Background and purpose Follow-up of intracranial aneurysms treated by flow diverter with MRI is complicated by imaging artifacts produced by these devices. This study compares the diagnostic accuracy of three-dimensional time-of-flight MR angiography (3D-TOF-MRA) and contrast-enhanced MRA (CE-MRA) at 3 T for the evaluation of aneurysm occlusion and parent artery patency after flow diversion treatment, with digital subtraction angiography (DSA) as the gold standard.
Materials and methods Patients treated with flow diverters between January 2009 and January 2013 followed by MRA at 3 T (3D-TOF-MRA and CE-MRA) and DSA within a 48 h period were included in a prospective single-center study. Aneurysm occlusion was assessed with full and simplified Montreal scales and parent artery patency with three-grade and two-grade scales.
Results Twenty-two patients harboring 23 treated aneurysms were included. Interobserver agreement using simplified scales for occlusion (Montreal) and parent artery patency were higher for DSA (0.88 and 0.61) and CE-MRA (0.74 and 0.55) than for 3D-TOF-MRA (0.51 and 0.02). Intermodality agreement was higher for CE-MRA (0.88 and 0.32) than for 3D-TOF-MRA (0.59 and 0.11). CE-MRA yielded better accuracy than 3D-TOF-MRA for aneurysm remnant detection (sensitivity 83% vs 50%; specificity 100% vs 100%) and for the status of the parent artery (specificity 63% vs 32%; sensitivity 100% vs 100%).
Conclusions At 3 T, CE-MRA is superior to 3D-TOF-MRA for the evaluation of aneurysm occlusion and parent artery patency after flow diversion treatment. However, intraluminal evaluation remains difficult with MRA regardless of the sequence used.
- Flow Diverter
- Magnetic Resonance Angiography
Statistics from Altmetric.com
Endovascular treatment is now the first-line treatment for the management of ruptured and unruptured intracranial aneurysms.1–3 As wide-necked, large, giant and fusiform aneurysms are difficult to treat with standard coiling, alternative techniques such as balloon-assisted coiling, stent-assisted coiling, flow diverters (FDs), and flow disrupters have been proposed.4–7 Aneurysm recurrence is a major concern after endovascular treatment, necessitating regular imaging follow-up.8 Follow-up of coiled aneurysms was initially based on digital subtraction angiography (DSA) and magnetic resonance angiography (MRA), which have proved to be good tools for the follow-up of coiled aneurysms.9–13 Three-dimensional time-of-flight MRA (3D-TOF-MRA) at 3 T is probably the most effective sequence to evaluate occlusion of coiled aneurysms.14 ,15 Follow-up of intracranial aneurysms treated with FD is challenging as these devices produce imaging artifacts. The potential for non-invasive techniques (CT angiography (CTA) or MRA) for follow-up of FD-treated intracranial aneurysms has been little evaluated.16 As MRA is widely used for follow-up of coiled aneurysms, the first step is to evaluate this technique. The aim of this prospective study is to compare the diagnostic accuracy of 3D-TOF-MRA and contrast-enhanced MRA (CE-MRA) at 3 T for the evaluation of aneurysm occlusion and parent artery patency after FD treatment, with DSA as the gold standard.
Materials and methods
All patients treated with FD were prospectively included in a database. Patients treated with FD and followed with both MRA and DSA were included. Patients did not have CTA. Both DSA and 3 T MRA performed within a period of <48 h were inclusion criteria. Exclusion criteria were patients with a contraindication to MRI, claustrophobia, refusal to pass examinations (MRI or DSA), or those aged <18 years. Also, as the study aimed to compare 3D-TOF-MRA and CE-MRA, any patients who had not received these two sequences during follow-up were excluded. Patients with multiple aneurysms (treated or untreated) were not excluded.
Imaging technique for intra-arterial DSA
Intra-arterial DSA was performed with a biplane angiographic system (Axiom Artis; Siemens, Erlangen, Germany). Using transfemoral catheterization, selective injections of the internal carotid artery (ICA) or vertebral artery (VA) were performed according to aneurysm location. The following standard projections were obtained: anteroposterior, lateral, and working views. For ICA, 8 mL of non-ionic contrast agent (iodixanol, Visipaque; GE Healthcare, Oslo, Norway) was injected with a velocity of 4 mL/s. For the VA, 8–10 mL was injected with a velocity of 4–5 mL/s.
