29 Failed Mechanical Thrombectomy: What to Do Next
29.1 Case Description
29.1.1 Clinical Presentation
A 73-year-old male with history of atrial fibrillation presented initially to the emergency department (ED) in an outside hospital following acute kidney injury and urinary retention. He was off his dabigatran for 2 to 3 weeks and was scheduled for transurethral resection of the prostate. While in hospital, the patient subsequently developed global aphasia and right-sided weakness. He was last seen normal at 7:10 p.m.
Noncontrast computed tomography (NCCT) of the brain was performed, with no evidence of acute large vascular territory infarct, acute intracranial hemorrhage, mass, or mass effect. Initial NCCT of the head recorded an ASPECTS score of 10. CT angiography was also performed at the outside hospital, which revealed a left supraclinoid ICA occlusion.
A telestroke consult was requested. The patient was transferred to our stroke center for further evaluation and treatment.
At time of arrival and assessment at our hospital, he was 4 hours from symptom onset. His National Institutes of Health Stroke Scale (NIHSS) score was 26.
29.1.2 Imaging Workup and Investigations
A repeat NCCT of the brain demonstrated no early or established ischemic changes and an ASPECTS score of 10 (Fig. 29.1a). There was no evidence of intracranial hemorrhage. CT angiography performed from the level of the aortic arch to the vertex confirmed occlusion of the left supraclinoid internal carotid artery (ICA) and left posterior cerebral artery (PCA) origin (Fig. 29.1b) with filling of the left middle cerebral artery (MCA) via patent anterior communicating artery (AComm). CT perfusion (CTP) scan was also performed, which revealed no core infarct and a very large area at risk (penumbra) involving the entire left anterior cerebral artery (ACA), MCA, and PCA territories (Fig. 29.1c–e).
Intracranial left ICA occlusion.
As he was within 4.5 hours of symptom onset, the patient received intravenous tissue plasminogen activator (IV-tPA) in the ED. Emergency multidisciplinary discussion took place with clear indications for emergent mechanical thrombectomy due to high NIHSS, the presence of a large vessel occlusion (LVO), and the absence of acute infarct of NCCT. The details of the procedure, risks, and benefits were discussed with the patient’s next of kin and written informed consent was obtained.
8-Fr short angiographic sheath.
5-Fr vert catheter 125 cm.
0.035 angled Terumo hydrophilic wire.
Fathom 16 microguidewire.
ACE 068 aspiration catheter.
Solitaire Platinum 6 × 40 mm stent retriever.
Trevo XP 4 × 40 mm stent retriever.
ACE 060 aspiration catheter.
8-Fr FlowGate balloon guide catheter (BGC).
8-Fr Angio-Seal closure device.
6-Fr Neuron Max 088.
The procedure was performed with conscious sedation. Local anesthetic (1% lidocaine) was administered at the left groin puncture site. A 19-gauge, single-wall, left common femoral arterial puncture was performed by palpation. An 8-Fr short angiographic sheath was placed for vascular access. A 6-Fr Neuron Max 088 was advanced to the left ICA over a 125-cm 5-Fr Vert catheter over an angled Terumo guidewire. Angiography confirmed the presence of a left ICA terminus thrombo-occlusion (Fig. 29.2a).
A Penumbra ACE068 aspiration catheter was then prepared with Velocity microcatheter and Fathom 016 microwire.
Under fluoroscopic and roadmap mask guidance, a Penumbra ACE068 aspiration catheter was advanced to the face of the clot in the supraclinoid ICA over a Velocity microcatheter and a Fathom 016 microwire. Aspiration was applied using the Penumbra pump for 4 minutes. During aspiration, no blood flow was seen in the catheter. Aspiration thrombectomy catheter was unsuccessful. Control angiography showed persisting occlusion at the supraclinoid ICA. Aspiration thrombectomy with the ACE068 catheter was repeated a second time, again unsuccessfully. Control angiography showed persisting occlusion at the supraclinoid ICA.
