Acute ischemic stroke with tandem/terminal ICA occlusion - CT perfusion based case selection for mechanical recanalization
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.158211
Source of Support: None, Conflict of Interest: None
Background: Rapid reperfusion in a patient with a favorable penumbral pattern is crucial to achieving a good outcome in acute ischemic stroke. Recanalization rates for tandem and terminal internal carotid artery (ICA) occlusion are better with endovascular management as compared with intravenous tissue plasminogen activator (IV-tPA) alone. We hypothesize that tissue-based selection would enable the identification of the ideal patient most suited for reperfusion therapy. We present our series of patients who developed tandem or terminal ICA occlusion and were selected for endovascular management based on their computed tomography (CT) perfusion (CTP) imaging.
Keywords: Acute ischemic stroke; CT perfusion; internal carotid artery; mechanical thrombectomy
Proportional recanalization rates are better with endovascular management as compared with intravenous thrombolysis irrespective of the site of occlusion in the anterior or posterior circulation. This is particularly the case with large vessel occlusion, where recanalization rates with endovascular management are high at 39.1% and 52% for terminal internal carotid artery (T-ICA) and basilar artery (BA) occlusion, respectively.  This is significantly better than the reported 4% achieved with intravenous tissue plasminogen activator (IV-tPA) in both the groups, and is precisely the argument in support of the hypothesis favoring endovascular management in acute stroke. ,,, In contrast, the open - labeled, randomized, phase 3, interventional management of stroke (IMS) III trial was stopped prematurely because of its futility. Although thrombolysis in cerebral infarction (TICI) 2b/3 reperfusion was achieved in 38% and 44% of patients with T-ICA and M1 segment of middle cerebral artery (MCA) occlusion, respectively, this did not translate into a good clinical outcome. The investigators concluded that patients who underwent endovascular treatment were not more likely to be functionally independent as compared with those who received IV-tPA alone.  A similar conclusion was drawn from the MR RESCUE and the SYNTHESIS trials. , Failure to demonstrate good outcomes in the endovascular trials was attributed to three key reasons, namely, lack of stringent tissue-based patient inclusion criteria, a delay in reperfusion therapy, and, the limited use of the newer stent retrievers. Nonetheless, post hoc analysis of a subset of patients with baseline and 24-h computer tomography angiography (CTA) from the IMS III trial revealed significantly higher recanalization rates and better clinical outcomes with endovascular management in the select group of patients with tandem ICA-M1 occlusion (T-ICAM-o) and terminal carotid L- or T-type occlusion.
Recent randomized trials, however, have shown that stent retrievers achieved substantially better angiographic, safety and clinical outcomes when compared with the Merci retrieval system.  MR CLEAN, an open-labeled, randomized trial has shown an absolute benefit of 14% with endovascular management. Although TICI 2b/3 reperfusion was achieved in 59%, a good outcome was observed only in 33% of the patients. This can largely be explained by a delay in the onset-to-groin puncture by an average duration of 3 h and lack of rigorous tissue-based patient selection criteria.  Therefore, the key to achieving a good outcome is rapid reperfusion in a patient with a small ischemic core and a large penumbra. However, the challenge rests in determining the patient group that will benefit from early reperfusion therapy. Henceforth, an imaging modality that can reliably determine the infarct core and penumbral volume, a reflection of collateral perfusion, is crucial to achieving a good outcome. Perfusion imaging, and in particular core evaluation based on cerebral blood volume (CBV) assessment, has shown to reliably predict the final infarct core and outcome. We hypothesize that a tissue-based selection would enable the identification of the ideal patient most suited for endovascular reperfusion therapy. We present our series of patients with tandem or terminal ICA occlusion who were selected for the endovascular management based on their computed tomography (CT) perfusion (CTP) imaging.
The data was collected prospectively on consecutive patients admitted to the emergency department within 12 h of the symptom onset and diagnosed with acute ischemic stroke from T-ICAM-o occlusion or terminal ICA occlusion over a period of 3 years from January 2011 to March 2014. All patients had a non-contrast CT (NCCT) head followed by CTP and CTA of the head and neck. The institutional review board has approved the study. The patients were evaluated by a team of neurologists, interventionists and intensivists and commenced on IV-tPA if eligible; and, emergently shifted to a dedicated biplanar neuro-angiography unit as per the pre-defined protocol [Figure 1].
