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

The BAND score: A simple model for upfront prediction of poor outcomes despite successful stroke thrombectomy

BACKGROUND

While endovascular thrombectomy (EVT) is beneficial for patients with acute large vessel occlusion ischemic strokes, a significant portion of patients still do poorly despite successful recanalization. Identifying patients at high risk for poor outcomes can be helpful for future clinical trial design and optimizing acute stroke triage.

METHODS

Consecutive EVT patients were identified from 2016 to 2021 at a Comprehensive Stroke Center, and clinical information was recorded. Poor outcome was defined as a 90-day modified Rankin Scale (mRS) of 4 or greater despite achieving a modified thrombolysis in cerebral infarction (mTICI) score of 2b or greater. Multivariable regression analyses were used to identify risk factors for poor outcomes, and a scoring system was constructed.

RESULTS

483 patients with successful recanalization were identified. From a randomly selected training cohort (n = 357), the 10-point BAND score was constructed from independent risk factors for poor outcomes: baseline disability (1 point: baseline mRS ≥ 2), age (1 point: 60-69 years, 2 points: 70-79 years, 3 points: 80-84 years, 4 points: 85 years or older), NIHSS (2 points: 13-17, 3 points: 18-22, and 4 points: ≥ 23), and delay from last known normal (1 point: ≥ 6 h). The BAND score was significantly associated with rates of poor outcomes (p < 0.001), and it achieved an area under the receiver-operating characteristic curve (AUC) of 0.80 (95 %CI 0.76-0.85) in our training cohort and 0.78 (95 %CI 0.70-0.86) in our validation cohort (n = 126). Overall, the BAND score had a significantly higher AUC value than the widely validated THRIVE score and the THRIVE-EVT calculation (p = 0.001 and 0.029, respectively). Among patients with high BAND scores (7 or higher), 88.2 % had poor outcomes.

CONCLUSION

The BAND score is a simple tool to predict poor outcomes despite successful recanalization. Future studies are needed to confirm the BAND score's external validity.

A Clinical and Imaging Fused Deep Learning Model Matches Expert Clinician Prediction of 90-Day Stroke Outcomes

BACKGROUND AND PURPOSE

Predicting long-term clinical outcome in acute ischemic stroke is beneficial for prognosis, clinical trial design, resource management, and patient expectations. This study used a deep learning-based predictive model (DLPD) to predict 90-day mRS outcomes and compared its predictions with those made by physicians.

MATERIALS AND METHODS

A previously developed DLPD that incorporated DWI and clinical data from the acute period was used to predict 90-day mRS outcomes in 80 consecutive patients with acute ischemic stroke from a single-center registry. We assessed the predictions of the model alongside those of 5 physicians (2 stroke neurologists and 3 neuroradiologists provided with the same imaging and clinical information). The primary analysis was the agreement between the ordinal mRS predictions of the model or physician and the ground truth using the Gwet Agreement Coefficient. We also evaluated the ability to identify unfavorable outcomes (mRS >2) using the area under the curve, sensitivity, and specificity. Noninferiority analyses were undertaken using limits of 0.1 for the Gwet Agreement Coefficient and 0.05 for the area under the curve analysis. The accuracy of prediction was also assessed using the mean absolute error for prediction, percentage of predictions ±1 categories away from the ground truth (±1 accuracy [ACC]), and percentage of exact predictions (ACC).

RESULTS

To predict the specific mRS score, the DLPD yielded a Gwet Agreement Coefficient score of 0.79 (95% CI, 0.71-0.86), surpassing the physicians' score of 0.76 (95% CI, 0.67-0.84), and was noninferior to the readers (P < .001). For identifying unfavorable outcome, the model achieved an area under the curve of 0.81 (95% CI, 0.72-0.89), again noninferior to the readers' area under the curve of 0.79 (95% CI, 0.69-0.87) (P < .005). The mean absolute error, ±1ACC, and ACC were 0.89, 81%, and 36% for the DLPD.

CONCLUSIONS

A deep learning method using acute clinical and imaging data for long-term functional outcome prediction in patients with acute ischemic stroke, the DLPD, was noninferior to that of clinical readers.

Interventions to Support the Return to Work for Individuals with Stroke: A Systematic Review and Meta-analysis

PURPOSE

An increasing number of individuals with stroke are having difficulties in returning to work, having a significant impact on both individuals and society. The aims of this meta-analysis were to summarize the interventions to support the return to work (RTW) for individuals with stroke and to quantitatively evaluate the efficacy of each type of intervention.

