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.

Trends in age and sex-specific prevalence of cancer and cancer subtypes in acute ischemic stroke from 2007-2019.

OBJECTIVE: To describe age and sex-specific prevalence of cancer in acute ischemic stroke (AIS) hospitalizations in the United States over the last decade. METHODS: We conducted a retrospective serial cross-sectional study using all primary AIS discharges (weighted n=5,748,358) with and without cancer in the 2007-2019 National Inpatient Sample. Admissions with primary central nervous system cancers were excluded. Joinpoint regression was used to compute the average annualized percentage change (AAPC) in cancer prevalence over time. RESULTS: Across the study period, 12.7% of AIS admissions had previous/active cancer, while 4.4% had active cancer. Of these, 18.8% were hematologic cancers, 47.2% were solid cancers without metastasis and 34.0% were metastatic cancers of any type. Age-adjusted active cancer prevalence differed by sex (males:4.8%; females:4.0%) and increased with age up to age 70-79 years (30-39 years 1.4%; 70-79 years:5.7%). Amongst cancer admissions, lung (18.7%) and prostate (17.8%) were the most common solid cancers in men, while lung (19.6%) and breast (13.7%) were the most prevalent in women. Active cancer prevalence increased over time (AAPC 1.7%, p<0.05) but the pace of increase was significantly faster in women (AAPC 2.8%) compared to men (AAPC 1.1%) (p-comparison =0.003). Fastest pace of increased prevalence was seen for genitourinary cancers in women and for gastrointestinal cancers in both sexes. Genitourinary cancers in men declined over time (AAPC -2.5%, p<0.05). Lung cancer prevalence increased in women (AAPC 1.8%, p<0.05) but remained constant in men. Prevalence of head/neck, skin/bone, gastrointestinal, hematological and metastatic cancers increased over time at similar pace in both sexes. CONCLUSION: Prevalence of cancer in AIS admissions increased in the US over the last decade but the pace of this increase was faster in women compared to men. Gastrointestinal cancers in both sexes and genitourinary cancers in women are increasing at the fastest pace. Additional studies are needed to determine whether this increase is from co-occurrence or causation of AIS by cancer.

Outcomes after endovascular embolization for the treatment of nasal and oropharyngeal hemorrhage: safety, efficacy, and rebleeding.

BACKGROUND: Intractable nasal and oropharyngeal hemorrhage may be treated with endovascular embolization, but limited data are available. We sought to evaluate the efficacy, safety, and factors associated with rebleeding. METHODS: A retrospective analysis of consecutive embolizations for nasal and oropharyngeal hemorrhage over a 10-year period at a single institution was performed. Outcomes included procedural success (defined as cessation of hemorrhage in the immediate postoperative period), rebleeding requiring an additional intervention, and procedural complications. RESULTS: A total of 47 embolizations on 39 patients were included. The mean age was 60 years (standard deviation 16.1), 23.1% of patients were women, and 21 (53.8%) patients had a previously diagnosed head/neck malignancy. Bleeding sites were the nose in 20 patients and oropharynx in 21 (two patients presented with both nasal and oral bleeding). Immediate procedural success was achieved in 45 (95.7%) embolizations. Rebleeding requiring an additional intervention occurred after 11 (23.4%) embolizations at a median of one day after the procedure. In the multivariate analysis, preoperative hypotension (odds ratio 4.78, 95% confidence interval 1.04-24.61) and the use of coils (odds ratio 6.09, 95% confidence interval 1.24-46.69) were associated with rebleeding requiring repeat intervention. Complications included two watershed strokes that were anticipated due to occlusion of the internal carotid artery. CONCLUSIONS: In our experience endovascular embolization was a safe and effective treatment option for stopping oral and nasal hemorrhage. However, rebleeding was present after 23.4% of treatments and was associated with preoperative hypotension and the use of coils. Further study in a large multi-institutional cohort is warranted.

Incidence of Aneurysmal Subarachnoid Hemorrhage with Procedures Requiring General Anesthesia in Patients with Unruptured Intracranial Aneurysms.

BACKGROUND: The role of general anesthesia in precipitating aneurysm rupture is not clearly defined. In this study, we aimed to assess the natural history of unruptured aneurysms in patients undergoing non-aneurysm-related procedures requiring general anesthesia. METHODS: Retrospective review of consecutive patients with untreated intracranial aneurysms that underwent unrelated surgery with operative note documentation of general anesthesia. Events of intraoperative and postoperative subarachnoid hemorrhage were recorded to determine the incidence of rupture. RESULTS: A total of 110 patients harboring 134 unsecured aneurysms were studied. The mean age was 56.5 years (range, 17-92), and 68% were women (n = 75/110). Mean aneurysm size was 3.5 mm (range 1.5-17). A total of 208 procedures were performed under general anesthesia. There were no events of subarachnoid hemorrhage in 5.7 years of follow-up. CONCLUSION: In our study, general anesthesia did not precipitate aneurysm rupture, and there were no instances of subarachnoid hemorrhage during the follow-up period. Our results suggest a benign natural history for aneurysms undergoing unrelated general anesthesia. However, this should be interpreted with caution given limitations related to our small sample size and retrospective study design.

Multicenter assessment of morbidity associated with cerebral arteriovenous malformation hemorrhages.

BACKGROUND: The optimal management strategy for unruptured cerebral arteriovenous malformations (AVMs) is controversial since the ARUBA trial (A Randomized trial of Unruptured Brain AVMs). An accurate understanding of the morbidity associated with AVM hemorrhages may help clinicians to formulate the best treatment strategy for unruptured AVMs. OBJECTIVE: To determine the morbidity associated with initial cerebral AVM rupture in patients presenting to tertiary medical centers. METHODS: Retrospective chart reviews from three tertiary academic medical centers were performed for the period between 2008 and 2014. All patients admitted with intracranial hemorrhage due to untreated AVMs were included in this study. Patient-specific variables, including demographics, imaging characteristics, neurologic examination results, and clinical outcome, were analyzed and recorded. RESULTS: 101 Patients met the inclusion criteria. Admission National Institutes of Health Stroke Scale (NIHSS) scores were 0, 1-9, and ≥10 in 26%, 29%, and 45% of patients, respectively. Hematoma locations were subarachnoid, intraventricular, intraparenchymal, and combined in 5%, 11%, 32%, and 52% of patients, respectively. Deep venous drainage was present in 43% of AVMs; AVM-associated aneurysms were present in 44% of patients. Emergent hematoma evacuations were performed in 37% of patients and 8% of patients died while in hospital. At discharge, of those who survived, NIHSS scores of ≥1 and ≥10 were found in 69% and 23%, respectively. At the 90-day follow-up, 34% had a modified Rankin Scale (mRS) score >2. Patients with admission NIHSS score ≥10 had significantly higher rates of midline shift, surgical hematoma evacuation, and follow-up mRS ≥3 (p<0.05). CONCLUSIONS: The morbidity associated with cerebral AVM rupture appeared to be higher in our study than previously reported. Morbidity from AVM rupture should be considered as an important factor, together with variables such as risk of AVM rupture and procedural risk, in determining the optimal treatment strategy for unruptured cerebral AVMs.