J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic

Original research Rebleeding and bleeding in the year following intracranial coiling: analysis of a large prospective multicenter cohort of 1140 patients— Analysis of Recanalization after Endovascular Treatment of (ARETA) Study Laurent Pierot ‍ ‍ ,1 Coralie Barbe,2 Denis Herbreteau,3 Jean-­Yves Gauvrit,4 Anne-­Christine Januel,5 Fouzi Bala,6 Frédéric Ricolfi,7 Hubert Desal,8 Stéphane Velasco,9 Mohamed Aggour,10 Emmanuel Chabert,11 Jacques Sedat,12 Denis Trystram,13 Gaultier Marnat ‍ ‍ ,14 Sophie Gallas,15 Georges Rodesch,16 Frédéric Clarençon,17 Chrysanthi Papagiannaki ‍ ‍ ,18 Phil White,19,20 Laurent Spelle21

For numbered affiliations see Abstract Introduction end of article. Background Endovascular treatment is the first Since the initial publication of the International line therapy for the management of ruptured and Subarachnoid Aneurysm Trial (ISAT), endovascular Correspondence to treatment (EVT) in the management of intracranial Dr Laurent Pierot, Radiology, unruptured intracranial , but delayed CHU Reims, Reims, Champagne-­ aneurysm rupture leading to bleeding/rebleeding can aneurysms (IA) has progressively become the first-­ 1–3 Ardenne, France; lpierot@​ ​gmail.​ occur subsequently. ARETA (Analysis of Recanalization line treatment. Several limitations of aneurysm com after Endovascular Treatment of intracranial Aneurysm) coiling were rapidly identified by interventional neuroradiologists, such as the difficulty of treating Received 5 March 2020 is a prospective, multicenter study conducted to Revised 12 April 2020 analyze aneurysm recanalization. We analyzed delayed wide-­neck aneurysms and the relatively high risk of Accepted 18 April 2020 bleeding and rebleeding in this large cohort. aneurysm recurrence, leading to the development Methods 16 neurointerventional departments of new endovascular techniques and devices, for example, balloon-­assisted coiling, stent-assisted­ prospectively enrolled patients treated for ruptured 4 5 and unruptured aneurysms between December 2013 coiling, flow diversion and flow disruption. and May 2015 (​ClinicalTrials.​gov: NCT01942512). The primary and most important role of aneu- Participant demographics, aneurysm characteristics rysm treatment is to prevent aneurysm rebleeding in cases of ruptured intracranial aneurysms (RIA)

and endovascular techniques were recorded. Data http://jnis.bmj.com/ and bleeding in cases of unruptured intracranial were analyzed from participants with ruptured or aneurysms (UIA). Based on initial analyses of the unruptured aneurysms treated by coiling or balloon-­ ISAT population, the number of patients suffering assisted coiling. Rates of bleeding and rebleeding were from rebleeding in the first year after treatment was analyzed and associated factors were studied using slightly higher in the endovascular group (26/801, univariable and multivariable analyses. 3.2%) compared with the surgical group (10/793, Results The bleeding rate was 0.0% in patients 1.3%).1 The risk of rebleeding after 1 year was also with unruptured aneurysms and 1.0% (95% CI 0.3% analyzed in subsequent ISAT publications, showing on September 27, 2021 by guest. Protected copyright. to 1.7%) in patients with ruptured aneurysms. In a progressive decrease over time.6 In contrast to multivariate analysis, two factors were associated with ruptured aneurysms, relatively little is known rebleeding occurrence: incomplete aneurysm occlusion regarding the protection afforded by aneurysm after initial treatment (2.0% in incomplete aneurysm coiling in preventing bleeding in unruptured aneu- occlusion vs 0.2% in complete aneurysm occlusion, OR rysms. This phenomenon was not analyzed in the 10.2, 95% CI 1.2 to 83.3; p=0.03) and dome-­to-neck­ Analysis of Treatment by Endovascular Approach of ratio (1.5±0.5 with rebleeding vs 2.2±0.9 without Non Ruptured Aneurysms (ATENA) study, where © Author(s) (or their rebleeding, OR 0.2, 95% CI 0.04 to 0.8; p=0.03). 1 year and long-­term follow-­up was not available, employer(s)) 2020. No Modalities of management of aneurysm rebleeding as and in the meta-analysis­ conducted by Naggara et commercial re-­use. See rights well as clinical outcomes are described. al.2 7 and permissions. Published by BMJ. Conclusions Aneurysm coiling affords good The Analysis of Recanalization after Endovas- protection against bleeding (for unruptured cular Treatment of intracranial Aneurysm (ARETA) To cite: Pierot L, Barbe C, aneurysms) and rebleeding (for ruptured aneurysms) study is a French, prospective, multicenter study, Herbreteau D, et al. which aims to determine the factors that affect J NeuroIntervent Surg Epub at 1 year with rates of 0.0% and 1.0%, respectively. ahead of print: [please Aneurysm occlusion and dome-­to-neck­ ratio are the aneurysm recanalization after endovascular treat- 8 include Day Month Year]. two factors that appear to play a role in the occurrence ment. Several baseline patient and aneurysm char- doi:10.1136/ of rebleeding. acteristics were collected as well as complications neurintsurg-2020-015971 that occurred during and after treatment .9 10 Given

Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 1 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke the large sample (n=1289), this study offers the opportunity to ►► Determine the rate of aneurysm rebleeding in the year conduct a detailed analysis of intra- and postoperative events following coiling among patients with ruptured aneurysms occurring in relation to endovascular treatment of intracranial ►► Determine the factors that are associated with the risk of aneurysms (ruptured or unruptured). Findings from analyses bleeding/rebleeding of intraoperative complications in the ARETA study occurring ►► Analyze the management used and clinical outcome in cases during aneurysm treatment have already been published.10 of bleeding/rebleeding. However, several complications occur after endovascular treat- ment of intracranial aneurysms, including: delayed aneurysm Data management rupture inducing rebleeding in ruptured aneurysms, bleeding Participating centers reported participant, aneurysm and treat- in unruptured aneurysms, thromboembolic events, and delayed ment characteristics, and complications on a standardized remote hematomas.4 While delayed bleeding of UIA is rare form. Clinical research assistants controlled the accuracy of the during the year following aneurysm treatment, rebleeding of collected data. A core laboratory centrally adjudicated the aneu- RIA can occur within 3 days (acute bleeding), 30 days (early rysm characteristics, treatment modalities and complications of rebleeding), or later (late rebleeding) following aneurysm treat- all the participants. ment.11 The goal of the present analysis is to determine the rate Rebleeding or bleeding was suspected when sudden changes of postoperative bleeding or rebleeding after coiling (including in clinical status occurred and were confirmed by the detection balloon-­assisted coiling) of unruptured and ruptured aneurysms, of fresh blood on CT in the vicinity of the aneurysm initially the clinical outcome of these events, and factors associated with treated. Clinical outcome of bleeding/rebleeding was evaluated these complications. by modified Rankin Scale (mRS) at hospital discharge. Poor clin- ical outcome was defined by mRS 3–6 at hospital discharge. Centers also collected preoperative digital subtraction angiog- Methods raphy (DSA) and immediate postoperative DSA, and transferred The ARETA study anonymized results to Reims Hospital. Aneurysm characteristics, ARETA was primarily designed to analyze factors that affect treatment modalities and complications of all participants were aneurysm recanalization after endovascular treatment. The reviewed, checked for accuracy and, if necessary, revised by the French Ministry of Health in a PHRC (Programme Hospi- core laboratory. talier de Recherche Clinique, No. 12-001-0372) provided funding. ARETA was registered on www.clinicaltrials.​ gov​ Study participants (NCT01942512), and received national regulatory authoriza- The participants were prospectively enrolled in 16 centers in tion from the Consultative Committee of Information Processing France between December 2013 and May 2015. Within the in Healthcare Research Program and the National Commission overall ARETA cohort, participants with at least one IA treated for Data Processing and Freedom. The study objective and its with a technique other than coiling or BAC (ie, stent-assisted­ protocol, including inclusion and exclusion criteria, have previ- coiling, flow diversion, flow disruption) were excluded from 8 ously been described. According to French law, the study did analysis. Participating study sites reported the following baseline not require approval by institutional review board, nor written participant characteristics: age, sex, current use of cigarettes, informed consent. regular consumption of alcohol, diabetes mellitus (glycemia Inclusion criteria were: age >18 years, saccular IA, ruptured >6 mmol/L), elevated blood pressure (defined as blood pressure or unruptured IA, and IA treated by any endovascular technique >140/90 mmHg uncorrected by medical treatment), polycystic http://jnis.bmj.com/ (coiling, balloon-assisted­ coiling (BAC), stent-assisted­ coiling kidney disease, and familial history of aneurysm. Family history (SAC), flow diversion, flow disruption). Exclusion criteria of aneurysm was defined as the presence of two or more family included dissecting or fusiform IA, IA associated with a brain members among first- and second-­degree relatives with proven arteriovenous malformation, and IA already treated by clips or aneurysmal or incidental aneurysms. coils. Clinical status before treatment was evaluated based on World Sample size was calculated based on a potential association Federation of Neurosurgical Societies (WFNS) grade for partici- between tobacco use and recanalization at 12 months. Based on

pants with RIA, and mRS for participants with UIA. on September 27, 2021 by guest. Protected copyright. estimated tobacco use in 40% of participants without recanal- Recorded aneurysm characteristics were: aneurysm sac diam- ization, and 55% with recanalization, a recanalization rate of eter (dichotomized into <7 mm and ≥7 mm); neck size and 25%, with an alpha of 5%, power of 95%, and a two-sided­ test, dome-­to-­neck ratio (wide neck defined as neck size ≥4 mm the number of participants required was 760 (NQuery software and/or dome to neck ratio <2); aneurysm location (extradural 3 version 4.0, Cork, Ireland). Assuming an estimated 40% rate of internal carotid artery (ICA), intradural ICA including the loss to follow-up­ or death at 12 months, a sample size of 1275 posterior communicating artery, middle cerebral artery (MCA), participants was considered necessary. anterior communicating/anterior cerebral artery (ACA/Acom) Four ARETA manuscripts have been previously published. or vertebrobasilar (VB)); aneurysm rupture status (ruptured or 8 One described study background and protocol. The remaining unruptured); aneurysm shape (regular or irregular); and number three studies described the population and modalities of treat- of IA (single or multiple). Aneurysm was classified as regular ment, analyzed patient and aneurysm risk factors associated with when there was a single sac with a smooth margin, and irregular aneurysm rupture, and analyzed intraoperative complications if it was a single sac with an irregular margin or a daughter sac or occurring during aneurysm coiling (including BAC).9 10 12 a multilobulated aneurysm. Treatment modalities including perioperative medications were at the discretion of the treating interventional neuroradiol- Study objectives ogist. Treatment modalities were categorized into coils, BAC, The study objectives were as follows: SAC, flow diversion, intrasaccular flow disruption, and parent ►► Determine the rate of aneurysm bleeding in the year following vessel sacrifice. According to the primary objective of ARETA, coiling among patients with unruptured aneurysms postoperative medications were not collected in the database. 2 Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke

