The Pattern of Leptomeningeal Collaterals on Computed Tomography Angiography is a Strong Predictor of Long Term Functional Outcome in Stroke Patients with Large Vessel Intracranial Occlusion

Abstract

Background and purpose

The role of non-invasive methods in the evaluation of collateral circulation has yet to be defined. We hypothesized that a favorable pattern of leptomeningeal collaterals, as identified by computed tomography angiography (CTA), correlates with improved outcomes.

Methods

Data from a prospective cohort study at two university based hospitals where CTA was systematically performed in the acute phase of ischemic stroke were analyzed. Patients with complete occlusion of the intracranial internal carotid artery (ICA) and/or the middle cerebral artery (MCA-M1 or M2 segments) were selected. Leptomeningeal collateral pattern was graded as a three category ordinal variable (less, equal, or greater than the unaffected contralateral hemisphere). Univariate and multivariate analyses were performed to define the independent predictors of good outcome at 6 months (mRS≤2).

Results

196 patients were selected. The mean age was 69 ±17 years and the median NIHSS score was 13 (IQR 6-17). In the univariate analysis, age, baseline NIHSS, pre-stroke mRS, ASPECT score, admission blood glucose, history of hypertension, coronary artery disease, congestive heart failure, atrial fibrillation, site of occlusion, and collateral pattern were predictors of outcome. In the multivariate analysis, age (OR 0.95; 95%CI [0.93-0.98], p=0.001), baseline NIHSS (OR 0.75; [0.69-0.83], p<0.001), pre-stroke mRS (OR 0.41; [0.22-0.76], p=0.01), intravenous r-tPA (OR 4.92; [1.83-13.25], p=0.01), diabetes( OR 0.31; [0.01-0.98], p=0.046) and leptomeningeal collaterals (OR 1.93; [1.06-3.34], p=0.03) were identified as independent predictors of good outcome.

Conclusion

Consistent with angiographic studies, leptomeningeal collaterals on CTA are also a reliable marker of good outcome in ischemic stroke.

INTRODUCTION

Proximal intracranial arterial occlusion (PIAO) is independently associated with poor functional outcomes and high mortality rates^1-3. Age, baseline NIHSS score, and the initial volume of CT hypodensity have been identified as important prognostic variables in patients suffering this devastating disease⁴. The presence of robust collateral flow on conventional angiography has been linked to improved clinical outcomes and reduced infarcts volumes^5-7. Although frequently evoked in clinical discussions, little is known about the pathophysiology of the cerebral collateral circulation and its relation with other important stroke predictors.

Conventional angiography provides the most complete and reliable information about Circle of Willis and leptomeningeal collaterals and is, therefore, considered the gold standard for collateral flow assessment. However, the applicability of catheter angiography is limited by its invasive nature and associated risks. Indirect assessment of collaterals can be accomplished by non-invasive methods including transcranial doppler (TCD), computed tomography angiography (CTA), and magnetic resonance angiography (MRA)⁸. TCD and MRA may provide important information about the status of the Circle of Willis collaterals but do not possess enough spatial resolution to evaluate the more distal leptomeningeal vascular bed. CTA, alternatively, encompasses a higher degree of anatomic resolution and can, more accurately, depict the leptomeningeal collaterals (Figure 1). Few previous studies have used CTA to assess the degree of collateral circulation to the ischemic tissue ^8-11.

Figure 1.

A : 51 years old male, initial NIHSS of 12, mRS at 6 months: 1. B: 47 years old female, initial NIHSS of 13, mRS at 6 months: 6.

There are advantages of using CTA in the triage of acute stroke patients¹². CTA is widely available, non-invasive, and provides a rapid assessment of the intra and extracranial vasculature depicting with high accuracy vessel stenoses and occlusions. We hypothesized that a favorable pattern of leptomeningeal collaterals as visualized on CTA correlates with improved functional outcomes in acute ischemic stroke patients.

