Original Research
Phase-Contrast MR Angiography of Vascular Malformations of the Spinal Cord at 0.5 T1 Franqoise Gelbert, MD Jean-Pierre Guichard, MD Klaus L. Mourier, MD Daniel Reizine, MD Armand Aymard, MD Pierre Gobin, MD Bernard George, MD Jean Cophignon, MD Jean-Jacques Merland, MD Preliminary experience with phase-contrast magnetic resonance (MR)angiographyat 0.5 T applied in 12 cases of vascular malformations of the spinal cord is reported. There were six intramedullary arteriovenous malformations (AVMs),four perimedullary fistulas. and two dural arteriovenous fistulas with perimedullary drainage, all proved with x-ray angiography. The small size of the vessels and their location within a bony structure presented a technical challenge. Serpentine vascular signal patterns were identified within the spinal canal in all cases, showing good correlation with the x-ray angiographic pattern. Relative to spin-echo images, MR angiograms allowed better visualization of the venous drainage. The nidus of intramedullary AVMs was more difficult to recognize. The ability to manipulate the velocity-encodingvalue allows better characterization of flow speed. The results underline the two dimensions of the phase-contrast technique, which provides both anatomic images and dynamic information about vascular malformations. MR angiography does not replace selective x-ray angiography, which is indispensable for therapeutic strategy (endovascularprocedure or surgery),but it can be considered a valuable alternative to x-ray angiography during follow-up.
PREVIOUS STUDlES HAVE SHOWN magnetic resonance (MR)imaging characteristics of different types of vascular malformations of the spinal cord ( 1-3). These malformations can be classified according to the type of feeding artery (radiculomedullary or meningeal) and vascular architecture (fistula or nidus). This classification commonly recognizes three types of malformations (4):(a) arteriovenous malformations (AVMs) issuing from a spinal artery, with a completely or partially intramedullary nidus (5); ( b )direct perimedullary fistulas between a spinal artery and the perimedullary veins (three types-I, 11, and 111-have been identified, according to the size of the fistula and vessels); and ( c )dural arteriovenous fistulas (AVFs) between a meningeal branch of a radicular artery and the perimedullary veins (6.7). AVMs and types I1 and 111perimedullary fistulas are high-flow malformations; type I perimedullary fistulas and dural AVFs are slow-flow malformations. We discuss and give examples of our initial experience with phase-contrast MR angiography of vascular malformations of the spinal cord. We also attempt to determine the clinical protocols and technical parameters for the future application of this method.
Index term% Angioma. central nervous system, 37.149 * Arteriovenous malformations. dural. 379.149 * Arteriovenous malformations. spinal, 37.149 * Comparative studies * Phase imaging Vascular studles
MATERIALS AND METHODS Twelve patients with proven vascular malformations of the spinal cord were referred for phase-contrast MR angiography. There were six intramedullary AVMs, four perimedullary fistulas, and two dural AVFs with perimedullary drainage. All MR imaging was performed with a 0.5-T magnet (MR Max;GE Medical Systems, Milwaukee). Routine MR studies were performed in all patients. They included sagittal- and axial-plane T 1-weighted (TR msec/TE msec = 500/50)and T2-weighted (2.000/60, 120)spin-echo images. MR angiography was performed in a sagittal orientation with a twodimensional gradient-echo sequence ( 160/25, 20" flip angle) and a section thickness of 20 mm. The image matrix was 160 x 160. The number of signals averaged was 10, with an imaging time of approximately
JMRI 1992: 2:631-636 Abbreviations: A V F = arteriovenous fistula. AVM = arteriovenous malformation. CSF = cerebrospinal fluid. TOF = time offlight.
I From the Deparlments of Neuroradioloey (F.G.. J.P.G.. D.R.. A.A.. P.G.. J.J.M.) and Neurosurgery lK.I,.M.. B.G.. J.C.I. HBpital Larihoisikre. 2 Rue Amhroisr Pare. 75010 Paris. France. Rereived February 26. 1992: revision rrquestrd April 6: revision received and acceptrd July 28. Address reprint requests to F.G.
