Objective:To explore the value of a model combining 18F-FDG PET/CT radiomics and clinical factors in prediction of lymphovascular invasion (LVI) in rectal cancer. Methods:This retrospective cohort study was conducted on 120 patients (86 males and 34 females; age (62.2±11.6) years) with rectal adenocarcinoma from the Affiliated Hospital of Qingdao University between January 2017 and November 2023. Patients were divided into a training set ( n=96) and testing set ( n=24) at the ratio of 8∶2 using simple random sampling without replacement with a fixed random seed. An external validation cohort consisted of 31 patients (17 males and 14 females; age (61.2±8.2) years) with rectal adenocarcinoma from Affiliated Hospital of Jining Medical University and Linyi Cancer Hospital between January 2020 and June 2024 was obtained. PET/CT-derived features were selected to build radiomics model. The χ2 test and logistic regression were used to identify clinical predictors of LVI for clinical modeling. A combined radiomics-clinical nomogram was developed, after that ROC analysis was conducted to evaluate the predictive performance. Results:Significant differences were found between LVI-positive ( n=40) and LVI-negative ( n=56) subgroups in body weight, carbohydrate antigen (CA) 19-9, metabolic tumor volume (MTV), and peak of SUV (SUV peak) in the training set ( χ2 values: 4.01-13.64, all P<0.05). Binary logistic regression identified body weight (odds ratio ( OR)=0.320, 95% CI: 0.095-0.906, P=0.033), CA19-9 ( OR=0.402, 95% CI: 0.120-0.917, P=0.033), and MTV ( OR=0.192, 95% CI: 0.090-0.575, P=0.002) as independent predictors of LVI, forming the clinical model. Thirteen PET features and fifteen CT features were selected and a radiomics model was built. ROC curve analysis showed that AUCs for the clinical model in the training, testing, and external validation sets were 0.765, 0.567, and 0.777, respectively; AUCs for the radiomics model were 0.925, 0.881, and 0.823; AUCs for the joint model were 0.938, 0.889, and 0.841. Conclusion:The joint model of 18F-FDG PET/CT radiomics and clinical factors can effectively predict LVI in rectal cancer, guiding preoperative therapy and surgical planning.

Objective:To explore the value of a model combining 18F-FDG PET/CT radiomics and clinical factors in prediction of lymphovascular invasion (LVI) in rectal cancer. Methods:This retrospective cohort study was conducted on 120 patients (86 males and 34 females; age (62.2±11.6) years) with rectal adenocarcinoma from the Affiliated Hospital of Qingdao University between January 2017 and November 2023. Patients were divided into a training set ( n=96) and testing set ( n=24) at the ratio of 8∶2 using simple random sampling without replacement with a fixed random seed. An external validation cohort consisted of 31 patients (17 males and 14 females; age (61.2±8.2) years) with rectal adenocarcinoma from Affiliated Hospital of Jining Medical University and Linyi Cancer Hospital between January 2020 and June 2024 was obtained. PET/CT-derived features were selected to build radiomics model. The χ2 test and logistic regression were used to identify clinical predictors of LVI for clinical modeling. A combined radiomics-clinical nomogram was developed, after that ROC analysis was conducted to evaluate the predictive performance. Results:Significant differences were found between LVI-positive ( n=40) and LVI-negative ( n=56) subgroups in body weight, carbohydrate antigen (CA) 19-9, metabolic tumor volume (MTV), and peak of SUV (SUV peak) in the training set ( χ2 values: 4.01-13.64, all P<0.05). Binary logistic regression identified body weight (odds ratio ( OR)=0.320, 95% CI: 0.095-0.906, P=0.033), CA19-9 ( OR=0.402, 95% CI: 0.120-0.917, P=0.033), and MTV ( OR=0.192, 95% CI: 0.090-0.575, P=0.002) as independent predictors of LVI, forming the clinical model. Thirteen PET features and fifteen CT features were selected and a radiomics model was built. ROC curve analysis showed that AUCs for the clinical model in the training, testing, and external validation sets were 0.765, 0.567, and 0.777, respectively; AUCs for the radiomics model were 0.925, 0.881, and 0.823; AUCs for the joint model were 0.938, 0.889, and 0.841. Conclusion:The joint model of 18F-FDG PET/CT radiomics and clinical factors can effectively predict LVI in rectal cancer, guiding preoperative therapy and surgical planning.