Abstract
Background
Recurrent intracranial aneurysm carries a risk of rupture and retreatment is often necessary. However, there is no consensus on the best retreatment modality of choice. Flow diverter has emerged as a promising option for this population in recent years. Given its high cost, patient selection to optimize outcomes is very important.
Purpose
To identify patient factors predisposing to failure of flow diverter retreatment.
Method
We conducted a systematic search on PubMed, Cochrane Library, Embase, Ovid/Medline, and ClinicalTrial.gov from 2000 to 2021. Studies regarding flow diverter retreatment of recurrent aneurysms were analyzed if they meet the inclusion criteria.
Results
A total of twenty-six studies were identified. Among 374 patients retreated with flow diverters, about 0.86 [0.81; 0.92] were successfully occluded and only 0.06 [0.02; 0.10] had unfavorable neurological outcomes. Major complications included intracranial hemorrhage (n = 7), ischemic stroke or thromboembolic event (n = 12), and death (n = 2). In-stent stenosis was reported in 10 of the cases. Saccular aneurysms are associated with a higher occlusion rate while aneurysm location, size, status, and prior treatment modality have no significant impact on retreatment efficacy.
Conclusions
We demonstrated that flow diverter is an effective retreatment strategy except in patients with non-saccular aneurysms. It should be considered as a first-line option for patients with recurrent intracranial aneurysm.
Introduction
Microsurgical clipping and endovascular coil embolization are the two main treatment options for intracranial aneurysms in the past decades. As novel endovascular technology develops, flow diverter (FD) has emerged as a promising alternative. FD works by promoting flow stasis and is particularly useful to treat traditionally difficult-to-treat aneurysm types such as large, broad-necked aneurysm arising from the petrosal carotid artery to the supraclinoid internal carotid artery (ICA). 1 The recurrence rates of intracranial aneurysms treated with standard coils or stent-assisted coiling (SAC) ranges from 13% to 48%, and around 4.5% of patients will require retreatment after microsurgical clipping.2–4 Re-surgery and endovascular reconstruction or deconstruction are options for retreatment. However, it is often technically difficult due to post-operative fibrosis and aberrant aneurysm anatomy. 5 Thus, endovascular FD placement has gained popularity as a salvage therapy for these patients in recent literature.1,6–12 Preliminary studies have shown that FD is a safe and effective option for recurrent aneurysms, but the scarcity of quality randomized control trials prevents risk profiling for patient selection. 11 As a result, how to select patients for either modality remains controversial. To our knowledge, this is the first meta-analysis to systematically evaluate the feasibility of FD in recurrent intracranial aneurysms and to discuss potential factors for suboptimal results.
Methods
Protocol
This study was conducted under the guidance of PRISMA guidelines. 13
Data sources
Two authors systematically searched five electronic databases, including PubMed, Cochrane Library, Embase, Ovid/Medline, and ClinicalTrial.gov., with the search terms [recurrent, reoperation, retreatment], [diverter, diversion, flow diverter, PED, silk, disruption] and [aneurysm]. Relevant studies, excluding non-English ones and those published before 2000/01/01, were collected and duplicates were removed by Zotero (version 5.0.97). Additional studies were identified from references.
Study selection
All the studies were reviewed independently by the two authors with predetermined inclusion and exclusion criteria (Table S1). FD retreatment is defined as patients who received FDs for the index aneurysm at least 6 months after initial treatment. Conflicts of opinions were resolved by consensus.
Data extraction
The two authors independently extracted data from the selected literature. The primary endpoint was successful occlusion of the index aneurysm, which was defined as modified Raymond Roy scale (mRR) I and II. A C or D on the O’Kelly-Marotta grading scale (OKM) or 90–100% occlusion was considered successful. On the contrary, patients with mRR III, OKM A or B, or <90% occlusion, were labeled as retreatment failure. The secondary endpoint was short-term (≤6 months) or long-term (>6 months) neurological outcome defined by modified Rankin scale (mRS) ≤2 (good) or >2 (poor). We also extracted additional information including author name, publication year, age, gender, and aneurysm location, size, and anti-platelet regimens. Any inconsistency was resolved by consensus.
Missing data
Missing data was handled independently by two authors based on the following approach: When patient information was not listed explicitly, we first conducted a complete case analysis by cross-referencing text and images to make an evidence-based inference. When this was not satisfactory, we then reached out to the corresponding author. If no reply was received, we assumed none of the participants with missing data had the event of interest For subgroup analysis, if the original paper used a different categorizing system in its outcome report, we would use percentage or split up (if proportion of each subgroup was not provided) the number of events and cases to each subgroup for analysis.