SMRI. 1992
631
Summary of Patient Data Patient/ Age (yj/Sex 1139lM 2/44/M 3/231M 4/30/F 512 1/M 6/19/M 7/26/M 8156IM 9/19/M 10/31/M 11/41/F
12/50/M
Location
Malformation
Velocity-Encoding Values (cm/sec)*
Medullary cone Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Cervical Thoracic Medullary cone
Dural AVF Intramedullary AVM lntramedullary AVM' Perimedullary fistula lntramedullary AVM Intramedullary AVM Intramedullary AVM Perimedullary fistula Perimedullary fistula lntramedullary AVM Perimedullary fistula Dural AVF
5. 10. 15 18 18 5.10 18 18.20 18,ZO 5 . 10, 15. 20 20 20 10, 15 5.10
;Best choice in italics. Cobb syndrome.
a. b. C. Figure 1. Patient 3. Thoracic intramedullary angioma. (a) X-ray angiogram in the lateral projection shows the nidus within the spinal canal, issuing from the anterior spinal artery (arrows).with ascending and descending venous drainage (arrowheads). Intramedullary angiomas are high-flowmalformations. (b)Sagittal T2-weighted image demonstrates serpentine signal voids (arrow. arrowhead) contrasting with the high signal intensity of cerebrospinal fluid (CSF).( c ) Phase-contrast MR angiogram. The velocity-encodingvalue was 18 cmlsec. slightly less than that used for cerebral vascular studies. Behind the vertebral bodies (outlinedl. a continuous serpentine vascular structure can be seen. The pattern of venous drainage (arrowheads)is better identified than on spin-echo images, as is the area with a more compact pattern that indicates the nidus (solid arrows). More anteriorly. intense signal from the thoracic aorta is responsible for the artifact (open arrow).
15 minutes, or four, which reduced the imaging time to approximately 5 minutes. The flow velocity varied according to the type of malformation. All patients underwent selective x-ray angiography within 24 hours before MR angiography. Four patients underwent MR angiography and x-ray angiography before and after treatment (embolization and/or surgery).
RESULTS The results were classified according to the type of malformation (Table].MR angiography showed an
632
JMRl
November/December 1992
abnormal vascular pattern within the spinal canal in all cases.
IntramedullaryAVMs (Figs 1,2) Spin-echo MR images showed an intramedullary area of low signal intensity associated with a local enlargement of the spinal cord, corresponding to the nidus. The intramedullary site of the nidus was well demonstrated on axial images. Vascular signal voids corresponding to the arterial feeder vessels and venous drainage were seen superior and inferior to the nidus and were better visualized on T2-weighted
Figure 2. Patient 10. Cervical intramedullary angioma. (a) Sagittal T2-weghted image shows local enlargement of the spinal cord, containing signal voids that correspond to the nidus. (b)Frontal-projection MR angiogram with a velocityencoding value of 20 cml sec. Cervical arteries and veins are clearly seen (arrowheads). Medially, the intramedullary nidus (arrow) can be seen. a.
b.
images because of the contrast with CSF hyperintensity (Fig Ibl. MR angiography demonstrated a serpentine anterior or posterior vessel within the spinal canal, which corresponded to the x-ray angiographic pattern. No difference was seen between arteries and veins. A more heterogeneous area corresponding to the nidus was observed (Figs Ic, 2b). When compared with spin-echo images, MR angiograms provided better visualization of the venous drainage: however, spin-echo images allowed a better characterization of the nidus and its intramedullary site. In one case in which MR imaging was performed before and after treatment, the remaining portion of the angioma was underestimated on spin-echo images, with no identification of signal voids. MR angiography more closely approximated the conventional angiographic control images by showing the small residual vessels. One patient had a complex intramedullary malformation associated with a metameric angiomatosis that included a vertebral angioma (Cobb syndrome). The vertebral angioma was better visualized at MR angiography (as an area of low signal intensity)than at spin-echo imaging.
PerimedullaryFistulas (Fig 3) In types I1 and 111 perimedullary fistulas, spin-echo sequences easily showed the large signal voids within hyperintense CSF on T2-weighted (Fig3b) images. The spinal cord appeared normal. MR angiography clearly demonstrated the serpentine vascular pattern and again better demonstrated the length of abnormal vessels than did spin-echo imaging. We studied a type I case characterized by a small shunt and slow flow within the abnormal venous drainage. Spin-echo images did not demonstrate
clearly the vascular abnormality, nor did MR angiography with velocity-encodingvalues of 20 cm/sec. By reducing velocity-encoding values to 15, 10, 8, and 5 cmfsec,we obtained increasingly more satisfymg images of the abnormal vessels within the spinal canal (Fig3c, 3d).
