The Risk Factors Associated With Cerebral Parenchymal Lesions In Cerebral Venous Sinus Thrombosis Analysis And Establishment Of A Prediction Model

Objectives: The purpose of this study was to investigate risk factors,to establish a predictive nomogram,and to provide a reference for customized clinical decisions in a cohort of patients with cerebral venous sinus thrombosis(CVST)accompanied by cerebral parenchymal lesions.Methods: A total of 112 patients with CVST in the First Affiliated Hospital and the Tenth Affiliated Hospital of Guangxi Medical University were enrolled from January 2014 to August 2020 in this retrospective study.Patients were subdivided into two groups based on the MRI and/or CT examination;we included fifty-nine patients with cerebral parenchymal abnormalities and fifty-three without cerebral parenchymal abnormalities.Clinical information was collected including demographic information(age,sex,and disease duration),clinical characteristics(headache,emesis,epilepsy,focal neurological impairment and consciousness disorder),clinical assessment scale(MRS score and ADL score on admission),neuroimaging data(MRI,CT,MRV,CTV and DSA)and laboratory indexes(fibrinogen、RPR、PLR、LMR、NLR、SII).The independent predictors of cerebral parenchymal lesions in CVST were analysed through univariate and multivariate logistic regression analyses,and then a nomogram predictive model was established.The area under the ROC curve(AUC-ROC)and calibration curve demonstrated the ability of the nomogram.Then,the bootstrap method was used for internal verification.Results: Among the 112 patients with CVST,59(52.7%)patients had cerebral parenchymal lesions,but the other 53(47.3%)did not.(1)A one-way analysis showed that factors including age,duration of disease,headache,epilepsy,neurological impairment,ADL score and MRS score on admission,involvement of the superior sagittal sinus alone,fibrinogen,LMR,NLR,and SII were all related to cerebral parenchymal lesions in patients with CVST.(2)Binary logistic regression analysis with all of the risk factors analysed as independent variables demonstrated that age(OR=1.129,95% CI: 1.049-1.214),disease duration(OR=0.149,95% CI: 0.028-0.794),epilepsy(OR=7.954,95%CI:1.283-49.322),neurological impairment(OR=11.792,95% CI: 2.397-58.018),fibrinogen(OR=3.186,95% CI: 1.694-5.992)and NLR(OR=1.717,95% CI:1.089-2.710)were independent risk factor in patients with cerebral parenchymal lesions.(3)The nomogram prediction model has a good predictive capability,consistency and discrimination(AUC-ROC: 0.894,95% CI: 0.821～0.944).Conclusions:(1)Age,duration of disease,epilepsy,neurological impairment,fibrinogen,and NLR are independent predictors of CVST accompanied by cerebral parenchymal lesions.(2)The forecast model of the nomogram had a good predictive value in CVST patients with cerebral parenchymal lesions and could be applied for disease prediction in early stages.