Abstract:

Objective To explore the influencing factors of early death （within 3 months） in adult glioma patients， and to construct a risk prediction model. Methods Retrospective analysis was performed on the clinical data of 228 adult glioma patients admitted to the 909th Hospital （Dongnan Hospital of Xiamen University） from June 2020 to June 2024. Patients were divided into a death group （n=32） and a survival group （n=196） based on whether death occurred within 3 months， and the clinical data between the two groups were compared. Multivariate logistic regression was used to analyze the influencing factors of death within 3 months， a logistic regression prediction model was constructed， and receiver operator characteristic （ROC） curve was plotted to analyze the predictive value of the model. Results There were no statistically significant differences between the two groups in age， gender， hypertension， diabetes， tumor location， tumor involvement， neurological impairment， maximum tumor diameter， chemotherapy， or radiotherapy （all P>0.05）. The death group showed higher proportions of cerebral herniation （χ²=20.74， P<0.001）， hospital admission Karnofsky performance status （KPS） score ≤70 （χ²=26.66， P<0.001）， tumor grade Ⅲ-Ⅳ （χ²=28.70， P<0.001）， MGMT promoter unmethylation （χ²=10.25， P=0.001）， IDH wild-type （χ²=6.18， P=0.013）， and incomplete tumor resection （χ²=10.37， P=0.001） compared with the survival group. Multivariate analysis revealed that cerebral herniation （OR=19.78， 95%CI： 5.33-73.41， P<0.001）， hospital admission KPS score ≤70 （OR=19.64， 95%CI： 5.54-69.59， P<0.001）， tumor grade Ⅲ-Ⅳ （OR=9.40， 95%CI： 3.02-29.27， P<0.001）， MGMT promoter unmethylation （OR=4.28， 95%CI： 1.18-15.54， P=0.027）， and incomplete tumor resection （OR=9.50， 95%CI： 2.72-33.23， P<0.001） were independent risk factors for early death in glioma patients. The risk prediction model for early death in glioma patients constructed based on these indicators was logit（P）=-18.04+2.96×cerebral herniation （with=1， without=0）+2.98×hospital admission KPS score （≤70=1， >70=0）+2.24×tumor grade （Ⅲ-Ⅳ=1， Ⅰ-Ⅱ=0）+1.45×MGMT promoter methylation （no=1， yes=0）+2.25×complete tumor resection （no=1， yes=0）. ROC curve analysis demonstrated that this model had predictive value for early death in glioma patients， with an area under the curve of 0.920 （95%CI： 0.868-0.972）， a sensitivity of 0.842， and a specificity of 0.906. Conclusions Cerebral herniation， hospital admission KPS score ≤70， tumor grade Ⅲ-Ⅳ， MGMT promoter unmethylation， and incomplete tumor resection are independent risk factors for early death in adult glioma patients. The risk prediction model constructed based on these indicators has good predictive value.

Key words: Glioma, Death, Prognosis, Encephalocele