Article Text


Original research
Intraprocedural predictors of post-stent retriever thrombectomy subarachnoid hemorrhage in middle cerebral artery stroke
  1. Perry P Ng1,2,
  2. Theodore C Larson1,
  3. Christopher W Nichols1,
  4. Mark M Murray1,
  5. Karen L Salzman2,
  6. Richard H Smith1
  1. 1 Centura Health Neurosciences and Spine, Lakewood, Colorado, USA
  2. 2 University of Utah Health Sciences Center, Salt Lake City, Utah, USA
  1. Correspondence to Dr Perry P Ng, CHPG Neurosciences and Spine, St Anthony Hospital, Lakewood, CO 80228-1726, USA; perryng{at}


Background Stent retriever thrombectomy (SRT) in acute thromboembolic stroke can result in post-thrombectomy subarachnoid hemorrhage (PTSAH). Intraprocedural findings associated with PTSAH are not well defined.

Objective To identify angiographic findings and procedural factors during SRT that are associated with PTSAH.

Materials and methods This was a retrospective, observational cohort study of consecutive patients with middle cerebral artery (MCA) acute ischemic stroke treated with SRT. Inclusion criteria were: (1) age ≥18 years; (2) thromboembolic occlusion of the MCA; (3) at least one stent retriever pass beginning in an M2 branch; (4) postprocedural CT or MRI scan within 24 hours; (5) non-enhanced CT Alberta Stroke Program Early CT Score >5. Exclusion criteria included multi-territory stroke before SRT.

Results Eighty-five patients were enrolled; eight patients had PTSAH (group 1) and 77 did not (group 2). Baseline demographic and clinical characteristics were comparable between the two groups. In group 1, a significantly greater proportion of patients had more than two stent retriever passes (62.5% vs 18.2%, P=0.01), a stent retriever positioned ≥2 cm along an M2 branch (100% vs 30.2%, P=0.002), and the presence of severe iatrogenic vasospasm before SRT pass (37.5% vs 5.2%, P=0.02). One patient with PTSAH and associated mass effect deteriorated clinically.

Conclusions An increased number of stent retriever passes, distal device positioning, and presence of severe vasospasm were associated with PTSAH. Neurological deterioration with PTSAH can occur.

  • angiography
  • complication
  • hemorrhage
  • stroke
  • subarachnoid

Statistics from


Stent retriever thrombectomy (SRT) has revolutionized the treatment of acute ischemic stroke (AIS) due to large vessel occlusion, with higher revascularization rates and better safety profile than with first-generation devices and better clinical outcomes than with best medical therapy.1–4

Complications with SRT include embolization to a new territory, vessel dissection, and frank vessel perforation, with a reported incidence of 1–7% for all device- or procedure-related serious adverse events.3 5–8 Stent retrievers also result in more extensive endothelial injury than with thromboaspiration catheters during retrieval of the device as they exert a continuous radial force against the vessel wall.9 It has been suggested that traction forces applied to the vessel and deformation of the adjacent anatomy cause stretching and rupture of small arterioles, resulting in post-thrombectomy subarachnoid hemorrhage (PTSAH) without frank vessel perforation or contrast extravasation seen on digital subtraction angiography (DSA).10 11 Although previous reports of PTSAH have shown a universally benign prognosis,10–14 one patient in our cohort had symptomatic PTSAH after developing a Sylvian fissure hematoma. PTSAH may result in clinical sequelae from a mass effect secondary to focal hematoma and/or subsequent vasospasm with delayed cerebral ischemia as observed in patients with aneurysmal subarachnoid hemorrhage. The possibility of ensuing vasospasm may necessitate a longer hospital stay for clinical and imaging follow-up and manifest as subsequent neurological decline in a patient with successful thrombectomy, emphasizing the importance of identifying intraprocedural risk factors that might help to mitigate PTSAH.

Apart from a previous report suggesting an association of PTSAH with adjunctive angioplasty after failed SRT,11 other procedural risk factors that increase the incidence of PTSAH have not been elucidated. We sought to identify angiographic findings and modifiable procedural factors related to the occurrence of PTSAH.