Imaging technique for MRA
MRA examinations were performed on 3T Philips (Achieva; Philips Healthcare, Best, The Netherlands). Examinations were performed with the following optimized parameters. For 3D-TOF-MRA: TE, 3.45 ms; TR, 18 ms; flip angle, 20°; total acquisition time, 4:59 min; number of sections, 140; section thickness, 0.55 mm; FOV, 210 mm; rectangular field of view, 90%; acquisition matrix, 464; reconstruction matrix, 512; reconstructed voxel size, 0.41×0.41×0.55 mm. For the CE-MRA: TE, 1.96 ms; TR, 5.4 ms; flip angle, 30°; total acquisition time 0:52 min; number of sections, 110; section thickness, 0.5 mm; FOV, 210 mm; rectangular field of view, 85%; acquisition matrix, 480; reconstruction matrix, 512; reconstructed voxel size, 0.41×0.41×0.50 mm. CE-MRA randomly sampled the central k-space during venous injection of a gadolinium-based contrast agent (meglumine gadorate, Dotarem; Guerbet, Aulnay-sous-Bois, France). A bolus of 20 mL was used, followed by 30 mL saline with a scopic-based detection of the bolus (phase contrast survey).
Clinical and anatomic data regarding patient (sex, age) and aneurysm (number, localization, type, size, aneurysm status) were collected. The interval time between aneurysm treatment and anatomic evaluation was also collected. Aneurysm location was classified into four groups: anterior cerebral artery/anterior communicating artery; ICA; middle cerebral artery; and posterior circulation/vertebrobasilar.
All examinations (DSA and MRA) were anonymised by different-number random assignment by series. All images were independently evaluated in random order by two experienced interventional neuroradiologists (LP: 20 years; AB: 2 years) and, in case of disagreement, consensus was found between the two radiologists. DSA, 3D-TOF-MRA, and CE-MRA were evaluated separately without knowledge of the MRA or DSA examination results. The pre-treatment DSA was unavailable, but the location of the aneurysms to be evaluated was provided to the readers. For both 3D-TOF-MRA and CE-MRA, source images and maximum intensity projection (MIP) reconstructions were analyzed. Aneurysm occlusion was evaluated using the three-grade Montreal scale17 (total occlusion, neck remnant, and aneurysm remnant). The modified two-grade scale (adequate occlusion (total occlusion + neck remnant) and aneurysm remnant) was also used.18 The patency status of the parent artery was evaluated as: no change in the parent artery diameter (patent), narrowing of the parent artery (stenosis), or parent artery occlusion. A simplified two-grade scale was used to assess patency of the parent artery, in which the parent artery was classified as normal or pathological (stenosis or occlusion).
Quantitative variables are reported as mean±SD, median (range) while qualitative variables were reported as number and percentage. We used Wilcoxon signed rank tests to compare the observed artifacts and visibility of FDs between 3D-TOF-MRA and MRA. p Values <0.05 were considered statistically significant. Weighted κ statistics were used to evaluate interobserver and intermodality agreement for each technique. According to Landis and Koch,19 the interpretation of κ was as follows: κ<0, no agreement; κ=0–0.19, poor agreement; κ=0.20–0.39, fair agreement; κ=0.40–0.59, moderate agreement; κ=0.60–0.79, substantial agreement; and κ=0.80–1.00, almost perfect agreement. Using the consensus evaluation of intra-arterial DSA as a reference test to evaluate the degree of aneurysm occlusion and patency of parent artery, the sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated for 3D-TOF-MRA and CE-MRA with corresponding 95% CI. All analyses were performed using SPSS V.20 (Chicago, Illinois, USA).
Between January 2009 and January 2013, 22 patients were included (age 26–77 years; mean 54.5±13.5 years; median 54 years), including 11 women (50%) and 11 men (50%). Among these 22 patients, 16 (73%) had one aneurysm and 6 (27%) had multiple (two) aneurysms. Twenty-one of the 22 patients (95.5%) were treated for one aneurysm by FDs and one patient (4.5%) had two aneurysms treated by FDs. Detailed characteristics of the aneurysms are shown in table 1. The circumstances of discovery were rupture in 6/22 patients (27.2%), mass effect in 6/22 patients (27.2%), and fortuitous in 10/22 patients (45.5%). The ruptured aneurysms were not treated with FDs at the acute phase of bleeding, but in a second-step treatment or in case of recanalization.