In a third attempt at thrombectomy the left M1 segment MCA was selectively catheterized by the Velocity microcatheter over the Fathom microwire. The Penumbra ACE068 aspiration catheter was brought to the face of the clot over the Fathom microwire and Velocity microcatheter. The wire was removed and a Solitaire Platinum 6 × 40 mm stent was unsheathed into the left M1 MCA and supraclinoid ICA. After 5 minutes had elapsed for clot integration into the stent retriever, aspiration pump suction was applied to the aspiration catheter, and the stent retriever and aspiration catheter were withdrawn as a unit from the patient, while applying suction using a 60-mL syringe to the guide catheter in the ICA. No clot was noted outside the patient. Control angiography showed persisting occlusion at the supraclinoid ICA.
A fourth attempt at thrombectomy was performed, this time using a Trevo stent retriever. The left M1 segment MCA was selectively catheterized by the Velocity microcatheter over the Fathom microwire. The Penumbra ACE068 aspiration catheter was brought to the face of the clot over the Fathom microwire and Velocity microcatheter. The wire was removed and a Trevo XP 4 × 40 stent retriever was unsheathed into the left M1 MCA and supraclinoid ICA using the “push and fluff” technique. After 5 minutes had elapsed for clot integration into the stent retriever, aspiration pump suction was applied to the aspiration catheter, and the stent retriever and aspiration catheter were withdrawn as a unit from the patient, while applying suction using a 60-mL syringe to the guide catheter in ICA. No clot was noted outside the patient. Control angiography showed persisting occlusion at the supraclinoid ICA.
The Neuron Max 088 guide catheter was then removed.
An 8-Fr Flow Gate was advanced to the left ICA over a 125-cm 5-Fr Vert catheter over an angled Terumo guidewire.
A fifth attempt at thrombectomy was performed using a stent retriever and BGC. Under roadmap mask guidance, the left M1 segment MCA was selectively catheterized with a Velocity microcatheter over a Fathom 016 wire. The wire was removed. A Solitaire Platinum 6 × 40 stent retriever was then unsheathed into the left M1 MCA and supraclinoid ICA. After 5 minutes, the balloon on the below the guide catheter was inflated and the stent retriever was withdrawn into the guide catheter, while applying suction using a 60-mL syringe to the guide catheter in the ICA. No clot was noted outside the patient. The balloon catheter was deflated. No clot was noted outside the patient. Control angiography showed persisting occlusion at the supraclinoid ICA.
A sixth attempt at thrombectomy was performed using an aspiration catheter and BGC. Under roadmap mask guidance, an ACE060 aspiration catheter was brought to the face of the clot in the supraclinoid ICA over a Terumo guidewire. The wire was removed. With the aspiration catheter remaining in this position, aspiration was applied using the penumbra pump for 5 minutes. Aspiration thrombectomy was then attempted during inflation of the BGC in the cervical ICA, and a 60-mL syringe suction applied to the BGC during aspiration catheter withdrawal into the guide catheter. No clot was noted outside the patient. The balloon catheter was deflated. Control angiography showed persisting occlusion at the supraclinoid ICA.
Finally, a seventh attempt at mechanical thrombectomy was made. The left M1 segment MCA was selectively catheterized by the Velocity microcatheter over the Fathom microwire. The Penumbra ACE060 aspiration catheter was brought to the face of the clot over the Fathom microwire and Velocity microcatheter. The wire was removed and a Solitaire Platinum 6 × 40 mm stent was unsheathed into the left M1 MCA and supraclinoid ICA. After 5 minutes had elapsed for clot integration into the stent retriever, aspiration pump suction was applied to the aspiration catheter. The balloon catheter was inflated, and the stent retriever and aspiration catheter were withdrawn as a unit from the patient, while applying suction using a 60-mL syringe to the guide catheter in the ICA. No clot was noted outside the patient. The balloon-guided catheter was deflated. Control angiography showed persisting occlusion at the supraclinoid ICA and modified treatment in cerebral infarction (mTICI) score of 0 (Fig. 29.2b).
At this point, after seven failed attempts at mechanical thrombectomy 3.5 hours into the procedure and approximately 8 hours after symptom onset, further attempts at mechanical thrombectomy were considered to be approaching futility and increased risk. Placement of a detachable stent retriever was considered; however, there was a concern regarding the risk of dual antiplatelet therapy in the setting of potential for hemorrhagic infarct, given the large penumbra on the CTP scan. After discussion with the patient’s family, the procedure was aborted.
After an angiographic run of the left common femoral artery (CFA), hemostasis of the left groin was achieved with 8-Fr Angio-Seal closure device.