METHODS

A systematic review and meta-analysis were conducted according to PRISMA guidelines. PubMed, Embase, Cochrane Library, CINAHL, and PsycINFO were searched until 26 June 2023, and the list of references of the initially included articles was also searched. Two researchers independently performed the search, screening, selection, and data extraction. The primary outcome was RTW rate (the RTW rate was defined as the proportion of individuals who returned to work in each group (intervention and control) at the endpoint). Pooled risk ratio (RR) was estimated using a random-effects model with 95% confidence intervals (CIs).

RESULTS

A total of 13 studies representing 4,282 individuals with stroke were included in our study. Results showed that physiological interventions could improve the RTW rate of individuals with stroke (RR: 1.19, 95% CI: 1.01 to 1.42, I2 = 72%). And receiving intravenous thrombolytic therapy was beneficial in promoting the RTW in individuals with stroke. Subgroup analysis and meta-regression analysis showed that the individuals' functional status during hospitalization was the only source of heterogeneity. Psychological interventions had little or no effect on the RTW rate of individuals with stroke (RR: 1.20, 95% CI: 0.58 to 2.51, I2 = 30%). Work-related interventions had little or no effect on the RTW rate of the individuals with stroke (RR:1.36,95%CI: 0.99 to 1.88, I2 = 73%). The subgroup analysis showed that country, age, and follow-up method were the sources of heterogeneity.

CONCLUSION

Physiological intervention promoted the RTW of individuals with stroke. But, the effect of psychological and work-related interventions in promoting the RTW of individuals with stroke was not significant. We anticipate that these findings may inform the design of future interventions. For future research, we recommend that more high-quality randomized controlled trials be conducted to further promote the RTW of individuals with stroke.

SYSTEMATIC REVIEW REGISTRATION

PROSPERO Registration Number, CRD42023443668.

Long-Term Effect of Mechanical Thrombectomy in Stroke Patients According to Advanced Imaging Characteristics

PURPOSE

Data on long-term effect of mechanical thrombectomy (MT) in patients with large ischemic cores (≥ 70 ml) are scarce. Our study aimed to assess the long-term outcomes in MT-patients according to baseline advanced imaging parameters.

METHODS

We performed a single-centre retrospective cohort study of stroke patients receiving MT between January 1, 2010 and December 31, 2018. We assessed baseline imaging to determine core and mismatch volumes and hypoperfusion intensity ratio (with low ratio reflecting good collateral status) using RAPID automated post-processing software. Main outcomes were cross-sectional long-term mortality, functional outcome and quality of life by May 2020. Analysis were stratified by the final reperfusion status.

RESULTS

In total 519 patients were included of whom 288 (55.5%) have deceased at follow-up (median follow-up time 28 months, interquartile range 1-55). Successful reperfusion was associated with lower long-term mortality in patients with ischemic core volumes ≥ 70 ml (adjusted hazard ratio (aHR) 0.20; 95% confidence interval (95% CI) 0.10-0.44) and ≥ 100 ml (aHR 0.26; 95% CI 0.08-0.87). The effect of successful reperfusion on long-term mortality was significant only in the presence of relevant mismatch (aHR 0.17; 95% CI 0.01-0.44). Increasing reperfusion grade was associated with a higher rate of favorable outcomes (mRS 0-3) also in patients with ischemic core volume ≥ 70 ml (aOR 3.58, 95% CI 1.64-7.83).

CONCLUSION

Our study demonstrated a sustainable benefit of better reperfusion status in patients with large ischemic core volumes. Our results suggest that patient deselection based on large ischemic cores alone is not advisable.

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.

Patients' perceptions on outcomes after mechanical thrombectomy in acute ischemic stroke

BACKGROUND

The modified Rankin Scale (mRS) is a clinician-reported scale that measures the degree of disability in patients who suffered a stroke. Patients' perception of a meaningful recovery from severe stroke, expected value of a stroke intervention, and the effect of disparities are largely unknown.

METHODS

We conducted a survey of patients, their family members, and accompanying visitors to understand their personal preferences and expectations for acute strokes potentially eligible for acute endovascular intervention using a hypothetical scenario of a severe stroke in a standardized questionnaire.

