Posterior communicating and anterior communicating arteries on pre-thrombectomy computed tomography scans are associated with good outcomes irrespective of leptomeningeal collateral status

Abstract

Background

Collateral blood flow is known to be an important factor that sustains the penumbra during acute stroke. We looked at both the leptomeningeal collateral circulation and the presence of anterior and posterior communicating arteries to determine the factors associated with good outcomes and mortality.

Methods

We included all patients with acute ischaemic stroke in the anterior circulation, who underwent thrombectomy with the same thrombectomy device from 2013 to 2016. We assessed the leptomeningeal circulation by the Tan, Miteff and Maas validated scoring systems on pre-treatment computed tomographic angiography scans and looked at collateral flow through anterior and posterior communicating arteries. The results were good functional outcomes at 3 months (modified Rankin scale 0–2) and mortality.

Results

A total of 147 consecutive acute stroke patients treated with the Embotrap device were included with a median National Institutes of Health stroke scale of 15 (range 2–26). On multivariate analysis only younger age (odds ratio (OR) 0.96/year, 95% confidence interval (CI) 0.94–0.99, P = 0.026), lower National Institutes of Health stroke scale score (OR 0.87/point, 95% CI 0.80–0.93, P < 0.001), number of attempts (OR 0.80/attempt, 95% CI 0.65–0.99, P = 0.043) and the presence of a patent anterior communicating artery (OR 14.03, 95% CI 1.42–139.07, P = 0.024) were associated with good functional outcomes. The number of attempts (OR 1.66/attempt, 95% CI 1.21–2.29, P = 0.002) was significantly associated with mortality and the presence of a patent posterior communicating artery (OR 0.098, 95% CI 0.016–0.59, P = 0.011) was inversely associated with mortality.

Conclusions

Our study shows that the presence of anterior and posterior communicating arteries is significantly associated with good functional outcomes and reduced mortality, respectively, independent of the leptomeningeal circulation status.

Introduction

The outcome of the endovascular treatment of patients with large vessel occlusive (LVO) stroke is influenced by clinical and radiological parameters, both of which may be important for interdisciplinary decision making.

The impact of the cerebral collateral status, as assessed by pre-treatment imaging, is known to be a predictor of a favourable outcome in LVO of the anterior circulation. It nonetheless remains a dynamic and complex variable which has yet to be conclusively and satisfactorily evaluated with the current techniques.¹

Computed tomography perfusion (CTP) or even magnetic resonance perfusion scans are useful aids that can give surrogate markers of the collateral status. However, there are many stroke centres that still rely solely on the basic requirement of a computed tomographic angiogram (CTA) or at most a multiphasic CTA before the decision for a thrombectomy procedure is reached.

We studied the impact of the circle of Willis and its associated collateral assessments and how they affect the outcome of thrombectomy in a cohort of acute ischaemic stroke patients treated with the Embotrap revascularisation device.

Methods

A total of 147 consecutive patients with anterior circulation occlusions were treated with the Embotrap revascularisation device between October 2013 and November 2016 (Figure 1). For each patient we recorded the gender, age and neurological examination at presentation using the National Institutes of Health Stroke Scale (NIHSS) score and modified Rankin scale (mRS) score were documented. The timings and the intervals from onset of symptoms to imaging, groin puncture and reperfusion were collected. All data were prospectively stored in the registry.

Figure 1.

Flow chart showing the patient selection for this study. Consecutive acute stroke patients treated with the Embotrap thrombectomy device from October 2013 until November 2016 who had anterior circulation occlusions were included.

Imaging assessment

The completeness of the circle of Willis was assessed by two experienced readers (LLLY and SC), who used arterial maximum intensity projections (MIP) from the CTP data and who were blinded to the clinical syndromes and outcomes. Arterial MIP reconstructions were created from CTP images with the same rank as the arterial peak. More specifically, the circle of Willis and leptomeningeal collaterals were assessed retrospectively from reconstructed CTP scans which were performed at baseline before thrombectomy.

