Decision and prognosis of mechanical thrombectomy: Stroke treatment Evaluation before mechanical intervention with thrombectomy (SATIN) study

Mechanical thrombectomy decision making and prognostication: Stroke treatment Assessments prior to Thrombectomy In Neurointervention (SATIN) study

Backdrop

Acute large vessel occlusion (ELVO) accounts for only one-third of acute ischemic strokes, but accounts for nearly 60% of paralysis and 90% of deaths after stroke. Mechanical thrombectomy (MT) is a powerful intervention that improves functional outcomes in suitably selected ELVO patients. Five landmark studies on MT, published in 2015, established the role of vascular intervention in stroke management and provided a basis for subsequent guideline rewrite.

With advances in imaging equipment and technology, there is an increasing likelihood that patients will be able to benefit from timely MT. But treatment options for stroke patients still depend on whether neurointerventionists can reasonably predict the success of the technique and who is more likely to benefit from MT. In actual clinical practice, the surgeon's decision is limited by time, priority clinical information, or the intervener's cognitive reserve. As a result, doctors often rely on implicit decision-making and all the cognitive biases that go along with it to choose treatment options. The objective of this study was to prospectively describe decision-making in MT selection and to assess the accuracy of prediction of patient outcomes by pre-MT neurointerventionists.

Method

A prospective multicenter cohort study of patients receiving MT was conducted at 11 experienced neurointerventional centers in the United States. The attending interventional neurologist was required to complete a preoperative questionnaire during the arterial puncture and identified key decision factors and the most likely imaging and clinical outcomes at 90 days. A post hoc review was then conducted to document the course and outcome of hospitalization.

(table1)

Result

A total of 299 patients were selected, among which 27 had participated in at least one operation, 62% had 6-10 years of experience, 21% had > 10 years, 16% had 3-5 years, and only 1 had less than 2 years of experience (> 80% of the interviewed physicians had independent practice experience > 5 years). Good clinical outcomes were achieved in 38% of patients (modified Rankin Scale (mRS) score 0-2). The most common factors influencing the decision to perform thrombus resection were occlusion site (81%), NIHSS score (74%), and perfusion imaging mismatch (43%). Only 140 patients (46.8%) had preoperative mRS Scores that accurately matched data collected retrospectively at admission. In 17.9% of cases, the scores were worse than predicted by the preoperative interventionists, while 34.8% were better than predicted.

A total of 299 patients were selected, among which 27 had participated in at least one operation, 62% had 6-10 years of experience, 21% had > 10 years, 16% had 3-5 years, and only 1 had less than 2 years of experience (> 80% of the interviewed physicians had independent practice experience > 5 years). Good clinical outcomes were achieved in 38% of patients (modified Rankin Scale (mRS) score 0-2). The most common factors influencing the decision to perform thrombus resection were occlusion site (81%), NIHSS score (74%), and perfusion imaging mismatch (43%). Only 140 patients (46.8%) had preoperative mRS Scores that accurately matched data collected retrospectively at admission. In 17.9% of cases, the scores were worse than predicted by the preoperative interventionists, while 34.8% were better than predicted.

As shown in the figure below, doctors predicted that MT patients' mRS Scores would be 0-2/150 (50.2%), 3-4/130 (43.5%), and 5-6/19 (6.4%). Overall, physicians accurately predicted preoperative scores in the 90-day mRS Group (0-2, 3-4, 5-6) 44.2 percent of the time.

Doctors incorrectly predicted that patients would do better in one group 28.1% of the time and in both groups 15.0% of the time; Patients achieved a better than expected mRS Score 12.7% of the time. Additional information was helpful: When doctors said they had all the patient information they needed to confidently decide to proceed with MT, 45.8 percent of patients predicted correctly, while 36.7 percent predicted correctly when not all information was available. Clinicians tend to overestimate the impact of occlusive site and perfusion imaging on the outcome, and underestimate the importance of pre-disease mRS On the outcome.

（table2）

（table3）

（table4）

As shown in the table above, NIHSS score, ASPECTS score and age were considered by interventional physicians to have an impact on prediction accuracy: With the increase of NIHSS score, accuracy decreased (p=0.048); The accuracy of ASPECTS increased with the increase of scores (p=0.037). However, the accuracy decreased with increasing age (p=0.023).

Sum up

SSATIN is the first prospective study to assess the ability of neurointerventionists to accurately predict clinical outcomes after ELVO MT. Overall, in addition to highlighting the inaccuracy of predictions, further development and improvement of machine learning algorithms for marginal MT candidates whose best management strategies are not clear, It may help reduce the risk of neurointerventional patients in making decisions about MT. In addition, prioritizing the acquisition and accurate recording of preoperative mRS Scores at the time of patient visit may also be another way to enhance the accuracy of MT decision-making and improve prognosis.

http://