A retrospective review of prospectively collected data was undertaken for consecutive patients with AIS due to unilateral middle cerebral artery (MCA; M1 and/or M2 segment) thromboembolic occlusions, between January, 2015 and October, 2017, treated with SRT at two acute care hospitals staffed by the same group of four neurointerventionalists. Inclusion criteria included at least one stent retriever pass beginning in an M2 branch. Adjunctive endovascular techniques were used at the discretion of the operator and included thromboaspiration, guidewire probing of thrombus, proximal internal carotid artery angioplasty and/or stenting, and administration of IA thrombolytic agents, antiplatelet agents, and/or verapamil. Patients were excluded from the analysis if they had concurrent AIS in another vascular territory before SRT or if they did not have postprocedural CT or MRI imaging within 24 hours of SRT for detection of PTSAH. Patients with parenchymal hematoma (PH1 or PH2, classified according to Fiorelli et al 15) with small adjacent PTSAH (assumed to be secondary to rupture of the parenchymal hematoma into the subarachnoid space) were not included in the PTSAH group for analysis.

IV tissue plasminogen activator (tPA) was received by all patients who were eligible according to the American Stroke Association guidelines current at the time of treatment. Patients had preprocedural CT imaging including non-enhanced CT (NECT) for determination of an Alberta Stroke Program Early CT Score (ASPECTS),16 and CT angiography to detect large vessel occlusion. Time from last known well to groin puncture was recorded as were baseline comorbidities, including history of hypertension, diabetes mellitus, atrial fibrillation, current smoking status, and blood sugar level before SRT.

Endovascular treatment was performed on flat-detector biplane neuroangiography systems, either Allura XPER FD20 (Philips Healthcare, Best, Netherlands) or Artis Q (Siemens Healthcare GmbH, Munich, Germany). Hand injections of contrast were used for angiographic acquisitions. Procedural images were reviewed on a Fujifilm Synapse picture archiving and communication system (PACS; Tokyo, Japan), and measurements of vessel angulation and stent retriever position made using the integrated tools palette. For M1 occlusions, clot location was considered ‘proximal’ if the occlusion was along the proximal third of the M1 segment, ‘mid’ if along the mid-third, and ‘distal’ if along the distal third to the point where the MCA turns upward at the M1/M2 junction. All vessel angles were measured using frontal plane two-dimensional DSA images. The M1–M1 angle was measured as the largest angle along the horizontal MCA (M1) segment, extending from the carotid terminus to the limen insulae with lines drawn along the center of the vessel on each side of the angulation. The M1–M2 angle was defined as the angle from the M1 segment as it turns superiorly (at the M1/M2 junction) to become the M2 segment into which the stent retriever is positioned as the vessel courses toward the circular sulcus. All angles were measured without a stent retriever in situ to avoid straightening of the vessel anatomy. The PACS software measured a straight line as having an angle of 180 degrees. Thus ‘calculated MCA angulation’ was obtained by subtracting the measured angles from 180 degrees to better reflect the angles through which the stent retriever was withdrawn, with higher numbers indicating greater vessel curvature. Total calculated vessel angle (TCVA) was the sum of the calculated M1–M1 and M1–M2 angles (figure 1).

Figure 1

The M1–M1 angle is measured as 125 degrees (A) and the M1–M2 angle 113 degrees (B) on frontal digital subtraction angiography projections. The calculated M1–M1 (CM1–M1) angle is 180–125=55 degrees. The calculated M1–M2 angle (CM1–M2) is 180–113=67 degrees. The Total Calculated Vessel Angle (TCVA) is therefore (CM1–M1)+(CM1–M2) (55+67)=122 degrees. In another patient the M1–M1 angle is 109 degrees (C) giving a CM1–M1 angle of 71 degrees. Occlusion of the inferior division at the M1/M2 junction is seen (arrow). In the late arterial phase (D) the proximal M2 segment is visualized with contained thrombus (arrow) allowing measurement of the M1–M2 angle (95 degrees) and calculation of CM1–M2 and TCVA of 85 and 156 degrees, respectively). In a different patient with large clot burden visualized within an M2 branch (arrows in E), the M1–M1 angle is 131 degrees (E) and M1–M2 angle 76 degrees (F), giving a TCVA of 153 degrees.