The anterior communicating artery was rated incomplete if the anterior communicating artery or A1 segment(s) ipsilateral to the stroke was hypoplastic or absent. Similarly, the posterior communicating artery was graded as incomplete if the ipsilateral posterior communicating arteries or P1 segments were hypoplastic or absent.²

Maas, Tan and Miteff grading systems were used to assess the leptomeningeal collateral system in different ways.^3–5 In particular, the Maas system compares the middle cerebral artery (MCA) area with the contralateral normal side, as an internal control, on a five-point grading scale.³ The Tan is the simplest grading system that determines if the leptomeningeal collaterals are 0, less than 50% or greater than 50% of the affected area.⁴ Finally, the Miteff system is based on the quality of the vasculature in the Sylvian fissure⁵ (see example images in the Supplementary file). We defined good collaterals as greater than 50% in the Tan grading, moderate or better vasculature in the Sylvian fissure in the Miteff grading, and equal or better collaterals than the contralateral normal side on the Maas grading.

Multimodal computed tomography protocol

All stroke examinations are performed on a 64-multidetector computed tomography (CT) scanner (Discovery CT750HD; GE Healthcare, Waukesha, WI, USA). Whole brain non-enhanced CT is performed with axial 5 mm slice thickness from the skull base to the vertex, using parameters of 120 kVp, smart AutomAs 200–300 (NI 7.6), 0.5 second rotation, 32*0.625 mm slice (collimation 20 mm) and single field of vision (SFOV) 32 cm with double field of vision (DFOV) 21 cm, bowtiefilter for ‘head’ MPR with 5 mm in axial, coronal and sagittal plane.

CTP is performed covering an 8 cm slab in 16 adjacent 5 mm slices. Patient centring is performed using non-enhanced CT anatomical references, including always basal ganglia and supraganglionar levels. The scan begins after some seconds after the administration of 65 mL of contrast agent at a rate of 5 mL/second followed by 80 mL saline flush. The parameters used are 80 kVp, 175 mA, 16 images of 5 mm thickness in two sets of eight images of 5 mm thickness (using axial shuttle technique, eight images of 5 mm per rotation), rotation time 0.4 seconds, detector coverage 40 mm, and 22 acquisitions, with a total exposure time of 17.6 seconds. These protocols are what is done generally, although they may be subject to change based on the age of the patient as well as the required area of scanning (see Supplementary file) The dataset is transferred to an image processing workstation for postprocessing after scanning. We then performed analysis of the circle of Willis and leptomeningeal collaterals by reconstructing the arterial vasculature from the CTP scans.

Clinical assessment

The primary outcomes looked at were: (a) good functional outcomes as measured by the mRS at 3 months and defined as less than or equal to 2, independently assessed for each patient by a neurologist not associated with the acute care; and (b) mortality within the 3 months as recorded by a comprehensive nationwide registry of deaths.

Results

A total of 147 consecutive patients were included from October 2013 until November 2016. Eighty-two patients were men (55.8%) and the median NIHSS at onset was 15 (range 2–26) with 71 patients (49.7%) receiving intravenous tissue plasminogen activator (tPA) before thrombectomy. The location of the occlusion was in the MCA in the majority of patients (67.3%) and in the proximal internal carotid artery (ICA) in 20 (13.6%) patients. Terminal ICA occlusions were seen in 20 (13.6%) patients and tandem occlusions in six (4.1%) patients. The onset-to-puncture time was a mean of 300.4 minutes (SD 171.2 minutes) while the puncture-to-reperfusion time was a mean of 64 minutes (SD 46.3 minutes). The median number of thrombectomy attempts was two (range 1–10). mTICI2b/3 recanalisation was seen in 125 (85.0%) patients. Good functional outcomes at 3 months were achieved in 79 patients (53.7%) and 13 patients (8.8%) were deceased at 3 months follow-up. (Table 1).

Table 1.

Baseline characteristics and results of the study population.

Characteristic	N = 147
Mean age (SD) (years)	67 (12.7)
Male sex	82 (55.8%)
Median pre-treatment NIHSS score (range)	15 (2–26)
General anaesthesia	21 (14.3%)
Intravenous tPA	71 (49.7%)
TOAST criteria
Large vessel atherosclerosis	44 (29.9%)
Cardioembolic	100 (68.0%)
Small vessel	0 (0%)
Others	3 (2.0%)
Undetermined	0 (0%)
Recent stroke	3 (2.1%)
Anterior cerebral artery (ACA) occlusion	2 (1.4%)
Internal carotid artery (ICA) occlusion	20 (13.6%)
Middle cerebral artery (MCA) occlusion	99 (67.3%)
Terminal internal cerebral artery (TICA) occlusion	20 (13.6%)
Tandem occlusion	6 (4.1%)
Onset to puncture time (SD) (minutes)	300.4 (171.2)
Puncture to reperfusion time (SD) (minutes)	64.0 (46.3%)
Tan grading good collaterals: >50% affected area	107 (73.3%)
Miteff grading good collaterals: moderate Sylvian fissure vessels	122 (83.0%)
Maas grading good collaterals: equal to contralateral side or better	46 (31.7%)
Anterior communicating artery (ACOMM) seen	139 (94.6%)
Posterior communicating artery (PCOMM) seen	68 (46.6%)
Complete circle of Willis seen	65 (44.2%)
Distal emboli	13 (8.9%)
mTICI 2 b/3	125 (85.0%)
Mortality	13 (8.8%)
Modified Rankin scale ≤2 at 3 months	79 (53.7%)
Symptomatic intracranial haemorrhage	3 (2.0%)
Number of attempts (median) (range)	2 (1–10)

