Discussion

The present study suggests that partial proximal flow restoration after endovascular therapy for acute concomitant ICA–MCA occlusions enhances further post-procedure endogenous recanalization and does so mostly within 5 h. The success rate of ICA recanalization alone in concomitant ICA–MCA lesions was 70.6% (12/17). However, it is important to remember that a criterion for inclusion in this analysis was a partial or complete residual M1 thrombus which implies overall incomplete technical success. The presence of a residual M1 thrombus at the conclusion of an acute stroke intervention would in many cases be considered suboptimal. The purpose of selecting this specific post-intervention outcome was to determine whether the phenomenon of delayed recanalization seen in other methods of acute stroke therapy is seen in the endovascular therapy cohort.

We intentionally did not apply the term ‘spontaneous’ to describe delayed recanalization following endovascular treatment because the primary treatment has a direct effect on the perfusion pressure contributing to flow across pial collaterals and on the circulation of endogenous plasminogen activator and fresh substrate across the residual thrombus.11 12 Instead, the term facilitated endogenous recanalization (FER) is intended to reflect this.

In an in vitro study by Sakharov and colleagues,11 fibrin specific lysis was accelerated 10-fold by an increased arterial shear rate, as compared with no flow. They concluded that flow accelerates lysis by enhancing plasminogen circulation across the surface of the clot and thus has a positive relationship with the rate of fibrinolysis.

Stein et al confirmed in healthy humans that flow directly increases plasminogen activator release. They studied a singular vascular bed (brachial artery) in healthy human subjects and showed that infusion of various pressors, which increased the flow and arterial shear stress in that vascular bed, led to increased tissue plasminogen activator release from the endothelium.12 This study further confirmed the in vivo evidence shown nearly a decade previously which first demonstrated that cultured endothelial cells released plasminogen activator in response to the wall shear stress generated by blood flow.13

In this study of five tandem extracranial ICA–MCA occlusions, two had spontaneous MCA clot dissolution immediately following cervical CAS. This best illustrates how proximal perfusion augmentation, likely through shear stress and epithelial plasminogen release, is at least one mechanism by which endogenous recanalization occurs.

The concept of spontaneous recanalization is not novel. A previous study using serial TCD revealed that early (<6 h) and delayed (>6 h) spontaneous recanalization occurred, respectively, in 18% and 52% of patients with cardioembolic stroke in the MCA territory without thrombolytic therapy until 36 h after stroke onset.14 In the PROACT study, angiographically confirmed spontaneous recanalization at 6–8 h after stroke onset occurred in approximately 18% of placebo patients with acute MCA occlusion receiving intravenous heparin.1 15 A recent MR angiography study in 82 patients with MCA occlusion showed that there was a higher delayed recanalization rate at 24 h after symptom onset in the intravenous thrombolysis group than the control group using the Thrombolysis in Myocardial Infarction (TIMI) grade. The rates of partial (TIMI2) and complete (TIMI3) reperfusion in the no thrombolysis group were 24% and 0%, respectively, while they were 21.2% and 17.3% in the intravenous thrombolysis group.16

In our study, FER within 24 h after stroke onset occurred in more than half of all patients with a residual thrombus in the proximal MCA after experiencing successful endovascular therapy of the proximal aspect of a complex ICA thrombus. Our study differs from previous studies of spontaneous recanalization by its exclusive selection of patients not only undergoing endovascular therapy but also ineligible for or failing intravenous thrombolytic therapy.3 17–20

Another significant difference from the aforementioned studies is that our study population is inherently more heterogenous. The mechanism of embolization to the MCA in patients with tandem MCA–ICA occlusions is most likely atheroembolic which may involve an entirely different pathway of thrombus evolution and subsequent composition than that seen in cardioembolism. However, our group has not demonstrated a histopathological difference in a series of clots retrieved with the Merci device.21

Another unavoidable difference between the two different subtypes of concomitant ICA–MCA occlusions is their primary method of treatment. Carotid angioplasty and stenting was exclusively utilized in patients with tandem extracranial ICA–MCA occlusions although only half of all studied patients with such tandem lesions were treated with CAS. Implantation of a stent in the cervical ICA is obviously a markedly different modality from clot extraction with a retriever or directed intra-arterial thrombolytic infusion. There are aspects of flow dynamics and endogenous thrombolysis that we are neither equipped to analyze nor able to control for in such a small cohort of acutely ill patients.21 However, the commonality between these two subgroups of patients with combined ICA–MCA occlusions is involvement of two sequential vessels with or without an intervening, angiographically patent vessel segment. Analysis of this specific situation allows us to observe how the recanalization of a proximal lesion (or aspect of a lesion in cases of intracranial ICA–MCA occlusions since they are most often contiguous) affects a distal residual one. Analysis of such a specific group allows us to gain insight into the possible mechanisms of endogenous recanalization.