The presence or absence of iatrogenic vasospasm before SRT along the vessel in which the stent retriever was positioned and subsequently withdrawn was documented. Vasospasm was considered ‘mild’ if there was a reduction of less than one-third in diameter of the affected vessel segment compared with an adjacent normal segment, ‘moderate’ if there was a reduction of between one-third and two-thirds in vessel diameter, and ‘severe’ if there was more than two-thirds reduction in vessel diameter. Stent retriever position along the M2 segment was measured from the M1/M2 junction to the distal markers of the device using a posteroanterior or lateral projection DSA or roadmap image. In cases where distance measurements on PACS were erroneous owing to incorrect calibration, corrected measurements were calculated by multiplying the ratio of the measured length of the device within an M2 branch by the known length of the stent retriever used divided by the total device length measured (figure 2). The final Thrombolysis in Cerebral Infarction (TICI) score17 was recorded.

Figure 2

Unsubtracted frontal image (A) shows a 3×20 mm stent retriever positioned in a proximal M2 branch with 17 mm of device extending into the M2 segment. Frontal digital subtraction angiography image (B) in a different patient shows a 3×23 mm stent retriever with distal marker in an M2 branch (black arrow) and proximal marker along the distal M1 segment (white arrow). The picture archiving and communication system image is uncalibrated so the total device length measured as 38.8+11.2 = 50 mm is incorrect. The actual length of the device in the M2 is calculated by multiplying the known length of the stent retriever (23 mm) by the measured length of the device in the M2 (38.8 mm) divided by the total device length measured (50 mm)=23×38.8 ÷ 50 ≈ 18 mm.

Follow-up NECT or MRI scans within 24 hours of SRT were reviewed to determine the presence of intracranial hemorrhage, including PTSAH. Choice of follow-up imaging modality was at the discretion of the treatment team. Patients who had contraindications to MRI, including cardiac pacemakers and other incompatible implants, or who required close physiological monitoring due to clinical instability, had follow-up CT. In patients who had both CT and MRI within 24 hours of SRT, both examinations were reviewed for the presence of PTSAH. The criteria of Yoon et al 11 were adopted such that hyperdensity on the CT scan within cortical sulci of the MCA territory treated with SRT was considered purely PTSAH if the Hounsfield unit (HU) measurement was <90, and PTSAH mixed with contrast extravasation if HU was ≥90. Subarachnoid hyperintensity on fluid-attenuated inversion recovery (FLAIR) MRI and/or hypointensity/”blooming’ on susceptibility weighted images localized to the treated MCA territory was considered positive for PTSAH.

Immediate clinical outcome was considered good if the patient was discharged home or to an inpatient rehabilitation facility, and poor if death occurred in hospital or the patient was discharged to a skilled nursing facility or hospice. National Institutes of Health Stroke Scale (NIHSS) score at discharge was considered good if ≤6 and poor if >6.

The study protocol was reviewed and approved by the local institutional review board.

Statistical analyses

Patients with and without PTSAH were compared by univariate analysis using Fisher’s exact tests for categorical variables, Wilcoxon rank sum tests for non-normal continuous data, and Student’s t-tests for normally distributed continuous data. SAS 9.4 (Cary, North Carolina, USA) was used for all analyses, supervised by a statistician. Two-tailed tests with an α of 0.05 were used for all tests.


Eighty-five patients were included in our analysis. On follow-up imaging, eight patients had PTSAH (group 1) and 77 patients did not (group 2). One patient with a 5 cm frontal hematoma with mass effect and small adjacent SAH was included in group 2. No patient had subarachnoid hyperdensity >90 HU to indicate contrast extravasation. There was no difference in baseline demographics, stroke severity, CT ASPECTS, and administration of IV tPA between the two groups before SRT (table 1).

