatment.
Positron emission tomography studies
J. Bob Blacklock, Edward H. Oldfield, Giovanni Di Chiro, Dung Tran, William Theodore, Donald C. Wright and Steven M. Larson
✓ Glucose utilization by normal and neoplastic cerebral tissue can be measured in humans using positron emission tomography (PET) with fluorine-18-labeled 2-deoxy-D-glucose (FDG). Malignant gliomas are known to exhibit hypermetabolic glucose consumption compared to normal brain. Barbiturate-sensitive cerebral glucose utilization is coupled to neuronal activity, and lesions lacking neuronal activity should be relatively insensitive to barbiturate suppression of glucose utilization. In a study to examine this phenomenon, three patients with cerebral gliomas underwent FDG-PET while awake and during deep barbiturate coma. Cerebral glucose utilization was measured in normal brain, tumor, and a homologous, non-neoplastic control site in the contralateral hemisphere. A glucose utilization ratio for tumor/control tissue was calculated.
The mean reduction of glucose utilization during barbiturate coma was: gray matter 67%, white matter 47%, basal ganglia 66%, thalamus 57%, cerebellar cortex 55%, tumor 32%, and the contralateral control site 64%. The mean tumor glucose utilization ratio was 1.48:1 in the awake state and 2.69:1 during barbiturate coma. The changes in gray matter, basal ganglia, thalamus, cerebellar cortex, and tumor/control tissue ratio were significant (p < 0.05). In one patient, deep tumor invasion not evident on computerized tomography, magnetic resonance imaging, or baseline FDG-PET was apparent during barbiturate-enhanced FDG-PET scanning.
The study findings suggest that gliomas resist suppression of glucose utilization by barbiturates; this supports the hypothesis that barbiturates reduce neuronal metabolism by blocking synaptic activity. This differential effect on normal brain and gliomas enhances the capability to assess the extent of neoplastic tissue in brain and may represent the basis for novel therapeutic strategies.
John L. Doppman, Giovanni Di Chiro, Andrew J. Dwyer, Joseph L. Frank and Edward H. Oldfield
✓ Magnetic resonance imaging (MRI) was performed on 12 patients with spinal arteriovenous malformations (AVM's). Six lesions were intramedullary, five were dural, and one was in a posterior extramedullary location. Serpentine filling defects similar to the classic myelographic findings were demonstrated within the high-signal cerebrospinal fluid on T2-weighted coronal scans. The intramedullary nidus was identified by MRI as an area of low-signal intensity within the cord in all six intramedullary AVM's. Neither the dural nor the posterior extramedullary lesions showed intramedullary components. It is concluded that MRI may noninvasively provide the initial diagnosis of a spinal AVM and distinguish intramedullary from dural and extramedullary lesions.
Edward H. Oldfield, Giovanni Di Chiro, Eugene A. Quindlen, Kenneth G. Rieth and John L. Doppman
✓ As demonstrated by selective spinal cord arteriography, over 80% of spinal cord arteriovenous malformations (AVM's) occupy a predominantly extramedullary position. Current therapy frequently requires surgical stripping of the long dorsal intradural vessel(s) from the underlying spinal cord over many cord segments. The authors report six patients with a dural arteriovenous fistula fed by a cluster of abnormal epidural arteries. These vessels, which surrounded and were embedded into the dural covering of a thoracic nerve root, drained into a long sinuous intrathecal paramedullary vein(s). The angiographic and surgical appearance of the intradural component of these lesions was identical to that of lesions previously classified as Type I AVM's of the spinal cord. All patients had symptoms and signs of myelopathy. In five patients, surgery was limited to coagulation and excision of the extradural vessels and division of the intradural arterialized vein. Progressive improvement began within days following surgery. No residual abnormality was demonstrated by postoperative selective spinal cord arteriography, which was performed in all five patients.
The findings support those of Kendall and Logue, that surgery restricted to elimination of the arteriovenous fistula at the intervertebral foramen is curative, and that more extensive surgery is unnecessary for this subgroup of AVM's of the spinal cord. These lesions comprise a sizable percent of all spinal AVM's. Resolution of myelopathy in these patients supports the hypothesis that venous hypertension causes chronic progressive myelopathy.