MRI-Based Algorithm May Help Identify Stroke Patients Who Will Benefit From Thrombectomy

Helene Raoult, MD

Study results show that a hierarchic algorithm based on diffusion-weighted imaging (DWI) has high predictive value for disability at 3 months after thrombectomy in patients with acute ischemic stroke (AIS).

The retrospective, observational study included 149 patients admitted to Rennes University Hospital or referred from 3 other general hospitals between January 2015 and May 2017. Participants were at least 18 years of age, had AIS with proximal arterial occlusion of the anterior circulation confirmed on 1.5 or 3T MRI, currently undergoing endovascular treatment with or without intravenous alteplase, and had been admitted to the center within 6 hours of symptom onset.

Within 4.5 hours of stroke onset, intravenous thrombolysis was performed with 0.9 mg/kg of alteplase (max, 90 mg) with an initial bolus of 10% of the total dose followed by an infusion of the remaining dose over 60 minutes.

At baseline, demographic and other characteristics, including history of treated diabetes, hypertension and/or atrial fibrillation, and history of ischemic stroke and transient attack were recorded. Blood glucose level, systolic blood pressure, and NIH Stroke Score (NIHSS) range of 0-42 were also recorded.

The primary endpoint of the study was the Modified Rankin Score (mRS) at 3 months after stroke onset, with poor prognosis rated as mRS >2 and good prognosis as mRS <2.

Among the 149 patients included in the analysis, 80 (53.7%; 95% CI, 45.3%-61.9%) had a poor prognosis at 3 months (20 with mRS = 3; 17 with mRS = 4; 5 with mRS = 5; and 38 with mRS = 6). Patients with poor prognosis were more likely to have a history of hypertension and atrial fibrillation, higher blood glucose levels, a higher NIHSS score at baseline, a higher DWI lesion volume, and more terminal internal carotid artery occlusions compared to those with good prognosis.

In terms of treatment, patients with poor prognoses at 3 months were less often treated with intravenous alteplase, had general anesthesia more often, and had a longer time to treatment with intravenous alteplase as well as a higher rate of failed recanalization. Overall, the rate of poor prognosis in the study population was 51.4%

Results of the multivariate analysis demonstrated that a DWI lesion volume >80 ml was the most important of 4 independent risk factors of poor prognosis at 3 months, followed by baseline NIHSS score of >14, age >75, and time from stroke onset to groin puncture of >4 hours.

Notably, the hierarchic algorithm predicted up to 100% of patients with a poor prognosis who had a DWI lesion volume >80 mL and were older than 75, as well as in those with a DWI lesion volume of >80 mL and time to groin puncture of >4 hours (area under the curve, 0.87). In the case that lesion volume was ≤80 mL, the addition of an NIHSS score of >14 increased the predictive value for poor prognosis from 19.1% to 56.3%. Notably, this increased to 100% when age >75 and time to puncture of >4 hours were added.

“The proposed hierarchic algorithm is based on the DWI lesion volume and 3 clinical predictors with cutoff values and is very easy to use in clinical practice to predict the lack of clinical benefit of thrombectomy. Its predictive value is higher than in most previous scores proposed, with an area under the curve of 0.87, and 0.75 for the HIAT2 score and 0.79 for the Pittsburgh Response to Endovascular Therapy score,” the study investigators concluded. “…This superior result may be explained by the use of MR imaging instead of CT, given that MR imaging is the most accurate and validated method for assessing acute stroke lesions.”

REFERENCE

Raoult H, Lassalle MV, Parat B, et al. DWI-based algorithm to predict disability in patients treated with thrombectomy for acute stroke. Am J Neuroradiol. Published online January 2020. doi: 10.3174/ajnr.A6379.