MRI and MR Angiography Findings to Differentiate Jugular Venous Reflux From Cavernous Dural Arteriovenous Fistula

Transcription

1 Neuroradiology/Head and Neck Imaging Original Research Kim et al. MRI and MRA to Differentiate JVR From Cavernous DAVF Neuroradiology/Head and Neck Imaging Original Research Eunhee Kim 1 Jae Hyoung Kim 1 Byung Se Choi 1 Cheolkyu Jung 1 Dong Hoon Lee 2 Kim E, Kim JH, Choi BS, Jung C, Lee DH Keywords: cavernous dural arteriovenous fistula, jugular venous reflux, MR angiography, MRI DOI: /AJR Received April 2, 2013; accepted after revision June 17, This work was supported by grant from the Seoul National University Bundang Hospital Research Fund. 1 Department of Radiology, Seoul National University College of Medicine, Seoul National University Bundang Hospital, 166 Gumi-ro, Bundang-gu, Seongnam-si, Gyeonggi-do , Korea. Address correspondence to J. H. Kim (jaehkim@snu.ac.kr). 2 Department of Radiology, Seoul Medical Center, Seoul, Korea. AJR 2014; 202: X/14/ American Roentgen Ray Society MRI and MR Angiography Findings to Differentiate Jugular Venous Reflux From Cavernous Dural Arteriovenous Fistula OBJECTIVE. Both jugular venous reflux (JVR) and cavernous dural arteriovenous fistula (DAVF) manifest as abnormal venous signal intensities on time-of-flight (TOF) MR angiography (MRA). We investigated brain MRI and MRA findings that might differentiate JVR from cavernous DAVF. MATERIALS AND METHODS. Forty-one patients with abnormal venous signal intensities on TOF MRA in the cavernous sinus and its vicinity were selected from 1508 patients who had undergone TOF MRA over the previous 6 months. For comparison, the examinations of 26 patients with cavernous DAVF who had undergone imaging over the previous 8 years were collected. The following findings were assessed: the side and location of abnormal venous signal intensities on intracranial TOF MRA; the signal intensity of the proximal jugular vein on T2-weighted imaging; whether there was early opacification of the cavernous sinus in the arterial phase of contrast-enhanced MRA (CE-MRA); the side of jugular venous drainage in the arteriovenous phase of CE-MRA; and whether retrograde jugular venous flow was seen on neck TOF MRA. RESULTS. Abnormal venous signal intensities were seen on the left side in 73% of patients with JVR and 58% of patients with cavernous DAVF; involvement of the cavernous sinus was found in 12% of patients with JVR and 100% of patients with cavernous DAVF. Increased signal intensity in the ipsilateral jugular vein on T2-weighted imaging was found in 73% of JVR patients and 4% of cavernous DAVF patients. Early opacification of the cavernous sinus in the arterial phase of CE-MRA, ipsilateral jugular venous drainage in the arteriovenous phase of CE-MRA, and ipsilateral retrograde jugular venous flow on neck TOF MRA were found in 0%, 0%, and 63%, respectively, of JVR patients and in 100%, 100%, and 0%, respectively, of cavernous DAVF patients. CONCLUSION. JVR and cavernous DAVF can be differentiated from one another using MRI and MRA. I n intracranial time-of-flight (TOF) MR angiography (MRA), the presaturation pulse is applied above the excited slab to remove venous flow, which runs in the craniocaudal direction. Thus, intracranial TOF MRA images are expected to show only arterial structures, which run in the caudocranial direction. Therefore, visualization of venous signals on intracranial TOF MRA suggests the possibility of an arteriovenous shunt or dural arteriovenous fistula (DAVF) [1 4]. However, venous sinuses, such as the cavernous sinus, inferior petrosal sinus, and sigmoid sinus, are occasionally observed on intracranial TOF MRA in patients who do not have an arteriovenous shunt or DAVF [5 10]. Several reports have drawn attention to the unusual visualization of venous signals in the absence of an arteriovenous shunt and have shown that some of these cases were due to the inflow effect of retrograde venous flow in a caudocranial direction caused by compression of the left brachiocephalic vein between the aorta and sternum [5 8, 11, 12]. Some authors have called this retrograde flow [5, 6, 11], whereas others have used the term jugular venous reflux (JVR) [8, 13, 14]. The incidence of JVR that is, abnormal venous signal intensities on intracranial TOF MRA has been reported to range from 1.3% to 6.2% [6, 7, 11]. Among the various types of arteriovenous shunt or DAVF, cavernous DAVF may show abnormal ve- AJR:202, April