Table 1 Patient and aneurysm characteristics Patients with at least Patients without All patients one ruptured aneurysm ruptured aneurysm Patient characteristics (n=1140) (n=794) (n=346) Sex Female, n (%) 757 (66.4%) 520 (65.5%) 237 (68.5%) Male, n (%) 383 (33.6%) 274 (34.5%) 109 (31.5%) Age (years), mean±SD 53.8±12.8 54.0±13.1 53.3±12.1 Elevated blood pressure* Yes, n (%) 112 (10.2%) 88 (11.6%) 24 (7.0%) No , n (%) 990 (89.8%) 673 (88.4%) 317 (93.0%) Current smoking† Yes, n (%) 511 (45.7%) 366 (47.2%) 145 (42.3%) No , n (%) 607 (54.3%) 409 (52.8%) 198 (57.7%) Regular alcohol consumption‡ Y es, n (%) 235 (21.0%) 164 (21.2%) 71 (20.6%) No, n (%) 882 (79.0%) 609 (78.8%) 273 (79.4%) Diabetes mellitus§ Yes, n (%) 57 (5.0%) 32 (4.1%) 25 (7.3%) No, n (%) 1074 (95.0%) 756 (95.9%) 318 (92.7%) Polycystic kidney disease¶ Yes, n (%) 16 (1.4%) 6 (0.8%) 10 (2.9%) Figure 1 Flow chart of the study population. No, n (%) 1116 (98.6%) 782 (99.2%) 334 (97.1%) Family history** Y es, n (%) 79 (7.1%) 37 (4.8%) 42 (12.3%) Statistical analysis No, n (%) 1033 (92.9%) 733 (95.2%) 300 (87.7%) The Department of Research and Public Health of Reims Aneurysm All aneurysms Ruptured aneurysms Unruptured Hospital was responsible for data management and statistical characteristics (n=1195) (n=817) aneurysms analysis. Data were described using mean±SD for continuous (n=378) variables and number and percentage for categorical variables. Aneurysm size (maximum Patient and aneurysm factors associated with aneurysm bleeding/ diameter) rebleeding in the year following treatment were assessed using Mean±SD (mm) 6.5±3.1 6.3±3.0 6.9±3.2 univariate analysis (Wilcoxon tests or Fisher exact tests, as <7 mm, n (%) 728 (60.9%) 520 (63.6%) 208 (55.0%) http://jnis.bmj.com/ appropriate) and multivariate analysis (logistic regressions with ≥7 mm, n (%) 467 (38.1%) 297 (36.4%) 170 (45.0%) stepwise selection, with an exit threshold of 0.20 and variables Neck size (mm), 3.2±1.5 3.0±1.3 3.6±1.6 significant at p=0.10 included). A p value <0.05 was considered mean±SD statistically significant. All analyses were performed using SAS Dome to neck ratio 2.1±0.9 2.2±0.9 2.0±0.8 version 9.4 (SAS Institute Inc, Cary, NC, USA). Wide neck†† Yes, n (%) 694 (58.1%) 440 (53.9%) 254 (67.2%) Results No, n (%) 501 (41.9%) 377 (46.1%) 124 (32.8%) on September 27, 2021 by guest. Protected copyright. Study population Aneurysm location After applying study inclusion criteria to the overall cohort A CA/Acom, n (%) 466 (39.0%) 366 (44.8%) 100 (26.5%) of 1289 participants (harboring 1761 aneurysms), the study MCA, n (%) 247 (20.7%) 165 (20.2%) 82 (21.7%) population for this analysis consisted of 1140 participants (53.8±12.8 years, 757 women (66.4%)) with 1195 aneurysms ICA intradural, n (%) 376 (31.5%) 230 (28.1%) 146 (38.6%) (figure 1). In this population, 794 patients (69.6%) presented ICA extradural, n (%) 14 (1.2%) 1 (0.1%) 13 (3.4%) with at least one RIA (54.0±13.1 years, 520 women (65.5%)) VB , n (%) 92 (7.7%) 55 (6.7%) 37 (9.8%) and 346 patients (30.4%) presented only UIA (53.3±12.1 years, Aneurysm shape 237 women (68.5%)) (table 1). Among the 794 patients with at Regular, n (%) 325 (27.2%) 164 (20.1%) 161 (42.6%) least one RIA, 753 had a single treated IA and 41 had multiple Irregular, n (%) 870 (72.8%) 653 (79.9%) 217 (57.4%) treated IAs, corresponding to a total of 838 treated IAs (817 RIAs and 21 UIAs). Among the 346 patients with only UIA, 335 Treatment had a single treated IA and 11 had multiple treated IAs, corre- Coiling, n (%) 648 (54.2%) 461 (56.4%) 187 (49.5%) sponding to a total of 357 treated UIAs. Thus, among the 1195 B AC, n (%) 547 (45.8%) 356 (43.6%) 191 (50.5%) IAs, 817 (68.4%) were RIAs treated by coiling (461/817, 56.4%) Occlusion‡‡ or BAC (356/817, 43.6%) during the initial procedure, while Complete , n (%) 687 (57.8%) 465 (57.2%) 222 (59.0%) 378 (31.6%) were UIAs treated by coiling (187/378, 49.5%) or Continued BAC (191/378, 50.5%) during the initial procedure (table 1).

Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 3 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke

RIA (1.00%, 95% CI 0.3% to 1.7%). Among the RIAs, 8/817 Table 1 Continued aneurysms (0.98%) rebled (table 2). Patients with at least Patients without Rebleeding occurred within 30 days (early rebleeding) All patients one ruptured aneurysm ruptured aneurysm Patient characteristics (n=1140) (n=794) (n=346) following the initial procedure in five patients (1, 3, 8, 13 and 26 days, respectively), and after 30 days (late rebleeding) in three Residual neck, n (%) 409 (34.4%) 290 (35.7%) 119 (31.7%) patients (151, 207 and 296 days, respectively). Residual aneurysm, 93 (7.8%) 58 (7.1%) 35 (9.3%) n (%) *38 missing data. †22 missing data. Characteristics of patients with ruptured aneurysms and ‡23 missing data. rebleeding §9 missing data. ¶8 missing data. Six of the eight patients with ruptured aneurysms who expe- **28 missing data. rienced rebleeding were female. Age was between 18 and 40 ††Wide neck defined as neck size ≥4 mm and/or dome to neck ratio <2. ‡‡6 missing data. years in two patients, between 41 and 60 years in four patients, ACA/Acom, anterior communicating/anterior cerebral artery ; BAC, balloon-assisted­ coiling; ICA, internal carotid artery; MCA, middle cerebral artery; VB, vertebrobasilar. and above 61 years in two patients. Two patients were current smokers and one patient regularly consumed alcohol. Aneurysm location was supraclinoid ICA in four patients, ACA/Acom in two Rate of bleeding in patients with unruptured aneurysms patients, MCA in one patient, and VB in one patient. Aneurysm At 12 month follow-­up (mean follow-­up 14.2±5.2 months), we size (maximum diameter) was <7 mm in six patients and ≥7 mm observed no episodes of bleeding (0.0%) in the cohort of 346 in two patients. The aneurysm neck was narrow (<4 mm) in five patients treated only for unruptured aneurysms. patients and wide (≥4 mm) in three patients. The dome-­to-neck­ ratio was between 0.7 and 2.2 (mean 1.5±0.5). All aneurysms Rate and delay of rebleeding in patients with ruptured had an irregular shape. Aneurysm occlusion after the initial aneurysms procedure was complete aneurysm occlusion in one patient, At 12 month follow-­up (mean follow-up­ 12.2±6.3 months), neck remnant in six cases, and aneurysm remnant in one patient. rebleeding was observed in 8/794 patients treated for at least one At the time of rebleeding, one patient (patient 8) rebled from the

Table 2 Characteristics of patients presenting with rebleeding after initial endovascular treatment of a ruptured aneurysm 1 2 3 4 5 6 7 8 Patient characteristics Diabetes mellitus No No No No No No No Missing Elev ated blood pressure No No No No No No No Missing Current smoking No No No Yes No Yes No Missing Regular alcohol consumption No No No Yes No No No Missing P olycystic kidney disease No No No No No No No Missing Family history No No No No No No No Missing

Aneurysm characteristics http://jnis.bmj.com/ Location* MCA VB Acom ID ICA ID ICA ID ICA ID ICA Acom Size (mm) 5.8 7.0 2.5 6.5 16.0 6.0 5.0 3.0 Neck size (mm) 3.8 6.5 3.5 2.9 8.0 4.7 3.2 1.5 Dome to neck ratio 1.5 1.1 0.7 2.2 2.0 1.3 1.6 2.0 Wide neck† Yes Yes Yes No Yes Yes Yes No

Aneurysm shape Irregular Irregular Irregular Irregular Irregular Irregular Irregular Irregular on September 27, 2021 by guest. Protected copyright. Initial procedure Technique‡ BAC BAC BAC BAC C BAC BAC BAC Occlusion§ NR NR NR NR AR NR AR CO Rebleeding Delay (days) 13 296 207 8 151 26 1 3 Retreatment Occlusion§ AR AR AR – – NR AR – Technique¶ FD SAC BAC SAC Clip SAC SAC – Outcome Initial treatment discharge mRS** – 1 1 – 0 – – – Retreatment discharge mRS** 4 1 1 1 6 3 5 6 12 month mRS** 0 0 1 1 6 0 0 6 *MCA, middle cerebral artery; VB, vertebrobasilar; Acom, anterior communicating artery; ID ICA, intradural internal carotid artery. †Wide neck defined as neck size ≥4 mm and/or dome to neck ratio <2. ‡BAC, balloon-­assisted coiling; C, coiling. §Occlusion, aneurysm occlusion; CO, complete occlusion; NR, neck remnant; AR, aneurysm remnant. ¶ FD, flow diversion; SAC, stent-­assisted coiling; BAC, balloon-­assisted coiling; Clip, clipping. **mRS, modified Rankin Scale.