The patient’s clinical condition remained unchanged throughout the procedure, with no change in neurological status pre and immediately postprocedure.
During his hospital stay, the patients physical examination revealed gaze preference, global aphasia, homonymous hemianopia, and facial and upper as well as lower extremity right-sided weakness.
NCCT of the brain was performed 24 hours postprocedure (Fig. 29.2c). This demonstrated a large MCA territory infarct involving most of the MCA territory as well as partial ACA infarcts of the paramedian frontal lobe, and partial PCA territory infarction of the occipital lobe and left cerebral peduncle. No hemorrhagic conversion was noted. The patient showed further neurological deterioration in the next 1 to 2 days with development of sluggish left pupil, without new intracranial hemorrhage or significant midline shift. Supportive care was provided to the patient, with the family electing to pursue palliative care shortly afterward.
Much of the clinical benefit of mechanical thrombectomy is accounted for by the effectiveness of the latest generation of stent retrievers and aspiration catheters. Successful reperfusion (defined as mTICI 2b or greater) can routinely be achieved up to 85% of the time with these devices. In recent randomized controlled trials of mechanical thrombectomy, however, 15 to 40% of stent-retriever mechanical thrombectomies failed to achieve substantial reperfusion (mTICI 2b or 3), and 8 to 18% had minimal or no reperfusion (mTICI 0–1). 1 , 2 , 3 , 4 , 5 Moreover, poor outcomes were still seen in some patients with substantial reperfusion due to delays and/or difficulties encountered during endovascular treatment. These issues highlight the need for further improvements in devices, techniques, and workflows.
From a technical standpoint, mechanical thrombectomy may fail for a variety of reasons: difficult vascular access, thrombus composition, device–thrombus interactions, or the formation of distal emboli.
Anatomical challenges may be encountered anywhere along the course of a catheter in accessing the intracranial clot. Femoral artery access is the traditional method of approach for mechanical thrombectomy procedures. The presence of peripheral artery disease may render femoral artery access more difficult or impossible (aortoiliac occlusive disease). The angle of the origin of the great vessels and the aortic arch often becomes more acute with age. In addition, substantial cervical arterial tortuosity may develop with age, especially in the setting of chronic hypertension. 6 Such anatomical difficulties may render selective catheterization of the great vessels and/or make catheter placement more difficult, requiring different catheter shapes, wires, and exchange techniques, or impossible. Furthermore, in such anatomical configurations, the forces required to navigate intermediate catheters and microcatheters intracranially may cause guide catheter herniation and loss of access. 6 In a study by Ribo et al, 6 predictors of difficult carotid access were older than 75 years, hypertension, dyslipidemia, and left carotid catheterization. Difficult access can cause significant delays in time to revascularization, decreased revascularization rates, and worse outcomes.
Various approaches have been promoted to bypass such scenarios, including direct carotid puncture and radial artery access. Direct carotid access can be safely achieved using ultrasound guidance. In a series of seven patients undergoing direct carotid puncture after multiple unsuccessful femoral attempts lasting from 20 to 90 minutes, carotid access was achieved in 15 minutes or less, and mTICI 2b–3 reperfusion was achieved in 6 (86%) patients within 7 to 49 minutes of access. 7 Achieving hemostasis after thrombectomy, however, is the major drawback, as a closure device for carotid puncture does not exist. Manual compression of a carotid puncture can cause a new thromboembolus to form, and the formation of a large neck hematoma may result in respiratory compromise or cranial nerve injury. Another drawback is the absence of devices with shorter lengths. 8
Radial artery puncture has been used with increasing frequency; however, this approach may not always allow improved access to aortic arch vessels and does not permit use of larger guide catheters.
Another anatomical obstacle to thrombectomy not infrequently encountered is the presence of a tandem steno-occlusive lesion in the carotid artery. This situation is covered in depth in another chapter of this book. In a retrospective single-center study by Kaesmacher et al, M2 MCA occlusions were at higher risk of reperfusion failure than other clot locations (adjusted OR = 3.36; 95% CI, 1.82–6.21). 9
Mechanical thrombectomy procedures can also fail due to device–thrombus interactions. Up to 20 to 30% of thrombi may be resistant to removal using current generation retrieval devices. There are a number of forces interacting during mechanical thrombectomy between the device, the thrombus, and the vessel wall, as well as procedural techniques that can influence these forces.