2 Kim et al. nous signal intensities in only the cavernous sinus and its vicinity on intracranial TOF MRA without prominent extracranial feeding vessels. Abnormal venous signal intensities in JVR are also frequently observed in the cavernous sinus and inferior petrosal sinus. Therefore, imaging findings of JVR can be similar to those of cavernous DAVF [6]. Several imaging clues for distinguishing between JVR and cavernous DAVF have been proposed [5 7]. Abnormal venous signal intensities persisting until the caudal end of the slab of intracranial TOF MRA with an intensity that gradually decreased in the caudocranial direction (i.e., the signal intensity in the cavernous sinus was weaker than that in the inferior petrosal sinus or sigmoid sinus) have been reported in JVR [6, 7]. These findings were attributed to the caudocranial direction of jugular reflux flow. Paksoy et al. [5] also reported that T2-weighted imaging showed high signal intensity in the ipsilateral jugular vein due to sluggish flow and that the delayed phase of contrast-enhanced MRA (CE-MRA) directly revealed stenosis in the left brachiocephalic vein in JVR. However, none of these reports provided detailed ways of distinguishing between JVR and cavernous DAVF using routine brain MRI and MRA. In this study, we investigated routine MRI and MRA findings to clarify which imaging findings are clues that can be used to differentiate JVR from cavernous DAVF. Materials and Methods This retrospective study was approved by the hospital institutional review board, and the requirement to obtain informed consent was waived. Study Population Patients with JVR During the 6-month period before the study, brain MRI and MRA examinations were performed of 3028 subjects in our hospital. Among these patients, the brain MRI and MRA studies of 1508 subjects included all the sequences that we needed to analyze (i.e., axial T2-weighted imaging and contrast-enhanced T1-weighted imaging sequences; intracranial TOF MRA sequences; and neck MRA sequences, either CE-MRA or TOF MRA). When we retrospectively reviewed the brain MR images and MR angiograms of these 1508 subjects, 44 patients showed abnormal venous signal intensities in the cavernous sinus, inferior petrosal sinus, sigmoid sinus, or transverse sinus on source images or maximum-intensity-projection images of intracranial TOF MRA. Of these 44 patients, three received a diagnosis of DAVF in the cavernous sinus (n = 1), in the sigmoid sinus (n = 1), or around the foramen magnum (n = 1) on the basis of conventional catheter angiography; the one patient with cavernous DAVF was included in the group with cavernous DAVF, and the other two were excluded. The other 41 patients had no neurologic symptoms suggesting cavernous DAVF and were undergoing MRI and MRA for a medical checkup or for the evaluation of stroke or nonspecific symptoms, such as headache or dizziness. Therefore, the final diagnosis was JVR. Twenty-four men and 17 women were included in the group of patients with JVR. The mean age of the JVR patients was 65 years (range, years; median, 64 years). Five underwent conventional angiography, but abnormal shunt flow was not found in any of these five patients. All of the intracranial studies of these 41 patients were TOF MRA, and the neck studies were CE-MRA (n = 33) or TOF MRA (n = 8). Patients with cavernous DAVF We found that a total of 55 patients had received a diagnosis of cavernous DAVF using conventional catheter angiography through a review of the electronic medical records for the most recent 8-year period before the study. Of these 55 patients, 26 who underwent brain MRI and MRA before treatment were included in the study. Four men and 22 women were included in cavernous DAVF group. The mean age of the patients with cavernous DAVF was 64 years (range, years; median, 62 years). All of the intracranial examinations of these 26 patients were TOF MRA, and the neck examinations were CE-MRA (n = 19) or TOF MRA (n = 7). In all 26 patients with cavernous DAVF, intracranial TOF MRA showed abnormal venous signal intensities in the cavernous sinus, inferior petrosal sinus, sigmoid sinus, or transverse sinus. Acquisition of MRI and MR Angiography MRI was conducted using two 1.5-T systems and two 3-T systems (Intera Achieva, Philips Healthcare). The protocols for brain MRI and MRA included axial T1-weighted imaging, T2-weighted imaging, FLAIR imaging, gradient-echo imaging, MRA, and