4 Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke coiled Acom aneurysm 3 days post-­procedure and quickly clini- Aneurysm factors associated with the risk of rebleeding in cally deteriorated leading to death before any retreatment. One treated ruptured aneurysms patient (patient 5) was treated by surgery (without DSA before In univariate analysis, aneurysm location (p=0.31) and size treatment) due to a compressive parenchymal hematoma, and (p=0.80) were not significantly associated with rebleeding in the aneurysm was clipped. The patient subsequently died. Six treated ruptured aneurysms. Mean neck size was larger in treated patients had DSA, which was not collected for analysis in one ruptured aneurysms that rebled than in those that did not rebleed patient. In the five remaining patients, aneurysm occlusion at the (4.3±2.1 mm vs 3.0±1.3 mm, respectively; p=0.049). Dome-­to-­ time of rebleeding was neck remnant in one patient and aneu- neck ratio was smaller in treated ruptured aneurysms that rebled rysm remnant in four patients (table 2). than in those that did not rebleed (1.5±0.5 vs 2.2±0.9, respec- tively; p=0.03). Risk of rebleeding was not associated with aneurysm shape (0.0% in regular aneurysms vs 1.0% in irregular Management and outcome of rebleeding in patients with aneurysms; p=0.37). Postoperative complete occlusion was less ruptured aneurysms frequent in treated ruptured aneurysms that rebled than in those Modalities of retreatment included clipping in one patient that did not rebleed (0.2% vs 2.0%, respectively; p=0.02). Risk (large temporal hematoma) and EVT in six patients (additional of rebleeding was not significantly different between aneurysms coiling with BAC in one patient, SAC in four patients, and flow with wide neck and aneurysms without wide neck (1.4% and diversion in one patient). One patient was left untreated due to 0.5%, respectively; p=0.30). In multivariate analysis, dome-­to-­ rapid clinical deterioration and died before any retreatment. At neck ratio (OR 0.2, 95% CI 0.04 to 0.8; p=0.03) and incom- hospital discharge after rebleeding, clinical outcome was good plete occlusion (OR 10.2, 1.2 to 83.3; p=0.03) were significantly (mRS 0 to 2) in three patients and poor (mRS 3 to 6) in five associated with rebleeding (table 4). patients, including death in two patients. At 12 month follow-­up (mean follow-up­ 11.1±5.6 months), clinical outcome was good in six patients (mRS 0 in four patients, mRS 1 in two patients) Discussion and two patients died (table 2). Our analysis of this recent, large, prospective, multicenter cohort shows that aneurysm coiling provides good protection against bleeding in UIAs (bleeding rate 0.0%) and rebleeding Patient factors associated with the risk of rebleeding in in RIAs (rebleeding rate 1.0%) in the year following the initial patients with ruptured aneurysms procedure. In this series, rebleeding occurred primarily in the We found no significant association between age (p=0.48), sex month following the initial bleeding and aneurysm treatment (p=0.72), elevated blood pressure (at the time of the initial (early rebleeding) in 5/8 patients, and between 1 month and procedure) (p=0.99), current smoking (p=0.46), regular 1 year after the initial procedure (late rebleeding) in 3/8 patients. alcohol consumption (p=0.99), diabetes mellitus (p=0.99), Two factors potentially play a role in rebleeding based on our polycystic kidney disease (p=0.99), family history (p=0.99) and findings, namely, postoperative aneurysm occlusion and dome-­ risk of rebleeding in patients with at least one treated ruptured to-­neck ratio. Rebleeding is a severe clinical event, which was aneurysm (table 3). fatal in 2/8 patients in this series, and can be managed using several treatment modalities (including clipping).

Table 3 Patient factors associated with rebleeding in the population Rate of bleeding (for UIA) and rebleeding (for RIA) of patients with ruptured aneurysms The main goal of IA treatment is to prevent bleeding in UIAs http://jnis.bmj.com/ Univariate and rebleeding in RIAs. According to the present series, in the Patient characteristics Rebleeding (8/794) analysis (p) year following initial aneurysm treatment, no bleeding was Age (mean±SD) Rebleeding: 49.4±19.9 years 0.48 observed after unruptured aneurysm treatment. Indeed, the risk No rebleeding: 54.0±13.0 years of bleeding of coiled UIAs has to be evaluated in the long-term­ Sex Women 6/520 (1.1%) 0.72 knowing the relatively low risk of rupture of untreated UIAs.13  Men 2/274 (0.7%) In a recent analysis conducted using a nationwide retrospective on September 27, 2021 by guest. Protected copyright. Elevated blood pressure* Yes 0/88 (0.0%) 0.99 cohort in South Korea (14 634 patients with UIA treated by  No 7/673 (1.0%) EVT, and 11 777 treated by clipping), the adjusted probability of rupture at 7 years was very low (0.9% in the endovascular group Current smoking† Yes 2/366 (0.6%) 0.46 and 0.7% in the surgical group).14  No 5/409 (1.2%) In RIAs, the risk of rerupture after aneurysm coiling has Regular alcohol Yes 1/164 (0.6%) 0.99 been evaluated in several studies. In ISAT, among the 801 consumption‡ patients with RIA treated by coiling, 40 (0.5%) presented with  No 6/609 (1.0%) rebleeding, including 14 that occurred before aneurysm treat- Diabetes mellitus§ Yes 0/32 (0.0%) 0.99 ment, leading to a rate of rebleeding after the first EVT of 3.2%  No 7/756 (0.9%) (26/801 patients). Early rebleeding (within 30 days of initial Polycystic kidney disease¶ Yes 0/6 (0.0%) 0.99 bleeding) was more frequent (20/801, 2.5%) than late rebleeding (31 days to 1 year; 6/801, 0.7%). In the large Sluzewski and van  No 7/782 (0.9%) Rooij series (431 patients with RIA), the rate of early rebleeding Family history** Yes 0/37 (0.0%) 0.99 was 1.4%.15 The same group also reported the occurrence of  No 7/733 (1.0%) late rebleeding (after 30 days) in four patients, including two *33 missing data. in the year following initial treatment (2/393, 0.5%).16 In two †19 missing data. ‡21 missing data. other large studies of patients with RIA treated by coiling, the §6 missing data. rate of early rebleeding was 0.9% and 1.1%.17 18 In our series, ¶6 missing data. **24 missing data. the rate of early rebleeding was 0.6% (5/794 patients), whereas

Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 5 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke

In a recent meta-­analysis, several risk factors were identified: Table 4 Aneurysm factors associated with rebleeding in the incomplete aneurysm occlusion, dual antiplatelet therapy, pres- population of patients with ruptured aneurysms ence of an adjacent intracranial hematoma, presence of an Multivariate analysis¶ Aneurysm Univariate aneurysmal bleb, Acom aneurysm location, small aneurysms characteristics analysis (p) OR (95% CI) p (<6 mm), and intraoperative rupture.19 In the univariate and Aneurysm size* multivariate analysis conducted in ARETA, no patient factors Mean±SD Rebleeding: 0.80 – – were associated with rebleeding, whereas two aneurysm factors 6.5±4.2 mm were identified: dome-­to-­neck ratio and postoperative occlusion. No rebleeding: 6.3±3.0 mm Aneurysm morphology seems to play an important role in aneu- rysm rebleeding. If the neck size was significantly associated with <7 mm 6/520 (1.2%) 0.71 – – rebleeding in univariate analysis (4.3±2.1 mm vs 3.0±1.3 mm; ≥7 mm 2/297 (0.7%) p=0.049), it is no longer the case in multivariate analysis. On Neck size the contrary, dome-­to-­neck ratio is significantly lower in patients Mean±SD Rebleeding: 0.049 – – with rebleeding compared with patients without rebleeding 4.3±2.1 mm (1.5±0.5 vs 2.2±0.9; p =0.03, p =0.03). It prob- No rebleeding: univariate multivariate 3.0±1.3 mm ably shows that more than the neck size by itself, it is more Dome-to-­ ­neck ratio the global morphology of the aneurysm that plays a role in the occurrence of rebleeding as previously reported.20 Postoperative Mean±SD Rebleeding: 0.03 0.2 [0.04-0.8], 0.03 1.5±0.5 aneurysm occlusion is an important factor to consider, as it can No rebleeding: be important for the therapeutic strategy. The rate of rebleeding 2.2±0.9 is 0.2% (1/465 patients) in postoperative complete occlusion, Wide neck† whereas it is 2.0% (7/348 patients) in postoperative incomplete Yes 6/440 (1.4%) 0.30 – – occlusion (punivariate=0.02; pmultivariate=0.06). Interestingly, among No 2/377 (0.5%) seven patients with incomplete post-­operative aneurysm occlu- Aneurysm location‡ sion, only two had an aneurysm remnant while the remaining five had a neck remnant. The Cerebral Aneurysm Rerupture After ACA/Acom 2/366 (0.5%) 0.31 – – Treatment (CARAT) study has previously shown that the degree MCA 1/165 (0.6%) of aneurysm occlusion after initial coiling is a strong predictor ID ICA 4/230 (1.7%) of the risk of subsequent rupture, with a 2.9% risk in cases with ED ICA 0/1 (0.0%) 91–99% aneurysm occlusion (that can be considered as a neck VB 1/55 (1.8%) remnant) and a higher risk when aneurysm occlusion is <90% 21 Aneurysm shape (5.9% to 17.6%). The fact that 5/8 patients (three with early rebleeding and two with delayed rebleeding) in our series had a Regular 0/164 (0.0%) 0.37 – – neck remnant immediately after the initial procedure suggests Irregular 8/653 (1.0%) that this occlusion type is not really sufficient to completely Multiple aneurysms prevent rebleeding. It has two consequences: (1) during the first Yes 7/634 (1.1%) 0.69 – – procedure, the goal is to obtain complete aneurysm occlusion, No 1/183 (0.5%) which offers the best protection against rebleeding; and (2) when http://jnis.bmj.com/ Postoperative complete occlusion§ a neck remnant is the final occlusion of the initial procedure, a careful follow-up­ or even a retreatment must be considered. Yes 1/465 (0.2%) Early and late rebleeding probably involves different mech- No 7/348 (2.0%) 0.02 10.2 (1.2 to 83.3) 0.03 anisms. According to Cho et al, early rebleeding can be related *Maximum diameter. †Wide neck defined as neck size ≥4 mm and/or dome to neck ratio <2. to intraprocedural administration of antiplatelet or thrombolytic ‡ACA, anterior cerebral/communicating artery; MCA, middle cerebral artery; ID ICA, intradural internal medications, maintenance of anticoagulation after the proce- carotid artery; ED ICA, extradural internal carotid artery; VB, vertebrobasilar. dure, incomplete treatment of the aneurysm, or presence of