CT perfusion protocol
Scanning was performed using a Siemens SOMATOM Definition AS + scanner (Erlangen Germany; scanning length 96 mm, duration 45 s). Thirty-five milliliters of Iomeprol 350 was injected at a flow rate of 6 mL/s with a scan delay time of 4 s. An extended scanning time of 60 s was used in patients with a reduced cardiac output. The perfusion software uses an optimized deconvolution analysis to post-process raw data and generate color maps for flow, volume, time to drain and mean transit time. A favorable penumbral pattern was defined as an infarct core less than 70% of the total tissue at risk and a CBV ASPECTS of >5.
CTA of the head and neck
CTA of the head and neck was obtained following an injection of 90 cc of non-ionic contrast in the antecubital vein at a rate of 3-5 cc/s. T-ICAM-o was defined as an occlusion or severe stenosis (>90%) involving the proximal ICA accompanied by a proximal M1 MCA occlusion. Two patterns of terminal ICA occlusion were defined: L-type, in which the occlusion starts in the terminal ICA proximal to the bifurcation and extends either into the proximal M1 MCA or A1-A2 segment of ACA; and, the T-type in which occlusion starts in the terminal carotid and extends to the M1 MCA and the A1-A2 segment of the ACA.
Procedure details and emergency carotid stenting protocol
General anesthesia was administered only in patients with an airway compromise and a distal-to-proximal approach was undertaken in patients with tandem occlusion. ,,,, With its long sheath (Cook medical, Bloomington, USA.) in the common carotid artery to provide stability, the guiding catheter (Penumbra reperfusion catheter/Neuron, Penumbra, Alameda, California, USA.) was navigated to the distal cervical ICA following mechanical thrombo-aspiration of the clot in the proximal ICA. Thereafter, a mechanical thrombectomy was performed using either SOLITAIRE 4 mm × 40 mm (ev3 Inc, Irvine, California, USA.) stent or Penumbra aspiration microcatheter as per the standard technique. Balloon guiding was not used due to its non-availability during the period of the study. Carotid artery stenting was performed if there was a significant residual stenosis before or after the procedure, depending on the operator's preference. The anti-platelet regimen varied depending on whether IV-tPA was administered. The initial and terminal half-life for recombination tPA is 4-5 min and 40 min, respectively. In patients who received IV-tPA, aspirin 300 mg was given at the time of stent delivery. Clopidogrel 300 mg was administered if post-procedure Dyna CT showed no evidence of hemorrhagic transformation. In the IV-tPA-naïve group, aspirin 300 mg and clopidogrel 600 mg were administered at the time when carotid stenting was considered.
The key clinical and imaging parameters as described in [Table 1] were documented. Symptomatic ICH was defined as hemorrhagic transformation associated with a drop in National Institute of Health Stroke Scale NIHSS by four points. A modified Rankin scale (mRS) score of 2 or less at 3 months was considered as a good outcome.
Fourteen (29.16%) of the 48 patients treated endovascularly for acute ischemic stroke had either T-ICAM-o or terminal ICA occlusion. Majority were women (57.14%), and the median age was 56 years (age range: 33-70 years). The key clinical and imaging parameters are presented in [Table 1]. Stent retrievers were used in most (85.7%, n = 14) of the patients and carotid stenting was performed in four (28.5%, n = 14) patients. A distal to proximal approach, wherein cerebral perfusion was established prior to stenting, was undertaken in all but one patient (representative cases: [Figure 2],[Figure 3],[Figure 4] and [Figure 5]).
The ASPECTS was lower than the initial NCCT ASPECTS by a median 2 points. In the TICI 2b/3 reperfusion cohort, the median ASPECTS on the initial NCCT, CBV map and 24-h NCCT were 9, 7.5 and 6.5, respectively.
Intravenous tPA administration
Six patients (42.8%, n = 14) received IV tPA; two (28.5%, n = 7) in the terminal ICA group and four (54.2%, n = 14) in the tandem ICA group. In both patients with terminal ICA occlusion, meaningful reperfusion was not achieved and they died. However, all patients in the tandem ICA group had a good outcome and TICI 2b/3 reperfusion was observed in three (75%, n = 4) patients. The median imaging-to-IV-tPA administration time was 135 (90-170) min.
Reperfusion and outcome
In the T-ICAM-o group, TICI 2b/3 reperfusion and a good outcome were observed in five (71.42%, n = 7) and six patients (85.71%, n = 7), respectively. One patient with TICI 2a reperfusion had a good outcome owing to prominent leptomeningeal collaterals from the ACA. In the terminal ICA occlusion group, TICI 2b/3 reperfusion and good outcomes were observed in three (42.8%, n = 7) and two patients (28.5%, n = 7), respectively. Reperfusion was the single most important outcome predictor, and, overall, good outcome was seen in 57.1% (n = 14) of the patients. General anesthesia was used in 12 (85.7%, n = 14) patients.
