From the Research
Prognosis of Brainstem Infarction
The prognosis of patients with brainstem infarction varies on an individual basis, with several factors influencing the outcome.
- A study published in 2018 1 found that patients with brainstem infarction located in the territory of multiple perforating arteries, who were discharged without administration of statins, showed a poor 1-year prognosis.
- Another study published in 2022 2 used deep neuroimaging features to predict functional outcomes in patients with acute brainstem infarction, achieving a high accuracy of 0.975 in predicting unfavorable outcomes at 3 months.
- A retrospective study from 1992 3 identified diabetes, bilateral lesions, and combination of these two factors as indicators of unfavorable outcome, associated with a short-term mortality rate greater than 30%.
- A study from 1975 4 found that soft palate paralysis, disturbed sensorium, need of special treatment measures, abnormalities in the ECG, fasting blood glucose 100 mg/100 ml or higher, and an abnormal EEG were all associated with high case fatality.
- Additionally, a review from 2013 5 noted that hemorrhagic complications are associated with an especially poor prognosis, and advanced neuroimaging techniques are essential in the decision-making process of brainstem stroke management.
Factors Influencing Prognosis
Several factors have been identified as influencing the prognosis of brainstem infarction, including:
- Location and size of the infarction 1
- Presence of diabetes or bilateral lesions 3
- Need of special treatment measures during the acute phase 4
- Abnormalities in the ECG or EEG 4
- Administration of statins at discharge 1
- Deep neuroimaging features extracted by convolutional neural networks (CNNs) 2
Outcome Prediction
The use of deep neuroimaging features has been shown to accurately predict functional outcomes in patients with acute brainstem infarction 2.
- A model based solely on 14 deep neuroimaging features achieved an area under the receiver operating characteristic curve (AUC) of 0.975 in predicting unfavorable outcomes at 3 months.
- The deep neuroimaging model outperformed traditional prognostic scores and demonstrated significant improvement over models combining clinical, laboratory, and conventional imaging features.