Table 1

Baseline demographic, clinical and CT imaging characteristics (n=85)

More baseline M2 occlusions were found in patients in group 1 than in group 2 (62.5% vs 37.7%); this was not statistically significant. SRT was performed with Solitaire (Medtronic Neurovascular, Irvine, California, USA) in 36 patients, Trevo (Stryker Neurovascular, Fremont, California, USA) in 54 patients and MindFrame (Medtronic Neurovascular) in five patients. A combination of these stent retrievers was used in 10 patients. There was no difference in the use of different stent retriever designs, diameters or lengths between the two groups. In 22 patients, no adjunctive technique was used. Adjunctive techniques in the remaining 63 patients consisted of thromboaspiration (n=63), guidewire disruption of thrombus (n=2), IA administration of thrombolytic agents and/or platelet receptor IIb/IIIa inhibitors (n=8), IA verapamil (n=12), and extracranial internal carotid artery stenting or angioplasty (n=5). Intracranial angioplasty or stenting was not used. There was no statistically significant difference in the frequency of use of adjunctive techniques between the two groups. Specifically, IA thrombolytic or antiplatelet agents were used in one patient in group 1, and seven patients in group 2 (12.5% vs 9%, P=0.56). No patients in either group showed evidence of intraprocedural arterial perforation or extravasation of contrast during SRT.

Six patients in group 1 (75%) and 43 patients in group 2 (56%) had PACS images that allowed measurement of stent retriever position within the treated M2 branch. A significantly greater proportion of patients in group 1 had stent retrievers placed ≥2 cm along an M2 branch compared with patients in group 2 (100% vs 30.2%, P=0.002). The average number of stent retriever passes made was 1.9 (range 1–6) for the entire cohort. In group 1, the mean number of passes was 3.3 (range 1–6) and in group 2 it was 1.8 (range 1–6). A significantly greater proportion of patients in group one had more than two stent retriever passes compared with those in group 2 (62.5% vs 18.2%, P=0.01). When TCVA was dichotomized at 110 degrees (median of TCVA values), there was no difference in the occurrence of PTSAH. However, there was a trend toward the occurrence of PTSAH when dichotomizing by the top 10% of the most extreme TCVAs (≥160 degrees), although it was not statistically significant (P=0.06). All patients in group 1 had iatrogenic vasospasm identified on DSA after a stent retriever pass. Moreover, there was a significant difference between the presence of severe vasospasm between the two groups (37.5% in group 1 vs 5.2% in group 2, P=0.02). These results are summarized in table 2.

Table 2

Angiographic findings and procedural variables (n=85)

There was no statistically significant difference in good clinical outcomes based on discharge disposition (P=0.70) or NIHSS score (P=0.71) between group 1 and group 2. However, one patient in group 1 with a large PTSAH had neurological decline several hours after thrombectomy with an increase in NIHSS score of 9 points but returned to an NIHSS of 1 after 3 days (figure 3).

Figure 3

Non-enhanced CT (NECT) obtained 3 hours after left M2 thrombectomy (A) with single pass of a 3×23 mm stent retriever (same patient as in figure 2B, who had worsening aphasia, headache, and right arm weakness after stent retriever thrombectomy) shows post-thrombectomy subarachnoid hemorrhage in the left frontal lobe sulci (arrows) and Sylvian fissure hematoma (*). Repeat NECT the next morning (B) shows increased size of the hematoma (*) with mass effect evidenced by mild midline shift.