§3 missing data. on September 27, 2021 by guest. Protected copyright. ¶Factors included in the multivariable analysis: neck size as quantitative variable, dome to neck ratio as intracranial hematoma, while delayed rebleeding is likely related quantitative variable and complete occlusion. to aneurysm recanalization with increased aneurysm size and/or coil compaction.18 late rebleeding (31 days to 1 year) was 0.4% (3/794 patients), leading to an overall 1 year rebleeding rate of 1.0%. These find- Management and clinical outcome of patients with ings highlight two important points: (1) there appears to have rebleeding been a progressive decrease in the rate of early rebleeding from Management of patients initially treated for RIA presenting with 2.5% in ISAT to 0.6% in our series, associated with a global rebleeding depends on the clinical status of the patient after decrease of aneurysm rebleeding in the first year (from 3.2% to rebleeding and location of the rebleeding (brain parenchyma 1.0%), which can potentially be attributed to technical improve- or subarachnoid space). In our series, one patient experienced ment of the coils and adjunctive devices and improved physician rapid clinical deterioration and died without receiving treat- skills (learning curve); and (2) rebleeding in the first year after ment. Another patient had a large temporal hematoma and was initial treatment typically occurs early (<30 days), whereas late treated surgically with removal of the hematoma, exploration rebleeding can occur several weeks or months after initial treat- and aneurysm clipping; despite this management, the patient ment and is rare. died. Retreatment in the remaining six patients was performed by an endovascular approach using several techniques: BAC in Risk factors for rebleeding of RIA one patient, SAC in four patients, and flow diversion in one Identifying risk factors associated with rebleeding of coiled patient. In the eight patients with rebleeding, clinical outcome at ruptured aneurysms is not easy due to low occurrence of events. discharge was good (mRS 1) in 3/8 patients (37.5%), poor (mRS 6 Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 J NeuroIntervent Surg: first published as 10.1136/neurintsurg-2020-015971 on 16 June 2020. Downloaded from Hemorrhagic stroke

3, 4 and 5) in 3/8 patients (37.5%), and death in 2/8 patients Competing interests None declared. (25.0%). At 12-months­ follow-­up, clinical status was better with Patient consent for publication Not required. good clinical outcome (mRS 0 in four patients and mRS 1 in Provenance and peer review Not commissioned; externally peer reviewed. two patients) in 6/8 patients (75.0%) and death in 2/8 patients Data availability statement The data related to this manuscript are available (25.0%). upon reasonable request.