axial and coronal contrast-enhanced T1- weighted imaging. The MRA study comprised intracranial TOF MRA plus either neck CE-MRA (n = 52) or neck TOF MRA (n = 15). Axial T1-weighted images of the brain were obtained with or without contrast injection using a spinecho sequence with the following parameters: TR range/te range, /9 12; FOV, mm; acquisition matrix, ; and slice thickness, 5 mm. Axial T2-weighted images were obtained using a turbo spin-echo sequence with the following parameters: TR range/te range, /80 100; echo-train length, 15 17; and the same geometric parameters as the T1-weighted sequence. Intracranial TOF MRA was performed in the axial plane using the following parameters: TR range/te range, 20 25/3 7; flip angle, 20 ; FOV, mm; matrix, ; slice thickness, 1.2 mm (0.6-mm overlap with adjacent section); and section slices, 140. Neck TOF MRA was performed in the axial plane using the following parameters: TR range/te range, 23 24/4 5; flip angle, 16 ; FOV, mm; matrix, ; slice thickness, 1.8 mm (0.9-mm overlap with adjacent section); and section slices, 120. A 20-mm presaturation band was applied 10 mm above the imaging volume to saturate the craniocaudal venous blood on intracranial and neck TOF MRA. CE-MRA was performed in the coronal plane, and scanning included the brain, neck, and aortic arch inferiorly. CE-MRA was performed with a 3D fast-field echo sequence and timing-robust centric k-space ordering using the following parameters [15]: TR range/te, 4 6/2; flip angle, 40 ; FOV, mm; matrix, (1.5-T system) or (3-T system); slice thickness, 1.2 mm (0.6-mm overlap with adjacent section); and section slices, 130. As contrast medium, 0.1 mmol/kg of gadobutrol (Gadovist, Bayer HealthCare) or gadodiamide (Omniscan, GE Healthcare) was injected at a rate of 2 ml/s with a power injector. Each bolus was immediately followed by a 20-mL saline flush. The contrast medium was usually injected through the right antecubital vein [16, 17]. MR fluoroscopic manual triggering was used to ensure proper timing in detecting the arrival of the contrast medium and initiating CE-MRA scanning. Analysis of MRI and MR Angiography Findings Two neuroradiologists blinded to the diagnosis independently reviewed the MRI and MRA examinations, which were presented in random order, of each patient. Radiologists completed imaging analysis including brain MRI, intracranial TOF MRA, and neck MRA (either CE-MRA or TOF MRA) altogether in a single session for each patient. Then they reached agreement by consensus. On intracranial TOF MRA, the side (right or left) and location (cavernous sinus, inferior petrosal sinus, sigmoid sinus, or transverse sinus) of abnormal venous signal intensities were assessed. The signal intensity (signal void or increased signal intensity) in the ipsilateral proximal jugular vein and sigmoid sinus on axial T2-weighted imaging and the presence or absence of a flow-void cluster (i.e., a cluster of vessels with flow void), exophthalmos, dilatation of the superior ophthalmic vein, and enlargement of the cavernous sinus on T2-weighted imaging and contrast-enhanced T1-weighted imaging were assessed. 840 AJR:202, April 2014

3 MRI and MRA to Differentiate JVR From Cavernous DAVF Owing to the centric k-space ordering of the CE-MRA technique, a slight delay in starting the scanning can cause opacification of the internal jugular vein. Therefore, we arbitrarily classified the 52 patients who underwent CE-MRA into two subgroups: arterial phase (n = 21) and arteriovenous phase (n = 31). The arterial phase was defined as the interval when the intracranial and cervical arteries were well visualized without opacification of the internal jugular vein. The arteriovenous phase was defined as the interval when the internal jugular vein was well visualized along with arterial structures. The following findings were then assessed in the CE-MRA subgroups and the neck TOF MRA examinations: the presence or absence of early opacification of the cavernous sinus in the arterial phase of CE- MRA (n = 21); the side (ipsilateral or contralateral to abnormal venous signal intensities on intracranial TOF MRA) of jugular venous drainage in the arteriovenous phase of CE-MRA (n = 31); and the presence or absence of retrograde jugular venous flow in neck TOF MRA (n = 15). Statistical Analysis All statistical analyses were performed using statistics software (PASW, version 18, SPSS). Statistical differences were examined using the