Fourteen of the 22 patients (63.7%) were first treated with FD and eight (36.3%) were retreated with FD. Fifteen of the 23 aneurysms (65.2%) were first treated with FD and eight (34.8%) were retreated with FD. Among the 15 aneurysms first treated with FD, 10 were treated with FD exclusively, four with coils and FD, and one with FD and stent. Among the eight aneurysms retreated with FD, six had initial treatment by coils, one was first treated with a stent, and one with stent and coils.
Nineteen of the 22 patients (86.4%) were treated with one FD and three patients (13.6%) with two partially overlapping FD in the same treatment. Fourteen of the 22 patients (63.6%) were treated with a Pipeline FD (Covidien/ev3; Plymouth, Minnesota, USA), four (18.2%) with a Silk FD (Balt Extrusion; Montmorency, France), and four (18.2%) with a Surpass FD (Stryker Neurovascular; Fremont, California, USA). The time interval between aneurysm treatment and follow-up examinations was 5–13 months (mean 9±3 months; median 11 months). Twelve of the 22 patients (54.6%) had imaging follow-up at 1 year after FD treatment, one (4.5%) at 8 months, eight (36.4%) at 6 months, and one (4.5%) at 5 months.
Consensus was found between the two radiologists in five of the 22 cases (22.7%) for aneurysm occlusion and in four cases (18.2%) for parent artery patency.
When evaluating aneurysm occlusion using the Montreal scale, the κ value was 0.62 for DSA, 0.35 for 3D-TOF-MRA, and 0.65 for CE-MRA. Using the simplified two-grade scale, the κ value was 0.88 for DSA, 0.51 for 3D-TOF-MRA, and 0.74 for CE-MRA. Evaluating the patency status of the parent vessel, κ was 0.62 for DSA, 0.13 for 3D-TOF-MRA, and 0.38 for CE-MRA. Using the simplified two-grade scale for parent artery patency, κ was 0.61 for DSA, 0.02 for 3D-TOF-MRA, and 0.55 for CE-MRA.
The results of aneurysm occlusion with the Montreal scale and simplified two-grade scale for DSA, 3D-TOF-MRA, and CE-MRA are shown in table 2.
3D-TOF-MRA and DSA were discordant for six of the 23 aneurysms: three neck remnants and two aneurysm remnants on DSA were classified as occluded with 3D-TOF-MRA; and one aneurysm remnant on DSA was classified as neck remnant with 3D-TOF-MRA. CE-MRA and DSA were discordant for four of the 23 aneurysms: three neck remnants and one aneurysm remnant on DSA were classified as occluded with CE-MRA. An example of disagreement for the degree of aneurysm occlusion between 3D-TOF-MRA and CE-MRA is shown in figure 1.
Patency of the parent vessel
The results of the status of the parent vessel artery with DSA, 3D-TOF-MRA, and CE-MRA are shown in table 3.
Using 3D-TOF-MRA, all patent parent arteries were patent on DSA. All cases of stenosis (9/23, 39.1%) were classified as normal with DSA, (ie, 9/9 (100%) false stenosis). Occlusion was seen in 8/23 aneurysms (34.8%), but the vessel was effectively occluded in one aneurysm on DSA. There were 7/8 (87.5%) false occlusions, which were in fact vessel stenosis in two cases and normal vessels in five cases on DSA. Using CE-MRA, all patent parent arteries were patent on DSA. All cases of stenosis (3/23, 13.0%) were classified as normal arteries on DSA (ie, 3/3 (100%) false stenosis). An example of stenosis as determined by 3D-TOF-MRA and CE-MRA but which did not exist on DSA (false positive on MR sequences) is shown in figure 2.