Good leptomeningeal collaterals by the Tan grading system on the pre-thrombectomy scans were seen in 107 patients (73.3%). Good collaterals by the Miteff and Mass grading systems were seen in 122 (83.0%) and 46 (31.7%) patients, respectively. A visible patent anterior communicating artery was seen in 139 (94.6%) of the cohort and a patent posterior communicating artery on the affected side was seen in 68 (46.6%) patients.

On univariate analysis, younger age, low pre-treatment NIHSS scores, the resence of a patent anterior communicating artery and the number of attempts were associated with good functional outcomes (mRS 0–2) at 3 months. (Table 2) When we looked at mortality as the outcome, only a patent posterior communicating artery and the number of attempts were associated with mortality on univariate analysis (Table 3).

Table 2.

Results of univariate and multivariate analysis: predictors of good functional outcomes.

	Univariate analysis			Multivariate analysis
	mRS 0–2 (n = 79)	mRS 3–6 (n = 68)	P value	Adjusted odds ratio (95% CI)	P value
Age, mean (SD), year	64.7 (13.6)	69.6 (11.0)	0.010	0.96 per year (0.94–0.99)	0.026
Men	46 (58.2%)	36 (52.9%)	0.317
Intravenous tPA	38 (48.1%)	33 (48.5%)	0.470
Recent stroke	3 (3.8%)	0 (0%)	0.150
General anaesthesia	9 (11.4%)	12 (17.6%)	0.168
NIHSS at onset, median (range)	13 (3–24)	16 (2–26)	<0.001	0.87 (0.80–093)	<0.001
Tan good collaterals	62 (78.5%)	45 (66.2%)	0.088
Miteff good collaterals	68 (86.1%)	54 (79.4%)	0.197
Maas good collaterals	27 (34.2%)	19 (27.9%)	0.265
Patent ACOMM	78 (98.7%)	61 (89.7%)	0.019	14.03 (1.42–139.07)	0.024
Patent PCOMM	41 (51.9%)	27 (39.7%)	0.109
mTICI 2 b/3	70 (88.6%)	55 (80.1%)	0.141
Number of attempts, median (range)	2 (1–10)	3 (1–8)	0.050	0.80 (0.65–0.99)	0.043
Onset-to-puncture time, mean (SD) minutes	205 (145.5)	317 (192.4)	0.285
Puncture-to-reperfusion time, mean (SD) minutes	41 (4.4)	51 (6.1)	0.069

tPA: tissue plasminogen activator; NIHSS: National Institutes of Health stoke scale; TICI: thrombolysis in cerebral infarction scale; ACOMM: anterior communicating artery; PCOMM: posterior communicating artery. Values in bold are statistically significant (P<0.05).

Table 3.

Results of univariate and multivariate analysis: predictors of mortality.

	Univariate analysis			Multivariate analysis
	Mortality (n = 13)	Alive (n = 134)	P value	Adjusted odds ratio (95% CI)	P value
Age, mean (SD) year	66.8 (13.0)	67.9 (13.9)	0.951
Men	7 (53.8%)	75 (56.0%)	0.554
Intravenous tPA	4 (30.8%)	67 (50.0%)	0.190
Recent stroke	0 (0%)	3 (2.2%)	0.754
General anaesthesia	4 (30.8%)	17 (12.7%)	0.318
Tan good collaterals	11 (84.6%)	96 (71.6%)	0.271
Miteff good collaterals	10 (76.9%)	112 (83.6%)	0.385
Maas good collaterals	3 (23.1%)	43 (32.1%)	0.360
Patent ACOMM	12 (92.3%)	127 (94.8%)	0.532
Patent PCOMM	2 (15.4%)	66 (49.3%)	0.016	0.098 (0.016–0.588)	0.011
NIHSS at onset, median (range)	14.5 (2–21)	15 (2–30)	0.860
mTICI 2b/3	9 (69.2%)	116 (86.6%)	0.108
Number of attempts, median (range)	4 (1–8)	2 (1–10)	0.004	1.66 (1.21–2.29)	0.002
Onset-to-puncture, mean (SD) minutes	308 (169.5)	306 (203.2)	0.626
Puncture-to-reperfusion time, mean (SD) minutes	63 (46.1)	88 (47.4)	0.100

