Iatrogenic Vessel Dissection in Endovascular Treatment of Acute Ischemic Stroke

Acute arterial ischemic stroke (AIS) is one of the leading causes of mortality and morbidity worldwide [1, 2]. Immediate and efficacious revascularization of the occluded vessel is crucial to improve outcome [3, 4]. Over the past decade endovascular management of AIS has rapidly improved and several catheter-based treatment systems have been developed. To date, 5 randomized controlled trials have shown that endovascular clot retrieval in addition to best medical treatment with or without intravenous recombinant tissue plasminogen activator (rtPA) improves outcome in patients with stroke occurring in the anterior circulation with proximal vessel occlusion [5‐9]; however, despite minimally invasive device systems, endovascular procedures still carry a risk of iatrogenic complications, potentially worsening the outcome. Manipulation of the catheter, guidewire or thrombectomy devices could cause vessel wall injury. Studies reporting ID rates during cerebral angiography mainly include diagnostic angiography only [10‐12] or interventional angiography for indications other than AIS, often in an elective, non-urgent setting [13‐15]. Only a few studies have reported complication rates in endovascular AIS treatment (ranging from 1.7% to 3.9%), but clinical characteristics, risk factors, outcomes and morphological descriptions are lacking [5, 9, 16, 17]. We aimed to assess the localization and morphology of ID, the clinical characteristics and risk factors for ID and outcomes in patients suffering from ID.

Our study is based on the Bernese stroke registry, a prospective database on patients with AIS [18‐20]. Baseline and 3‑month outcome data on consecutive AIS patients admitted to the Neurological Department of the University Hospital of Bern are collected. For this study, we reviewed clinical and imaging data on all AIS patients aged ≥16 years who underwent endovascular AIS treatment between January 2000 and March 2012. All images of confirmed or assumed IDs were retrieved from our PACS (picture archiving and communicating system) and the digital subtraction angiography (DSA) series were carefully re-evaluated for this study. An ID was defined as the presence of any of the following signs on DSA: intimal flap, fusiform or irregular aneurysmal dilation at a non-bifurcation site, double lumen or a string-and-bead sign, or if signs of a wall hematoma were present on fat-saturated T1-weighted magnetic resonance imaging (MRI).

Baseline clinical data, such as age, sex, pre-stroke score on the modified Rankin scale (mRS), vascular risk factors, National Institutes of Health Stroke score (NIHSS) on admission, and stroke etiology according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification [21] were collected. Imaging data, such as vessel occlusion site and grade of occlusion, were assessed at baseline, and after endovascular treatment. Furthermore, imaging in ID cases was reviewed for ID characteristics on DSA, such as localization and stenosis caused by an ID in the acute phase DSA and recanalization on follow-up imaging by magnetic resonance angiography (MRA) or computed tomography angiography (CTA). Stenosis caused by an ID was defined as high grade if ≥80% of the vessel lumen was obstructed, moderate for 50–79%, and mild for <50% local degree of stenosis. The length of the stenosis was classified as either short (≤1 cm) or long (>1 cm).

Follow-up imaging was performed in all patients between 12 and 24 h after the intervention or sooner in the case of any clinical deterioration. The remaining local arterial constriction was graded according to the thrombolysis in myocardial ischemia (TIMI) classification [22]. All interventions were performed in our 24-h/7-day stroke unit by 5 fully trained and experienced neurointerventionalists in accordance with institutional guidelines. Endovascular treatment was based on a diagnostic transfemoral 4‑vessel DSA and consisted of (1) mechanical thrombectomy (MT), (2) local intra-arterial thrombolysis using urokinase, (3) combined intra-arterial lysis with urokinase and systemic intravenous thrombolysis using rtPA or (4) by a combination of MT and pharmacological thrombolysis. Until the introduction of stent retriever technology in 2010, MT was performed by aspiration via a 5 French aspiration catheter (Tracker 35, Boston Scientific; VASCO-ASP, BALT [Montmorency, France]), percutaneous transluminal angioplasty (PTA), stenting or gentle mechanical disruption of the thrombus in combination with local injection of urokinase into the thrombus for up to 2 h. The Merci device (Concentric Medical, Mountain View, California, USA) has not been used. Based on the localization, grade of stenosis and hemodynamic relevance of the lesion, the responsible interventional neuroradiologist and neurologist decided on endovascular treatment of ID. Clinical outcome at 3 months was assessed using the mRS, with favorable outcome defined as a mRS score ≤ 2.

Out of 869 consecutive AIS patients who underwent an acute endovascular stroke intervention, 866 were included in this study with a median age of 68 years (interquartile range IQR 57–76 years) and 385 (45%) women. Of the patients three had to be excluded from the study because the intervention was not documented in our PACS system.