When evaluating aneurysm occlusion using the Montreal scale, inter-technique agreement values were 0.48 and 0.69 for 3D-TOF-MRA/DSA and CE-MRA/DSA, respectively. Using the simplified two-grade scale (adequate occlusion/aneurysm remnant), inter-technique agreement values were 0.59 and 0.88 for 3D-TOF-MRA/DSA and CE-MRA/DSA, respectively. Evaluating the status of the parent vessel, inter-technique agreement was 0.04 for 3D-TOF-MRA/DSA and 0.19 for CE-MRA/DSA. Using the simplified two-grade scale for the patency of the parent artery, inter-technique agreement was 0.11 and 0.32 for 3D-TOF-MRA/DSA and CE-MRA/DSA, respectively.
Diagnostic accuracies for aneurysm remnant
Using the simplified Montreal scale, CE-MRA yielded better sensitivity (0.83, 95% CI 0.54 to 1.00) than 3D-TOF-MRA (0.50, 95% CI 1.10 to 0.90). The two sequences had an excellent specificity and PPV (1, 95% CI 1.00 to 1.00). CE-MRA also had a better NPV (0.94, 95% CI 0.83 to 1.00) than 3D-TOF-MRA (0.84, 95% CI 0.69 to 1.00).
Diagnostic accuracies for parent vessel patency
CE-MRA had a poor specificity and PPV (0.63, 95% CI 0.41 to 0.85 and 0.30, 95% CI 0.02 to 0.58, respectively), but better than 3D-TOF-MRA (0.32, 95% CI 0.11 to 0.52 and 0.19, 95% CI 0.00 to 0.38, respectively). The two sequences had an excellent sensitivity and NPV (1, 95% CI 1.00 to 1.00). CE-MRA also had a better NPV (0.94, 95% CI 0.83 to 1.00) than 3D-TOF-MRA (0.84, 95% CI 0.69 to 1.00).
In our series of patients with aneurysms treated with FD, CE-MRA was superior to 3D-TOF-MRA for the evaluation of aneurysm occlusion and parent vessel patency. Aneurysm occlusion after FD treatment is a progressive phenomenon resulting from flow modifications inside the aneurysm that lead to intra-aneurysmal thrombosis modulated by antiplatelet treatment. As a result, aneurysm occlusion after FD treatment can take days or weeks to complete. In addition, the presence of metal inside the parent artery can potentially induce intra-stent stenosis or thrombosis. For these reasons, imaging follow-up is mandatory after FD treatment to evaluate aneurysm occlusion and depict potential intra-stent stenosis or thrombosis.
DSA is an appropriate imaging modality for the follow-up of aneurysms treated with FD, capable of clearly detecting any residual intra-aneurysmal flow and the patency of the parent vessel. However, DSA is an invasive and irradiating technique associated with a risk of thromboembolic complications as well as complications related to contrast media injection, raising the need for evaluation of non-invasive imaging techniques (MRA or CTA) for aneurysm follow-up after FD treatment. The first step was to evaluate MRA as several studies have demonstrated the diagnostic accuracy of MRA for the evaluation of coiled intracranial aneurysms.11–15 This technique is currently used in clinical practice but no study has yet precisely evaluated MRA for the follow-up of aneurysms treated with FD.
In the follow-up of FD-treated aneurysms, aneurysm occlusion and parent artery status must be assessed. The Montreal scale was used for the evaluation of aneurysm occlusion in our series. Two other scales, the KAMRAN scale and the SMART scale, were recently described for the follow-up of aneurysms treated with FD.20 ,21 The KAMRAN scale has five grades and the distinction between grades is relatively subjective, making it difficult to apply to MRA data. The SMART scale has dynamic items and cannot be applied to our MRA data. Moreover, evaluation of aneurysm occlusion with the KAMRAN or SMART scales precludes comparison with previous series of aneurysms treated with coils or other devices in which the Montreal scale was generally used.
As previously reported,22 when applying the Montreal scale, our interobserver agreement was relatively low regardless of the imaging technique used. Interobserver agreement improved with the two-grade scale derived from the Montreal scale. It is noteworthy that interobserver and intermodality agreement was always better with CE-MRA than with 3D-TOF-MRA for evaluation of both aneurysm occlusion and patency of the parent vessel.