Student t test or Fisher exact test. In all cases, statistical significance was defined as a p value less than Results There was a significant statistical difference between the group with JVR and the group with cavernous DAVF with regard to sex (p < 0.001, Student t test) but not age (p = 0.744, Student t test). Intracranial Time-of-Flight MR Angiography In the patients with JVR, the abnormal venous signal intensities were on the right side in 11 patients and on the left in 30. For the patients with cavernous DAVF, the respective numbers were 15 and 11 (p < 0.05). The locations of the abnormal venous signal intensities are summarized in Table 1. Involvement of the cavernous sinus, regardless of involvement of the inferior petrosal sinus, sigmoid sinus, or transverse sinus, was found in 12% (5/41) of the patients with JVR (Figs. 1A, 2A, and 3A) and in 100% (26/26) of the patients with cavernous DAVF (Fig. 4A) (p < 0.001). MRI Findings The brain MRI findings are summarized in Table 2. Increased signal intensity in the TABLE 1: Location of Abnormal Venous Signal Intensities on Intracranial Time-of-Flight (TOF) MR Angiography (MRA) in Patients With Jugular Venous Reflux (JVR) and Patients With Cavernous Dural Arteriovenous Fistula (DAVF) Location of Abnormal Venous Signal Intensities Patients with CS involvement ipsilateral proximal jugular vein and sigmoid sinus on T2-weighted imaging was found in 73% (30/41) of the patients with JVR (Figs. 1B and 2B) and 4% (1/26) of the patients with cavernous DAVF (p < 0.001). One or more of the following findings were observed on T2-weighted imaging or contrastenhanced T1-weighted imaging (Figs. 4D 4G) in none of the patients with JVR and in 81% (21/26) of the patients with cavernous DAVF (p < 0.001): flow-void cluster, exophthalmos, dilatation of the superior ophthalmic vein, and enlargement of the cavernous sinus. Contrast-Enhanced MR Angiography The CE-MRA findings are summarized in Table 3. During the arterial phase of CE- MRA, early opacification of the cavernous sinus was found in none of the patients with JVR (Fig. 1C) and in all the patients with JVR (n = 41) Cavernous DAVF (n = 26) CS 0 (0) 17 (65) CS and IPS 4 (10) 7 (27) CS, IPS, and SS 1 (2) 2 (8) Patients with no CS involvement IPS 24 (59) 0 (0) IPS, SS, and TS 7 (17) 0 (0) SS and TS 5 (12) 0 (0) Note CS = cavernous sinus, IPS = inferior petrosal sinus, SS = sigmoid sinus, TS = transverse sinus. cavernous DAVF. During the arteriovenous phase of CE-MRA, all patients with JVR had contralateral jugular venous drainage to abnormal venous signal intensities on intracranial TOF MRA (i.e., none had ipsilateral jugular venous drainage) (Fig. 2C), whereas all patients with cavernous DAVF had ipsilateral or bilateral jugular venous drainage (Fig. 4C) (p < 0.001). Neck Time-of-Flight MR Angiography The neck TOF MRA findings are summarized in Table 3. Retrograde jugular flow on neck TOF MRA was identified in 63% (5/8) patients with JVR (Fig. 3C) and in none of the patients with cavernous DAVF (p = 0.026). Discussion JVR is known to be caused by physiologic compression of the brachiocephalic vein that TABLE 2: Brain MRI Findings in Patients With Jugular Venous Reflux (JVR) and Patients With Cavernous Dural Arteriovenous Fistula (DAVF) MRI Finding JVR (n = 41) Cavernous DAVF (n = 26) p Signal intensity at the < ipsilateral proximal jugular vein and sigmoid sinus on T2-weighted imaging Increased 30 (73) 1 (4) Low 11 (27) 25 (96) MRI findings suggestive of cavernous DAVF a < None 41 (100) 5 (19) One or more 0 (0) 21 (81) a Findings include flow-void cluster, exophthalmos, dilatation of the superior ophthalmic vein, and enlargement of the cavernous sinus on T2-weighted imaging or contrast-enhanced T1-weighted imaging. AJR:202, April

4 Kim et al. TABLE 3: Neck MR Angiography (MRA) Findings in Patients With Jugular Venous Reflux (JVR) and Patients With Cavernous Dural Arteriovenous Fistula (DAVF) Subgroups and Neck MRA Findings JVR (n = 41) Cavernous DAVF (n = 26) p Contrast-enhanced MRA, arterial phase group (n = 21) Early opacification of the < cavernous sinus Yes 0 (0) 6 (100) No 15 (100) 0 (0) Contrast-enhanced MRA, arteriovenous phase group (n = 31) Ipsilateral jugular venous drainage a < Yes 0 (0) 13 (100) No 18 (100) 0 (0) TOF MRA group (n = 15) Ipsilateral retrograde jugular venous flow a Yes 5 (63) 0 (0) No 3 (38) 7 (100) Note TOF = time-of-flight. a Ipsilateral means on the same side as abnormal venous