Dural AVFs We applied the same imaging techniques to the two dural AVFs, also characterized by their very slow flow, and succeeded in depicting small serpentine vessels corresponding to venous drainage (Fig 4). 0
DISCUSSION
Vascular malformations of the spinal cord are rare but severe (4).Therapeutic strategy depends on the type of malformation and combines surgical and endovascular procedures. Owing to their evolution, these malformations require long-term follow-up and treatment and a large number of angiographic studies. Spin-echo MR imaging has already dramatically changed the study of this abnormality by almost replacing myelography as a first-step examination (3). The interpretation of conventional spin-echo and gradient-echo images of vascular disease can be complex and ambiguous because of the variable effects that flowing blood and thrombosis have on image in tensity (2.8).
The two basic strategies that have emerged for imaging blood flow are the phase-contrast and time-offlight (TOF)methods. The phase-contrast method (913) uses a bipolar phase-encoding gradient applied along the three orthogonal axes to separate stationary and moving protons by means of phase shift. Protons Volume2
Number6
JMRl
633
Figure 3. Patient 4. Thoracic perimedullary fistula. (a] Lateral x-ray angiogram shows a direct fistula (type I, slow flow] between the anterior spinal artery (arrow) and the dilated perimedullary veins (arrowheads). (b)Sagittal TP-weighted image shows vascular signal voids with a discontinuous pattern (arrows).(c) Lateralprojection phase-contrast MR angiogram with a velocity-encoding value of 15 cm/ sec. A slightly hypointense. discontinuous vascular signal within the spinal canal (arrows)can be seen. Vertebral bodies are outlined. (d)Same projection as in c , with velocity-encoding value reduced to 5 cm/sec after consideration of x-ray angiographic data. The vascular pattern is more intense and similar to x-ray angiographic projection. C.
in stationary tissues acquire no phase changes with application of the bipolar gradient pulse; however, flowing protons accumulate phase as they move along the gradient field. Two radio-frequency pulses are applied, and a vector subtraction technique is used to eliminate background signal. Velocity encoding can be selected for arteries (40-50-cm/sec flow) or veins (I 20-cm/sec flow). The data are collected with a se634
JMRl
November/December 1992
d.
ries of 2-cm-thick sections. In the spinal cord, the sagittal plane appeared to be appropriate for both the anatomy of vessels and direction of flow. The TOF method ( 14) uses a two-dimensional sequential section approach and a three-dimensional volumetric method. In TOF studies, the influx of fully relaxed spins into the excited volume is used to advantage: the unsaturated inflowing blood will have
a.
b.
C.
Figure 4. Patient 1. Dural fistula with perimedullary drainage. (a) Frontal selective x-ray angiogram of radicular artery shows small fistula (arrow] between one branch of the radicular artery and the perimedullary veins (arrowheads).In this type of malformation, flow within the fistula and veins is very slow, with a delay of several seconds between arterial injection of contrast material and the appearance of the fistula and veins. (b)Sagittal T1-weighted image shows a normal spinal cord (arrow), with no abnormal signal voids. (c) MR angiogram with a reduced velocity-encoding value (5cm/sec]. The vascular pattern (arrowheads) is not prominent but can be followed for a satisfying length within the spinal canal. Vertebral bodies are outlined.
detectably higher signal intensity relative to stationary tissue in steady-state saturation. The differential saturation will depend on the T1 and T2 of tissues surrounding the vessel. Signal loss due to pulsatile and complex flow is less problematic. The TOF method therefore appears more limited in the case of slow flow, since signal from slow flow and/or saturated flow cannot be separated from that of stationary tissue. The phase-contrast method has advantages over the TOF method in detection of slow flow and allows better definition of small vessels. Vascular malformations of the spinal cord illustrate these problems and possibilities and represent a difficult technical challenge. The angiographic architecture in the spinal cord consists of very small vessels, some with remarkably slow flow. In addition, the malformations are small and contained in a bony structure surrounded by CSF pulsation and close to the cardiac cavity and aortic vessels. Despite this, our preliminary results are promising and can be summarized as follows. lntramedullary AVMs occur in young patients. The arterial feeder vessel is always the anterior spinal artery, with some participation by the posterior spinal artery. AVMs are high-flow malformations, with a high risk of subarachnoid hemorrhage. The venous drainage involves the perimedullary veins, which are dilated and have an anterior and/or posterior ascend-