SUBJECTS AND METHODS

We analyzed data from 741 consecutive patients enrolled in a prospective cohort study at two university-based hospitals, the Screening Technology and Outcomes Project in Stroke (STOPStroke), in which admission non-enhanced CT scans (NCCT) and CTA were obtained in all patients suspected of having ischemic stroke (stroke, transient ischemic attack, or stroke mimics) in the first 24 hours of symptom onset. Patients were excluded if iodinated contrast agent administration was contraindicated (i.e., history of contrast agent allergy, pregnancy, congestive heart failure, increased creatinine level) or if there was evidence of intracranial hemorrhage on NCCT. The STOPStroke study received institutional review board approval at both participating institutions and was Health Insurance Portability and Accountability Act compliant.

Data on clinical history, laboratory results, demographics, stroke risk factors, and pre-stroke modified Rankin Score (mRS) were collected on all patients at baseline by direct interview or by review of the medical chart by trained staff. NIHSS scores were obtained at baseline as part of patient admission work-up. Time to hospital arrival was calculated as the amount of time elapsed between the onset of symptoms (last time seen normal for non witnessed symptom onset patients) and the time of arrival to the emergency medicine department. Time to CTA was calculated in a similar fashion. Follow-up mRS was obtained at 6 months. For the present study, patients with acute complete occlusion of the intracranial internal carotid artery (intracranial ICA) and/or M1 and/or M2 segments of the middle cerebral artery (MCA) were selected. Patients with bilateral, and/or posterior circulation strokes were excluded from the analysis.

Image Protocol and Review

The STOPStroke NCCT and CT angiographic protocol is described elsewhere¹³. Image review was independently performed on a picture archiving and communication system workstation (Impax; AGFA Technical Imaging Systems, Richfield Park, NJ) by a board-certified neuroradiologist and a clinical neurologist experienced in stroke imaging interpretation. Disagreements in readings were resolved by consensus. Reviewers were blinded to follow-up clinical and imaging findings but had information in regard to the patients’ age, sex, and presenting clinical symptoms. Neither of the reviewers had participated in the selection of the patients. Variable window width and center level settings were used for optimal ischemic hypoattenuation detection with NCCT and CTA images¹⁴. In all cases, NCCT images obtained for acute stroke were reviewed first, followed by CTA images. Reviewers rated the ischemic lesion on the NCCT scan images according to ASPECTS¹⁵. For every image, each of 10 regions were reviewed for the presence or absence of ischemic lesions according to a five point level of certainty score (score 5, definitely present; score 4, probably present; score 3, equivocal; score 2, probably absent; score 1, definitely absent).For the ASPECTS, just the regions assigned with scores 4 or 5 were used. After the CTA, readers assessed the presence of complete occlusion of the ICA and MCA by thrombus according to the same certainty scale. Just scores 4 or 5 were used as evidence of complete ICA or MCA occlusion as well. Leptomeningeal vascularity was graded in the following scores: 1, absent; 2, less than the contralateral unaffected side; 3, equal to the contralateral unaffected side; 4, more than the contralateral unaffected side; 5, exuberant. Because of the very small number of patients in the extreme scores, the scale was collapsed into three ordinal groups: less than contralateral unaffected side (score 1-2), equal to contralateral unaffected side (score 3) and greater than contralateral unaffected side (scores 4-5).

Statistical Analysis

All statistical analysis was performed using SPSS software (version 17.0, SPSS Inc., Chicago, IL). The following baseline clinical variables were included: age, gender, comorbidities (previous history of hypertension, congestive heart failure, coronary artery disease, diabetes, atrial fibrillation and prior history of stroke or transient ischemic attack), smoking and alcohol consumption, baseline NIHSS score, pre-stroke mRS score (continuous and dichotomized: ≤2 vs. >2), systolic blood pressure (continuous and dichotomized: <150 vs. ≥150 mmHg), glucose levels (continuous and dichotomized: <140 vs. ≥140 mg/dL), ASPECT score (dichotomized: ≤ 7 and > 7), IV thrombolytic use, IA therapy, time to hospital arrival and time to CTA. Good clinical outcome was defined as an mRS≤2 at 6 months follow-up.