The results of our study indicate that the risk for ID in AIS patients undergoing endovascular treatment is low. Mechanical thrombectomy and smoking status are associated with a higher risk for ID. The ID did not have a significant impact on clinical outcome compared to patients without ID despite severe complications possibly related to ID in 2 patients. To date, arterial dissection as a complication of endovascular AIS treatment has been systematically reported by the MR CLEAN, REVASCAT, SWIFT and STAR investigators [5, 9, 16, 17]. Dissection occurred in 1% of STAR patients and 3.9% of REVASCAT patients as a complication of AIS treatment with either Solitaire® (Medtronic, Dublin, Ireland) flow restoration thrombectomy or combined with medical management alone [9, 17]. The MR CLEAN trial found a procedure-related dissection in 1.7% and vessel perforation in 0.9% of patients who received intra-arterial treatment, but provided no detailed information on the type of treatment administered (e.g. intra-arterial thrombolysis, MT or both) [5]. In the SWIFT trial, ID was observed in 5 (3.5%) of the 144 patients treated with MT (Merci® and Solitaire® devices) [16]. These numbers are in line with the ID frequency of 2% (18/866) overall and 3.4% (18/524) seen in MT patients in our study. One of the IDs occurring in SWIFT was adjudicated to be a serious adverse event; however, detailed information on clinical and radiological characteristics and outcome of patients with ID are lacking in all of the abovementioned studies. In a study on interventional cerebral angiographies with indications other than AIS, ID occurred in 0.26–0.7% of patients, all being asymptomatic [13, 15]. In purely diagnostic cerebral angiography, ID occurred in 0–0.4% of patients, and again all were described as being asymptomatic [10‐13].

In this study 12 out of 16 dissections in the anterior circulation occurred during recanalization of a former occluded ICA. The fact that this occlusion location is the most difficult to recanalyze, has to be performed without imaging-guided navigation, normally needs the most passes and largest catheters and devices presumably leads to a higher number of IDs in these patients. Moreover, movement of the patient during or after passage of the carotid occlusion was observed in 9 patients (see selected case Fig. 2). It is well known that turning the head and neck changes the shape and curves of the carotid arteries and the turning the head maneuver has been a valuable technique in interventional neuroradiology for many years if passage of a constricted ICA is difficult.

Elongation of the ICA, especially in combination with kinking and coiling, which has been shown to be associated with carotid dissections [23], became visible after recanalization of the ICA in 10 of the 16 IDs in the anterior circulation. Considering that the damage to the vessel wall is invisible until successful recanalization of the artery, we cannot establish the exact time and reason for the dissection in all our patients; however, in our experience of more than 200 carotid artery revascularizations in the setting of AIS [24], damage to the wall can happen during the passage of the wire, transit of the catheter (Fig. 3), PTA, stent placement or retrieval of instruments used to recanalize the distal tandem occlusion. Stent placement, the last step in tandem occlusion recanalization using the upside-down technique [25], was the most likely reason for carotid dissection in 2 patients. The occluded ICA appeared to be straight with the wire or distal protection material in place, but returned to the original shape with kinking or narrow looping after removal of the aspiration or distal access catheter (Fig. 4). The linear stretching of the proximal ICA segment by the stent may aggravate the kinking at the distal end of the stent, resulting in a dissection, which was treated in our 2 cases by insertion of a second, more flexible stent.

All patients with an ID-related high-grade stenosis and 3 patients with a stenosis of <80% were treated with a stent, achieving TIMI 2–3 recanalization in all but one. Stent placement in initially low-grade stenosis was performed only if control DSA indicated progression of the dissection. During the waiting period of 15–30 min, a microwire with the tip in the true lumen distal to the ID secured rapid access and stenting in the case of an increase of the false lumen and subsequent complete occlusion of the artery. Similarly, therapy in the five SWIFT ID patients consisted of conservative management in three, and stent placement and balloon angioplasty in one each [16]. The rationale for choosing the different treatment modes, recanalization rates and patient outcomes was not reported.

In our study, patients with ID were significantly more often smokers compared to patients without ID. Smoking has not been described as a risk factor for spontaneous cervicocerebral artery dissection [26]; however, in patients with spontaneous VA dissection, smokers had cerebral ischemia more often than non-smokers [27]. In smokers endothelial function is impaired, the coagulation cascade is activated, and inflammatory cytokines are upregulated [28]. Thus, an increased vulnerability of the arterial wall to mechanical injuries in smokers is possible. Patients with ID had different AIS etiologies than patients without ID. Cardioembolic strokes were less frequent in ID patients. We hypothesize that a cardioembolic occlusion of a previously normal artery may be easier to reopen than in other subtypes of stroke. Cervicocerebral arterial dissection carries the risk for serious complications, such as cerebral ischemia or subarachnoid hemorrhage. Comparing clinical outcome and mortality in the endovascularly treated AIS patients included in this study, patients with ID did not significantly differ from patients without ID. Nevertheless, two patients died as a result of AIS complications possibly related to the ID, indicating that ID is a clinically relevant complication of endovascular treatment with mechanical devices; therefore, endovascular treatment for AIS should only be performed in institutions where such complications can be managed by staff experienced in weighing the risks and benefits of this intervention and skilled in how to manage ID.

This study is mainly limited by the single center design and the retrospective identification of ID cases based on the initial report of the interventional neuroradiologist; however, the DSA series are stored in full length in our PAC System, and two senior interventional neuroradiologists independently reviewed the DSAs of all ID patients. Another limitation may be that different endovascular recanalization techniques and devices have been used since the start of this study in 2000. On the other hand, by comparing those different techniques it became evident that most IDs occur during gaining access to the thrombus, especially if an occluded, torturous internal carotid artery has to be passed without biplane roadmap navigation. Surprisingly, no ID was observed during withdrawal of a stent retriever.

This study does not allow conclusions to be drawn on how to best treat an ID. Taking into account the benign course of IDs in most patients, “watchful waiting” seems justified prior to stent implantation.

In conclusion, ID is rare in endovascular treatment of acute ischemic stroke; however, a high index of suspicion for ID is key to its detection during the endovascular intervention, to immediately provide appropriate treatment, for instance in case of progression, and thus increase the chances of a good outcome.