RESULTS

Of 164 survey respondents, 65 (39.6%) were the patient involved, 93 (56.7%) were a family member, and six (3.7%) were accompanied visitors (friends, other). Minimally acceptable disability after a stroke intervention was considered as mRS 2 by 42 respondents (25.6%), as mRS 3 by 79 (48.2%), and as mRS 4 by 43 (26.2%) respondents. Race was associated with different views on this question (p < 0.001; Hispanic and Black patients being more likely to accept disability than Caucasian and Asian patients), while sex (p = 0.333) and age (p = 0.560) were not. Sixty-three respondents (38.4%) viewed minimally acceptable probability of improvement with an intervention as over 50%, 57 (34.8%) as 10-50%, and 44 (26.8%) as less than 10%.

CONCLUSIONS

A wide range of acceptable outcomes were reported regardless of gender or age. However, race was associated with different acceptable outcome. This is an important finding to demonstrate because of the persistent racial and ethnic disparities in the utilization of endovascular therapy for acute stroke in the United States.

Functional Outcome Prediction in Acute Ischemic Stroke Using a Fused Imaging and Clinical Deep Learning Model

BACKGROUND

Predicting long-term clinical outcome based on the early acute ischemic stroke information is valuable for prognostication, resource management, clinical trials, and patient expectations. Current methods require subjective decisions about which imaging features to assess and may require time-consuming postprocessing. This study's goal was to predict ordinal 90-day modified Rankin Scale (mRS) score in acute ischemic stroke patients by fusing a Deep Learning model of diffusion-weighted imaging images and clinical information from the acute period.

METHODS

A total of 640 acute ischemic stroke patients who underwent magnetic resonance imaging within 1 to 7 days poststroke and had 90-day mRS follow-up data were randomly divided into 70% (n=448) for model training, 15% (n=96) for validation, and 15% (n=96) for internal testing. Additionally, external testing on a cohort from Lausanne University Hospital (n=280) was performed to further evaluate model generalization. Accuracy for ordinal mRS, accuracy within ±1 mRS category, mean absolute prediction error, and determination of unfavorable outcome (mRS score >2) were evaluated for clinical only, imaging only, and 2 fused clinical-imaging models.

RESULTS

The fused models demonstrated superior performance in predicting ordinal mRS score and unfavorable outcome in both internal and external test cohorts when compared with the clinical and imaging models. For the internal test cohort, the top fused model had the highest area under the curve of 0.92 for unfavorable outcome prediction and the lowest mean absolute error (0.96 [95% CI, 0.77-1.16]), with the highest proportion of mRS score predictions within ±1 category (79% [95% CI, 71%-88%]). On the external Lausanne University Hospital cohort, the best fused model had an area under the curve of 0.90 for unfavorable outcome prediction and outperformed other models with an mean absolute error of 0.90 (95% CI, 0.79-1.01), and the highest percentage of mRS score predictions within ±1 category (83% [95% CI, 78%-87%]).

CONCLUSIONS

A Deep Learning-based imaging model fused with clinical variables can be used to predict 90-day stroke outcome with reduced subjectivity and user burden.

Evaluation of computed tomography perfusion and angiogram use in stroke evaluation for thrombectomy at a community emergency department setting

PURPOSE

Evaluate concordance of provider practices with clinical guidelines for thrombectomy screening in an emergency department (ED) via computed tomography perfusion and angiogram (CT-P/A).

METHODS

A retrospective observational study was conducted for patients 18 years or older who received a CT-P/A of the head and neck in a US Midwestern ED between September 2019 through June 2021. Healthcare system records reviewed for patient information, CT-P/A findings, and treatment decisions.

RESULTS

During study period, 68,403 patients presented to the ED with 718 (1.1%) receiving a CT-P/A. Of these patients, 105 (14.6%) were transferred to a regional facility for potential thrombectomy, with 74 (70.5%) receiving procedure, 28 (26.7%) not receiving procedure, and 3 (2.9%) with insufficient follow-up information. Of patients receiving CT-P/A, 23 met DAWN criteria for thrombectomy, with 21 (91.3%) transferred for potential thrombectomy and 20 (95.2%) receiving the procedure; in comparison, 81 patients (11.7%) did not meet all DAWN criteria and were transferred for potential thrombectomy, with 52 (64.2%) receiving procedure. Lastly, 55 patients met DEFUSE-3 criteria for thrombectomy with 49 (89.1%) being transferred for potential thrombectomy and 45 (91.8%) receiving procedure. In comparison, 53 patients who did not meet all DEFUSE-3 criteria were transferred for potential thrombectomy, with 27 (50.9%) receiving procedure.

CONCLUSIONS

This study helps to understand CT-P/A usage, especially in patients that fall outside of treatment criteria in the current thrombectomy literature. Results may have value to institutions interested in using CT-P/A as a diagnostic tool as well as institutions already incorporating it in stroke assessments.