Continuous variables are reported as mean ± standard deviation (SD) or as median ± interquartile range (IQR). Categorical variables were reported as proportions. Baseline characteristics were compared among the different leptomeningeal collateral patterns. Differences in continuous variables were assessed by one-way ANOVA or the Kruskal-Wallis test in case of non-normally distributed data. Differences between proportions were assessed by the χ2 test. We also compared mortality, both in-hospital and at follow-up, according to the pattern leptomeningeal collaterals.

Univariate analysis was used to test the association between different variables and the outcome (follow-up mRS). Multivariate logistic regression with backward elimination (p-value for elimination of 0.1) was used to identify independent predictors for good outcome (follow-up mRS ≤2). Variables significantly associated with a favorable outcome in the univariate analysis (p<0.1) were included in the multivariable model. Given their strong clinical association with clinical outcomes, we a priori planned to force the following variables into the model: age, baseline NIHSS, IV r-tPA , IA therapy, pre-stroke mRS and the site of intracranial occlusion. The pattern of leptomeningeal collaterals was also forced into the model since it was the predictor of interest. The site of intracranial occlusion and the pattern of leptomeningeal collaterals were tested as ordinal variables in the regression analysis. In order to evaluate the importance of letpomeningeal collaterals in untreated patients, the same model was also used in a subset analysis including just cases with no IV r-tPA and/or IA intervention. A two-sided p-value < 0.05 was considered significant.

RESULTS

Of the 741 patients include in the database, 196 subjects were identified as having isolated anterior circulation PIAO involving a single hemisphere and follow-up at 6 months available. The mean age was 69 ±16.7 years. Eighty four (43%) were males. The majority of the study population was composed by Caucasians (158 subjects - 81%). The median NIHSS at admission was 13 (IQR 6-17). One hundred and three patients (53%) had an ASPECT score greater than 7 (Table 1).

TABLE 1.

Baseline characteristics of the population

Baseline Characteristics	N = 196
Age (years, mean±SD)	69 (±17)
Male sex	84 (43%)
Caucasian Race	158 (81%)
Baseline NIHSS (median, IQR)	13 (6–17)
Baseline mRS ≤2	177 (90%)
Baseline ASPECTS (median, IQR)	8 (6–10)
Baseline ASPECTS >7	103 (53%)
Systolic BP (mmHg, mean±SD)	155 (±29.8)
Systolic BP ≥150mmHg	105 (54%)
Admission glucose ( mg/dl, median, IQR)	122 (107–146)
Admission glucose ≥ 140mg/dl	60 (31%)
Anti-platelets	61 (31%)
IV TPA	61 (31%)
IA thrombolysis	22 (11%)
Hypertension	116 (59%)
Diabetes	32 (16%)
Dislipidemia	58 (30%)
Atrial fibrillation	63 (32%)
Congestive heart failure	21 (11%)
Coronary artery disease	45 (23%)
Alcohol	24 (12%)
Smoking	56 (29%)
Prior s trok e or TIA	32 (16%)
Site of intracranial occlusion
Intracranial ICA	43 (22%)
MCA-M1	89 (45%)
MCA-M2	64 (33%)
Pattern of leptomeningeal collaterals
Less than contralateral hemisphere	45 (23%)
Equal than contralateral hemisphere	96 (49%)
Greater than contralateral hemisphere	55 (28%)

Forty eight (24%) subjects were treated with intravenous r-tPA, 9 subjects (5%) received intra-arterial therapy, and 13 (7%) had combined intravenous and intra-arterial therapy. The median 6-month mRS of the study cohort was 3 (IQR 1-6). At six months, 82 subjects (42%) had achieved a good outcome (mRS≤2) and 49 subjects (28%) had died. Most of fatalities (67%, 33/49) occurred during the hospitalization.

