Prediction of delayed cerebral ischemia after cerebral aneurysm rupture using explainable machine learning approach.

BACKGROUND: Aneurysmal subarachnoid hemorrhage results in significant mortality and disability, which is worsened by the development of delayed cerebral ischemia. Tests to identify patients with delayed cerebral ischemia prospectively are of high interest. OBJECTIVE: We created a machine learning system based on clinical variables to predict delayed cerebral ischemia in aneurysmal subarachnoid hemorrhage patients. We also determined which variables have the most impact on delayed cerebral ischemia prediction using SHapley Additive exPlanations method. METHODS: 500 aneurysmal subarachnoid hemorrhage patients were identified and 369 met inclusion criteria: 70 patients developed delayed cerebral ischemia (delayed cerebral ischemia+) and 299 did not (delayed cerebral ischemia-). The algorithm was trained based upon age, sex, hypertension (HTN), diabetes, hyperlipidemia, congestive heart failure, coronary artery disease, smoking history, family history of aneurysm, Fisher Grade, Hunt and Hess score, and external ventricular drain placement. Random Forest was selected for this project, and prediction outcome of the algorithm was delayed cerebral ischemia+. SHapley Additive exPlanations was used to visualize each feature's contribution to the model prediction. RESULTS: The Random Forest machine learning algorithm predicted delayed cerebral ischemia: accuracy 80.65% (95% CI: 72.62-88.68), area under the curve 0.780 (95% CI: 0.696-0.864), sensitivity 12.5% (95% CI: -3.7 to 28.7), specificity 94.81% (95% CI: 89.85-99.77), PPV 33.3% (95% CI: -4.39 to 71.05), and NPV 84.1% (95% CI: 76.38-91.82). SHapley Additive exPlanations value demonstrated Age, external ventricular drain placement, Fisher Grade, and Hunt and Hess score, and HTN had the highest predictive values for delayed cerebral ischemia. Lower age, absence of hypertension, higher Hunt and Hess score, higher Fisher Grade, and external ventricular drain placement increased risk of delayed cerebral ischemia. CONCLUSION: Machine learning models based upon clinical variables predict delayed cerebral ischemia with high specificity and good accuracy.

Benefit of Intravenous Alteplase before Thrombectomy Depends on ASPECTS.

PURPOSE: Baseline variables could be used to guide the administration of additional intravenous alteplase (IVT) before mechanical thrombectomy (MT). The aim of this study was to determine how baseline imaging and demographic parameters modify the effect of IVT on clinical outcomes in patients with ischemic stroke due to large vessel occlusion. METHODS: Multicenter retrospective cohort study of ischemic stroke patients triaged by multimodal-CT undergoing MT treatment after direct admission to an MT-eligible center. Inverse-probability weighting analysis (IPW) was used to assess the treatment effect of IVT adjusted for baseline variables. Multivariable logistic regression analysis with IPW-weighting and interaction terms for IVT was performed to predict functional independence (mRS 0-2 at 90-days). RESULTS: 720 patients were included, of which 366 (51%) received IVT. In IPW, the treatment effect of IVT on outcome (mRS 0-2) distinctively varied according to the ASPECTS subgroup (ASPECTS 9-10: +15%, ASPECTS 6-8: +7%, ASPECTS <6: -11%). In multivariable logistic regression analysis, IVT was independently associated with functional independence (aOR: 1.57, 95% CI: 1.16-2.14, p = 0.003) and the interaction term was significant for ASPECTS and IVT revealing that IVT was only significantly associated with better outcomes in patients with higher ASPECTS. No other significant baseline variable interaction terms were identified. INTERPRETATION: ASPECTS was the only baseline variable that showed a significant interaction with IVT for outcome prediction. Use of IVT prior to MT in patients with an ASPECTS of <6 was not associated with a treatment benefit and should be considered carefully. ANN NEUROL 2022;92:588-595.

Intravenous tPA (Tissue-Type Plasminogen Activator) Correlates With Favorable Venous Outflow Profiles in Acute Ischemic Stroke.