ing and/or descending course. In the present study, MR angiography allowed good characterization of the vessels and nidus; however, spin-echo images better demonstrated the intramedullary component of the nidus by depicting a local enlargement of the spinal cord. Perimedullary fistulas occur in middle-aged and young patients. There is a direct fistula between a spinal artery and the perimedullary veins. Three types have been identified. Type I is characterized by a small shunt and very slow flow in a small-caliber perimedullary drainage. Types I1 and I11 are giant or large fistulas with dilated, high-flow venous drainage and a high-velocity vascular pattern similar to that of AVMs. The area of the shunt was not clearly identified. These fistulas could be distinguished from AVMs with spinecho images showing no enlargement of the spinal cord. Dural AVFs with perimedullary drainage usually involve older patients. The arterial feeder vessel is not a spinal artery but a meningeal branch of a radicular artery draining abnormally through a direct shunt into the perimedullary veins. These veins are dilated and characterized by their length and slow flow. We applied the same imaging parameters to dural AVFs and type I perimedullary fistulas. By reducing the velocity-encodingvalues from 20 to 5 cm/sec. we obtained progressively more satisfying images of these malformations. Volume2
Number6
JMRl
635
In conclusion, MR angiography provided better information than spin-echo imaging with regard to the effective flow pattern of the vascular malformations. Because of poor definition of the nidus, it was difficult in the present study to differentiate an angioma from a fistula. This can be achieved with spin-echo techSpinniques, as described in previous studies (2,3,6). echo MR imaging can also show complications such as subarachnoid or intramedullary hematoma and spinal cord atrophy ( 3 ) .By choosing various velocityencoding values, we obtained dynamic information about flow within the malformation, which can lead to further applications, especially in postoperative cases. Even high-flow malformations have an average velocity range (18-20 cm/sec) less than that of intracerebral vascular malformations: however, this parameter has to be considerably reduced for slow-flow malformations, as illustrated above. MR angiography can be regarded a s a useful part of the imaging examination of vascular malformations of the spinal cord. It provides information complementary to T I - and T2weighted images. It does not replace diagnostic x-ray angiography but can allow a longer time interval between angiographic studies.
1. Masaryk T J . Ross JS, Modic MT, Ruff RA, Selman WR, Ratcheson RA. Radiculomeningeal vascular malformations of the spine: MR imaging. Radiology 1987; 164:845849. 2. Modic MT. Masaryk T J , Paushter D. Magnetic resonance imaging of the spine. Radio1 Clin North Am 1986: 24:229245.
JMRl
7. 8.
9.
10.
11. 12.
References
636
6.
November/December 1992
13. 14.
Dormont D. Gelbert F. Assouline E, et al. MR imaging of spinal cord AV malformations at 0.5 tesla: study of 34 cases. AJNR 1988: 9:833-838. Hurth M, Houdart R, Djindjian R. Rey A, Djindjian M. Arteriovenous malformations of the spinal cord. Prog Neurol Surg 1978: 9:238-266. Gueguen B, Merland JJ, Riche MC, Rey A. Vascular malformations of the spinal cord: intrathecal perimedullary arteriovenous fistulas fed by medullary arteries. NeuroloRy 1987 : 37 :969-979. Mourier KL, Celbert F, Rey A. et al. Spinal dural arteriovenous malformations with perimedullary drainage. Acta Neurochir 1989: 100:136-141. Kendall BE, Logue V. Spinal epidural angiomatous malformation draining into intrathecal veins. Neuroradiology 1977; 13:181-189. Bradley WC. Flow phenomena in MR imaging. A J R 1988: 150~983-994. Vadel L, Brown IF, Kaaft KA. Fatouross PP, Laine IT. Intracranial vascular abnormalities: value of MR phase contrast imaging to distinguish thrombus from flowing blood. AJNR 1990: 11:1133-1140. Spritzer CE, Pelc N J , Lee J N , Evans AJ, Sostman HD, Riederer ST. Rapid MR imaging of blood flow with a phase-sensitive, limited-flip-angle, gradient recalled pulse sequence: preliminary experience. Radiology 1990; 176: 255-262. Masaryk TJ, Modic MT. Ross JS, et al. Intracranial circulation: preliminary clinical results with 3D (volume)MR angiography. Radiology 1989; 171:793-799. Dumoulin CL, Hart HR. Magnetic resonance angiography. Radiology 1986: 161:717-720. Ross JS, Masaryk T J . Modic MT. Ruggieri PM, Haacke EM, Selman WR. lntracranial aneurysms: evaluation by MR angiography. AJNR 1990: 11:449-456. Pernicone J R . Siebert J E , Potchen E J , Pera A. Dumoulin CL, Souza SP. Three-dimensional phase-contrast MR angiography in the head and neck: preliminary report. AJNR 1990: 11:457-466.
