The Influence of Body Mass Index on Outcomes in Patients Undergoing Mechanical Thrombectomy for Anterior Circulation Large Vessel Occlusion: Institutional Experience and Meta-analysis.

BACKGROUND: An obesity paradox, whereby patients with higher body mass index (BMI) experience improved outcomes, has been described for ischemic stroke. It is unclear whether this applies to patients undergoing mechanical thrombectomy (MT) for large vessel occlusion (LVO). METHODS: Mechanical thrombectomies for anterior circulation LVO between 2015 and 2021 at a single institution were reviewed. Multivariable logistic regressions were used to determine the association between BMI and favorable functional outcome (90-day modified Rankin Scale 0-2), intracranial hemorrhage, and malignant middle cerebral infarction. A systematic review was performed to identify studies reporting the effect of BMI on outcomes among patients receiving MT for LVO. The data from the systematic review were combined with the institutional data by using a random effects model. RESULTS: The institutional cohort comprised 390 patients with a median BMI of 27 kg/m2. Most patients were obese [36.7% (BMI ≥ 30 kg/m2)], followed by overweight [30.5% (BMI ≥ 25 and < 30 kg/m2)], normal [27.9% (BMI ≥ 18.5 and < 25 kg/m2)], and underweight [4.9% (BMI < 18.5 kg/m2)]. As a continuous variable, BMI was not associated with any of the outcomes. When analyzing BMI ordinally, obesity was associated with lower odds of favorable 90-day modified Rankin Scale (odds ratio 0.42, 95% confidence interval 0.20-0.86). The systematic review identified three eligible studies comprising 1,348 patients for a total of 1,738 patients. In the random effects model, there was no association between obesity and favorable outcome (odds ratio 0.89, 95% confidence interval 0.63-1.24). CONCLUSIONS: Obesity is not associated with favorable outcomes in patients undergoing MT for LVO.

Development and Internal Validation of Machine Learning Models to Predict Mortality and Disability After Mechanical Thrombectomy for Acute Anterior Circulation Large Vessel Occlusion.

OBJECTIVE: Mechanical thrombectomy (MT) improves outcomes in patients with LVO but many still experience mortality or severe disability. We sought to develop machine learning (ML) models that predict 90-day outcomes after MT for LVO. METHODS: Consecutive patients who underwent MT for LVO between 2015-2021 at a Comprehensive Stroke Center were reviewed. Outcomes included 90-day favorable functional status (mRS 0-2), severe disability (mRS 4-6), and mortality. ML models were trained for each outcome using prethrombectomy data (pre) and with thrombectomy data (post). RESULTS: Three hundred and fifty seven patients met the inclusion criteria. After model screening and hyperparameter tuning the top performing ML model for each outcome and timepoint was random forest (RF). Using only prethrombectomy features, the AUCs for the RFpre models were 0.73 (95% CI 0.62-0.85) for favorable functional status, 0.77 (95% CI 0.65-0.86) for severe disability, and 0.78 (95% CI 0.64-0.88) for mortality. All of these were better than a standard statistical model except for favorable functional status. Each RF model outperformed Pre, SPAN-100, THRIVE, and HIAT scores (P < 0.0001 for all). The most predictive features were premorbid mRS, age, and NIHSS. Incorporating MT data, the AUCs for the RFpost models were 0.80 (95% CI 0.67-0.90) for favorable functional status, 0.82 (95% CI 0.69-0.91) for severe disability, and 0.71 (95% CI 0.55-0.84) for mortality. CONCLUSIONS: RF models accurately predicted 90-day outcomes after MT and performed better than standard statistical and clinical prediction models.

The influence of pre-reperfusion blood pressure on outcomes following mechanical thrombectomy for anterior circulation large vessel occlusion.

