Evaluation Of IDH1 Genotype In High-grade Glioma Based On MRI Signs And Radiomics

Objective:1.To investigate the preoperative evaluation of isocitrate dehydrogenase 1（IDH1）genotype of high-grade gliomas by conventional magnetic resonance image（MRI）and diffusion weighted imaging（DWI）value.2.To explore the value of radiomics model in predicting IDH1genotype of high-grade gliomas before operation.Materials and methods:1.The clinical and imaging data of 82 patients with high-grade astrocytoma confirmed by surgery and pathology in the second hospital of Lanzhou University from March 2016 to August 2019 were analyzed retrospectively,including 29 cases of IDH1mutation and 53 cases of IDH1 wild type.All patients can obtain the results of IDH1 gene detection and complete MRI scanning sequence before operation,Including T1WI,T2WI,FLAIR,CE-T1WI and DWI.The patients’gender,age,tumor location,number of lesions,tumor maximum diameter,peritumoral edema maximum diameter,boundary,involving insular lobe and corpus callosum,crossing midline,cystic necrosis,hemorrhage,contrast-enhanced enhancement degree and DWI quantitative parameters were compared and analyzed ROC)curve was used to evaluate the diagnostic efficacy of each parameter for IDH1 genotype of high grade glioma.2.From March2016 to August 2019,113 patients with high-grade gliomas confirmed by surgery and pathology were collected retrospectively,IDH1 gene test results and T1WI,T2WI,FLAIR and CE-T1WI sequences during preoperative MRI were obtained in all patients.According to the proportion of7:3,they were divided into training set（n=80）and verification set（n=33）.The model of IDH1genotype prediction of high-grade glioma was constructed by using radiomics features,and multivariate logistic regression algorithm was further combined with magnetic resonance phenotypic features to construct radiomics nomograms.The area under the curve AUC and the decision curve analysis（DCA）were used to evaluate the performance of the radiomics nomogram.Results:1.1 The average age of IDH1 mutant patients in high-grade gliomas is less than the IDH1 wild-type patients（42.76±10.86 years old）vs（48.98±14.21 years old）,the difference between the groups is statistically significant（P<0.05）;IDH1 Mutant tumors are more common in the frontal lobe,IDH1 wild type is more common in the parietal occipital lobe and other parts,the difference between the groups is statistically significant（P<0.05）;IDH1 mutant tumors are easy to invade the island lobes,accounting for about 41.4%,IDH1 wild type accounted for about17.0%,the difference between the groups was statistically significant（P<0.05）;IDH1 mutant tumors were single,accounting for about 100%,IDH1 wild type single accounting for about84.9%,the difference between the groups is statistically significant（P<0.05）;the cystic necrosis of IDH1 mutant tumor accounted for about 69.0%,and the wild type accounted for about 90.6%.The difference between the groups was statistically significant（P<0.05）;The level of moderate enhancement was none or mild enhancement,moderate enhancement,and severe enhancement accounted for 72.4%,24.1%,and 3.1%,respectively,and IDH1 wild type accounted for 13.2%,41.3%,and 45.3%,respectively.The difference between the groups is statistically significant（P<0.05）.1.2 The average maximum diameter of IDH1 mutant tumors is greater than IDH1 wild type（67.36±18.35mm）vs（50.02±17.15mm）,and the difference between the groups is statistically significant（P<0.05）.The ADCmean,ADCmin,rADCmean,and rADCmin of IDH1mutant patients were larger than those of IDH1 wild type patients,which were（0.94±0.42）×10^-3mm²/s vs（0.87±0.47）×10^-33 mm²/s,（0.90±0.42）×10^-33 mm²/s vs（0.82±0.52）×10^-3mm²/s,（1.22±0.88）vs（1.13±0.80）,（1.16±0.87）vs（1.06±0.84）,the differences between the groups are statistical Significance（P<0.05）.1.3 The ROC curve results show that the ADCmin value has the highest efficacy in diagnosing the two groups of tumors.The AUC value is 0.891,with the maximum ADCmin=0.88×10^-33 mm²/s as the threshold.The sensitivity for differential diagnosis of the two groups of tumors is 86.24%,specific The degree is 80.16%and the accuracy is 81.70%.2.1 A multisequence radiomics model for predicting IDH1 genotype of high-grade gliomas consists of 9 radiomics features.The AUCs obtained in the training and validation sets are0.953 and 0.873,respectively.2.2 The performance of the radiomics normogram constructed based on the combination of radiomics features and image phenotypic features was better than that of the radiomics feature model and image phenotypic feature model.The AUC of the radiomics normogram vs radioomics feature model vs image phenotypic feature model in training concentration was 0.970 vs 0.953 vs 0.898.The AUC of the verification set of the radiomics normogram vs radioomics feature model vs image phenotypic feature model was 0.889 vs 0.873vs 0.861.Conclusion:1.Magnetic resonance signs（tumor occurrence site,tumor maximum diameter,whether it invades the insula,number of lesions,cystic necrosis,and degree of enhancement）and DWI quantitative parameters have important reference value for preoperative evaluation of high-grade glioma IDH1 gene status.2.The construction of radiomics normogram can distinguish IDH1 genotype of high-grade glioma well,and show good diagnostic efficiency in training set and verification set,which can provide non-invasive imaging strategy for the prediction of IDH1genotype of high-grade glioma patients before operation.