Early risk stratification and temporal biomarker patterns of trousseau syndrome-related cerebral infarction in lung cancer

Background Trousseau syndrome-related cerebral infarction (TSCI) is a severe complication of lung cancer, but tools for early risk stratification remain limited. Methods In this retrospective case-control study, we enrolled 94 patients with newly diagnosed lung cancer and TSCI and 94 matched lung cancer controls without stroke at the First Affiliated Hospital of Zhengzhou University between 2021 and 2025. Using cases in which TSCI occurred at diagnosis or within 6 months thereafter, we developed and internally validated a five-variable model based on serum magnesium, international normalized ratio, prothrombin time, albumin-to-globulin ratio, and a history of hypertension. Concurrent-onset (Fulminant) and delayed-onset TSCI subgroups were compared across coagulation, inflammatory, tumor-burden, and nutritional domains. Temporal biomarker patterns were examined using locally estimated scatterplot smoothing (LOESS). Results The model showed good discrimination (AUC = 0.861) and sample-specific calibration within the selected case-control sample. Fulminant TSCI showed pronounced abnormalities across coagulation/fibrinolysis, inflammatory, tumor-burden, and nutritional domains. Temporal analyses suggested heterogeneity in biomarker patterns: early-onset TSCI was associated with higher Ki67, neuron-specific enolase, and lactate dehydrogenase, whereas late-onset TSCI was more closely associated with higher triglyceride levels. Conclusions These findings identify a set of routinely available clinical variables associated with early high-risk TSCI in patients with lung cancer. The proposed nomogram demonstrated favorable discriminative performance and may provide a useful framework for early risk stratification and clinical surveillance. Further validation in independent prospective cohorts would help to confirm its generalizability and support future clinical application.