For the depiction of remnant aneurysm using the simplified Montreal scale, both CE-MRA and 3D-TOF-MRA have excellent specificity and PPV (1.00 in all cases), meaning that the remnant aneurysms detected by both techniques were always confirmed by DSA. However, CE-MRA had a better sensitivity than 3D-TOF-MRA (0.83 and 0.50, respectively), meaning that CE-MRA has a better accuracy for detecting aneurysm remnant. This finding is in contrast to that reported by Pierot et al for coiled aneurysms where CE-MRA and 3D-TOF-MRA had relatively similar sensitivity (0.80 and 0.74, respectively).15 For aneurysms treated with FD, CE-MRA is clearly a better sequence for the evaluation of aneurysm occlusion.
CE-MRA is also better than 3D-TOF-MRA for the evaluation of parent artery patency, having a higher specificity (0.63 and 0.32, respectively). Susceptibility artifacts are probably increased for 3D-TOF-MRA compared with CE-MRA, as already shown for nitinol stents.23 ,24 Notably, false stenosis was less frequently observed with CE-MRA than with 3D-TOF-MRA (13.0% and 39.1%, respectively) whereas false occlusion and stenosis falsely interpreted as occlusion were observed in a similar percentage of aneurysms with CE-MRA and 3D-TOF-MRA. Parent artery status is difficult to evaluate with MRA because FDs create imaging artifacts that blur visualization of the parent artery singularly inside the FD, resulting in poor PPV. However, CE-MRA and 3D-TOF-MRA had an excellent NPV (100%) for vessel patency evaluation, indicating that all normal parent vessels on these sequences are effectively patent on DSA. This suggests that CE-MRA, besides having good accuracy to depict aneurysm remnant, can be sufficient for follow-up when showing a patent parent vessel. Otherwise, DSA remains mandatory for the evaluation of vessel patency.
In our study parent artery patency was evaluated using both MRA native images and reconstructions. For both readers, native images were more useful in the evaluation of parent artery patency, suggesting that native images are less sensitive to the artifacts apparent in MRA sequences. MIP reconstructions showed increased signal loss due to the metallic artifact, excluding a precise analysis of the vessel lumen. As both MRA sequences evaluated in the present series do not permit a correct analysis of the parent artery status, technical improvement of the current MRA sequences or use of different MRA sequences should be pursued. Choi et al25 compared 4D-MRA and 3D-TOF-MRA in a group of 26 patients with aneurysms treated with stents and found a higher quality of visualization of stented arteries with 4D-MRA. This sequence should therefore be evaluated in patients treated with FDs.
Our study has some limitations. First, a limited number of patients were included. However, this is the largest series of patients used to address the question of interest. To our knowledge, only one other small series has previously addressed this topic; it used CTA as the gold standard, which has not been validated as a gold standard for the follow-up of aneurysms treated by an endovascular approach.16 Second, CTA was not evaluated in our series. It was necessary to compare the two different MRA sequences before the next potential step which will be direct comparison of MRA (CE-MRA) and CTA. As the optimal non-invasive technique for the follow-up of aneurysms treated with FD is not known, it is probably too early to propose precise modalities of follow-up. A third limitation of the current study is that the studied population was inhomogeneous as some patients were treated with FD and coils and various FDs were used. Finally, despite the use of three different FDs, we were unable to determine if the results of MRA sequences differed between devices because of the small patient numbers in each group.
At 3 T, CE-MRA is superior to 3D-TOF-MRA for the evaluation of aneurysm occlusion and the analysis of parent artery patency after FD treatment. Intraluminal evaluation remains difficult with MRA, regardless of the sequence used. Further work using optimized or different MRA sequences (eg, 4D-MRA) and including a larger number of patients will be necessary to determine the utility of MRA in the follow up of FD-treated aneurysms.
Contributors All the authors drafted the manuscript, gave their final approval for publication and agreed to be accountable for all aspects of the work. JA and LP contributed to the conception of the work and to the acquisition, analysis and interpretation of data. AB, SS, KK and CP contributed to the design of the work and to the acquisition of data.
Competing interests None.
Ethics approval Approval of the institutional review board of Reims Hospital was obtained for this study. The study conformed to the STARD guidelines.
Patient consent Informed consent was obtained for all patients.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional unpublished data from the study are available to researchers by contacting the corresponding author.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.