According to the pattern of leptomeningeal collaterals, 45 subjects (23%) were graded as less (score 1-2), 96 subjects (49%) were graded as equal (score 3) and 55 subjects (28%) were graded as greater (score 4-5) when compared to contralateral unaffected hemisphere. There was no significant difference in the time from stroke onset to CT scan across the different patterns of leptomeningeal collaterals (p=0.6). Subjects with equal or greater CTA leptomeningeal collaterals had higher baseline ASPECT score as compared to subjects with less CTA leptomeningeal collaterals (p=0.02). Subjects with greater leptomeningeal collaterals had lower systolic blood pressure and lower prevalence of hypertension. In general, subjects with greater leptomeningeal collaterals had lower prevalence of other cardiovascular risk factors as well, but none of these reached statistical significance (Table 2).

TABLE 2.

Comparison of patients according to the patterns of collaterals

Characteristics	Pattern of Collaterals			P
	Less than contralateral hemisphere (N = 45)	Equal than contralateral hemisphere (N = 96)	Greater than contralateral hemisphere (N = 55)
Age (mean ±SD)	70 (±16)	71 (±16)	64 (±17)	0.07†
Gender (male)	23 (52%)	39 (41%)	22 (40%)	0.37*
Race (caucasians)	31 (70%)	81 (84%)	45 (82%)	0.15*
Baseline NIHSS (median-IQR)	15 (11-18)	12 (5-17)	12 (5-16)	0.06‡
Baseline ASPECTS (>7)	15 (34%)	55 (57%)	32 (58%)	0.02*
Syst. BP mmHg (mean ±SD)	157 (±31)	159 (±30)	145 (±26)	0.01†
Syst. BP (≥ 150mmHg)	25 (57%)	56 (58%)	23 (42%)	0.13*
Admission glucose mg/dl (median -IQR)	125 (108-145)	124 (107-151)	114 (104-137)	0.15‡
Admission glucose (≥ 140mg/dl)	14 (32%)	33 (34%)	12 (22%)	0.26*
Antiplatelets	9 (20%)	37 (38%)	15 (27%)	0.08*
Hypertension	28 (64%)	63 (66%)	24 (44%)	0.02*
Diabetes	9 (20%)	18 (19%)	5 (9%)	0.22*
Dislipidemia	15 (34%)	28 (29%)	15 (27%)	0.75*
Atrial fibrilation	16 (36%)	33 (34%)	14 (25%)	0.43*
Congestive heart failure	6 (14%)	13 (13%)	2 (4%)	0.13*
Coronary artery disease	9 (20%)	27 (28%)	9 (16%)	0.23*
Alcohol	6 (14%)	14 (15%)	4 (7%)	0.40*
Smoking	12 (27%)	34 (35%)	10 (18%)	0.08*
Prior stroke or TIA	8 (18%)	14 (15%)	10 (18%)	0.79*
Time to CTA - hours (median-IQR)	3 (1-9)	4 (2-11)	4 (1-12)	0.60‡

^*Chi-square test

^†One-way ANOVA

^‡Kruskal-Wallis test;

There was a statistically significant difference in the rates of in-hospital mortality: 33% (14 subjects), 13% (12 subjects), and 13% (7 subjects) in patients graded as less, equal, and greater leptomeningeal collaterals, respectively (p=0.01). At 6 months, subjects with equal and greater leptomeningeal collaterals tended to have lower mortality (22% and 20% respectively) when compared with subjects with less leptomeningeal collaterals (39%, p=0.06).