signal intensities on intracranial TOF MRA. leads to stagnation or reversal of the internal jugular vein flow. Although most reported cases of JVR were detected on the left side using MRI and were caused by compression of the left brachiocephalic vein between the aorta and sternum [5, 8, 11, 18], JVR has also been reported on the right side in which the right brachiocephalic vein was being compressed by the brachiocephalic artery and resulted in severe stenosis just proximal to the superior vena cava [6]. Because the left brachiocephalic vein has a long course, it tends to be more easily compressed than does the right brachiocephalic vein. Uchino et al. [6] A suggested that atherosclerosis in middle-aged and elderly people could cause dilatation and tortuosity of the aortic arch and result in stenosis of the left brachiocephalic vein. JVR is also known to be influenced by respiration; compression of the brachiocephalic vein can be relieved by full inspiration, because the distance between the aorta and sternum increases with elevation of the anterior chest wall [5, 8, 11]. Several special imaging techniques have been used to diagnose JVR such as phase-contrast MRA to show the flow direction [11] and Doppler ultrasound to reveal retrograde flow [13, 14]; in addition, by applying a presaturation pulse below the excited slab, which suppressed flow signals of the caudocranial direction, investigators reported that no signals were found in the veins that were previously enhanced on intracranial TOF MRA with a presaturation pulse above the excited slab [5, 8]. JVR has been regarded as a physiologic stenosis; however, the results of recent studies suggest that JVR might be related to some neurologic diseases or symptoms, such as transient global amnesia and transient monocular blindness [8, 13, 14]. JVR may impede venous outlet flow and cause back pressure into the ocular and cerebral circulations, thereby causing transient cerebral venous congestion or an alteration of the effective ocular perfusion pressure. Intracranial DAVFs are pathologic shunts between dural arteries and dural venous sinuses or meningeal veins [4]. The MRI findings of DAVF include flow-void clusters, engorged ophthalmic vein or proptosis, white matter hyperintensity, intracranial hemorrhage, dilated leptomeningeal or medullary vessels, a venous pouch, and leptomeningeal or medullary vascular enlargement. The MRA findings of DAVF include an identifiable fistula, venous flow visualization, and Fig year-old woman with left jugular venous reflux detected by 3-T MR system. A, Intracranial time-of-flight MR angiography (MRA) image shows abnormal high signal intensities in left inferior petrosal sinus (arrows). B, T2-weighted image shows increased signal intensity of left internal jugular vein compared with dark signal intensity of right internal jugular vein (arrows). C, Early opacification of cavernous sinus is not observed on arterial phase of contrast-enhanced MRA image. B C 842 AJR:202, April 2014

5 MRI and MRA to Differentiate JVR From Cavernous DAVF prominent extracranial vessels [3]. Although false-negative cases on MRA have occasionally been reported, visualization of venous flow on MRA was found in 90% of cases of DAVF [3]. When prominent extracranial vessels and venous flow are observed on MRA, DAVF can easily be diagnosed. However, it is necessary to distinguish between JVR and DAVF if venous flow alone is present on intracranial TOF MRA. In this study, we attempted to distinguish between JVR and cavernous DAVF by analyzing routine brain MR images and MR angiograms without the aid of special imaging techniques or conventional angiography. Because JVR and cavernous DAVF had A Fig year-old man with left jugular venous reflux detected by 3-T MR system. A, Intracranial time-of-flight MR angiography (MRA) image shows abnormal high signal intensities in left inferior petrosal sinus (arrows). B, T2-weighted image shows increased signal intensity of left internal jugular vein in contrast to dark signal intensity of right internal jugular vein (arrows). C, Arteriovenous phase contrast-enhanced MRA image shows contralateral jugular drainage (arrows). A Fig year-old woman with left jugular venous reflux (JVR) detected by 3-T MR system. A, Intracranial time-of-flight (TOF) MR angiography (MRA) image shows abnormal high signal intensities in left sigmoid and transverse sinuses (arrows). B, Left internal jugular vein and right internal jugular