BACKGROUND: We evaluated how systolic blood pressure (SBP) and mean arterial pressure (MAP) parameters between presentation and reperfusion influence functional status and intracranial hemorrhage (ICH). METHODS: All patients who underwent MT for LVO at a single institution were reviewed. Independent variables included SBP and MAP measurements obtained on presentation, between presentation and reperfusion (pre-reperfusion), and between groin puncture and reperfusion (thrombectomy). Mean, minimum, maximum, and standard deviations (SD) for SBP and MAP were calculated. Outcomes included 90-day favorable functional status, radiographic ICH (rICH), and symptomatic ICH (sICH). RESULTS: 305 patients were included. Higher pre-reperfusion SBPmax was associated with rICH (OR 1.41, 95% CI 1.08-1.85) and sICH (OR 1.84, 95% CI 1.26-2.72). Higher SBPSD was also associated with rICH (OR 1.38, 95% CI 1.06-1.81) and sICH (OR 1.59, 95% CI 1.12-2.26). Greater SBPmax (OR 0.64, 95% CI 0.47-0.86), MAPmax (OR 0.72, 95% CI 0.52-0.97), SBPSD (OR 0.63, 95% CI 0.46-0.86), and MAPSD (0.63, 95% CI 0.45-0.84) during thrombectomy were associated with lower odds of 90-day favorable functional status. In a subgroup analysis, these associations were primarily limited to patients with intact collateral circulation. Optimal SBPmax cutoffs for predicting rICH were 171 (pre-reperfusion) and 179 mmHg (thrombectomy). Cutoffs for predicting sICH were 178 (pre-reperfusion) and 174 mmHg (thrombectomy). CONCLUSION: Greater maximum BP and variability in BP during the pre-reperfusion period are associated with unfavorable functional status and ICH after MT for anterior circulation LVO.

Machine learning prediction of malignant middle cerebral artery infarction after mechanical thrombectomy for anterior circulation large vessel occlusion.

OBJECTIVE: Prediction of malignant middle cerebral artery infarction (MMI) could identify patients for early intervention. We trained and internally validated a ML model that predicts MMI following mechanical thrombectomy (MT) for ACLVO. METHODS: All patients who underwent MT for ACLVO between 2015 - 2021 at a single institution were reviewed. Data was divided into 80% training and 20% test sets. 10 models were evaluated on the training set. The top 3 models underwent hyperparameter tuning using grid search with nested 5-fold CV to optimize the area under the receiver operating curve (AUROC). Tuned models were evaluated on the test set and compared to logistic regression. RESULTS: A total of 381 patients met the inclusion criteria. There were 50 (13.1%) patients who developed MMI. Out of the 10 ML models screened on the training set, the top 3 performing were neural network (median AUROC 0.78, IQR 0.72 - 0.83), support vector machine ([SVM] median AUROC 0.77, IQR 0.72 - 0.83), and random forest (median AUROC 0.75, IQR 0.68 - 0.81). On the test set, random forest (median AUROC 0.78, IQR 0.73 - 0.83) and neural network (median AUROC 0.78, IQR 0.73 - 0.83) were the top performing models, followed by SVM (median AUROC 0.77, IQR 0.70 - 0.83). These scores were significantly better than those for logistic regression (AUROC 0.72, IQR 0.66 - 0.78), individual risk factors, and the Malignant Brain Edema score (p < 0.001 for all). CONCLUSION: ML models predicted MMI with good discriminative ability. They outperformed standard statistical techniques and individual risk factors.

Comparison of the Safety, Efficacy, and Procedural Characteristics Associated with Proximal and Distal Radial Access for Diagnostic Cerebral Angiography.