PTSAH after mechanical thrombectomy has been reported to occur with the MERCI device,18 Penumbra aspiration system,19 and stent retrievers.3 11 20 Intraprocedural perforation during SRT with contrast extravasation on DSA is rare, with an incidence 1.0% in one series, although prognosis is poor with in-hospital mortality of almost 60%.21 Most PTSAH after SRT is less dramatic and found on postprocedural imaging. Machi et al 12 reported on 56 patients who had SRT with the Solitaire stent, in which two patients (3.6%) had PTSAH. Dorn et al 13 found an incidence of PTSAH in 5.6% (6 of 108 patients) treated with the Solitaire stent for AIS. Both these studies included patients treated for anterior and posterior circulation stroke. Yoon et al 11 studied 74 patients who had SRT with the Solitaire stent, including 44 patients treated for MCA stroke. Nine (20.5%) of these 44 patients had PTSAH and/or subarachnoid contrast extravasation on postprocedural CT scans performed immediately after SRT. Kurre et al 10 reported on 76 patients treated with the pREset LITE (Phenox GmbH, Bochum, Germany) stent retriever and of the 61 patients with M2 occlusions treated, PTSAH occurred in 13.3%. The incidence of PTSAH in this study of 9.4% is in line with previous studies of MCA thrombectomy.

Suggested mechanisms for PTSAH include stretching of arterioles and accompanying venules in the subarachnoid space during withdrawal of the stent retriever and disruption of cerebral microvascular permeability barriers secondary to contrast neurotoxicity, exogenous plasminogen activators, and/or reperfusion injury.11 22 Stent retrievers have been shown to exert greater radial force when deployed in 1.5 mm diameter compared with 3.5 mm diameter silicone tubes, causing increased friction with the tube wall and requiring a greater traction force to retrieve the device at a rate of 2 mm/s.23 Moreover, the traction force was directly proportional to the length of the deployed stent. Since M2 branches typically have a diameter of ≤2 mm we suspect that the higher traction force required during SRT increases the likelihood of stretch injury to M2 perforating arterioles and that this phenomenon is exacerbated by the length of stent retriever positioned along the M2 branch. This may explain the significantly higher incidence of PTSAH in patients who had a stent retriever pass beginning two or more centimeters in an M2 branch in this study.

Iatrogenic vasospasm is common with SRT, occurring in 35 of 85 patients (41%) in this study. Previously reported rates vary from 22.5% to 37.5%, with a higher incidence after treatment of smaller vessels.24 25 All patients with PTSAH in our study had vasospasm seen on DSA, suggesting that iatrogenic vasospasm along the occluded vessel may herald an increased risk of PTSAH. Moreover, the presence of severe vasospasm correlated with a higher incidence of PTSAH (37.5% vs 5.2%, P=0.02). We suspect that iatrogenic vasospasm increases the risk of PTSAH with subsequent passes of a stent retriever as the vessel constricts to a smaller diameter, thereby increasing friction between the device and vessel wall and therefore the traction force required for withdrawal. The effect of vasospasm may be exacerbated in M2 occlusions as these vessels have a small normal diameter so that any reduction in vessel caliber would result in greater friction and traction forces during SRT than in larger proximal vessels. The utility of IA verapamil for treating intraprocedural vasospasm during stroke thrombectomy is unclear. IA verapamil was administered in two patients who had PTSAH in this study compared with 10 patients without PTSAH, which was not statistically significant (20% vs 13%, P=0.31). When used during stroke thrombectomy, IA verapamil may not reach the spastic vessel owing to occlusive thrombus and/or it may not have adequate time to take effect given the time-critical nature of stroke intervention.

Schwaiger et al 26 showed that angulation of the MCA affects recanalization outcomes during SRT, with poor recanalization (TICI 0–2a) being associated with greater vessel angulation. They suggested that strongly curved vessels may increase friction between the vessel wall and the device, with increased tension forces during retrieval being transferred to the vessel causing folding of vessel segments proximally and elongation of the vessel distally, the same mechanisms proposed to cause PTSAH. Similar to that study, we also found no increased rate of PTSAH with vessel curvature, although we did detect a trend toward PTSAH occurring in patients with extreme MCA angulation (TCVA ≥160 degrees).