A NCCT brain was performed at 24 h to evaluate for hemorrhagic transformation. Parenchymal hematoma type 2 was seen in one patient (7.1%, n = 14) who received IV-tPA for right T-ICAM-o and was on dual anti-platelet therapy for carotid artery stenting.  This was, however, not associated with a drop in NIHSS by ≥4 points and the patient had an mRS of 2 at 6 months.
Two patients with TICI 2a reperfusion and decompressive hemicraniectomy (DH) had an mRS of 4 at 90 days. The third patient with terminal ICA occlusion developed a large infarct and died in spite of TICI 2b reperfusion. There was truncation in the tissue density curves and the observed CBV ASPECTS of 6 is likely to be erroneous. In fact, the infarct core could have been larger than presumed and the poor collaterals could have resulted in rapid progression of the infarct core prior to the reperfusion.
A median of 2 (range: 1-4) passes per device was performed. A single device was used in all and four passes were attempted in two patients. In one patient with terminal ICA occlusion, an accidental stent deployment occurred. TICI 2a flow was observed with an mRS of 4 at 3 months.
Failure to demonstrate benefit with endovascular management in acute stroke in the earlier endovascular trials, particularly the IMS III trial, can be attributed to many factors. First and foremost, patient selection was not based on the core and penumbral assessment from multimodal imaging. Secondly, a significant delay noted in the initiation of intra-arterial therapy was an important contributory factor as a 30-min deferral in the initiation of therapy can result in a 10% decrease in the functional independence. Finally, majority of the patients received intra-arterial tPA infusion as the standard endovascular intervention, with stent retrievers being used only in five patients. With evolution in technology and widespread use of stent retrievers, MR CLEAN, a recent randomized trial, has shown a significant absolute benefit of 14% with endovascular management in acute ischemic stroke.
Robustness of imaging predictors of outcome
Rapid reperfusion in a patient with a small infarct core is the key to achieving good outcomes. Various methods to assess infarct core volume have been described. An ASPECTS score of 0-7 on a NCCT head or CTA-source imaging (SI) strongly predicted a poor outcome.  However, NCCT is less sensitive to early ischemia and there is poor inter-rater agreement in the initial few hours after the symptom onset. This can be explained by an image noise ranging between 2 and 3.5 Hounsfield units (HU) even in high-quality scanners and a drop in CT Hounsfield unit (HU) of only 1.3 in the initial 2.5 h from stroke onset.  Similarly, estimation of the infarct core on CTA-source imaging (SI) can be misleading and erroneous, as the short scanning times with the modern multidetector row CT scanners do not allow for a steady state to be achieved between the arterial and tissue contrast and provide flow rather than volume-weighted images.  Moreover, a CTA that is not time resolved is insensitive to incomplete occlusion, resulting in imprecise assessment of collateral status. The collateral assessment in the distal occlusions can also be challenging. 
Perfusion imaging has emerged as a promising tool in reliably determining the core volume. The CBV ASPECTS performed better than the NCCT ASPECTS in determining outcomes, with a reported sensitivity and specificity of >80%. Similarly, with diffusion-weighted brain imaging as the gold standard, the sensitivity with which CBV maps determined a regional infarct was over 96% as compared to 80% with NCCT. Furthermore, concordance between the initial CBV and the final NCCT ASPECTS was >0.9 as compared to 0.4 with respect to the initial and the final NCCT ASPECTS. Additionally, there was a significant mean difference of 1.6 between the CBV and NCCT ASPECTS, and an excellent inter-rater agreement (interclass coefficient >0.9) was noted with CBV ASPECTS. ,
The median ASPECTS on CBV was lower than that on the initial NCCT by 2 in our cohort. A good concordance between the CBV and the 24-h NCCT ASPECTS, with a median difference of 1 was seen in the TICI 2b/3 reperfusion group. On the contrary, the median difference between the initial and final NCCT ASPECTS was high at 2.5. The CBV ASPECTS was a good tool in estimating the extent of core at presentation in our patient group.
