The Value Of PET/CT Texture Analysis In Predicting The Mid-term Efficacy And Prognosis Of Gastrointestinal Diffuse Large B-cell Lymphoma

Part I Texture analysis improved the role of pretreatment ¹⁸F-FDG PET/CT in interim response prediction of primary gastrointestinal diffuse large B cell lymphomaPurpose: To explore the role of pretreatment ¹⁸F-fluorodeoxyglucose(¹⁸F-FDG)positron emission tomography(PET)/computed tomography(CT)texture analysis(TA)in predicting the interim response of primary gastrointestinal diffuse large B cell lymphoma(PGIL-DLBCL),and compare it with metabolic and clinicopathological features.Methods: The clinicopathological features and pre-treatment ¹⁸F-FDG PET/CT images of 34 PGIL-DLBCL patients were studied retrospectively.The maximum standard uptake value(SUVmax),metabolic tumor volume(MTV)and texture features(TFs)were extracted from PET/CT images.TFs with intra-class correlation coefficients(ICCs)higher than 0.8 were selected for further analysis.The interim response was evaluated after 3-4 cycles of chemotherapy according to the Lugano criteria,and patients were divided into complete response(CR)group and non-CR group.The differences of categorical variables and continuous variables between groups were analyzed using Fisher's exact test and independent-sample T test or Mann-Whitney U test,respectively.The predictive efficacies and the optimal cut-off values of continuous variables were evaluated by receiver operating characteristic(ROC)curve analysis.A prediction probability was generated by binary logistic regression analysis.Results: There were 22(64.71%)patients achieved CR in interim response assessment.The group with intestinal involvement had significantly lower CR rate than that without intestinal involvement.The CR rates were not significantly different between groups of other clinicopathological features.The SUVmax,MTV and some of the PET/CT TFs were significantly lower in CR group(p values ranged from <0.001 to 0.048).SUVmax,PET-entropy and PET-volume were good predictor of interim response,with areas under the curves(AUCs)of ROC curve analysis to be 0.864,0.811 and 0.807,respectively(p = 0.001,0.003 and 0.004,respectively).The prediction probability generated from the combination of the SUVmax,entropy,volume and intestinal involvement had a higher AUC(0.913)than all single parameters(sensitivity = 0.92,specificity = 0.86,accuracy = 0.88).Conclusions: PET/CT TA has potential in predicting the interim response of PGILDLBCL while prospective studies with large sample sizes and validation analyses are needed to confirm the current results.The combination of TA with metabolic and clinicopathological features is conducive to the interim response prediction of PGILDLBCL.Part II The role of ¹⁸F-FDG PET/CT texture analysis in predicting progression free survival of primary gastrointestinal diffuse large B cell lymphomaPurpose: To explore the role of pretreatment ¹⁸F-fluorodeoxyglucose(¹⁸F-FDG)positron emission tomography(PET)/computed tomography(CT)texture analysis(TA)in predicting the progression free survival(PFS)of primary gastrointestinal diffuse large B cell lymphoma(PGIL-DLBCL),and compare it with metabolic and clinicopathological features.Methods: The clinicopathological features and pre-treatment ¹⁸F-FDG PET/CT images of 30 PGIL-DLBCL patients were studied retrospectively.The maximum standard uptake value(SUVmax),metabolic tumor volume(MTV)and texture features(TFs)were extracted from PET/CT images.TFs with intra-class correlation coefficients(ICCs)higher than 0.8 were selected for further analysis.The differences of categorical variables and continuous variables between groups were analyzed using Fisher's exact test and independent-sample T test or Mann-Whitney U test,respectively.The efficacies of continuous variables in predicting 3-year PFS were assessed by receiver operating characteristic(ROC)curve analysis.The Kaplan-Meier survival curves of PFS were plotted and compared using log-rank test.The potential predictors of PFS and their hazard ratios(HRs)were assessed by Cox regression analysis.Results: Ten(33.3%)patients progressed or recurred with a median PFS time of 5 months.The 3-year PFS rates were significantly lower in patients with intestinal involvement and with non-complete response(CR)in interim response assessment(p = 0.046 and 0.004,respectively).SUVmax and PET max-frequency were significantly lower in 3-year PFS group(p = 0.010 and 0.012,respectively),while CT GLCM inertia13 was significantly higher in 3-year PFS group(p = 0.043).The areas under the curves(AUCs)of SUVmax,PET max-frequency and CT GLCM inertia13 in ROC curve analysis were 0.796,0.788 and 0.730,respectively.The Kaplan-Meier survival curves found PFS time was significantly longer in interim-response-CR group,in intestinalnot-involved group,in SUVmax<20.2 group,in PET-max-frequency<4.5 group and in CT-GLCM-inertia13?4.4 group(?² = 5.23-10.39,p = 0.001-0.022).The univariate Cox regression analysis showed that the HRs of interim response non-CR,intestinal involvement and CT GLCM inertia were 7.00,5.37 and 0.62,respectively(p = 0.006,0.034 and 0.043,respectively).The multivariate Cox regression analysis showed nonCR in interim response assessment to be the only independent risk factor for PFS(HR = 5.50,95% confidence interval = 1.12-27.09,p = 0.036).Conclusions: PET/CT TA has potential in predicting the PFS of PGIL-DLBCL patients,and clinical features have important roles in the prognosis of PGIL-DLBCL.Comprehensively interpreting both the texture and clinical features is important in prognosis prediction.