tPA: tissue plasminogen activator; NIHSS: National Institutes of Health stoke scale; TICI: thrombolysis in cerebral infarction scale; ACOMM: anterior communicating artery; PCOMM: posterior communicating artery. Values in bold are statistically significant (P<0.05).

On multivariate analysis of factors for functional outcomes at 3 months, only younger age (odds ratio (OR) 0.97 per year, 95% confidence interval (CI) 0.94–0.99, P = 0.026), lower NIHSS score (OR 0.87 per point, 95% CI 0.80–0.93, P < 0.001), number of thrombectomy attempts (OR 0.80 per attempt, 95% CI 0.65–0.99, P = 0.043) and the presence of a patent anterior communicating artery (OR 14.03, 95% CI 1.42–139.07, P = 0.024) were associated with good functional outcomes (Table 2). On multivariate analysis of factors for mortality, only the number of thrombectomy attempts (OR 1.66 per attempt, 95% CI 1.21–2.29, P = 0.002) were significantly associated with mortality, the presence of a patent posterior communicating artery (OR 0.098, 95% CI 0.16–0.59, P = 0.01) was inversely associated with mortality (Table 3).

Discussion

Our study shows that the presence of anterior and posterior communicating arteries have a significant association with good functional outcomes and reduced mortality, respectively, independent of the state of the leptomeningeal circulation.

The cerebral collateral circulation is an important structure that serves to balance and sustain the perfusion to the brain.⁶ It does this by way of a secondary network of vessels that forms connections between the main cerebral arteries. If an obstruction occurs in a cerebral artery, these collateral vessels are able to stabilise and maintain the cerebral blood flow, delaying ischaemia at least temporarily.⁷ The converse is also true when impaired collateral circulation has been shown to be associated with an increased risk of ischaemic stroke.^8.9 The cerebral collateral circulation can be divided into: the intracranial and extracranial communicating branches; the circle of Willis; and the leptomeningeal collateral vessels.¹⁰

Collateral flow coming from the external carotid artery and the leptomeningeal collaterals are considered as secondary collateral routes, which are recruited only when the primary collaterals fail.⁷ While leptomeningeal collaterals are widely acknowledged to be beneficial in acute stroke, the optimal way of assessing it is still a matter of much debate. A multiphasic CTA score has also been used to select patients for thrombectomy,¹¹ and the MRCLEAN investigators have verified that a simple four-point method dividing the collaterals into more or less than 50%, very similar to the Tan system, was associated with good outcomes.¹² Nonetheless, there is still a lack of evidence on which CTA grading method is superior. We assessed three different methods of leptomeningeal collateral assessment which had previously been validated in acute stroke patients treated with intravenous recombinant tissue-type plasminogen activator.^3–5 In our study none of the three methods were associated with good outcomes at 3 months or mortality on multivariate analysis, we suspect this may be due to the moderate sample size or the exclusion of poor collaterals due to selection bias, i.e. only patients with good collaterals are transferred from other hospitals for thrombectomy.

The circle of Willis is the source of several important pathways for collateral blood flow to the ischaemic brain and is thought to protect against future ischaemic stroke by providing increased blood flow.¹³ However, it is unlikely that the circle of Willis exclusively evolved for the purpose of compensating for vessel occlusions, because pathological conditions that occur in the much later parts of life are less likely to be able to determine the path of evolution.¹³ Even from an anatomical haemodynamic perspective, the communicating arteries are too small for effective blood supply in many cases. It has been suggested that the circle of Willis is instead mainly used to equalise pressure in the brain and dissipate excess force.¹³

Nonetheless, asymmetry of the circle of Willis results in asymmetric blood flow and portions of the circle of Willis can frequently be hypoplastic or absent, which has been shown to increase the risk of stroke by 5.4–7.0-fold.^14–16 Studies suggest an association between the completeness of the circle of Willis and the potential risk of stroke for patients with symptomatic ICA stenosis or occlusion,^17–21 or even in patients without prior cerebrovascular disease.⁶ To our knowledge, the benefit of the completeness of a circle of Willis has not been reported in patients undergoing thrombectomy for acute stroke. In our study we show that the presence of both the anterior and posterior communicating vessels are important for better outcomes, regardless of how we measured the leptomeningeal collateral circulation.

