| Part Ⅰ Survival analysis and model establishment of gliomaBackground:Glioma is the most common tumor in the central nervous system.The current treatment principles are surgery,postoperative radiotherapy and adjuvant chemotherapy.After active comprehensive treatment,patients still relapse or even die.The median survival time for glioblastoma is about 14.6 months.Pathological examination is the gold standard for evaluating treatment response.There are many factors and complex mechanisms that affect the recurrence and prognosis of patients with glioma,such as age,gender,tumor size,pathological grade and degree of surgical resection,etc.Therefore,accurate assessment of the recurrence and prognosis of glioma patients in clinical work is the focus of individualized treatment.Machine learning is a branch of the field of artificial intelligence,which is the frontier and hotspot of prediction and translational research in the field of glioma.Glioma recurrence and survival models are constructed through radiomics,genomics or clinical information,which can be applied to make accurate and individualized prognosis judgments for patients.Purpose:The purpose of this study is to explore the prognostic factors of the recurrence and survival rate of glioma,and to establish a non-invasive and accurate assessment of recurrence and survival prognosis based on the clinical characteristics closely-related to therapeutic efficacy and treatment-associated information.Materials and Methods:A retrospective analysis of 270 patients undergoing postoperative radiotherapy for gliomas was conducted.Univariate and multivariate analysis were completed respectively according to the age,gender,KPS,tumor location,radiotherapy operation time,pathological grade,surgical method,and whether postoperative concurrent radiotherapy and chemotherapy were performed.The prognostic factors that affect the recurrence and survival rate of patients with glioma are studied here,With the help of the relevant software package of the R language,the clinical features and treatment-related information closely related to the therapeutic efficacy are used to construct a logistic regression recurrence prediction model and the Cox regression survival prediction model,and authenticating it.Results:The 1-year survival rate of the whole group patients was 79%,2-year survival rate was 55%,3-year survival rate was 46%,PFS was 20 months,OS was 30 months.univariate analysis showed KPS score(P=0.012),pathology Grading(P<0.001)and degree of surgical resection(P<0.001)affect the overall survival time after glioma surgery.Multivariate analysis suggested that KPS(P=0.002),pathological grade(P<0.001)and degree of surgical resection(P<0.001)were independent prognostic factors affecting the overall survival time of patients.Regarding the impact of progression-free survival(PFS),the results of univariate analysis showed that pathological grade(P<0.001)and degree of surgical resection(P<0.001)were factors affecting PFS in patients with postoperative radiotherapy for glioma.The results of Cox regression analysis showed that KPS(P=0.006),pathological grade(P<0.001),and degree of surgical resection(P<0.001)were independent prognostic factors that affected patients with local recurrence.Using clinical information,a logistic regression model was successfully established to predict tumor recurrence.The logistic regression model has a specificity of 84.6%,a sensitivity of 77%,and an accuracy rate(AUC)of 87.7%.The Cox regression prediction model for glioma survival was also successfully constructed using clinical information.The model showed that the patient’s pathological grade and tumor recurrence were the most important factors affecting the prognosis.The 5-year survival rate C-index value was 0.86.The clinical pathological grade of patients were higher,the shorter the lifetime.Conclusion:Better general condition of the patient,lower pathological grade and higher degree of surgical resection,the local recurrence is late,suggest better survival prognosis.Using clinically easily accessible parameters,a model for predicting tumor recurrence and a model for predicting glioma survival were established.They perform well in internal verification and has good practicability.Part Ⅱ Analysis of factors for the recurrence of glioblastomaBackground:The conclusions of the first part of this article suggest that pathological grading is not only a prognostic factor,but also a predictive factor.In the WHO Ⅰ-Ⅳclassification,glioblastoma(Glioblastoma,GBM)has the worst prognosis and the shortest survival period.Glioblastoma is an invasive growth and has no clear boundary with normal brain tissue.Almost all GBM patients will relapse or even die after radiotherapy.Due to the high heterogeneity of GBM,the length of time for patients