In the univariate analysis, younger age (p<0.001), lower pre-stroke mRS (p<0.001), lower baseline NIHSS, (p<0.001), ASPECTS >7 on admission CT (p=0.02), lower admission glucose (p=0.04), equal or greater (as opposed to less) leptomeningeal collaterals (p=0.002) as well as the absence of history of hypertension (p=0.01), atrial fibrillation (p=0.001), congestive heart failure (p=0.01), and coronary artery disease (p=0.04) were significantly associated with good outcome at 6 months. Patients with occlusion of the M2 segment of the MCA had also significantly higher odds for good outcome when compared to subjects with intracranial ICA occlusion (Table 3).

TABLE 3.

Univariate analysis of predictors of good functional outcome (mRS ≤ 2) at 6 months

Characteristics	Outcome		OR (95% CI)	P
	Good (N = 82)	Poor (N = 114)
Age - years (mean ±SD)	62 (±17)	74 (±15)	0.95 (0.94-0.97)	< 0.001†
Gender (male)	40 (49%)	44 (39%)	1.5 (0.8-2.7)	0.15*
Race (caucasians)	70 (85%)	88 (77%)	1.7 (0.8-3.6)	0.15*
Pre Stroke mRS (median-IQR)	0	0 (0-1)	0.4 (0.25-0.65)	<0.001‡
Pre Stroke mRS (≤ 2)	82 (100%)	95 (83%)	-	< 0.001*
Baseline NIHSS (median-IQR)	7 (4-12)	16 (12-19)	0.8 (0.75-0.86)	< 0.001‡
ASPECTS (>7)	51 (62%)	52 (46%)	1.9 (1.1-3.5)	0.02*
Syst. BP mmHg(mean ±SD)	151 (± 26)	157 (± 32)	0.99 (0.98-1.0)	0.13†
Syst. BP (≥ 150mmHg)	42 (51%)	63 (55%)	0.8 (0.5-1.5)	0.58*
Admission glucose mg/dl (median-IQR)	116 (102-140)	125 (110-152)	0.98 (0.98-1.0)	0.04‡
Admission glucose (≥ 140mg/dl)	22 (27%)	38 (33%)	0.7 (0.4-1.4)	0.33*
Anti-platelets	28 (34%)	33 (29%)	1.2 (0.7-2.3)	0.44*
IV TPA	25 (30%)	36 (32%)	0.9 (0.5-1.8)	0.87*
IA thrombolysis	7 (8%)	15 (13%)	0.6 (0.2-1.6)	0.31*
Hypertension	40 (49%)	76 (67%)	0.5 (0.3-0.8)	0.01*
Diabetes	9 (11%)	23 (20%)	0.5 (0.2-1.1)	0.09*
Dislipidemia	24 (29)	34 (30%)	1.0 (0.5-1.8)	0.93*
Atrial fibrillation	16 (19%)	47 (41%)	0.3 (0.2-0.7)	0.001*
Congestive heart failure	3 (4%)	18 (16%)	0.2 (0.1-0.7)	0.01*
Coronary artery disease	13 (16%)	32 (28%)	0.5 (0.2-0.9)	0.04*
Alcohol	12 (15%)	12 (10%)	1.5 (0.6-3.4)	0.39*
Smoking	29 (35%)	27 (24%)	1.8 (0.9-3.3)	0.07*
Prior stroke or TIA	9 (11%)	23 (20%)	0.5 (0.2-1.1)	0.09*
Time to hospital arrival - hours (median-IQR)	2 (0.5-6)	3 (1-10)	0.98 (0.95-1.01)	0.16‡
Site of intracranial occlusion
Intracranial ICA	12 (15%)	12 (15%)	Reference	0.06
MCA-M1	37 (45%)	37 (45%)	1.8 (0.8-4.0)	0.13
MCA-M2	31 (27%)	33 (40%)	2.7 (1.2-6.3)	0.02
Pattern of collaterals
Less than contralateral hemisphere	10 (12%)	34 (30%)	Reference	0.01
Equal than contralateral hemisphere	41 (50%)	55 (49%)	2.5 (1.1-5.7)	0.02
Greater than contralateral hemisphere	31 (38%)	24 (21%)	4.4 (1.8-10.6)	0.001