Phase-Contrast MR Angiography of Vascular Malformations of the Spinal Cord at 0.5 T1 Franqoise Gelbert, MD Jean-Pierre Guichard, MD Klaus L. Mourier, MD Daniel Reizine, MD Armand Aymard, MD Pierre Gobin, MD Bernard George, MD Jean Cophignon, MD Jean-Jacques Merland, MD Preliminary experience with phase-contrast magnetic resonance (MR)angiographyat 0.5 T applied in 12 cases of vascular malformations of the spinal cord is reported. There were six intramedullary arteriovenous malformations (AVMs),four perimedullary fistulas. and two dural arteriovenous fistulas with perimedullary drainage, all proved with x-ray angiography. The small size of the vessels and their location within a bony structure presented a technical challenge. Serpentine vascular signal patterns were identified within the spinal canal in all cases, showing good correlation with the x-ray angiographic pattern. Relative to spin-echo images, MR angiograms allowed better visualization of the venous drainage. The nidus of intramedullary AVMs was more difficult to recognize. The ability to manipulate the velocity-encodingvalue allows better characterization of flow speed. The results underline the two dimensions of the phase-contrast technique, which provides both anatomic images and dynamic information about vascular malformations. MR angiography does not replace selective x-ray angiography, which is indispensable for therapeutic strategy (endovascularprocedure or surgery),but it can be considered a valuable alternative to x-ray angiography during follow-up.
PREVIOUS STUDlES HAVE SHOWN magnetic resonance (MR)imaging characteristics of different types of vascular malformations of the spinal cord ( 1-3). These malformations can be classified according to the type of feeding artery (radiculomedullary or meningeal) and vascular architecture (fistula or nidus). This classification commonly recognizes three types of malformations (4):(a) arteriovenous malformations (AVMs) issuing from a spinal artery, with a completely or partially intramedullary nidus (5); ( b )direct perimedullary fistulas between a spinal artery and the perimedullary veins (three types-I, 11, and 111-have been identified, according to the size of the fistula and vessels); and ( c )dural arteriovenous fistulas (AVFs) between a meningeal branch of a radicular artery and the perimedullary veins (6.7). AVMs and types I1 and 111perimedullary fistulas are high-flow malformations; type I perimedullary fistulas and dural AVFs are slow-flow malformations. We discuss and give examples of our initial experience with phase-contrast MR angiography of vascular malformations of the spinal cord. We also attempt to determine the clinical protocols and technical parameters for the future application of this method.
Index term% Angioma. central nervous system, 37.149 * Arteriovenous malformations. dural. 379.149 * Arteriovenous malformations. spinal, 37.149 * Comparative studies * Phase imaging Vascular studles
MATERIALS AND METHODS Twelve patients with proven vascular malformations of the spinal cord were referred for phase-contrast MR angiography. There were six intramedullary AVMs, four perimedullary fistulas, and two dural AVFs with perimedullary drainage. All MR imaging was performed with a 0.5-T magnet (MR Max;GE Medical Systems, Milwaukee). Routine MR studies were performed in all patients. They included sagittal- and axial-plane T 1-weighted (TR msec/TE msec = 500/50)and T2-weighted (2.000/60, 120)spin-echo images. MR angiography was performed in a sagittal orientation with a twodimensional gradient-echo sequence ( 160/25, 20" flip angle) and a section thickness of 20 mm. The image matrix was 160 x 160. The number of signals averaged was 10, with an imaging time of approximately
JMRI 1992: 2:631-636 Abbreviations: A V F = arteriovenous fistula. AVM = arteriovenous malformation. CSF = cerebrospinal fluid. TOF = time offlight.
I From the Deparlments of Neuroradioloey (F.G.. J.P.G.. D.R.. A.A.. P.G.. J.J.M.) and Neurosurgery lK.I,.M.. B.G.. J.C.I. HBpital Larihoisikre. 2 Rue Amhroisr Pare. 75010 Paris. France. Rereived February 26. 1992: revision rrquestrd April 6: revision received and acceptrd July 28. Address reprint requests to F.G.