to relapse varies.What factors determine the recurrence of GBM?How to prevent and prolong the recurrence time of GBM is one of the hotspots and focuses of current research.Purpose:This study explored the analysis of factors related to the recurrence of glioblastoma.In terms of clinical and pathological factors,coupled with imaging and genetic information,it is hoped that a more accurate GBM recurrence model can be constructed to predict the time of GBM recurrence,and provide a theoretical basis for postoperative recurrence prevention,treatment and prognosis of GBM.Materials and Methods:Retrospective analysis of 176 patients with recurrence of glioblastoma,the age,gender,KPS,tumor location,whether it invaded the subventricular zone,tumor size,radiotherapy operation time,pathological grade,surgical method,whether accepting postoperative concurrent chemoradiotherapy,IDH1 mutation,MGMT methylation and other factors were analyzed by univariate and multivariate analysis.In addition to clinical pathological factors,imaging and genetic information are used to construct a recurrence prediction model.Results:The results of univariate analysis showed that the patient’s age,gender,KPS,tumor location,enhancement characteristics,tumor diameter,tumor volume,radiotherapy interval,concurrent chemoradiotherapy,adjuvant chemotherapy,adjuvant chemotherapy cycle had nothing to do with recurrence time(P>0.05).However,the degree of surgical resection(P=0.004),the sensitivity of tumor radiotherapy(P=0.001),the involvement of the subependymal area(P=0.004),IDH1 status(P=0.048),TERT C228T status(P=0.012),GFAP The expression(P=0.044)was related to the time of recurrence.The results of multivariate analysis showed that the degree of surgical resection(P=0.002),SVZ invasion(P=0.008),radiotherapy sensitivity(P=0.002),and TERT C228T mutation(P=0.023)are independent factors that affect the recurrence of GBM.The histogram shows the ranking of the contribution to GBM recurrence.The top 9 contributing factors are IDH1 mutation,radiotherapy interval,age,tumor diameter,KPS score,tumor volume,whether there is enhancement in MRI images,and judge the effect of each factor in recurrence classification,The discrimination is poor,and the model cannot be constructed.Conclusion:Patients with partial tumor resection,SVZ invasion,IDH1 wild-type,TERT C228T mutation,GFAP positive and radiotherapy resistance have a shorter recurrence time.The key factors affecting the recurrence time of patients with glioblastoma are the scope of surgical resection,tumor invasion SVZ,TERT C228T status and radiotherapy sensitivity.Combining clinical parameters,genomics and radiomics,etc.cannot successfully construct a recurrence prediction model.Part Ⅲ Study on the role and mechanism of AHIF in the resistance of glioblastoma to radiotherapyBackground:The second part of this article reports on the recurrence factors of GBM after radiotherapy.The most critical one is radiotherapy sensitivity.The median recurrence time of radiotherapy-resistant GBM is 6 months,while the median recurrence time of radiotherapy-sensitive GBM is 14 months.Elucidating the radiotherapy resistance mechanisms of GBM and increasing the sensitivity of radiotherapy are of great significance for improving the overall survival of GBM patients.LncRNAs have been found tu mediate the recurrence or progression of tumors after radiotherapy.Antisense hypoxia-inducible factor(AHIF)is a long non-coding RNA produced by the antisense strand of HIF-1α gene.Though there have been some literatures on the function and mechanism of AHIF in tumors,there are few studies on the mechanism of AHIF in glioblastoma radiotherapy resistance.Purpose:To study the expression and effects of AHIF on glioblastoma radiotherapy via the cellular and in vivo mouse models.Materials and Methods:The expression of AHIF were examined by RT-PCR and the cellular function have been analyzed by clonal formation,comet assay and apoptotic assay.Proteins changed in apoptosis have been detected by WB.Finally,in vivo GBM models were established to investigate the functions of AHIF.Results:After glioblastoma,U87-MG and T98G cell lines were irradiated,the expression of AHIF gene was up-regulated.After knocking down AHIF,the number and size of cell colonies were significantly reduced compared with the control group.The glioma intracranial tumor-bearing mouse model obtained consistent results,and the survival rate of the AHIF knockdown group mice was higher.In view of this,knocking down the AHIF gene improves the radiosensitivity of GBM in vivo and in vitro.Conclusion:After irradiation in glioblastoma,the expression of long non-coding RNAAHIF is increased in both in vivo and in vitro experiments;knocking out this gene increases the sensitivity of radiotherapy. |
PDF Full Text Request