^*chi-square test

^†independent samples t-test

^‡Mann-Whitney-U test;

In the multivariate analysis, age (OR 0.95; [0.93-0.98], p=0.001), baseline NIHSS score (OR 0.75 [0.69-0.83], p<0.001), intravenous r-tPA (OR 4.92, [1.83-13.25], p=0.002), pre-stroke mRS (OR 0.38 [0.22-0.76], p=0.01), previous history of diabetes (OR 0.31 [0.01-0.98], p=0.046) and the pattern of leptomeningeal collaterals (OR 1.93 [1.06-3.50], p=0.03) were significantly associated with good outcome at the 6-month follow-up (Table 4). When the multivariate analysis was applied to the untreated patients, only age (OR 0.94 [0.91-0.98], p=0.01), admission NIHSS score (OR 0.75 [0.67-0.84], p<0.001), and the pattern of leptomeningeal collaterals (OR 2.89 [1.25-6.67], p = 0.01) were significantly associated with a favorable outcome (Table 5).

TABLE 4.

Multivariate analysis of predictors for good outcome

Characteristics	OR (95% CI)	P
Age (years)	0.95 (0.93-0.98)	0.001
Baseline NIHSS	0.75 (0.69-0.83)	< 0.001
IV TPA	4.92 (1.83-13.25)	0.002
IA thrombolysis	1.39 (0.32-5.99)	0.66
Pre-stroke mRS	0.38 (0.20-0.74)	0.01
Diabetes	0.31 (0.01-0.98)	0.046
Site of intracranial occlusion*	1.82 (1.00-3.31)	0.05
Pattern of leptomeningeal collaterals*	1.93 (1.06-3.50)	0.03

^*tested as ordinal variables in the logistic regression model

TABLE 5.

Multivariate analysis of predictors for good outcome (no IV r-tPA and /or IA thrombolysis)

Characteristics	OR (95% CI)	P
Age (years)	0.94 (0.91-0.98)	0.01
Baseline NIHS (median)	0.75 (0.67-0.84)	<0.001
Pre Stroke mRS	0.48 (0.23-1.01)	0.05
Diabetes	0.20 (0.04-1.01)	0.06
Site of intracranial occlusion*	1.73 (0.82-3.65)	0.15
Pattern of leptomeningeal collaterals*	2.89 (1.25-6.67)	0.01

^*tested as ordinal variables in the logistic regression model

DISCUSSION

In the present study, a favorable pattern of leptomeningeal collaterals, as measured by CTA, was associated with improved functional outcomes at 6 months in a cohort of patients with acute anterior circulation PIAO. In a multivariate model, robust leptomeningeal collaterals remained an independent predictor of good long term outcomes along with younger age, lower pre-stroke mRS and baseline NIHSS scores, administration of IV r-tPA, and absence of diabetes. More proximal intracranial occlusions showed a strong trend towards worst outcomes. When the analysis was restricted to patients not treated with IV r-tPA and/or endovascular intervention, there was an even stronger association between the degree of leptomeningeal collaterals and good outcomes. Notably, the pattern of leptomeningeal collaterals on CTA was also associated with lower in-hospital mortality and a trend towards lower mortality at 6 months.

PIAO is associated with poor functional outcomes and high mortality rates^{1, 2}. Indeed a previous analysis of the STOPStroke database demonstrated that the presence of PIAO was one of the strongest predictors of good outcomes (OR 3.33 [0.24-0.45], p<0.001) and mortality (OR 4.5 [2.7-7.3], p<0.001) at 6 months³. Other studies have demonstrated that age and baseline NIHSS scores are amongst the most potent predictors of outcome in patients with PIAO^{4, 16, 17}. Even though there is no dispute about the importance of collateral circulation in the setting of PIAO, its physiology and relationship with other clinical variables remains largely unknown⁸. In our study, a history of hypertension was more frequently found amongst patients with fewer CTA leptomeningeal collaterals. We also found higher mean systolic blood pressure levels in patients with less or equal leptomeningeal collaterals. Liebeskind et al. reported similar results showing that angiographic collateral flow was inversely associated with pretreatment systolic blood pressure and history of hypertension in stroke patients undergoing thrombectomy¹⁸. Other factors, such as statins, have been also associated with increased collateralization in acute stroke but were not evaluated in the present study¹⁹.