Statistical analysis
Random effect model was used in this meta-analysis. Heterogeneity within studies was evaluated by Cochrane's Q test and I2. We also conducted subgroup analysis stratified by previous treatment type, aneurysm size, and whether the aneurysm ruptured at presentation. All statistical analysis was completed by R language (version 4.1.2) and R studio (version 2021.09.1 + 372) using the meta, metafor, and forestplot packages.14,15 P-value less than 0.05 was set as statistical significance.
Bias evaluation and quality assessment
For randomized control trials and observational studies, the Risk of Bias tool and the Newcastle-Ottawa Scale (NOS) were used to evaluate the quality of the literature, respectively. Publication bias was analyzed with funnel plots and Egger's tests.
Results
Literature search
A total of 1195 records were identified from the previously mentioned databases and 28 additional studies were identified in the referenced articles. After removal of the duplicates, 406 records underwent title/abstract screening, of which 96 proceeded to full-text screening with predetermined inclusion and exclusion criteria. 66 studies were excluded for reasons specified in the PRISMA flowchart. 30 studies were deemed eligible for qualitative synthesis and among them, 26 (21 retrospective studies and 5 prospective studies) were used for quantitative synthesis (Figure S1).
Characteristics of the included studies
A series of 374 patients from 44 institutes are included in the meta-analysis (2012–2021). The mean age was 52.6 years and 67% were women. Most studies gave dual anti-platelet (clopidogrel 75–300 mg/day, aspirin 81–320 mg/day) for 1–7 days before the treatment in unruptured aneurysms. For ruptured aneurysms, a loading dose of dual anti-platelet (clopidogrel 300–600 mg, aspirin 300–650 mg) was given. After the procedure, patients received dual anti-platelet therapy for 3–6 months followed by life-long aspirin monothrerapy in most cases. An average of 1.14 FDs were used for each aneurysm. Other characteristics of the included studies are summarized in online supplemental data (Table S2).
The proportion (PR) of successful FD retreatment following aneurysm recurrence was 0.89 [0.82–0.94] (Figure S2). DSA follow-ups were available at 3, 6, and 12 months after retreatment in 45, 140, and 190 cases, with a complete or near-complete occlusion rate of 57.8%, 90.7%, and 80.0%, respectively. Regarding composite complications, there were 7 intracranial hemorrhages, 12 ischemic stroke or thromboembolic events, 1 focal carotid artery dissection, 1 carotid-cavernous fistula with oculomotor nerve palsy, and 2 cases of death. Ischemic stroke due to acute in-stent thrombosis occurred at 0–15% despite anti-platelet prophylaxis. The incidence rate of long-term in-stent stenosis ranged from 0–57%, although most of them are mild-to-moderate with no neurological deficit. There is mild heterogeneity among the included studies (I2 = 35%, τ2 = 0.9051, p = 0.04).
Subgroup analysis
Aneurysm morphology
We analyzed whether different aneurysm morphology predicts retreatment failure (Figure S3). When a dichotomy classification is used, there is a significantly higher retreatment failure in morphologies other than ‘saccular’ (PR = 0.92 [0.85; 0.96] vs. 0.77 [0.64; 0.87], p = 0.01). When we subdivided ‘other’ into ‘fusiform’ and ‘other’ (including blister and dissecting types), the overall subgroup difference remains (p = 0.05); while direct comparison between the latter two showed no intergroup heterogeneity (p = 0.92), suggesting a true superiority of FD retreatment for saccular aneurysms comparing to other morphologies.
Aneurysm location
Depending on aneurysm location, the dynamic of blood flow and technical accessibility are known factors to consider during aneurysm treatment (Figure S4). 16 Therefore, we ran a subgroup analysis based on four major aneurysm locations: anterior cerebral artery (ACA), middle cerebral artery (MCA), ICA, and posterior circulation. FD retreatment performs similarly in these four subgroups, with the success rate ranging from 0.83–0.86 with no subgroup difference.
Prior treatment modality
We analyzed the success rate of FD retreatment following various treatment modalities (Figure S5). The success rate was consistently around 90% following clipping (PR = 0.87 [0.73; 0.95]), coiling (PR = 0.89 [0.75; 0.96]), or coiling + stenting (PR = 0.89 [0.65; 0.97]). Prior treatment with FD results in a 100% success rate (1.00 [0.00; 1.00]) while those undertaking stenting alone had a success rate of 62% (0.60 [0.30; 0.84]). However, the difference among subgroups is non-significant (I2 = 0%, p = 0.27).
