All the patients were consecutively enrolled from August 2013 to January 2016 to undergo catheter angiography and 3D HRMR VWI (pre- and post-contrast), when they fulfilled the following criteria: 1) clinical symptoms and findings on computed tomographic angiography (CTA) or MRA suggestive of an intracranial VBDA; 2) no contraindication for MR imaging and catheter angiography. Demographics, clinical findings, imaging data, and risk factors such as hypertension, hyperlipidemia, diabetes mellitus, and cigarette smoking were also collected prospectively. Atherosclerotic intraplaque hemorrhage (IPH), atherosclerotic ulceration and aneurysm with wall atherosclerosis that might mimic dissection were excluded from this study.
Catheter angiography protocol
All patients underwent catheter angiography on an Allura Xper FD20/10 (Philips Healthcare), Artis Zee Biplane, or Artis Zeego angiography unit (Siemens Healthcare). Catheter angiography on anteroposterior and lateral projections, and oblique projections as necessary were performed after placement of a diagnostic catheter in the V1 segment of the vertebral artery in question with injection of 5 ml of contrast (Iodixanol, GE healthcare) at a rate of 3ml per second at 300 Pa pressure. Then, 3D angiography was performed while injecting a total 18 ml contrast at a rate of 3ml per second at 300 Pa pressure.
Catheter angiography assessment
Direct and indirect signs of VBDAs were assessed by two experienced interventional neuroradiologists on both conventional and 3D catheter angiography images independently, blinded to the clinical and other imaging information including CTA, MRA and VWI. Direct signs included visualization of an intimal flap, or a double lumen. Indirect signs included long filiform or irregular stenosis, occlusion; or fusiform or serpentine aneurysmal dilation with focal, long filiform, or irregular stenosis (pearl-and-string sign); or fusiform or aneurysmal dilation at a non-branching site.2
MR exam was performed with a 3T MRI scanner (Siemens, Magnetom Skyra, Erlangen, Germany) using a 20-channel phased-array head and neck joint coil. A 3D time-of-flight (TOF) MRA was first acquired to localize the intracranial arteries with the following parameters：TR/TE = 20/3.4 ms, FOV = 192 ×240 mm, thickness = 1 mm, matrix = 460×640, and NEX = 1, the voxel volume was 0.4 × 0.4 × 1 mm on MRA. HRMR VWI was acquired with a 3D SPACE T1W sequence in the coronal plane (45-mm-thick slab) to cover the major intracranial vessels as identified on the TOF MRA with the following parameters: TR/TE = 600/20 ms, FOV = 182 × 220 × 45 mm, matrix= 212 × 256 ×50, average = 1.5, echo train duration = 118, the voxel volume was 0.9 × 0.9 ×0.9 mm on SPACE T1W image. SPACE post-contrast T1WI was obtained 5 minutes after gadolinium injection (0.1 mmol/kg of gadopentetate dimeglumine, Magnevist; Bayer Schering Pharma, Berlin, Germany) with the same parameters as pre-contrast T1WI.
Resource data of VWIs were reformatted to acquire long and short axial images of the vertebrobasilar artery, with voxel volume = 0.9 × 0.9 × 0.9 mm. Multiple lesions with varying locations and orientations were displayed with individually reconstructed projections (Figure 1-3).
Direct and indirect signs of VBDAs were assessed by 2 neuroradiologists independently with at least 5 years of experience in reading intracranial HRMR VWI, and were blinded to the clinical and other imaging information including CTA, MRA and catheter angiography images. The direct signs included intimal flap, double lumen and IMH. Intimal flap sign was defined as an abnormal linear structure separating a true lumen from a false lumen (Figure 1-3). To discriminate the intimal flap from the inflow artifact that can appear at the center of the lumen, we deemed only linear structures as an intimal flap when it extended to and from the arterial sidewall on SPACE pre- or post-contrast T1W images.2 Double lumen sign was defined as two lumens represented as two jets of flow void within one vessel on MRI. This is distinct from fenestration, which are separate vessels. IMH sign was defined as detection of eccentric or circumferential hyper-, iso-, or hypointense signal (corresponding to hemorrhage age) on pre-contrast T1WI within the arterial wall (Figure 2), 9 generally extending longitudinally and located in the proximal or distal segment of pseudolumen, or filled the pseudolumen entirely. Intramural, partially thrombosed haematoma (Mizutani type 3 dissection) was diagnosed on imaging when heterogenous, multi-aged hemorrhage in the vessel wall was identified11,12, which is classified into IMH.
IMH was differentiated from intraluminal thrombus by location and enhancement: IMH is separated from the lumen by the intimal flap with no or sometimes heterogeneous enhancement. Lesions were considered intraluminal thrombus when it was seen in a juxtaluminal location with obvious enhancement.10 Atherosclerotic IPH and ulceration which may mimic dissection was discriminated from IMH and excluded from this study with the following criteria: IPH and ulceration was still located within the atherosclerotic plaque with relatively small size, generally not associated with a focal enlargement of external diameter.3 Atherosclerosis within an aneurysmal wall which may mimic chronic IMH was determined and excluded with the following characteristics including: at sites of arterial bifurcation, not associated with intimal flap and double lumen, not associated with heterogeneous T1 hyperintense signal (recurrent hemorrhage).
The indirect signs were also assessed based on the luminal characteristics of VWI, similar to that of catheter angiography.
After assessment on catheter angiography and VWI, segments suspicious for VBDAs were given a final imaging diagnosis, individually. Based on catheter angiography, lesion was diagnosed as definite VBDAs by interventional neuroradiologists when intimal flap or double lumen was detected. On VWI, when IMH, intimal flap, or double lumen was found， it was considered a definite VBDA by neuroradiologists. If there were only indirect signs of dissection, the lesion was considered a possible VDBA. A lesion was considered segmental ectasia (Mizutani type 2), if there was only local dilation and normal wall thickness, without evidence of mural thrombosis.12
Continuous variables were described as mean ± standard deviation, or as interquartile range (if not normally distributed). Cohen’s k-statistic was computed to quantify the intra-observer and inter-oberver agreement for asessement of itimal flap, double lumen and IMH. A value of k > 0.75 was used to indicate a high level of agreement. After consensus, McNemar test was used to assess the difference for definitively diagnosing VBDAs and detecting direct dissection signs between VWI and catheter angiography. SPSS 11.5 (SPSS, Inc., Chicago, IL) was used for statistical analysis. All reported p values were 2-sided, and a p value of ＜0.05 was considered significant.