Collaterals and Intra-Arterial Treatment Effects

Study Questions:

What is the effect of collateral vessel status, measured by computed tomography angiography (CTA), on functional outcome after intra-arterial stroke treatment?


This was a post hoc, investigator initiated analysis of the MR CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment of Acute Ischemic Stroke in the Netherlands) study. MR CLEAN was a randomized controlled trial of intra-arterial treatment (IAT) compared with standard care after anterior circulation acute ischemic stroke in patients with occlusion seen on CTA. For the current study, two independent neuroradiologists blindly reviewed centrally stored CTA data to determine a collateral score, graded on a 4-point scale, ranging from no collaterals to good collaterals. The primary outcome was the modified Rankin (mRS) score (a measure of functional outcome) at 90 days. Multivariable ordinal logistic regression was used to determine an adjusted common odds ratio (acOR) for a shift toward a better outcome, as measured by the mRS. Adjustments were made for age; stroke severity; time from onset to randomization; history of prior stroke, diabetes, or atrial fibrillation; and carotid terminus occlusion. Agreement on collateral grading was measured by a weighted kappa.


This analysis includes 493 patients, representing 99% of the patients in MR CLEAN. There was moderate agreement (weighted kappa 0.60) among the neuroradiologists grading collateral status. Baseline characteristics were similar among patients with all four collateral grades except for the following groups having worse collaterals: older age, male sex, higher stroke severity, diabetes, hyperlipidemia, and more early ischemic changes on baseline noncontrast head CT. The effect of IAT was modified by collateral status (p = 0.038). There was a gradation of this effect, as it was most pronounced in patients with the best collaterals (acOR, 3.2; 95% confidence interval [CI], 1.7-6.2), followed by those with moderate collaterals (acOR, 1.6; 95% CI, 1.0-2.7). In patients with poor collaterals (acOR, 1.2; 95% CI, 0.7-2.3) and absent collaterals (acOR, 1.0; 95% CI, 0.1-8.7), there was no significant effect. Collateral status did not impact symptomatic hemorrhage rate. In patients treated with IAT, there was no association between collateral status and recanalization rate, but there was an association between collateral status and infarct volume—in that patients with better collaterals had less tissue infarcted. Patients with worse collateral status had a higher mortality rate than those with better collaterals.


Collateral status on CTA may modify the effect of IAT on functional outcome. Patients with good collaterals seem to benefit from IAT while, in this study, IAT did not benefit patients with absent or poor collaterals.


Ischemic stroke patients with effective collateral flow have smaller infarcts and better outcomes than patients with poor or absent collaterals. With the establishment of IAT as the standard of care, there is uncertainty about how collateral flow modifies the effect of IAT. While many of the contemporary studies of IAT used advanced imaging, such as CT perfusion, MR CLEAN used CTA, which is currently used at most hospitals across the United States. This generalizability strengthens the findings of this analysis. If patients who are unlikely to benefit from IAT could be prospectively identified, this population could be spared from an invasive procedure that does not improve their outcome. While this study is intriguing, at this point, it is premature to withhold IAT from patients with poor or absent collaterals. The effect of very rapid treatment in patients with poor collaterals may mitigate the lack of benefit seen in this analysis. More data, including prospective trials, are needed in this area. Additionally, intraobserver rating of collaterals was only modest in this study, suggesting that the ability to accurately define collateral status from CTA in clinical practice will be challenging.

< Back to Listings