vein (arrows) are seen as signal voids on T2-weighted image. C, Neck TOF MRA image shows ipsilateral jugular vein flow as well as arterial structures (arrows); these findings indicate presence of caudocephalad (retrograde) venous flow. different imaging findings, it was not difficult to differentiate one from the other. The JVR signal was predominantly on the left side in contrast to cavernous DAVF. In addition, JVR mainly involved the inferior petrosal sinus, sigmoid sinus, or transverse sinus without involvement of the cavernous sinus, whereas cavernous DAVF always involved the cavernous sinus. These findings are consistent with the pathogenetic mechanism of JVR in which the left brachiocephalic vein is more easily compressed than the right. Because the cavernous sinus is located more cephalad than the inferior petrosal sinus and sigmoid sinus, the retrograde venous signal in the cavernous sinus may be weaker. B B CE-MRA has been widely used for neck MRA because of its wide coverage and short scanning time. However, suboptimal synchronization between the arrival time of the contrast medium and CE-MRA scanning frequently causes venous opacification. Therefore, in this study the patients who underwent CE-MRA were divided into subgroups representing the arterial phase and the arteriovenous phase (i.e., venous opacification group). The presence or absence of early opacification of the cavernous sinus in the arterial phase of CE-MRA appeared to be an imaging finding for distinguishing between JVR and cavernous DAVF. This finding could not be assessed during the C C AJR:202, April

6 Kim et al. A B D F Fig year-old woman with left cavernous dural arteriovenous fistula (DAVF) detected by 1.5-T MR system. A, Intracranial time-of-flight (TOF) MR angiography (MRA) image shows abnormal high signal intensities in left cavernous sinus (large black arrow), inferior petrosal sinus (small black arrows), and sigmoid and transverse sinuses (white arrows). B, Right and left internal jugular veins (arrows) are seen as signal voids on T2-weighted image. C, Arteriovenous phase contrast-enhanced MRA image shows opacification of left cavernous sinus and inferior petrosal sinus (small arrows) and early drainage in ipsilateral internal jugular vein (large arrows). D, T2-weighted image shows abnormal flow void in left cavernous sinus and clivus (arrows). E G, Contrast-enhanced T1-weighted images show enlarged left superior ophthalmic vein (arrows, E), exophthalmos (arrows, F), and enlarged left cavernous sinus (arrow, G). H, Conventional catheter angiogram. Frontal view of left external carotid artery shows DAVF in left cavernous sinus (black arrow) and early venous drainage into left inferior petrosal sinus and internal jugular vein (white arrows). C E G H 844 AJR:202, April 2014

7 MRI and MRA to Differentiate JVR From Cavernous DAVF arteriovenous phase of CE-MRA because of venous contamination. However, the venous drainage patterns assessed in the arteriovenous phase showed significant differences between patients with JVR and those with cavernous DAVF. All patients with JVR had contralateral jugular vein drainage, whereas those with cavernous DAVF had ipsilateral or bilateral jugular drainage. It seems that the ipsilateral jugular vein in JVR has some degree of resistance, due to the retrograde flow and elevated venous pressure, so the intracranial venous blood drains more easily through the contralateral jugular vein where the venous pressure is low. Investigators have reported that a more delayed venous phase in CE-MRA can be useful for revealing stenosis in the left brachiocephalic vein [5]. Therefore, multiphase or time-resolved CE-MRA, although not routinely used in most hospitals, could be more helpful in distinguishing between JVR and cavernous DAVF by evaluating early opacification of the cavernous sinus, the side of jugular venous drainage, and any stenosis of the brachiocephalic vein in a single imaging sequence. In fact, recent studies have shown the usefulness of multiphase or time-resolved CE-MRA in patients with DAVF, although JVR was not included in those studies [19 22]. The presence of jugular venous flow on neck TOF MRA is direct evidence of JVR. In this study, however, retrograde jugular venous flow on neck TOF MRA was not observed as frequently as abnormal venous signal intensities from the inferior petrosal sinus or sigmoid sinus on intracranial TOF MRA. In other words, retrograde jugular venous