Value Of PET/MR-IVIM In Predicting Early Prognosis Of Patients With Concurrent Chemoradiotherapy For Cervical Cancer

Objective:To evaluate the value of PET/MR multiple parameters before treatment and MR-intravoxel incoherent motion（IVIM）parameters 4 weeks after concurrent chemoradiotherapy（CCRT）in predicting the prognosis of cervical cancer patients 6months after treatment.Material and Methods:From September 2017 to March 2019,58 cervical cancer patients（age:35-78 years）performed CCRT in Shengjing Hospital of China Medical University were retrospectively enrolled.PET/MR was performed before treatment,and MR-IVIM was performed at the end of the fourth week of treatment and 6 months after treatment.Patients were divided into complete response（CR）group and non-CR group according to the response evacuation criteria in solid tumors（RECIST）.PET imaging parameters（maximum standardized uptake value（SUVmax）,metabolic tumor volume（MTV）,total lesion glycolysis（TLG））,MR-IVIM parameters(slow diffusion coefficient（D）,fast diffusion coefficient（D*）,perfusion fraction（F）,gross tumor volume（GTV）,diffusion volume（DV）before（referred to as 1）and after（referred to as 2）treatment and their percentage changes（ΔD,ΔD*,ΔF,ΔGTV,ΔDV）were measured.The parameters of PET and MR-IVIM were analyzed by the receiver operating characteristic（ROC） curve.Multiple logistic regression was used to establish the combination prediction model of parameters before treatment and all parameters before and during treatment（respectively Combination₁and Combination₂）,and the two models were compared by De Long test.Results:Of 58 patients,37（63.79%）were CR,and 21（36.21%）were non-CR.MTV（area under curve（AUC）=0.764,P<0.01）,TLG（AUC=0.719,P<0.01）,GTV₁（AUC=0.761,P<0.01）and DV₁（AUC=0.785,P<0.01）before treatment had positive effects on the treatment response,and the valuable parameters after treatment were D₂（AUC=0.683,P=0.02）,GTV₂（AUC=0.861,P<0.01）,DV₂（AUC=0.891,P<0.01）,ΔD（AUC=0.723,P<0.01）,ΔGTV（AUC=0.768,P<0.01）andΔDV（AUC=0.865,P<0.01）.The best prediction model before treatment was DV₁（Combination₁;AUC=0.785,P=0.004）.The best prediction model for all parameters before and after treatment was the combination of MTV,DV₂andΔD（Combination₂;AUC=0.965,P<0.001）.Combination₂was superior to Combination₁（z=2.982,P=0.003）.Conclusion:In evaluating the early prognosis of cervical cancer patients treated with CCRT,DV₁has good predictive potential before treatment,while the combination of pre-treatment MTV with post-treatment DV₂andΔD has superior joint predictive potential and can provide help for patients’ personalized treatment.