Analysis Of Prognosis After Endovascular Embolization For High Altitude Aneurysmal Subarachnoid Hemorrhage

Objective:To investigate risk factors affecting long-term clinical outcomes in ruptured intracranial aneurysm（RIA）patients after interventional embolization in Qinghai Province and establish nomogram.Methods:From April 2012 to April 2019,371 RIA patients who underwent endovascular embolization in Affiliated Hospital of Qinghai University were consecutively and retrospectively enrolled.All patients were diagnosed as aneurysmal subarachnoid hemorrhage by head CT and/or DSA.The total of 371patients were all born in Qinghai Province and had lived locally for more than 1year.They were randomly divided into a training group（270 cases）and a validation group（101 cases）at a ratio of 7:3.For clinical follow-up at 24 months after interventional embolization,the Glasgow Prognostic Scale（GOS）score 4-5 was defined as good prognosis and 1-3 as poor prognosis.Demographic characteristics and clinical data were compared between patients with good prognosis（192 cases）and poor prognosis（78 cases）in the training group.Demographic features included age,gender,Tibetan,high altitude（altitude≥2500 m）,etc.Clinical data included hypertension,type 2 diabetes,smoking history,elevated hemoglobin level（hemoglobin>175 g/L）,increased intracranial pressure（cerebrospinal fluid pressure>180 mm H₂O）,Hunt-Hess scale score 4-5,modified Fisher scale score,postoperative complications（ischemic or hemorrhagic）in hospital,etc.The poor prognosis of patients was dependent variable in the training group,and parameters of P<0.1 in the univariate analysis were independent variables.The nomogram prediction model was constructed,according to risk factors selected by multivariate Logistic regression analysis.Bootstrap method was applied to verify the prediction model internally,and the model differentiation was determined by consistency index.The closer the consistency index was to 1.0,the better the model differentiation was.The decision curve analysis（DCA）was applied to judge the clinical practicability of the prediction model.The greater the critical probability range between the model curve and the net benefit curve of all patients receiving endovascular embolization,the better the clinical practicability is.The calibration curve was drawn,and the calibration of the prediction model was determined according to the Hosmer-Lemeshow test,with P>0.05 indicating a good calibration.Results:（1）There were no significant difference in demographic characteristics and clinical data between the training group and the validation group（all P>0.05）.（2）Among the 270 patients in the training group,192 had good prognosis and 78 had poor prognosis.Compared to good prognosis patients,the incidence of high altitude,Tibetan,increased intracranial pressure,Hunt-Hess scale 4-5,elevated hemoglobin and postoperative complications in hospital were higher in poor prognosis patients with statistically significant differences（74.4%[58/78]vs 41.1%[79/192],χ²=24.480;59.0%[46/78]vs 22.4%[43/192],χ²=33.585;34.6%[27/78]vs 12.0%[23/192],χ²=18.835;80.8%[63/78]vs 39.1%[75/192],χ²=38.619;55.1%[43/78]vs 20.3%[39/192],χ²=31.793;14.1%[11/78]vs 2.6%[5/192],χ²=13.155;all P<0.01）.There were no significant differences in the other demographic characteristics and clinical data between good prognosis and poor prognosis in the training group（all P>0.05）.（3）According to the multivariate Logistic regression,high altitude（OR,5.117,95%CI 2.453-10.676）,Tibetan（OR,3.615,95%CI 1.785-7.322）,increased intracranial pressure（OR,2.665,95%CI 1.130-6.286）,Hunt-Hess scale 4-5（OR,6.671,95%CI 3.137-14.186）,elevated hemoglobin（OR,3.032,95%CI 1.475-6.233）,postoperative complications（OR,4.763,95%CI 1.223-18.553）were independent risk factors of poor prognosis in RIA patients after interventional embolization（all P<0.05）.（4）The nomogram shows that the Hunt-Hess scale 4-5 was 100.0,increased intracranial pressure 51.3,High altitude85.0,Tibetan 68.2,elevated hemoglobin 59.0,and postoperative complications in hospital 82.0.The consistency indexes were 0.873 and 0.852 in training and validation group,respectively,which indicated good model discrimination.（5）The results of the clinical decision curve showed that the critical probability of the training group was 0.25-0.98,and the validation group was 0.25-0.95,indicating good clinical application effect in both the training and the validation group.（6）The results of the calibration analysis showed that in the training group（Hosmer-Lemeshow test,χ²=9.963,P=0.256）and validation group（Hosmer-Lemeshow test,χ²=8.256,P=0.214）,calibrated prediction is consistent and close to that of the uncalibrat-ed prediction,and the calibration curve performs well.Conclusions:This single study shows that high altitudes,Tibetan,intracranial hypertension,Hunt-Hess scale 4-5,elevated hemoglobin,and postoperative complications increased the risk of poor prognosis in RIA patients after interventional embolization in Qinghai Province.