flow was observed in approximately two thirds of patients with JVR. Because faster flow has higher signal intensity on TOF MRA, we can speculate that the velocity of retrograde flow may be faster in the inferior petrosal sinus and sigmoid sinus than in the internal jugular vein because the sinuses both have smaller diameters than the vein [6]. Other findings suggesting JVR were increased signal intensity in the ipsilateral jugular vein and sigmoid sinus on T2-weighted imaging, which seems to be caused by slow flow [5]. Tanaka et al. [18] reported that hemostasis due to compression of the left brachiocephalic vein could be the major cause of increased signal intensity in the ipsilateral internal jugular vein and sigmoid sinus. Increased signal intensity in the ipsilateral internal jugular vein and sigmoid sinus was also observed in 73% of patients with JVR and 4% of patients with cavernous DAVF; this increased signal intensity could have been the result of a slow flow velocity regardless of the flow direction [18]. Patients with JVR did not have any findings suggesting cavernous DAVF, such as flow-void cluster, exophthalmos, engorgement of the superior ophthalmic vein, or enlargement of the cavernous sinus; the lack of these findings can also be helpful for establishing the diagnosis of JVR. In contrast, 81% of patients with cavernous DAVF showed one or more of those MR findings suggesting cavernous DAVF, although falsenegative cases were present. Although the spectrum of MRI findings of DAVF was variable in one study, ranging from no visible lesion to intracranial hemorrhage or venous infarction, flow-void clustering was most frequently observed in 80% of patients with DAVF on MRI [3]. Another mechanism, apart from DAVF or JVR, that can cause abnormal venous signal intensities on TOF MRA is normal venous flow in the caudocranial direction. Abnormal venous signal intensities can arise from the sphenoparietal sinus, which flows upward from the bottom of the middle cranial fossa to the cavernous sinus [9]. Thus, the cavernous sinus can be visualized owing to the antegrade venous flow from the sphenoparietal sinus. Recently, flow signals in the pterygoid plexus, emissary vein, and cavernous sinus were reported in healthy subjects, and high signal intensity in the cavernous sinus can also arise from pterygoid plexus reversal through the sphenoid emissary vein [23]. However, we included venous signals in the major venous sinuses including the cavernous sinus, inferior petrosal sinus, sigmoid sinus, and transverse sinus regardless of visualization of the sphenoparietal sinus or pterygoid plexus because the aim of this study was to differentiate JVR from cavernous DAVF. There are several limitations to this study. First, conventional catheter angiography was performed in only five patients with JVR, so we cannot completely exclude the possibility that some patients with JVR had an asymptomatic small DAVF. However, the brain MRI and MRA findings for JVR were different from those for cavernous DAVF, and all patients with JVR had imaging findings consistent with JVR. Second, patients with DAVF other than the cavernous type were not included in this study. JVR also needs to be differentiated from other types of DAVF located in the sigmoid sinus or transverse sinus. When prominent extracranial vessels and abnormal venous signal intensities are visualized on TOF MRA, DAVF can be easily diagnosed. However, if prominent extracranial vessels are not visualized on TOF MRA, we believe that differentiation of JVR from cavernous DAVF based on the MRI findings investigated in this study might be possible. Third, although we used MR systems of both 1.5 and 3 T with different scanning parameters, we did not assess the differences between the results obtained using these two MR systems. The incidence of a high signal intensity in the dural sinuses was lower with the 3-T systems than with the 1.5-T systems, but the mechanism of abnormal venous signal intensities was jugular venous reflux in both cases [6, 7]. Different field strengths and scanning parameters might influence the detection of abnormal venous signal intensities but are unlikely to have affected our results regarding the differentiation of JVR and cavernous DAVF. In conclusion, routine MRI and MRA enabled us to differentiate JVR from cavernous DAVF. Increased signal intensity of the ipsilateral jugular vein on T2-weighted imaging, the absence of early