Résumé
Objective To noninvasively predict isocitrate dehydrogenase(IDH)status of glioma via combining imaging and clini-cal features before surgery,so as to provide basis for individualized clinical treatment decision.Methods A total of 47 patients with glioma confirmed by pathological and molecular genetic tests were included,including 20 with IDH mutant type and 27 with IDH wild type.After diffusion tensor imaging(DTI)scanning,fractional anisotropy(FA)and mean diffusivity(MD)values of tumor paren-chyma were calculated.Combining DTI parameters with MRI morphological features of tumor,blood neutrophil/lymphocyte ratio(NLR)and patient's age,binary logistic regression model was established to effectively predict IDH status of glioma patients before surgery.Results There were significant differences in FAmean/FANAWM,MDmin,NLR,tumor location and age between IDH mutant type and IDH wild type groups(P<0.05).The binary logistic regression model concluding,FAmean/FANAWM,MDmin,cystic degeneration,NLR and age,predicted IDH status of glioma with area under the curve(AUC)of 0.961 and 95%confidence interval(CI)of 0.914-1.00.Conclusion The regression model established via combining DTI,MRI morphological features and blood NLR has great performance in classifying IDH status of glioma,and can help predict IDH status noninvasively before surgery,so as to assist clinical individualized treatment.