MRI Imaging Feature Prediction PLGG Genotype Value Analysis

BackgroundPediatric Low Grade Gliomas(PLGG)are a group of primary central nervous system tumors that occur in the Pediatric population.The disease accounts for more than 30% of central nervous system tumors in children.The overall prognosis of PLGG was good,with a 5-year overall survival rate of about 95%.However,survivors often suffer neurological,endocrine,and other complications as a result of disease or treatment.In particular,patients with PLGG who cannot be completely resected by surgery may have the possibility of disease progression.Therefore,certain adjuvant therapy is needed for PLGG in the brain stem,diencephalon,visual pathway and other positions that can only be biopsied or partially resected,especially for the progression after resection.The high incidence of adverse reactions in traditional chemoradiotherapy limits its wide clinical application,while the new chemoradiotherapy has various shortcomings.Recent advances in genomics have provided many potential therapeutic targets for PLGG targeting.Moreover,molecular pathological typing based on gene phenotype is of great significance for the prognosis of PLGG.Therefore,noninvasive detection of PLGG common genotypes is of great significance.Multimodal MRI provides a feasible way for noninvasive genotype detection.In addition,no study on the prediction of PLGG genotype by MRI image features was found in the literature search.ObjectivesTo explore the predictive significance of MRI image features for some common genotypes of PLGG.MethodsThe relationship between genotype and MRI imaging characteristics of patients with PLGG who accepted surgical treatment in the first affiliated hospital of zhengzhou university from 2001 to 2016 were retrospectively analyzed.The genotype information was acquired through genetic testing of the retained formalin fixed paraffin embedding pathological tissue,the MRI features of image data was acquired through image feature analysis of the MRI electronic image data retained by hospital.The correlation between MRI image features and genotypes,such as descriptive image features and quantitative indexes of MRI image,was statistically analyzed,and the predictive significance of image features on genotypes was evaluated.ResultsPossible independent predictors of each genotype in descriptive imaging features include the tumor location to KIAA1549-BRAF fusion(P=0.045),the definition of tumor edge to BRAF mutation(P=0.035),the tumor location(P=0.038)and the signal strength of ADC(P=0.032)to MYB amplification.MRI quantitative index FLAIR sequence 1% threshold may be an independent predictor of KIAA1549-BRAF fusion(P=0.019).The AUC results showed that KIAA1549-BRAF fusion: combined descriptive image features and MRI image quantitative index =0.842(0.746-0.939)> combined descriptive image features and MRI image quantitative index that could be independently predicted =0.819(0.719-0.919)> overall descriptive image characteristics =0.809(0.708-0.909)and overall MRI image quantitative index =0.661(0.533-0.789).The overall quantitative index of MRI image =0.661(0.533-0.789)was not greater than the quantitative index of each individual MRI image(1% threshold of T1 enhanced sequence =0.651(0.515-0.786),1% threshold of FLAIR sequence =0.665(0.533-0.796),and the minimum value of FLAIR sequence =0.676(0.547-0.805)).BRAF mutation: combined with descriptive image characteristics and quantitative index of MRI image =0.954(0.889-1.000)> combined with possibly independently predicted descriptive image characteristics and quantitative index of MRI image =0.871(0.718-1.000)≥ overall descriptive image characteristics =0.768(0.556-0.981)and overall quantitative index of MRI image =0.871(0.718-1.000).The overall quantitative index of MRI image =0.871(0.718-1.000)is not greater than the quantitative index of each individual MRI image(FLAIR sequence maximum =0.871(0.718-1.000)).MYB amplification: combined descriptive image characteristics and MRI image quantification index =0.706(0.520-0.892)= combined descriptive image characteristics and MRI image quantification index that may be independently predicted =0.706(0.520-0.892)< overall descriptive image characteristics =0.903(0.788-1.000)and overall MRI image quantification index =0.753(0.601-0.906),The overall quantitative index of MRI image was not greater than that of each individual MRI image(variance of T1 enhanced sequence =0.741(0.582-0.901),and the 10% boundary value of ADC sequence =0.754(0.579-0.929)).ConclusionsTumor location and FLAIR sequence 1% boundary value may be independent predictors of KIAA1549-BRAF fusion,tumor edge sharpness may be independent predictors of BRAF mutation,and tumor location and ADC signal strength may be independent predictors of MYB amplification.Compared with cerebellar PLGG,telencephalic PLGG was less likely to undergo changes in KIAA1549-BRAF fusion.PLGG with clear edges was less likely to undergo BRAF mutation than PLGG with unclear edges.Telencephalic PLGG was more likely to undergo MYB amplification than cerebellar PLGG.PLGG with high ADC signal intensity was less prone to MYB amplification.The increase in the number of quantitative indicators of related MRI images alone could not improve the prediction effect of genotypes,and the increase in the number of relevant indicators in combination with descriptive image features and quantitative indicators of MRI images could improve the prediction effect of genes.There was no specific tendency for genotype prediction between purely descriptive imaging features or quantitative MRI imaging indicators.