Stroke is a devastating global health issue, ranking second in mortality and third in disability-adjusted life years (DALYs). Ischemic stroke, particularly large artery occlusion (LAO), is a significant contributor to these statistics, with a strong correlation to 90-day mortality. Despite mechanical thrombectomy (MT) being the standard treatment, disparities persist, especially in low- and middle-income countries (LMICs), which bear a disproportionate burden of stroke cases.
The post-ischemic stroke period is characterized by rapid neuroinflammation, with neutrophils playing a pivotal role in the thromboinflammatory cascade. These neutrophils interact critically with platelets, amplifying inflammation and disrupting the immune microenvironment, leading to secondary brain injury. The neutrophil-to-platelet ratio (NPR) reflects this inflammatory state and its potential impact on patient outcomes.
This study aims to address the need for improved risk stratification tools post-MT. By developing a novel NPR-based predictive model using machine learning, we can forecast early (90-day) post-MT mortality in LAO patients. This approach will enable healthcare providers to identify high-risk individuals promptly, allowing for targeted therapeutic strategies.
The study analyzed data from 320 LAO-AIS patients who underwent MT at the Affiliated Hospital of Guilin Medical University between 2023 and 2025. Patients were carefully selected based on specific criteria, and their data was processed to ensure accuracy and reliability.
The research team gathered comprehensive baseline clinical data, including age, gender, medical history, and laboratory results. Early mortality was defined as death occurring at discharge or within 90 days post-discharge. NPR was calculated as the ratio of neutrophils to platelets, and its standardized value (NPR_std) was used to account for potential scaling biases.
Statistical analysis revealed significant baseline differences in age, responsible artery, white blood cell count, neutrophil count, platelet count, NPR, and NPR_std. These variables were further analyzed using advanced statistical methods, including the Boruta feature selection algorithm and multiple imputation techniques.
The study identified six independent predictive factors: NPR_std, age, decompressive craniectomy (DC), responsible artery, lymphocyte count (LYM), and prothrombin time (PT). These factors were integrated into a predictive model, which was then validated internally. The model demonstrated strong predictive performance, with an AUC of 0.926 for the training set and 0.853 for the validation set.
To enhance the model's interpretability, SHAP analysis was employed, revealing that NPR_std holds the highest importance. The SHAP value plots provided a clear visualization of how different variable distributions influence the model's predictions.
The study's findings contribute to the growing body of research focused on post-MT prognosis in LAO-AIS. By directly incorporating NPR as a predictive factor, the study offers a clinically relevant approach, facilitating targeted interventions for patients post-thrombectomy.
The elevated neutrophil-to-platelet ratio (NPR) has significant clinical implications beyond stroke. It has been linked to adverse prognosis in sepsis patients, in-hospital mortality in burn patients, and perioperative complications in cardiovascular surgery. In AIS patients, NPR is associated with hemorrhage risk after intravenous thrombolysis, and in testicular germ cell tumors, it correlates with tumor staging and clinical characteristics.
To bridge the gap between research and clinical practice, a practical clinical workflow was proposed. Upon admission of an LAO-AIS patient eligible for MT, routine baseline data is collected, including NPR_std, LYM, and PT. These parameters are then inputted into the nomogram, generating a quantifiable probability of 90-day mortality. This risk stratification guides personalized clinical decision-making, ensuring timely and targeted interventions for high-risk patients.
While the study provides valuable insights, it is not without limitations. The single-center, retrospective design limits the generalizability of the findings, and further external validation is necessary. Additionally, the study did not systematically analyze key procedural variables and post-operative medications, which may independently influence mortality risk.
In conclusion, this study developed a predictive model for early mortality risk in LAO-AIS patients following thrombectomy. The model, driven by key factors like NPR_std and age, offers a clinically actionable tool to enhance personalized medicine in LAO-AIS. Further research and validation are needed to confirm the model's broad utility and clinical applicability.