opacification of the cavernous sinus in the arterial phase of CE-MRA, the absence of ipsilateral jugular venous drainage in the arteriovenous phase of CE-MRA, and the presence of ipsilateral retrograde jugular venous flow on neck TOF MRA were the most reliable findings to suggest JVR. Acknowledgment We thank the staff at the Medical Research Collaborating Center at Seoul National University Bundang Hospital for statistical analyses. References 1. Chen JC, Tsuruda JS, Halbach VV. Suspected dural arteriovenous fistula: results with screening MR angiography in seven patients. Radiology 1992; 183: Ikawa F, Uozumi T, Kiya K, Kurisu K, Arita K, Sumida M. Diagnosis of carotid-cavernous fistulas with magnetic resonance angiography: demonstrating the draining veins utilizing 3-D timeof-flight and 3-D phase-contrast techniques. Neurosurg Rev 1996; 19: Kwon BJ, Han MH, Kang HS, Chang KH. MR imaging findings of intracranial dural arteriovenous fistulas: relations with venous drainage patterns. AJNR 2005; 26: Gandhi D, Chen J, Pearl M, Huang J, Gemmete JJ, AJR:202, April

8 Kim et al. Kathuria S. Intracranial dural arteriovenous fistu- flow signal on three-dimensional time-of-flight phy from the left brachiocephalic venous stasis. J las: classification, imaging findings, and treat- MR angiography. AJNR 1999; 20: Magn Reson Imaging 1999; 10: ment. AJNR 2012; 33: Kudo K, Terae S, Ishii A, et al. Physiologic change 18. Tanaka T, Uemura K, Takahashi M, et al. Com- 5. Paksoy Y, Genc BO, Genc E. Retrograde flow in in flow velocity and direction of dural venous si- pression of the left brachiocephalic vein: cause of the left inferior petrosal sinus and blood steal of nuses with respiration: MR venography and flow high signal intensity of the left sigmoid sinus and the cavernous sinus associated with central vein stenosis: MR angiographic findings. AJNR 2003; 24: Uchino A, Nomiyama K, Takase Y, et al. Retrograde flow in the dural sinuses detected by threedimensional time-of-flight MR angiography. Neuroradiology 2007; 49: Inano S, Itoh D, Takao H, et al. High signal intensity in the dural sinuses on 3D-TOF MR angiography at 3.0 T. Clin Imaging 2010; 34: Chung CP, Hsu HY, Chao AC, Chang FC, Sheng WY, Hu HH. Detection of intracranial venous reflux in patients of transient global amnesia. Neurology 2006; 66: Sakamoto M, Taoka T, Iwasaki S, et al. Paradoxical parasellar high signals resembling shunt diseases on routine 3D time-of-flight MR angiography of the brain: mechanism for the signals and differential diagnosis from shunt diseases. Magn Reson Imaging 2004; 22: Ouanounou S, Tomsick TA, Heitsman C, Holland CK. Cavernous sinus and inferior petrosal sinus analysis. AJNR 2004; 25: Shieh CC, Hu HH, Yang ST, et al. Validation of the jugular venous reflux animal model by threedimensional time-of-flight MRA with a clinical scanner. Cerebrovasc Dis 2010; 30: Chung CP, Hsu HY, Chao AC, Cheng CY, Lin SJ, Hu HH. Jugular venous reflux affects ocular venous system in transient monocular blindness. Cerebrovasc Dis 2010; 29: Hsu HY, Chao AC, Chen YY, et al. Reflux of jugular and retrobulbar venous flow in transient monocular blindness. Ann Neurol 2008; 63: Kim SH, Lee JS, Kwon OK, Han MK, Kim JH. Prevalence study of proximal vertebral artery stenosis using high-resolution contrast-enhanced magnetic resonance angiography. Acta Radiol 2005; 46: McCarthy J, Solomon NA. The effects of injection site, age, and body position on cervical venous reflux. Radiology 1979; 130: Lee YJ, Chung TS, Joo JY, Chien D, Laub G. Suboptimal contrast-enhanced carotid MR angiogra- internal jugular vein on MR images. Radiology 1993; 188: Ziyeh S, Strecker R, Berlis A, Weber J, Klisch J, Mader I. Dynamic 3D MR angiography of intraand extracranial vascular malformations at 3T: a technical note. AJNR 2005; 26: Akiba H, Tamakawa M, Hyodoh H, et al. Assessment of dural arteriovenous fistulas of the cavernous sinuses on 3D dynamic MR angiography. AJNR 2008; 29: Farb RI, Agid R, Willinsky RA, Johnstone DM, Terbrugge KG. Cranial dural arteriovenous fistula: diagnosis and classification with time-resolved MR angiography at 3T. AJNR 2009; 30: Nishimura S, Hirai T, Sasao A, et al. Evaluation of dural arteriovenous fistulas with 4D contrast-enhanced MR angiography at 3T. AJNR 2010; 31: Watanabe K, Kakeda S, Watanabe R, Ohnari N, Korogi Y. Normal flow signal of the pterygoid plexus on 3T MRA in patients without DAVF of the cavernous sinus. AJNR 2013; 34: AJR:202, April 2014