Correlation Of Quantitative Paremeters In Dynamic Contrast-Enhanced Magnetic Resonance Imaging With The Tumor Microenvironment And Its Change After Neoadjuvant Chemotherapy In Osteosarcoma Patients

Part 1 The Relationship between DCE-MRI and Tumor Microcirculation and Hypoxia in OsteosarcomaObjective:To investigate whether the quantitative parameters of dynamic enhanced magnetic resonance imaging(DCE-MRI)can express the microcirculation perfusion and hypoxic metabolism in osteosarcoma.Methods:A total of ten subjects of Primary periknee osteosarcoma were involved according to the inclusion criteria.All patients accepted MRI examination before surgery.The scanning sequences included routine T1 WI,T2WI sequence,DCE-MRI and enhanced T1WI imaging sequence.Based on Sagittal position of knee joint,DCE-MRI scan included two flip angle scanning(5°;15°)with gradient recalled echo(GRE)T1WI imaging sequence and the fast GRE 3D DYNAMIC T1WI sequence.A total of 80 phases were scanned.A high pressure mass injection of Gd-DTPA 0.1mmol/kg(2.5ml/s)and the equivalent physiological saline was followed after the first 5 phases.There were 24 images in each period,and the total scan time was 6 minutes and 23 seconds.The femoral or tibial segments containing the intact tumor segment were obtained after surgery.Being the distal femur intercondylar fossa or the proximal tibial intercondylar crest marked as.anatomical markers,the tumor segments were cut open along the median sagittal plane.The sagittal plane specimens of the 4-5mm thickness were cut out on one side,which was divided into several 1cm x 1cm pieces like horizontal gridding.The DCE-MRI fuature of these small tissue pieces and pathological results were analyzed.comparatively.The tumor necrosis rate(TNR),microvascular density(MVD),vascular endothelial factor(VEGF)and hypoxia inducible factor-1α(HIF-1α)were picked as the pathological gold standard.On the DCE-MRI image which closest to the pathological section of the tumor segment by the anatomical markers,the horizontal gridding was divided into a 1cm*1cm size of a region of interest(ROI),which matched with the pathological samples of the corresponding position.The DCE-MRI data were processed using the hemodynamic software developed by GE Healthcare Omni Kinetics(version number 2017.01.22).Toff’s model used Extended Toft’s linear model.The quantitative parameters of each sample were measured,which includes the volume transfer constant(Ktrans),the rate constant(Kep),the extravascular extracellular volume fraction(Ve)and the fraction plasma volume(Vp).Regression between DCE-MRI quantitative parameters and MVD were tested using stepwise regression Poisson regression model.With Ktrans,Kep,Ve and Vp as the dependent variables,a covariance analysis was made with TNR,VEGF and HIF-la as independent variables.Using the regress gradually logistics regression,the regression relationship between the quantitative parameters and HIF-1α was inspected,whose efficiency of the prediction is evaluated by the receiver working characteristic curve(ROC curve).Association between VEGF and quantitative parameters were evaluated by back-propagation(BP)neural network,and the result was output ROC curve.Results:Ten osteosarcoma patients were obtained,including 460 tumor specimens.The statistical results are as follows:1.Ktrans Kep and Ve could predict the MVD value in tumor tissues of osteosarcoma based on a poisson regression equation between them.2.Ktrans and Ve were statistically significant in the tumor necrosis of osteosarcoma(P<0.01).3.VEGF expression was positive when 0.684<Ve<1(sensitivity,62.1%;specificity,54.4%).4.The Kep value greater than 0.155(sensitivity,44.1%;specificity,72.2%)was identified as threshold values for determining positive expression of HIF-la.Vp was correlated to HIF-la(P<0.05).Conclusions:The quantitative parameters of MRI can express the distribution of microcirculation perfusion,tumor necrosis and hypoxia metabolism in the tissue of osteosarcoma.This method is helpful to the assessment of the curative effect for osteosarcoma.Part 2 Application of DCE-MRI in Neoadjuvant Chemotherapy Evaluation in Osteosarcoma PatientsObjective:To explore the significance of the quantitative perfusion parameters of DCE-MRI in the evaluation of neoadjuvant chemotherapy for osteosarcoma.Methods:All patients accepted new adjuvant chemotherapy before operation and repeated MRI examinations were performed before and after the whole course of chemotherapy besides the group entry standard being same as in the Part 1,so were the other perfusion imaging methods,tumor tissue pathology and data measurement.Ensure that the locations of ROI in DCE-MIR imaging of before and after chemotherapy were detail matched with the pathological samples.The quantitative perfusion parameters before and after chemotherapy were measured,and the differences between before and after chemotherapy were calculated.The detection of tumor necrosis rate with pathological method,divided tumor remission level.Single factor analysis of variance was used to compare the quantitative perfusion parameters of the samples before and after the whole course of chemotherapy.The ROC curve was used to assess the quantitative parameters and their difference before and after neoadjuvant chemotherapy in predicting the efficacy of neoadjuvant chemotherapy.A randomized forest was used to analyze the accuracy of perfusion parameters in predicting the level of remission after neoadjuvant chemotherapy.Results:9 cases were enrolled with 309 obtained tumor samples.The statistical results are as following:1.The quantitative parameters of DCE-MRI were significant different between before and after neoadjuvant chemotherapy.2.To evaluate the effectiveness,quantitative parameters after chemotherapy was better than the disparity between before and after chemotherapy.When tumor tissues expressed sensitively after chemotherapy(TNR≧90%),diagnostic threshold of the perfusion parameters after chemotherapy respectively:Ktrans less than 0.663/min(sensitivity,89.8%;specificity,52.4%);Kep less than 0.34/min(sensitivity,66.1%;specificity,65.1%);Ve less than 0.508(sensiticity 81.4%;specificity,54%);Vp less than 0.109(sensitivity,54.2;specificity,61.9%).3.The accuracy of TNR based on random forest reached 98.7%,in which post-Ktrans,post-Ve and pre-Kep contributed the most influence.Conclusions:1.The quantitative parameters of DCE-MRI varied before and after neoadjuvant chemotherapy.It is of certain value for the evaluation of chemotherapy effect.2.The random forest has a high accuracy in predicting tumor necrosis of osteosarcoma after neoadjuvant chemotherapy,which suggests that machine learning has a prospect and value in clinical study.