The results in the Interventional Management of Stroke study showed that AOL and TIMI score had modest agreement on the prediction of clinical 3 month outcome and a stronger association between a good AOL score and favorable outcome.2 6 We thus based our definition of FER₂₄ on the AOL score and only assessed its impact on clinical outcome at the time of discharge. Although four of the seven patients with FER₂₄ were dead or severely disabled at the time of discharge, the other three were completely symptom free whereas none of those without recanalization had good outcomes.

Because of its stronger association with good clinical outcome, we based our definition of proximal recanalization on the AOL score and only assessed its impact on clinical outcome at the time of discharge. However, we did include the overall Thrombolysis in Cerebral Ischemia (TICI) reperfusion score (table 2)22 of the residual middle cerebral arterial occlusion lesion to help further clarify how individual lesions might lend themselves to FER. Two TICI 0 residual MCA occlusions were present in our cohort and two cases of TICI I occlusions. Reperfusion was classified as TICI IIa in four and TICI IIb in the remaining four. Despite persistent MCA thrombus, good TICI scores are more likely to predict FER, such as in the case shown in figure 1.

Residual MCA thrombus at the conclusion of endovascular therapy represents incomplete revascularization and thus a technically suboptimal procedure. It is possible that cases with proximal recanalization but residual MCA thrombi included in the study represent cases where the operator felt that partial reperfusion was angiographically adequate. In order to determine if this bias existed, we also determined the reason for aborting further endovascular therapy from the angiographic report (table 2). In all but three cases, the final angiographic result was considered unsatisfactory although further reperfusion could not be accomplished for procedural reasons: (1) technical difficulties (three), (2) the maximum dose of infusion medication was reached (two), (3) procedural complications (two) and (4) further thrombectomy attempts were deemed futile (two). In two cases, no reasons for procedural termination were specified in the angiographic report (one TICI IIa after three pulls, one TICI IIb after two pulls). In only one case, no further therapy was attempted because the angiographic result was considered acceptable (TICI IIb, figure 1).

Although 11 of the 17 patients in our study were stable enough to undergo neurovascular imaging, six did so at some interval afterwards but all within 24 h. However, inconsistency in this particular variable is somewhat irrelevant since the primary outcome, facilitated clot dissolution, was detected in six of the seven patients by 5 h. We were able to qualitatively determine complete or partial FER although we did not attempt to quantify the degree of FER since the post-intervention angiographic studies were all non-invasive CT angiography or MR angiography. TICI or AOL scores are difficult to assign for these imaging modalities.

In this study, one patient was too unstable for transport for follow-up imaging to determine HT with the immediate post-procedural CT which did not demonstrate HT. The only method to determine FER was bedside TCD which continued to show no flow in the MCA after 24 h. The patient subsequently died shortly after this study. Although TCD can be much less sensitive than MR angiography or CT angiography, the TCD results were unequivocal and thus this patient was not excluded in this comparison despite the non-routine vascular imaging modality.

Although not significant, the rate of any HT was higher in those patients with successful proximal recanalization (75% vs 25%; p=0.1). Although this likely did not reach significance because of the small study size, we suspect that this is because of the high proportion of M1a patients in this cohort. The M1a pre-intervention MR pattern indicates sufficiently prolonged occlusion of the lenticulostriate ostia arising from the proximal MCA, which results in deep MCA territory infarction.23 In fact, of patients that underwent pre-intervention MRI, 90% (9/10) of the successful proximal recanalizers and 50% (2/4) of the failed recanalizers had this M1a pattern. The basal ganglia are particularly permeable following infarction and thus reperfusion puts patients with the M1a pattern at particularly high risk for HT.22 24

This study has several limitations. An aspect we could not assess was the histologic degree of distal clot maceration. In the majority of cases, the procedure was not aborted after proximal recanalization. Attempts were almost invariably made to recanalize the distal residual MCA thrombus. Our cohort's angiographic degree of residual occlusion was homogenous, as it was an inclusion criterion for analysis, but it does not reflect the microscopic clot disruption that can occur after multiple passes. Even though they may fail to recanalize, retrieval attempts may fragment a lesion sufficiently to facilitate delayed recanalization. Secondly, the modest sample size impaired our ability to detect small between group differences. Although the probability of any HT, parenchymal HT and symptomatic HT was similar with and without FER₂₄, it is possible that any difference avoided detection due to our small study numbers. Lastly, we did not analyze the robustness of each patient's pial collaterals or the size and presence of relevant communicating arteries prior to the intervention. The potential relationship between pretreatment collateral flow and subsequent FER₂₄ should be addressed in future studies.

The finding in this study suggests that restoration of proximal perfusion enhances the endogenous dissolution of downstream occlusions. Although attempts at endovascular therapy for acute large intracranial vessel occlusions should strive for complete recanalization, an operator should not be discouraged if proximal flow is restored, but the overall anatomic result is suboptimal. In addition, post-interventional vascular imaging should be performed routinely in such scenarios, as it will help guide post-intervention care, to include blood pressure goals and anticoagulant usage. In light of these findings, the next logical question that arises is whether FER₂₄ can be achieved in the more commonly encountered singular MCA trunk occlusion.