The pressure differential between blood flow at the face of the clot and retrograde blood flow on the distal end of the clot is one such interaction, which may explain the decreased recanalization rates seen in patients with poor collaterals. 10 Another is the combined force of friction and adhesion between the thrombus and the vessel wall. Recent studies have demonstrated decreased revascularization rates of hypodense, fibrin-rich thrombi. 11 In vitro experiments using thrombi of varying fibrin and red blood cell (RBC) proportions have shown that fibrin-rich thrombi (< 20% RBC content) have a significantly higher coefficient of static friction. 12 Furthermore, longer thrombi would be expected to have increased friction and adhesion, given their larger surface area for thrombus–vessel interaction. Some thrombi can be quite hard, making passage of a microwire and/or microcatheter beyond the clot difficult or impossible. In one retrospective series of 72 consecutive patients with failed mechanical thrombectomy, this was seen in 15 of 72 (21%). 13
Techniques to overcome clot inertia (combined friction/adhesion and impaction forces) include maximizing clot integration by the stent retriever or aspiration catheter. For stent retrievers, device–thrombus interaction is maximized by positioning the device, so that the active element is deployed across the entire length of the thrombus (the proximal marker of a stent retriever should be positioned proximal to the clot). Clot–device interaction can be increased with closed-cell design stent retrievers (Trevo, Stryker) by using the “push and fluff” technique. This involves pushing the wire to deploy the device, rather than unsheathing it, as well as pushing forward on the catheter during deployment to cause the device to expand into the clot 14 . A higher rate of first-pass reperfusion and of complete reperfusion (mTICI 3) was seen using this technique compared to unsheathing of the Trevo. 15
The ASTER and COMPASS trials have shown that, as a first-line therapy, both ADAPT technique using aspiration catheters and stent retrievers achieve similar rates of successful reperfusion. In the ASTER study, there was a higher rate of conversion to another device or technique with the first-line ADAPT compared to stent retriever (32.8 vs. 23.8%, respectively, p =0.053). 16
Use of a cervical BGC, which markedly reduces the pressure head on the face of a thrombus, not only allows for more effective retrieval but also decreases thrombus fragmentation and distal embolization compared to a traditional cervical guide catheter. 17 , 18 A recent meta-analysis of five nonrandomized studies of 2,022 patients (1,083 BGC group and 939 non-BGC group) was reported by Brinjijki et al. 19 Patients treated with BGCs had higher odds of first-pass recanalization, mTICI 3, mTICI 2b/3, modified Rankin scale (mRS) score of 0 to 2, lower odds of mortality, lower mean number of passes, and shorter procedure times. 19 In ASTER, 92% of the stent-retriever cases utilized a BGC versus none of the ADAPT cases. 20 The ASTER trial results showed no difference in revascularization rates between primary ADAPT approach, primary stent retriever, and BGC; however, the better clinical outcomes for stent retrievers, although not statistically significant, may be related to the decreased frequency of distal emboli with BGC use. 16
Stent retrievers may also be used in combination with aspiration catheters to maximize clot retrieval forces at both the face of the clot and along the length of the clot. This technique is also advantageous in tortuous intracranial anatomy by optimizing the angle at which the stent retriever is withdrawn (“line of force”) and avoiding undesirable traction on the artery itself. 21 Recent studies have emerged combining stent-retriever thrombectomy, local aspiration, and BGC, with promising early results. 22
Thrombus–device interactions also occur during and after multiple attempts. Static friction is greater than kinetic friction; therefore, once the retrieval is begun, device withdrawal should be steady until the thrombus is removed from the body. Also, each thrombectomy attempt has the potential for causing compression of the thrombus, which also increases the friction between the thrombus and vessel wall, making subsequent retrieval more difficult. 20 In addition, with each subsequent pass, the cumulative risks of the procedure increase, with potential for endothelial damage and vessel perforation.
