Enhancement Effect of Pre-Treatment 18F-FDG PET/CT Radiomic Features in Prognostic Evaluation of Diffuse Large B-Cell Lymphoma / 中国医学影像学杂志

ABSTRACT

Purpose To investigate the predictive value of 18F-FDG PET/CT radiomic features before treatment for progression free survival(PFS)and overall survival(OS)in diffuse large B-cell lymphoma(DLBCL).Materials and Methods A total of 135 patients with pathologically proven DLBCL in Chinese PLA General Hospital from January 2016 to December 2018 and 18F-FDG PET/CT imaging prior to treatment were retrospectively collected.Patients were randomly divided into training set and test set using 8∶2,and then the training set was divided into training set and verification set using 8∶2 to construct the model.Semi-automatically delineated patients'lymphoma lesions as regions of interest and extracted the features.Univariate COX and least absolute shrinkage and selection operator regression were used to screen the features,and the non-zero radiomic features were obtained and the weight coefficients were used to calculate the Radscore value of each patient,and the predictive value of Radscore on PFS and OS was analyzed.Three models were established using traditional prognostic indicators(metabolic parameters and clinical factors),Radscore and combination.C-index,time-dependent area under the curve and decision curve were used to evaluate the model prediction efficiency.Finally,based on the optimal model,a column diagram was drawn,and the calibration curve was used to verify the efficiency of the column diagram.Results The combined model predicted PFS and OS at 3 and 5 years better than the traditional prognostic index model and Radscore model(Z=0.962 1-2.253 9,all P<0.05).Decision curve showed that the combined model achieved the greatest clinical net benefit.The calibration curve showed that the predicted values of the nomogram were in good agreement with the observed values.Conclusion Radscore is an independent prognostic factor for survival in DLBCL patients.The combined model has great application value in guiding the formulation of clinical individualized treatment plan.