The region supplied by the posterior circulation may receive blood from the anterior circulation through the posterior communicating arteries and vice versa.²⁰ However, the impact of a small posterior communicating artery may not simply be explained by an enhanced potential for collateral flow due to the presence of compensatory routes, as a study has shown that even in the absence of internal carotid occlusion, posterior communicating artery hypoplasia is an independent contributor to the risk of ischaemic stroke.²² The regions of the brain supplied by the anterior and posterior communicating arteries are different. For example, the anterior communicating artery tends to supply larger regions of the contralateral MCA territory, which is less often supplied by the posterior communicating artery and may explain why the presence of a patent posterior communicating artery was not associated with good functional outcome. Conversely, basilar strokes are known to have a greater association with mortality which explains why the presence of a posterior communicating artery but not the anterior communicating artery is inversely associated with mortality.²³

We acknowledge the limitation of this being a single centre study with its inherent bias, although the imaging was core-lab assessed. We did not attempt to quantify the degree of blood supply from the anterior or posterior communicating artery. The size of our sample may be responsible for the inability of the leptomeningeal collateral assessment or recanalisation status to reach significance, contrary to previous reports.¹² Similarly, the high rate of good collaterals suggests a bias in treating patients with favourable leptomeningeal collaterals which could also explain why leptomeningeal collaterals were not predictors of good outcomes. Our study had a higher than expected incidence of anterior communicating arteries present (see Supplementary file), this could be due to a selection bias whereby patients who were chosen for thrombectomy had better perfusion status. Finally, the limited sample size did not allow us to analyse if our findings could be extrapolated in patients with only MCA or only ICA occlusions. On the other hand, the strengths of our study are the core-lab assessment and the uniform thrombectomy device use which reduces the number of variables in the study.

Conclusions

Aside from known prognostic factors, such as baseline NIHSS and age and the number of attempts, the presence of good collateral flow based on the anatomical presence of anterior and posterior communicating arteries was a strong predictor for a favourable clinical outcome and reduced mortality, respectively.

Author contribution

Leonard Yeo was responsible for concept of the study, writing the draft, data analysis and editing of the manuscript. Ake Holmberg was responsible for data gathering and data analysis. Anastasios Mpotsaris was responsible for data analysis and editing of the manuscript. Michael Söderman was responsible for concept of the study, data gathering and editing of the manuscript. Staffan Holmin was responsible for data gathering and editing of the manuscript. Åsa Kuntze Söderqvist was responsible for data gathering and editing of the manuscript. Jens Kolloch was responsible for data gathering and editing of the manuscript. Pervinder Bhogal was responsible for data gathering and editing of the manuscript. Benjamin Tan was responsible for editing of the manuscript. Anil Gopinathan was responsible for editing of the manuscript. Yang Cunli was responsible for editing of the manuscript. Vamsi Gontu was responsible for data gathering and editing of the manuscript. Tommy Andersson was responsible for concept of the study, data gathering and editing of the manuscript. Patrick A Brouwer was responsible for concept of the study, data gathering and editing of the manuscript. Sandra Cornelissen was responsible for concept of the study, writing the draft, data analysis and editing of the manuscript.

Data sharing

Further patient data from the registry is available upon email request.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Tommy Andersson is a consultant for Ablynx, Amnis Therapeutics, Medtronic, Cerenovus/Neuravi, Rapid Medical and Stryker. Patrick Brouwer is a consultant for Medtronic, Stryker, Cerenovus/Neuravi and BALT. Michael Söderman is a consultant for Cerenovus and Archer Research and has a scientific collaboration with Philips Healthcare. Leonard Yeo has received substantial grant funding from the National Medical Research Council (NMRC), Singapore, substantial grant funding from Ministry of Health (MOH), Singapore, and a moderate grant funding from I²R, A-STAR, Singapore. Pervinder Bhogal is a consultant for Neurvana and Phenox.

Ethics approval

Institutional ethics review board approval for this study was obtained from the regional ethical committee in Stockholm (Regionala etikprövningsnämnden i Stockholm) Diarienr: 2016/1041-31/4.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.