Intra-arterial thrombolysis was one of the first strategies employed in stroke treatment. Currently, recombinant tPA is sometimes used in combination with mechanical thrombectomy, either initially to “soften” the thrombus or, more commonly, to recanalize distal fragments which were originally present or dislodged during mechanical thrombectomy. There is a theoretical increased risk of postprocedure hemorrhage after intra-arterial tPA (IA-tPA), especially if the patient has been treated with IV thrombolysis or if injecting into an already infarcted territory; however, at low dosages (< 10 mg), there have not been reports of significant increases in hemorrhage rates. Intracranial arterial administration of a GPIIb/IIIa inhibitor is most commonly used in the setting of thromboembolic complications of aneurysm coiling or intracranial stenting. 23 The efficacy, safety, timing, route of administration, and dosage in the setting of stroke thrombectomy needs to be evaluated in larger studies. 24
Currently, there is no clear data or consensus on which device should be used if the first one or more passes is unsuccessful during mechanical thrombectomy. More complicated techniques may take longer and may be problematic, especially for novice operators. There is no absolute cut-off for the number of passes during mechanical thrombectomy that should be performed prior to aborting the procedure. With each thrombectomy pass, however, there is potential for thrombus compression and increasing difficulty of subsequent retrieval, as well as increased risk of vessel dissection and perforation. Moreover, as the length of the procedure and number of thrombectomy attempts increases, the probability of a good outcome decreases. 25 This can be due to increased clot friction, collateral failure and infarct growth, or increased rate of complications.
Various rescue techniques have been described for refractory cases. Crossing Y-Solitaire thrombectomy technique is one rescue strategy described by Aydin et al in ten cases after three to five unsuccessful attempts. 26 Deployment of two crossing Solitaires into the limb of each M2 branch led to complete recanalization in 80% (after one pass) or 100% of cases (after two passes). Minor-moderate reversible spasm was noted in 50% of cases and minor hemorrhage was noted in 20%. At 90 days, the mRS score was 0 to 3 in all cases. The authors speculate that increased strut surface contact, improved thrombus integration, and engagement of thrombus may contribute to the success of this strategy. Permanent implantation of intracranial stents after multiple failed attempts has also been described in several reports. 27 , 28 , 29 This approach is discussed in further detail in a separate chapter in this book.
In some patients, a delay in successful reperfusion can have the same result as a technical failure of clot removal. Delays can be preprocedural or intraprocedural. Preprocedural delays can be caused by patient transfer (initial presentation to a center without thrombectomy capabilities), delayed stroke team activation, placement of a Foley catheter, and unnecessarily waiting for laboratories or next of kin for consent.
Intraprocedural delays may relate to use of general anesthesia, lack of standardized protocols and defined staff roles during mechanical thrombectomy procedures, poor coordination between technologist, nursing, anesthesiologist, neurologist, and/or neurointerventional team members, use of general anesthesia, poor device selection, and catheter incompatibility (see Chapter 6).
Patient outcomes after failed mechanical thrombectomy are poor. Supportive medical therapy should be administered to minimize infarct growth, and recognize and prevent hemorrhagic transformation and malignant edema. Younger patients pose a particular challenge after failed LVO mechanical thrombectomy due to the limited space for cerebral edema. It is also important to discuss the procedural outcome and prognosis with the patient’s family after a failed mechanical thrombectomy procedure.
29.2.1 Postprocedure Care
Blood pressure management after failed mechanical thrombectomy is critical to minimize infarct growth, while avoiding hemorrhagic conversion of infarcted brain. Generally, a systolic blood pressure (SBP) below 180 mm Hg is targeted. Patients require close monitoring, and should be admitted to an appropriate level of postprocedural care, such as a stroke unit, high-dependency unit or intensive care unit. After failed mechanical thrombectomy, patients may develop large vascular territory infarcts. In younger patients, a high level of vigilance is required for the possible development of malignant cytotoxic edema within 2 to 3 days, since this can result in life-threatening herniation syndromes. Depending on patient factors and preferences, consideration may be given to decompressive craniectomy in a subset of these patients. In older patients, the combination of more advanced cerebral volume loss and large MCA territory infarcts may not result in significant mass effect. Discussion with family regarding prognosis and resultant significant disability are important. Mortality rates following large vascular territory ischemic stroke are higher, due to increased frequency of complications related to poor airway protection, dysphagia, limited mobility, and institutionalization. The goals of care should be established with next of kin keeping in mind patient’s advanced directive or previously expressed wishes.