Thresholds for perfusion imaging
Newer deconvolutional softwares effectively correct for the delay and dispersion, and a long segment coverage in a short scanning time can be achieved with the modern multi-detector row CT [MDCT] scans. Various thresholds for relative blood flow and volume have been studied to accurately estimate the volume of tissue at risk as well as the ischemic core.  Likewise, values ranging from 1.2 to 3 for the ratio of the volume of tissue at risk to the core have been used to define a favorable penumbra. , A higher ratio would, however, result in exclusion of a considerable proportion of patients. Henceforth, in our center, we use a ratio of 1.4, which equates to a core volume of 70% or less as this corresponded to a favorable clinical response in approximately 55% of the patients.  There was an excellent agreement in the decision making and lesion volume estimation based on both the qualitative color maps and the quantitative measure. In none of the patients was the decision altered when the qualitative color maps were used.  Therefore, we used qualitative color maps to estimate the core and the penumbral volume.
Reperfusion and outcome
Early reperfusion is an independent predictor of a good outcome. The recanalization rate with IV-tPA was approximately 4% in patients with either terminal ICA or BA occlusion.  Likewise, in T-ICAM-o, the recanalization rates and good outcomes with IV-tPA was approximately 3/5 th and 2/5 th as that of the isolated M1 occlusion (IMO), and has been attributed to inadequate drug delivery to the M1 segment. Post hoc analysis of a subset with T-ICAM-o and CTA at baseline and 24 h in the IMS III trial revealed good outcomes in 42.86% of the patients managed with endovascular intervention as compared with none treated with IV-tPA. In our T-ICAM-o cohort, TICI 2b/3 reperfusion and a good outcome were seen in 71.4% and 85.7% of patients, respectively. The strikingly better results that we observed, in our opinion, is attributable to stringent perfusion imaging-based patient selection for treatment. TICI 2b/3 reperfusion and good outcomes were seen in 42.8% and 28.5% of patients with terminal ICA occlusion, and was comparable to that seen in the IMS III trial. A poor recanalization rate with carotid terminal occlusion has been attributed to inadequate leptomeningeal collateralization and mechanical clot compaction due to the higher perfusion pressure proximal to the occlusion. A large clot size and its burden, that are key determinants for the site of occlusion, along with lack of balloon-guiding catheters that allow for proximal flow arrest, are the likely explanations for the less than expected recanalization rate seen in our cohort.  In the tandem ICA group, our results contrast with that obtained in the MR CLEAN trial. Although the group of patients with extracranial ICA occlusion were likely to benefit from endovascular management as compared with standard medical therapy, this was not statistically significant in the trial. The likely explanation for this disparity and failure to demonstrate favorable results was the lack of stringent tissue-based selection criterion. In our observation, CBV ASPECTS, a reliable measure of infarct core at presentation, was crucial to the significantly better results observed in our cohort.
The median imaging-to-puncture time of 70 min in our series was shorter than that reported from high-volume centers, and we attribute this to strict adherence to the institutional stroke protocol.  Similarly, a short median puncture-to-first stent deployment of 20 min and a median puncture-to meaningful reperfusion of 68.5 min was observed. Even in the group where carotid artery stenting was performed, the median time to achieve meaningful reperfusion was 90 min. None of our patients had a symptomatic ICH. In our opinion, rapid reperfusion was an additional key factor that may have lead to good outcomes in our cohort.
Rapid reperfusion is the key to a good outcome, and this is evident from the recently published ESCAPE trial, wherein the prime inclusion factor was the ability to achieve reperfusion within 90 min of the study imaging.  A favorable outcome was seen commonly in patients who had received IV-tPA prior to intervention in the MR CLEAN trial. In our opinion, the benefits outweigh the serious adverse events with IV-tPA and we follow a "drip and ship" (commence IV-tPA in all eligible patients and immediately shift the patient to the neurovascular suite) policy to avoid delay in achieving recanalization.
The standard set protocol, delivering an effective dose of 5.3 mSv used for perfusion scanning is well below the dose that can cause local organ injury.  Post-processing was performed using optimized deconvolutional software to overcome the effects of delay and dispersion. A limitation of this study is that the results from our single-center, small sample cannot be generalized.
Perfusion imaging-based patient selection along with rapid reperfusion was the key to achieving significantly better outcomes in our ICA occlusion patient cohort as compared with that reported in previous studies. Proportional recanalization rates for proximal occlusion are significantly better with endovascular therapy as compared with IV thrombolytic therapy. Terminal ICA clots pose a unique problem attributable to a large clot burden and mechanical clot compaction, and better strategies to retrieve clots from this location need to be defined to achieve higher reperfusion rates. Finally, strict adherence to the institutional stroke protocol was the key to achieving rapid reperfusion in our cohort.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]