Establishment And Verification Of A Clinical Prediction Model For Lower-grade Gliomas With Autophagy-related Genes

Objective:Lower-grade gliomas(Lr GG)are composed of low-grade and intermediate-grade gliomas,and their clinical behaviors vary greatly.Some of them are lazy,while others quickly develop into glioblastoma.Recent studies have shown that autophagy is closely related to the prognosis of Lr GG.There is an urgent need for a survival nomogram containing autophagy-related genes to predict the prognosis of Lr GG,and to provide consultation and risk assessment for follow-up treatment.Methods:From the Chinese Glioma Genome Atlas(CGGA)and Cancer Genome Atlas(TCGA),the following variables of lower grade glioma were obtained: tumor grade(II or III),pathology,age,sex,radiotherapy status,chemotherapy status,and IDH1 subtype.Autophagy-related genes can be downloaded from the Human Autophagy Database(HADb).The autophagy risk scoring formula was constructed by survival filtering analysis,and then the Nomogram nomogram was constructed using Cox proportional hazard regression.These models were developed using CGGA data and used TCGA data for independent external verification.The model was internally verified using 10-fold cross-validation,and the calibration curve is used for external verification.Results:A nomogram model was developed to use 8 variables in the training set to predict the1-year,3-year,5-year,and 10-year overall survival rates of patients with lower-grade glioma:tumor grade,medical history,age,sex,radiation status,chemotherapy status,IDH subtypes and genetic risks related to autophagy.The C index used for internal verification and the time-dependent receiver operating characteristics(ROC)(the nomogram used to predict overall survival based on the training set)was 0.778,and the C index used for external verification(validation set)was 0.816.The results of the calibration chart show that the actual observed values and predicted values obtained through the nomogram have good agreement between the two groups.Conclusion:We have developed a ready-made nomogram model that includes clinical features that can predict overall survival.The nomogram can provide consultation and risk assessment for the follow-up treatment of Lr GG patients.A free online webpage for implementing this nomogram is also provided: https://cdd1152.shinyapps.io/Lr GGNomappC/.