SMRI. 1992
631
Summary of Patient Data Patient/ Age (yj/Sex 1139lM 2/44/M 3/231M 4/30/F 512 1/M 6/19/M 7/26/M 8156IM 9/19/M 10/31/M 11/41/F
12/50/M
Location
Malformation
Velocity-Encoding Values (cm/sec)*
Medullary cone Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Thoracic Cervical Thoracic Medullary cone
Dural AVF Intramedullary AVM lntramedullary AVM' Perimedullary fistula lntramedullary AVM Intramedullary AVM Intramedullary AVM Perimedullary fistula Perimedullary fistula lntramedullary AVM Perimedullary fistula Dural AVF
5. 10. 15 18 18 5.10 18 18.20 18,ZO 5 . 10, 15. 20 20 20 10, 15 5.10
;Best choice in italics. Cobb syndrome.
a. b. C. Figure 1. Patient 3. Thoracic intramedullary angioma. (a) X-ray angiogram in the lateral projection shows the nidus within the spinal canal, issuing from the anterior spinal artery (arrows).with ascending and descending venous drainage (arrowheads). Intramedullary angiomas are high-flowmalformations. (b)Sagittal T2-weighted image demonstrates serpentine signal voids (arrow. arrowhead) contrasting with the high signal intensity of cerebrospinal fluid (CSF).( c ) Phase-contrast MR angiogram. The velocity-encodingvalue was 18 cmlsec. slightly less than that used for cerebral vascular studies. Behind the vertebral bodies (outlinedl. a continuous serpentine vascular structure can be seen. The pattern of venous drainage (arrowheads)is better identified than on spin-echo images, as is the area with a more compact pattern that indicates the nidus (solid arrows). More anteriorly. intense signal from the thoracic aorta is responsible for the artifact (open arrow).
15 minutes, or four, which reduced the imaging time to approximately 5 minutes. The flow velocity varied according to the type of malformation. All patients underwent selective x-ray angiography within 24 hours before MR angiography. Four patients underwent MR angiography and x-ray angiography before and after treatment (embolization and/or surgery).
RESULTS The results were classified according to the type of malformation (Table].MR angiography showed an
632
JMRl
November/December 1992
abnormal vascular pattern within the spinal canal in all cases.
IntramedullaryAVMs (Figs 1,2) Spin-echo MR images showed an intramedullary area of low signal intensity associated with a local enlargement of the spinal cord, corresponding to the nidus. The intramedullary site of the nidus was well demonstrated on axial images. Vascular signal voids corresponding to the arterial feeder vessels and venous drainage were seen superior and inferior to the nidus and were better visualized on T2-weighted
Figure 2. Patient 10. Cervical intramedullary angioma. (a) Sagittal T2-weghted image shows local enlargement of the spinal cord, containing signal voids that correspond to the nidus. (b)Frontal-projection MR angiogram with a velocityencoding value of 20 cml sec. Cervical arteries and veins are clearly seen (arrowheads). Medially, the intramedullary nidus (arrow) can be seen. a.
b.
images because of the contrast with CSF hyperintensity (Fig Ibl. MR angiography demonstrated a serpentine anterior or posterior vessel within the spinal canal, which corresponded to the x-ray angiographic pattern. No difference was seen between arteries and veins. A more heterogeneous area corresponding to the nidus was observed (Figs Ic, 2b). When compared with spin-echo images, MR angiograms provided better visualization of the venous drainage: however, spin-echo images allowed a better characterization of the nidus and its intramedullary site. In one case in which MR imaging was performed before and after treatment, the remaining portion of the angioma was underestimated on spin-echo images, with no identification of signal voids. MR angiography more closely approximated the conventional angiographic control images by showing the small residual vessels. One patient had a complex intramedullary malformation associated with a metameric angiomatosis that included a vertebral angioma (Cobb syndrome). The vertebral angioma was better visualized at MR angiography (as an area of low signal intensity)than at spin-echo imaging.
PerimedullaryFistulas (Fig 3) In types I1 and 111 perimedullary fistulas, spin-echo sequences easily showed the large signal voids within hyperintense CSF on T2-weighted (Fig3b) images. The spinal cord appeared normal. MR angiography clearly demonstrated the serpentine vascular pattern and again better demonstrated the length of abnormal vessels than did spin-echo imaging. We studied a type I case characterized by a small shunt and slow flow within the abnormal venous drainage. Spin-echo images did not demonstrate
clearly the vascular abnormality, nor did MR angiography with velocity-encodingvalues of 20 cm/sec. By reducing velocity-encoding values to 15, 10, 8, and 5 cmfsec,we obtained increasingly more satisfymg images of the abnormal vessels within the spinal canal (Fig3c, 3d).