ORCID iDs Limitations Laurent Pierot http://orcid.​ ​org/0000-​ ​0002-6893-​ ​661X First, ARETA was designed to analyze recanalization after EVT Gaultier Marnat http://orcid.​ ​org/0000-​ ​0002-7611-​ ​7753 of intracranial aneurysms, not intra- and postoperative compli- Chrysanthi Papagiannaki http://orcid.​ ​org/0000-​ ​0002-9473-​ ​9644 cations. However, given that ARETA is, to the best our knowl- edge, the largest recent series of patients with aneurysm treated References by EVT, the cohort offers the opportunity to analyze the rate of 1 Molyneux A, Kerr R, Stratton I, et al. International Subarachnoid Aneurysm Trial (ISAT) bleeding/rebleeding during 1 year follow-up.­ Second, in relation of neurosurgical clipping versus in 2143 patients with ruptured intracranial aneurysms: a randomised trial. Lancet 2002;360:1267–74. to the previous limitation, and according to the study’ primary 2 Pierot L, Spelle L, Vitry F, et al. Immediate clinical outcome of patients harboring objective (recanalization), postoperative medications were not unruptured intracranial aneurysms treated by endovascular approach: results of the collected in the database knowing that postoperative antiplatelet ATENA study. Stroke 2008;39:2497–504. medication can potentially play a role in the occurrence of 3 cognard C, Pierot L, Anxionnat R, et al. Results of embolization used as the first rebleeding in patients with ruptured aneurysms. Third, patients treatment choice in a consecutive nonselected population of ruptured aneurysms: clinical results of the clarity GDC study. Neurosurgery 2011;69:837–41. who died at home or were lost to follow-­up after hospital 4 Pierot L, Wakhloo AK. Endovascular treatment of intracranial aneurysms: current discharge may have experienced undiagnosed rebleeding, status. Stroke 2013;44:2046–54. leading to an underestimation of the rebleeding rate. 5 Pierot L, Moret J, Barreau X, et al. Safety and efficacy of aneurysm treatment with WEB in the cumulative population of three prospective, multicenter series. J Neurointerv Surg 2018;10:553–9. Conclusion 6 Molyneux AJ, Birks J, Clarke A, et al. The durability of endovascular coiling versus Aneurysm coiling affords good protection against bleeding (for neurosurgical clipping of ruptured cerebral aneurysms: 18 year follow-up­ of unruptured aneurysms) and rebleeding (for ruptured aneurysms) the UK cohort of the International Subarachnoid Aneurysm Trial (ISAT). Lancet 2015;385:691–7. at 1 year with rates of 0.0% and 1.0%, respectively. Aneurysm 7 naggara ON, Lecler A, Oppenheim C, et al. Endovascular treatment of intracranial occlusion and dome-­to-­neck ratio are the two factors that appear unruptured aneurysms: a systematic review of the literature on safety with emphasis to play a role in the occurrence of rebleeding. Consequently, the on subgroup analyses. Radiology 2012;263:828–35. goal of initial treatment of a ruptured aneurysm is to obtain 8 Benaissa A, Barbe C, Pierot L. Analysis of recanalization after endovascular treatment of intracranial aneurysm (ARETA trial): presentation of a prospective multicenter complete aneurysm occlusion. If this outcome is not possible, study. J Neuroradiol 2015;42:80–5. close monitoring needs to be undertaken. 9 gawlitza M, Soize S, Barbe C, et al. Aneurysm characteristics, study population, and endovascular techniques for the treatment of intracranial aneurysms in a large, Author affiliations prospective, multicenter cohort: results of the Analysis of Recanalization after 1Radiology, CHU Reims, Reims, Champagne-­Ardenne, France Endovascular Treatment of intracranial Aneurysm study. AJNR Am J Neuroradiol 2Department of Research and Public Health, CHU Reims, Reims, Champagne-­ 2019;40:517–23. Ardenne, France 10 Pierot L, Barbe C, Nguyen HA, et al. Intraoperative complications of endovascular 3Neuroradiology, CHRU Tours, Tours, Centre, France treatment of intracranial aneurysms with coiling or balloon-assisted­ coiling in a 4Neuroradiology, CHU Rennes, Rennes, Bretagne, France prospective multicenter cohort of 1088 participants: Analysis of Recanalization after 5 Neuroradiology, CHU Toulouse, Toulouse, Midi-­Pyrénées, France Endovascular Treatment of intracranial Aneurysm (ARETA) study. Radiology. In Press http://jnis.bmj.com/ 6Interventional neuroradiology, CHRU Lille Pôle Spécialités Médicochirurgicales, Lille, 2020;295:381–9. Hauts-de-­ ­France, France 11 Jartti P, Isokangas J-M,­ Karttunen A, et al. Early rebleeding after coiling of ruptured 7Neuroradiology, CHU Dijon, Dijon, Bourgogne, France intracranial aneurysms. Acta Radiol 2010;51:1043–9. 8Neuroradiology, CHU Nantes, Nantes, Pays de la Loire, France 12 Pierot L, Barbe C, Ferré J-­C, et al. Patient and aneurysm factors associated with 9Radiology, CHU Poitiers, Poitiers, France aneurysm rupture in the population of the ARETA study. J Neuroradiol 2019:30440–7. 10Neuroradiology, CHU Saint-Étienne, Saint-­Etienne, Rhône-­Alpes, France 13 Wiebers DO, Whisnant JP, Huston J, et al. Unruptured intracranial aneurysms: natural 11Neuroradiologie, CHU Clermont-­Ferrand, Clermont-­Ferrand, France history, clinical outcome, and risks of surgical and endovascular treatment. Lancet 12Neurointerventionnel, CHU Nice, Nice, Provence-­Alpes-­Côte d’Azu, France 2003;362:103–10. on September 27, 2021 by guest. Protected copyright. 13neuroradiology, CH Sainte Anne, Paris, Île-­de-­France, France 14 Kim YD, Bang JS, Lee SU, et al. Long-­term outcomes of treatment for unruptured 14Interventional and Diagnostic Neuroradiology, CHU de Bordeaux, Bordeaux, intracranial aneurysms in South Korea: clipping versus coiling. J Neurointerv Surg Aquitaine, France 2018;10:1218–22. 15Neuroradiology, CHU Bicêtre, Le Kremlin-­Bicetre, Île-de-­ ­France, France 15 sluzewski M, van Rooij WJ. Early rebleeding after coiling of ruptured cerebral 16Neuroradiology, Hopital Foch, Suresnes, Île-­de-­France, France aneurysms: incidence, morbidity, and risk factors. AJNR Am J Neuroradiol 17Neuroradiology, IFR des Neurosciences CHU Pitie-­Salpetriere IFR 70, Paris, Île-­de-­ 2005;26:1739–43. France, France 16 sluzewski M, van Rooij WJ, Beute GN, et al. Late rebleeding of ruptured intracranial 18Interventional Neuroradiology, CHU Rouen, Rouen, Normandie, France aneurysms treated with detachable coils. AJNR Am J Neuroradiol 2005;26:2542–9. 19Institute for Ageing & Health, Newcastle University, Newcastle upon Tyne, UK 17 Fleming JB, Hoh BL, Simon SD, et al. Rebleeding risk after treatment of ruptured 20Neuroradiology, Newcastle upon Tyne Hospitals, Newcastle upon Tyne, UK intracranial aneurysms. J Neurosurg 2011;114:1778–84. 21Interventional Neuroradiology, CHU Bicêtre, Le Kremlin-­Bicetre, Île-­de-­France, 18 cho YD, Lee JY, Seo JH, et al. Early recurrent hemorrhage after coil embolization in France ruptured intracranial aneurysms. Neuroradiology 2012;54:719–26. 19 li K, Guo Y, Zhao Y, et al. Acute rerupture after coil embolization of ruptured intracranial saccular aneurysms: a literature review. Interv Neuroradiol Contributors All authors have: provided a substantial contribution to the 2018;24:117–24. conception and design of the studies and/or the acquisition and/or the analysis of 20 chang SH, Shin HS, Lee SH, et al. Rebleeding of ruptured intracranial aneurysms in the data and/or the interpretation of the data; drafted the work or revised it for the immediate postoperative period after coil embolization. J Cerebrovasc Endovasc significant intellectual content; approved the final version of the manuscript; agreed Neurosurg 2015;17:209–16. to be accountable for all aspects of the work, including its accuracy and integrity. 21 Johnston SC, Dowd CF, Higashida RT, et al. Predictors of rehemorrhage after treatment Funding The French Health Ministry has funded ARETA (Programme Hospitalier de of ruptured intracranial aneurysms: the Cerebral Aneurysm Rerupture After Treatment Recherche Clinique, No. 12-001-0372). (CARAT) study. Stroke 2008;39:120–5.

Pierot L, et al. J NeuroIntervent Surg 2020;0:1–7. doi:10.1136/neurintsurg-2020-015971 7