Collateral circulation likely improves neurological outcome by limiting the extent of brain infarction. As demonstrated by the strong association between CTA collateral pattern and baseline ASPECT scores, our results support the notion that adequate collateral perfusion is essential to the viability of the penumbral tissue. Good collateral circulation has been also linked to higher recanalization rates^{20, 21}. Although we might simply argue in favor of sparing of the penumbral tissue, collateral flow may increase availability of the fibrinolytics to the clot and facilitate its dissolution or, as recently proposed, it may affect revascularization by altering the degree of thrombus impaction into the cerebrovasculature^{16, 22}. Notably, we could not find a difference in time from stroke onset to CTA between patients with good and poor collaterals. This persistence of collateral flow might explain the findings of a recent MRI study which demonstrated similar occurrence rates of diffusion-perfusion mismatch in patients with large vessel strokes who were imaged within less than 9 hours versus 9-24 hours from stroke onset²³. Despite its high spatial resolution, CTA provides very little information about flow itself. As a consequence, CTA may lead to an overestimation of the strength of the collateral circulation. Indeed, a much higher percentage of patients were classified as having good leptomeningeal collaterals in our CTA study as compared to previous angiographic studies²⁴.

Only a few studies have previously used CTA leptomeningeal vascular pattern as a surrogate for collateral flow and even fewer have linked CTA leptomeningeal vascular pattern to long term outcomes¹⁷. In a univariate analysis using a subset of patients with MCA occlusions, the STOPStroke investigators have previously demonstrated a relationship between poor CTA collaterals and in hospital worsening¹¹. Rosenthal et al reported that CTA leptomeningeal collaterals had only a minimal positive impact in the outcomes of patients who did not achieve recanalization and no impact in the outcomes of patients who were completely recanalized.⁹ Tan et al. reported that good CTA collaterals correlated with improved outcomes in uni- but not in multivariate analyses¹⁰. In our study, the effect of the pattern of leptomeningeal collaterals on the outcome of untreated patients (presumably with a higher proportion of non-recanalized patients) was higher as compared to the entire cohort. Our study is the largest one to evaluate the association of leptomeningeal collaterals as identified by CTA and long term outcomes. Probably due to methodological differences (different level of occlusions, incorporation of complete occlusions only, higher number of patients, etc), our results differ from some of the previous studies.

Our study has some limitations. As a retrospective analysis there is a potential for bias in the selection of the subjects. The use of a prospective collected cohort contributed to minimize this problem. The status of recanalization, an important predictor of outcome in ischemic stroke, was not assessed in our study. Finally, our patient cohort was composed primarily by a Caucasian population and this may limit the generalizability of our findings. Conversely, the exclusion of patients with partial occlusions, bilateral and/or posterior strokes contributed to the homogeneity of the cohort and made the comparisons between groups more powerful.

As conventional angiography, an invasive approach, is impractical for most patients with ischemic stroke, the study of non-invasive methods, such as CTA, as a surrogate marker of collateral circulation may lead to a better understanding of its determinants and find potential treatments to enhance collateral flow to the ischemic tissue. Consistent with angiographic studies, leptomeningeal collaterals on CTA are also a reliable marker for good outcome in acute ischemic stroke patients presenting with PIAO. The acquisition of this information, with no extra processing time, may greatly assist in the selection of patients who are potential candidates for reperfusion therapies.