Aneurysm status at recurrence
The PR of successful retreatment for ruptured or unruptured aneurysms are 0.94 [0.75; 0.99] and 0.88 [0.80; 0.93], respectively, with no significant subgroup difference (p = 0.40) (Figure S6). For ruptured aneurysms, the PR of short-term poor neurological outcome is 0.08 [0.04; 0.16] while it is 0.03 [0.01; 0.10] in long-term follow-up. On the other hand, they are both 0.05 [0.01; 0.18] for unruptured aneurysms. There is no statistical difference between ruptured and unruptured aneurysms in terms of short-term (p = 0.59) and long-term (p = 0.64) neurological outcomes. The results suggest that short-term neurological sequelae, either due to aneurysm rupture or intraoperative complications, generally recovers overtime.
Aneurysm size
We compared the PR of successful FD retreatment between aneurysms of various sizes, with the cutoff set at small (<10 mm), large (≥10 mm and <25 mm), and giant (≥25 mm) (Figure S7). Interestingly, large aneurysms have a non-significantly higher success rate (0.92 [0.74; 0.98]) compared to the other two subgroups (small: 0.88 [0.73; 0.95]; giant: 0.70 [0.30; 0.93]), supporting the widely accepted indication of FD for large aneurysms. Again, these finds indicate that FD is a feasible retreatment option regardless of aneurysm size.
Quality assessment
Because we only include observational studies in this meta-analysis, we use only NOS for quality assessment. The NOS table in online supplemental data (Table S3) showed the risk of bias for these studies. As we can see, most of the studies have a score of more than eight.
Publication bias assessment
We evaluate the publication bias by Egger's test and visualize the result with a funnel plot (Figure S8). There is a gross asymmetry in study distribution, which is significant with a p-value of 0.0002. Close inspection of the funnel plot revealed skewing to the right at the lower part, while most studies at the upper part are leaning towards the middle or even on the left. This is possibly due to more careful patient selection in smaller studies, resulting in ascertainment bias.
Discussion
The best way to treat recurrent or residual intracranial aneurysms is not well established. Re-surgery carries the highest of scarring, anatomical distortion due to adhesions as well as morbidity and mortality rate among all others.2,17 Thus, endovascular treatment is preferred in most situations. However, even within endovascular treatment, each modality varies in its indications, contraindications, and risk profile, leaving optimal patient selection at issue. Despite its established role in primary wide-neck, fusiform, large aneurysms, 18 research on FD use in previously treated intracranial aneurysms is pending.
Our study demonstrated the success rate of FD retreatment is around 90%, regardless of prior treatment modalities. Interestingly, those receiving standing stent alone have a discernible but non-significant lower success rate of 62%. In the study by Chalouhi et al., 6 telescoping stents were used for the initial treatment in two stent-alone cases, resulting in one incomplete occlusion after FD retreatment. In our experience, the more overlapping stent in the initial treatment, the higher likelihood of insufficient vessel wall apposition with FD retreatment, which leads to endoleak and treatment failure. 19 Besides, stent overlapping also creates many technical problems to following interventions including deployment failure owing to stent catching and navigation failure of the microcatheter. Direct FD deployment instead of drag-and-drop technique has been proposed for the retreatment of previously stented aneurysms but the outcomes were still inferior to FD retreatment after other endovascular approaches.20,21 Nevertheless, our study demonstrated that FD for stented aneurysms is still a feasible option given proper patient selection and a single FD placement (mean 1.14) is sufficient in most cases.
The retrospective cohort by Mascitelli et al. 22 concluded smaller aneurysm size (<10 mm) is a favorable predictor for endovascular retreatment of intracranial aneurysms. However, most of the cases in the study received SAC or stand-alone stenting for retreatment, with only two FD cases. Another research team from Canada proved the feasibility of FD after primary microsurgical clipping in a very selected group of patients, similar to that for previously coiled aneurysms. Specifically, they recommended using FD for patients with small-sized aneurysm recurrences. 23 We demonstrated that FD is an effective retreatment choice across a wide range of aneurysm size, with no significant difference in success rate and neurological outcomes among small-, large-, and giant-sized aneurysms. Nevertheless, we suggest FD retreatment only for recurrent intracranial aneurysms less than 25 mm in size due to potentially worse results. This hypothesis was supported by Lubicz et al., 24 who reported a high delayed complication rate with Silk FD alone in the treatment of complex intracranial aneurysms.