Dural AVFs We applied the same imaging techniques to the two dural AVFs, also characterized by their very slow flow, and succeeded in depicting small serpentine vessels corresponding to venous drainage (Fig 4). 0
DISCUSSION
Vascular malformations of the spinal cord are rare but severe (4).Therapeutic strategy depends on the type of malformation and combines surgical and endovascular procedures. Owing to their evolution, these malformations require long-term follow-up and treatment and a large number of angiographic studies. Spin-echo MR imaging has already dramatically changed the study of this abnormality by almost replacing myelography as a first-step examination (3). The interpretation of conventional spin-echo and gradient-echo images of vascular disease can be complex and ambiguous because of the variable effects that flowing blood and thrombosis have on image in tensity (2.8).
The two basic strategies that have emerged for imaging blood flow are the phase-contrast and time-offlight (TOF)methods. The phase-contrast method (913) uses a bipolar phase-encoding gradient applied along the three orthogonal axes to separate stationary and moving protons by means of phase shift. Protons Volume2
Number6
JMRl
633
Figure 3. Patient 4. Thoracic perimedullary fistula. (a] Lateral x-ray angiogram shows a direct fistula (type I, slow flow] between the anterior spinal artery (arrow) and the dilated perimedullary veins (arrowheads). (b)Sagittal TP-weighted image shows vascular signal voids with a discontinuous pattern (arrows).(c) Lateralprojection phase-contrast MR angiogram with a velocity-encoding value of 15 cm/ sec. A slightly hypointense. discontinuous vascular signal within the spinal canal (arrows)can be seen. Vertebral bodies are outlined. (d)Same projection as in c , with velocity-encoding value reduced to 5 cm/sec after consideration of x-ray angiographic data. The vascular pattern is more intense and similar to x-ray angiographic projection. C.
in stationary tissues acquire no phase changes with application of the bipolar gradient pulse; however, flowing protons accumulate phase as they move along the gradient field. Two radio-frequency pulses are applied, and a vector subtraction technique is used to eliminate background signal. Velocity encoding can be selected for arteries (40-50-cm/sec flow) or veins (I 20-cm/sec flow). The data are collected with a se634
JMRl
November/December 1992
d.
ries of 2-cm-thick sections. In the spinal cord, the sagittal plane appeared to be appropriate for both the anatomy of vessels and direction of flow. The TOF method ( 14) uses a two-dimensional sequential section approach and a three-dimensional volumetric method. In TOF studies, the influx of fully relaxed spins into the excited volume is used to advantage: the unsaturated inflowing blood will have
a.
b.
C.
Figure 4. Patient 1. Dural fistula with perimedullary drainage. (a) Frontal selective x-ray angiogram of radicular artery shows small fistula (arrow] between one branch of the radicular artery and the perimedullary veins (arrowheads).In this type of malformation, flow within the fistula and veins is very slow, with a delay of several seconds between arterial injection of contrast material and the appearance of the fistula and veins. (b)Sagittal T1-weighted image shows a normal spinal cord (arrow), with no abnormal signal voids. (c) MR angiogram with a reduced velocity-encoding value (5cm/sec]. The vascular pattern (arrowheads) is not prominent but can be followed for a satisfying length within the spinal canal. Vertebral bodies are outlined.
detectably higher signal intensity relative to stationary tissue in steady-state saturation. The differential saturation will depend on the T1 and T2 of tissues surrounding the vessel. Signal loss due to pulsatile and complex flow is less problematic. The TOF method therefore appears more limited in the case of slow flow, since signal from slow flow and/or saturated flow cannot be separated from that of stationary tissue. The phase-contrast method has advantages over the TOF method in detection of slow flow and allows better definition of small vessels. Vascular malformations of the spinal cord illustrate these problems and possibilities and represent a difficult technical challenge. The angiographic architecture in the spinal cord consists of very small vessels, some with remarkably slow flow. In addition, the malformations are small and contained in a bony structure surrounded by CSF pulsation and close to the cardiac cavity and aortic vessels. Despite this, our preliminary results are promising and can be summarized as follows. lntramedullary AVMs occur in young patients. The arterial feeder vessel is always the anterior spinal artery, with some participation by the posterior spinal artery. AVMs are high-flow malformations, with a high risk of subarachnoid hemorrhage. The venous drainage involves the perimedullary veins, which are dilated and have an anterior and/or posterior ascend-