A greater number of stent retriever passes was associated with PTSAH in this study. This may occur because additional passes cause repeated traction injury to small perforating branches. We propose that an additional mechanism for PTSAH might be related to unrecognized microguidewire perforation or dissection when probing the thrombus, with delayed rupture of the occluded artery and/or its branches. Kurre et al 10 suspected iatrogenic dissection in 2 of 76 patients undergoing SRT. Blind probing of the thrombus is necessary given the inability to roadmap the occluded vessel and/or due to patient movement, as many centers, including ours, now perform SRT without general anesthesia. Microperforations may not result in immediate bleeding if the vessel remains occluded or if the vessel immediately develops vasospasm. Additional stent retriever passes necessitate repeated guidewire probing of the occluded vessel, increasing the potential for vessel injury. Subsequent recanalization of an occluded vessel with occult dissection or microperforation may lead to delayed, slow extravasation and PTSAH.

Previous reports of PTSAH after stent retriever thrombectomy have universally suggested a benign prognosis. In the study by Yoon et al,11 none of the 12 patients with PTSAH experienced postprocedural neurological deterioration, and their discharge NIHSS score, rate of good clinical outcome, and mortality did not differ from those who did not have PTSAH. Similar benign courses with PTSAH have been reported by other investigators.10 12–14 In our study 1 of 8 patients (12.5%) developed a sizeable left Sylvian fissure PTSAH with mass effect and had worsening of neurological symptoms with an increase in the NIHSS score of 9 points several hours after the procedure (figure 3). The decision to treat with SRT was based on fluctuating neurological symptoms despite a low NIHSS score. Cryoprecipitate and platelets were administered in an attempt to reverse the effect of tPA, and the patient gradually improved to an NIHSS score of 1 over the ensuing 72 hours without the need for surgical evacuation of the hematoma. This case demonstrates that PTSAH may not be as benign as previously thought. Had it occurred in a younger patient with fewer involutional changes into which a subarachnoid hematoma could expand and who might be more prone to SAH-related vasospasm, the outcome could have been different. This case also highlights the need for immediate postprocedural imaging in patients with worsening stroke symptoms after thrombectomy, which can be difficult to detect in patients with higher baseline NIHSS scores. Delaying the diagnosis of significant PTSAH may prevent timely reversal of thrombolytic, antiplatelet and/or anticoagulant medications with worse clinical outcome.

This study has several limitations. The retrospective review of procedural images on PACS was met with filming deficiencies that did not allow measurement of stent retriever position in some patients in each group. The measurement of vessel angles using frontal projection 2D DSA images provides only an approximation of vessel angulation, which can vary depending on the degree of caudal (Waters) and cranial (Towne) angulation of the tube. Moreover, vessel angulation in the anteroposterior plane which might have been better seen on a lateral projection was not accounted for. Finally, inclusion criteria involved having a postprocedural CT or MRI scan performed up to 24 hours after SRT. It is possible that a small PTSAH might have resolved or redistributed during this time and be missed, resulting in an underestimation of its incidence.


PTSAH after SRT for MCA stroke is associated with higher number of stent retriever passes, distal placement of the device along an M2 branch, and presence of severe iatrogenic vasospasm. There was a trend toward PTSAH occurring with extreme MCA curvature. Clinical deterioration can occur with PTSAH. Further work should be conducted to confirm our findings and to identify additional procedural factors that might lead to PTSAH.


The authors gratefully acknowledge Christy Casper, Alexandra Graves, and Janet Carlson, Kari Scaletta and Stephanie Hansen for their dedication in the care of these patients with stroke and their efforts in collecting and maintaining stroke data at their respective hospitals.


View Abstract


  • Contributors PPN: database creation, data mining, project planning, writing of first manuscript draft, guarantor. TCL, project planning, literature search, review of angiograms, proofreading of manuscript. CWN: met statisticians for data analysis, literature search, data mining. MMM: data mining, proofreading of manuscript. KLS: literature search, imaging CT and MRI review, proofreading of manuscript. RHS: institutional review board application and reviews, coordination of project with all authors, data mining, proofreading of manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval The research protocol was reviewed and approved by the Catholic Health Initiatives institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.