Construction and validation of a nomogram model to predict symptomatic intracranial hemorrhage after intravenous thrombolysis in severe white matter lesions.

Cerebral white matter lesions (WMLs) increase the risk of bleeding after intravenous thrombolysis (IVT) but are also considered to require IVT. Its risk factors and predictive models are still poorly studied. The aim of this study is to develop a clinically applicable model for post-IVT haemorrhage. It offers the possibility to prevent symptomatic intracranial hemorrhage (sICH) in patients with IVT in severe WMLs. A large single-center observational study conducted a retrospective analysis of IVT in patients with severe WMLs from January 2018 to December 2022. Univariate and multi-factor logistic regression results were used to construct nomogram model, and a series of validations were performed on the model. More than 2,000 patients with IVT were screened for inclusion in this study after cranial magnetic resonance imaging evaluation of 180 patients with severe WMLs, 28 of whom developed sICH. In univariate analysis, history of hypertension (OR 3.505 CI 2.257-4.752, p = 0.049), hyperlipidemia (OR 4.622 CI 3.761- 5.483, p < 0.001), the NIHSS score before IVT (OR 41.250 CI 39.212-43.288, p < 0.001), low-density lipoprotein levels (OR 1.995 CI 1.448-2.543, p = 0.013), cholesterol levels (OR 1.668 CI 1.246-2.090, p = 0.017), platelet count (OR 0.992 CI 0.985-0.999, p = 0.028), systolic blood pressure (OR 1.044 CI 1.022-1.066, p < 0.001), diastolic blood pressure (OR 1.047 CI 1.024-1.070, p < 0.001) were significantly associated with sICH. In a multifactorial analysis, the NIHSS score before IVT (OR 94.743 CI 92.311-97.175, p < 0.001), and diastolic blood pressure (OR 1.051 CI 1.005-1.097, p = 0.033) were considered to be significantly associated with sICH after IVT as risk factors for the occurrence of sICH. The four most significant factors from logistic regression are subsequently fitted to create a predictive model. The accuracy was verified using ROC curves, calibration curves, decision curves, and clinical impact curves, and the model was considered to have high accuracy (AUC 0.932, 95% 0.888-0.976). The NHISS score before IVT and diastolic blood pressure are independent risk factors for sICH after IVT in patients with severe WMLs. The models based on hyperlipidemia, the NIHSS score before IVT, low-density lipoprotein and diastolic blood pressure are highly accurate and can be applied clinically to provide a reliable predictive basis for IVT in patients with severe WMLs.