OBJECTIVES: Radial access is an increasingly popular approach for performing cerebral angiography. There are two sites for radial artery puncture: proximal transradial access (pTRA) in the wrist and distal transradial access (dTRA) in the snuffbox. These approaches have not been directly compared. MATERIALS AND METHODS: Consecutive diagnostic cerebral angiograms performed at a single institution were retrospectively reviewed. Outcomes included fluoroscopy time, radiation dose, contrast volume, time to obtain access, procedure duration, and time to discharge home. Success rates as well as minor and major complication rates associated with each approach were also compared. Multivariate linear regression models were used to determine the relationship between access site and outcomes while adjusting for covariates. RESULTS: A total of 287 angiograms on 244 patients met the inclusion criteria. pTRA was associated with shorter fluoroscopy time (ß -2.54, 95% CI -4.18 - -0.9, p = 0.003) and lower radiation dose (ß -242.89, 95% CI -351.55 - -134.24, p < 0.001), but not contrast volume. Time to obtain access, procedure duration, and time to discharge home were similar between approaches. A total of 10 minor complications occurred with similar rates for each approach (8 for dTRA, 2 for pTRA, p = 0.168) and there were no major complications. The conversion rate to femoral access was low (1.05% overall) and did not differ with approach. CONCLUSION: dTRA and pTRA are associated with similarly high rates of safety and efficacy. Procedure duration, time to obtain access, and time to discharge did not differ between approaches.

Transition from Proximal to Distal Radial Access for Diagnostic Cerebral Angiography: Learning Curve Analysis.

BACKGROUND: Distal transradial access (dTRA) has several advantages compared with proximal transradial access (pTRA) for cerebral angiography. The learning curve for transitioning from pTRA to dTRA has not been described. METHODS: Retrospective analysis of the first 75 diagnostic cerebral angiograms performed with dTRA by a single operator was performed. Outcomes included time for sheath insertion, sheath to first vessel time, procedure duration, fluoroscopy time, radiation dose, and contrast volume. Their associations with procedure number were evaluated with multivariate linear regressions, segmented linear regression, and locally weighted regression (LOESS). RESULTS: The mean age of patients was 56.1 years and 61.3% were female. Seventy-four of 75 angiograms were successfully completed with dTRA. There were 3 minor and no major complications. After adjusting for covariates, sheath to first vessel time (ß = -0.50, P < 0.001) and procedure duration (ß = -0.26, P = 0.002) were associated with procedure number. Time for sheath insertion, fluoroscopy time, radiation dose, and contrast volume were not associated with procedure number. Segmented linear regression identified break-points of 33 for sheath to first vessel time and 11 for procedure duration, which corresponded to the procedure number after which these outcomes trended down. LOESS models for time to sheath placement, procedure duration, fluoroscopy time, and radiation dose predicted minimum values between procedures 40-50. CONCLUSIONS: Transitioning from pTRA to dTRA for diagnostic cerebral angiography is feasible and safe. The learning curve is overcome between procedures 11 and 33, and further refinement in performance occurs through procedures 40-50.

Distinguishing between predictive and incentive value of uncertain gambling-like cues in a Pavlovian autoshaping task.

The flashing lights and celebratory sounds that dominate slot-machine gambling are believed to promote engagement and motivation to keep playing. However, these cues are often presented in the absence of reward, and previous research suggests that this reward uncertainty, which degrades their predictive value, also increases their incentive value. Here, we used autoshaping to tease apart the impact of reward uncertainty on the predictive and incentive value of a conditioned stimulus (CS) using serial cues. Each CS trial began with the presentation of a predictive CS1, followed by a CS2, holding primarily incentive value, because of its proximity to sucrose reward delivery, under Certain (100%-1) or Uncertain (50%-1- 2-3) reward conditions. Subsequently, we tested the impact of amphetamine and nicotine on cue attraction, and the ability of these cues to either serve as a conditioned reinforcer, or promote motivation for sucrose during a progressive ratio task. Finally, we measured anxiety behavior, and examined its interaction with each cue and uncertainty. Our results suggest that reward uncertainty increases attraction to the incentive CS2 and its ability to trigger motivation and reward-seeking. However, although the CS2 is largely ignored under Certain conditions, both CS1 and CS2 become conditioned reinforcers for both groups. Finally, amphetamine reduced the attraction of the CS1 for both groups but had no effect on the attraction of the CS2. These results suggest that reward uncertainty recruits and increases the incentive value of cues with limited predictive value and highlights the distinction between cue attraction, reward-seeking and conditioned reinforcer properties.