ing and/or descending course. In the present study, MR angiography allowed good characterization of the vessels and nidus; however, spin-echo images better demonstrated the intramedullary component of the nidus by depicting a local enlargement of the spinal cord. Perimedullary fistulas occur in middle-aged and young patients. There is a direct fistula between a spinal artery and the perimedullary veins. Three types have been identified. Type I is characterized by a small shunt and very slow flow in a small-caliber perimedullary drainage. Types I1 and I11 are giant or large fistulas with dilated, high-flow venous drainage and a high-velocity vascular pattern similar to that of AVMs. The area of the shunt was not clearly identified. These fistulas could be distinguished from AVMs with spinecho images showing no enlargement of the spinal cord. Dural AVFs with perimedullary drainage usually involve older patients. The arterial feeder vessel is not a spinal artery but a meningeal branch of a radicular artery draining abnormally through a direct shunt into the perimedullary veins. These veins are dilated and characterized by their length and slow flow. We applied the same imaging parameters to dural AVFs and type I perimedullary fistulas. By reducing the velocity-encodingvalues from 20 to 5 cm/sec. we obtained progressively more satisfying images of these malformations. Volume2
Number6
JMRl
635
In conclusion, MR angiography provided better information than spin-echo imaging with regard to the effective flow pattern of the vascular malformations. Because of poor definition of the nidus, it was difficult in the present study to differentiate an angioma from a fistula. This can be achieved with spin-echo techSpinniques, as described in previous studies (2,3,6). echo MR imaging can also show complications such as subarachnoid or intramedullary hematoma and spinal cord atrophy ( 3 ) .By choosing various velocityencoding values, we obtained dynamic information about flow within the malformation, which can lead to further applications, especially in postoperative cases. Even high-flow malformations have an average velocity range (18-20 cm/sec) less than that of intracerebral vascular malformations: however, this parameter has to be considerably reduced for slow-flow malformations, as illustrated above. MR angiography can be regarded a s a useful part of the imaging examination of vascular malformations of the spinal cord. It provides information complementary to T I - and T2weighted images. It does not replace diagnostic x-ray angiography but can allow a longer time interval between angiographic studies.
1. Masaryk T J . Ross JS, Modic MT, Ruff RA, Selman WR, Ratcheson RA. Radiculomeningeal vascular malformations of the spine: MR imaging. Radiology 1987; 164:845849. 2. Modic MT. Masaryk T J , Paushter D. Magnetic resonance imaging of the spine. Radio1 Clin North Am 1986: 24:229245.
JMRl
7. 8.
9.
10.
11. 12.
References
636
6.
November/December 1992
13. 14.
Dormont D. Gelbert F. Assouline E, et al. MR imaging of spinal cord AV malformations at 0.5 tesla: study of 34 cases. AJNR 1988: 9:833-838. Hurth M, Houdart R, Djindjian R. Rey A, Djindjian M. Arteriovenous malformations of the spinal cord. Prog Neurol Surg 1978: 9:238-266. Gueguen B, Merland JJ, Riche MC, Rey A. Vascular malformations of the spinal cord: intrathecal perimedullary arteriovenous fistulas fed by medullary arteries. NeuroloRy 1987 : 37 :969-979. Mourier KL, Celbert F, Rey A. et al. Spinal dural arteriovenous malformations with perimedullary drainage. Acta Neurochir 1989: 100:136-141. Kendall BE, Logue V. Spinal epidural angiomatous malformation draining into intrathecal veins. Neuroradiology 1977; 13:181-189. Bradley WC. Flow phenomena in MR imaging. A J R 1988: 150~983-994. Vadel L, Brown IF, Kaaft KA. Fatouross PP, Laine IT. Intracranial vascular abnormalities: value of MR phase contrast imaging to distinguish thrombus from flowing blood. AJNR 1990: 11:1133-1140. Spritzer CE, Pelc N J , Lee J N , Evans AJ, Sostman HD, Riederer ST. Rapid MR imaging of blood flow with a phase-sensitive, limited-flip-angle, gradient recalled pulse sequence: preliminary experience. Radiology 1990; 176: 255-262. Masaryk TJ, Modic MT. Ross JS, et al. Intracranial circulation: preliminary clinical results with 3D (volume)MR angiography. Radiology 1989; 171:793-799. Dumoulin CL, Hart HR. Magnetic resonance angiography. Radiology 1986: 161:717-720. Ross JS, Masaryk T J . Modic MT. Ruggieri PM, Haacke EM, Selman WR. lntracranial aneurysms: evaluation by MR angiography. AJNR 1990: 11:449-456. Pernicone J R . Siebert J E , Potchen E J , Pera A. Dumoulin CL, Souza SP. Three-dimensional phase-contrast MR angiography in the head and neck: preliminary report. AJNR 1990: 11:457-466.