Although there is no statistical difference between ruptured and unruptured recurrent aneurysms in successful occlusion or follow-up neurological outcomes in our study, it should be noted that pre- and post-procedural anti-platelet medications are necessary when FD is planned. A fatal intracranial hemorrhage because anti-platelet therapy was not suspended. 22 Indeed, all the cases with hemorrhagic complications identified in this meta-analysis happened in ruptured aneurysms.6,25–27
Details on FD retreatment failure were surprisingly scarce. Park et al. 11 reported a case of initially ruptured dissecting MCA aneurysm. The patient underwent FD placement but the residual sac was found to have enlarged in the subsequent follow-up. An additional FD was given but a follow-up angiography was not available at the time of this meta-analysis. Heiferman et al. 28 reported one FD failure case with a giant right cavernous fusiform aneurysm previously treated with two overlapping stents and coils. They stated that the FD could not achieve complete apposition against the vessel wall, resulting in endoleak. A recent multicenter study suggested fusiform morphology as the strongest predictor of retreatment failure. 29 This is in consistency with our finding that morphologies other than saccular, including fusiform, blister, and dissecting aneurysms, bears a lower successful rate. Besides, we should keep in mind that deformation of the mesh structure in FD deployed in arteries with discrepant diameters or tortuous arteries does add risk in retreatment.
The literature review showed that FD retreatment is generally feasible for aneurysms arising from the anterior communication region and distal ACA;25,26,30 consistent with our result that there is no statistical difference between subgroups by location. While those arising from basilar apex should be addressed more selectively in one study but the case number is limited. 27
Of note, there were 28 cases of coil + FD combined treatment in these 374 patients. Given the small percentage and unclear indication for such a combination, we did not exclude these cases from the meta-analysis. We hypothesize that FD still takes the dominant role to induce endothelialization and aneurysm thrombosis with a minor contribution from coils under this circumstance.
Pipeline Embolization Device had been the only FD available until 2018, when Surpass Streamline Flow Diverter was added to the ammunition in July 2018. The success rate for FD retreatment was more heterogeneous in the years prior to 2019. Therefore, despite a shared mechanism of action, the two devices might have some intrinsic differences and each has a unique niche in clinic practice. Although we could not analyze the success rate for each type of FD, one in vitro study showed a significantly superior flow diversion effect of the Surpass device compared with the Pipeline device (p = 0.03), 31 a possible explanation for the three studies conducted in 2015, 2014, and 2017 that have the lowest success rates (0.46, 0.50, 0.50).
There are some limitations to this study. First, we could not calculate the absolute change in residual sac size and mRS before and after FD retreatment. Post-operative DSA and mRS are reasonable surrogates but may fail to represent the underlying patient condition, which leans toward a more conservative result. Second, some of the included studies reported a spontaneous reduction in residual sac size from the first angiography follow-up. Whether improved aneurysm occlusion in these cases was due to de-escalation of anti-platelet medication or FD retreatment cannot be confidently distinguished.7,28 Third, there is considerable heterogeneity among the studies included, variations in institutional protocols, and physician familiarity with FD placement. Fourth, the retrospective nature of most referenced studies could not infer causality. Lastly, as a single-arm analysis without direct comparison to other modalities, we could not demonstrate superiority or non-inferiority of FD to other modalities.
Conclusion
FD is effective in treating recurrent intracranial aneurysms. Prior treatment modality, aneurysm size, and status of the aneurysm at recurrence do not affect its success rate and neurological outcomes. To our knowledge, this is the first meta-analysis attempting to stratify patient outcomes for retreatment of aneurysms with FD. Randomized control trials are needed to compare FD to other modalities in patients with recurrent intracranial aneurysms and elucidate possible unfavorable patient factors to treatment success in order to optimize the cost-effectiveness of FD retreatment.
Supplemental Material
sj-docx-1-ine-10.1177_15910199221095972 - Supplemental material for Flow diverter retreatment for intracranial aneurysms: A meta-analysis of efficacy and feasibility
Supplemental material, sj-docx-1-ine-10.1177_15910199221095972 for Flow diverter retreatment for intracranial aneurysms: A meta-analysis of efficacy and feasibility by Irene (Tai-Lin) Lee, Yung-Shuo Kao, Yen-Jun Lai and Ho-Hsian Yen in Interventional Neuroradiology
Footnotes
Abbreviations
Author contribution
Irene (T.L.) Lee and H.H. Yen provided the idea, conducted the study, analyzed and interpreted, wrote and revised the manuscript; Y.S. Kao helped with data analysis and interpretation. Y.J. Lai revised and supervised the study.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