BACKGROUND: Intravenous tPA (tissue-type plasminogen activator) is often administered before endovascular thrombectomy (EVT). Recent studies have questioned whether tPA is necessary given the high rates of arterial recanalization achieved by EVT, but whether tPA impacts venous outflow (VO) is unknown. We investigated whether tPA improves VO profiles on baseline computed tomography (CT) angiography (CTA) images before EVT. METHODS: Retrospective multicenter cohort study of patients with acute ischemic stroke due to large vessel occlusion undergoing EVT triage. Included patients underwent CT, CTA, and CT perfusion before EVT. VO profiles were determined by opacification of the vein of Labbé, sphenoparietal sinus, and superficial middle cerebral vein on CTA as 0, not visible; 1, moderate opacification; and 2, full. Pial arterial collaterals were graded on CTA, and tissue-level collaterals were assessed on CT perfusion using the hypoperfusion intensity ratio. Clinical and demographic data were determined from the electronic medical record. Using multivariable regression analysis, we determined the correlation between tPA administration and favorable VO profiles. RESULTS: Seven hundred seventeen patients met inclusion criteria. Three hundred sixty-five patients received tPA (tPA+), while 352 patients were not treated with tPA (tPA-). Fewer tPA+ patients had atrial fibrillation (n=128 [35%] versus n=156 [44%]; P=0.012) and anticoagulants/antiplatelet treatment before acute ischemic stroke due to large vessel occlusion onset (n=130 [36%] versus n=178 [52%]; P<0.001) compared with tPA- patients. One hundred eighty-five patients (51%) in the tPA+ and 100 patients (28%) in the tPA- group exhibited favorable VO (P<0.001). Multivariable regression analysis showed that tPA administration was a strong independent predictor of favorable VO profiles (OR, 2.6 [95% CI, 1.7-4.0]; P<0.001) after control for favorable pial arterial CTA collaterals, favorable tissue-level collaterals on CT perfusion, age, presentation National Institutes of Health Stroke Scale, antiplatelet/anticoagulant treatment, history of atrial fibrillation and time from symptom onset to imaging. CONCLUSIONS: In patients with acute ischemic stroke due to large vessel occlusion undergoing thrombectomy triage, tPA administration was strongly associated with the presence of favorable VO profiles.

Favorable Venous Outflow Profiles Correlate With Favorable Tissue-Level Collaterals and Clinical Outcome.

BACKGROUND AND PURPOSE: Patients with acute ischemic stroke due to large vessel occlusion and favorable tissue-level collaterals (TLCs) likely have robust cortical venous outflow (VO). We hypothesized that favorable VO predicts robust TLC and good clinical outcomes. METHODS: Multicenter retrospective cohort study of consecutive acute ischemic stroke due to large vessel occlusion patients who underwent thrombectomy triage. Included patients had interpretable prethrombectomy computed tomography, computed tomography angiography, and cerebral perfusion imaging. TLCs were measured on cerebral perfusion studies using the hypoperfusion intensity ratio (volume ratio of brain tissue with [Tmax >10 s/Tmax >6 s]). VO was determined by opacification of the vein of Labbé, sphenoparietal sinus, and superficial middle cerebral vein on computed tomography angiography as 0, not visible; 1, moderate opacification; and 2, full. Clinical and demographic data were determined from the electronic medical record. Using multivariable regression analyses, we determined the association between VO and (1) favorable TLC status (defined as hypoperfusion intensity ratio ≤0.4) and (2) good functional outcome (modified Rankin Scale score, 0-2). RESULTS: Six hundred forty-nine patients met inclusion criteria. Patients with favorable VO were younger (median age, 72 [interquartile range (IQR), 62-80] versus 77 [IQR, 66-84] years), had a lower baseline National Institutes of Health Stroke Scale (median, 12 [IQR, 7-17] versus 19 [IQR, 13-20]), and had a higher Alberta Stroke Program Early Computed Tomography Score (median, 9 [IQR, 7-10] versus 7 [IQR, 6-9]). Favorable VO strongly predicted favorable TLC (odds ratio, 4.5 [95% CI, 3.1-6.5]; P<0.001) in an adjusted regression analysis. Favorable VO also predicted good clinical outcome (odds ratio, 10 [95% CI, 6.2-16.0]; P<0.001), while controlling for favorable TLC, age, glucose, baseline National Institutes of Health Stroke Scale, and good vessel reperfusion status. CONCLUSIONS: In this selective retrospective cohort study of acute ischemic stroke due to large vessel occlusion patients undergoing thrombectomy triage, favorable VO profiles correlated with favorable TLC and were associated with good functional outcomes after treatment. Future prospective studies should independently validate our findings.