Construction And Validation Of Predictive Models For Accessory Renal Artery And Multiple Tumor-feeding Segmental Renal Arteries Based On Three-Dimensional Computed Tomography Image Reconstruction In RCC Patients

Objective1.To construct and validate a predictive model for assessing the presence of an accessory renal artery(aRa)in renal cell carcinoma(RCC)patients based on data measured on three-dimensional(3D)sphere.2.To develop and validate a predictive model with tumor size measured in 3D sphere for multiply tumor feeding segmental arteries(TFSRA)in RCC patients.3.Exploring the guiding significance and application mode of 3D model in retroperitoneal laparoscopic kidney surgery.Methods1.Construction and validation of a predictive model for assessing the presence of ARa.A total of 170 RCC patients from December 2015 to October 2020 presented to our hospital for pre-operative high-resolution abdominal CT examinations were enrolled.Then images were processed by dedicated Medical Imaging Three Divisional Visualization System.Constructing the prediction model using the data measured on 3D reconstructions and then,imaging data in 2D sphere of those patients were measured to evaluate the accuracy of the formula in 2D sphere.Meanwhile,75 patients’CT data with cT1-cT3 stage renal cell carcinoma downloaded from TCIA imaging data base were used to complete the reconstruction to validate the formula’s predictive efficacy with third-party data.2.Development and validation of a predictive model for multiply TFSAs in RCC patients.Among the 170 RCC patients,the TFSRA of 132 patients with stage cT1 could be clearly identified in 3D reconstructions.Parameters related were measured to develop a predictive model for multiply TFSAs and then,tumor data in 2D sphere of those patients were measured to evaluate the accuracy of the predictive formula in 2D sphere.Meanwhile,downloading 75 patients’ CT data from TCIA imaging data with cT1 stage and validating the formula’s predictive efficacy using parameters measured in their reconstructions as third-party data.3.Application of 3D reconstruction models in laparoscopic RCC surgery.3.1 Laparoscopic Radical Nephrectomy(LRN):Clinical data of 84 RCC patients who underwent LRN from December 2015 to October 2020 were enrolled.Among them,3D reconstruction of 42 patients were completed preoperatively to guide the renal surgery(experimental group),while 42 patients without performing 3D reconstruction and CT images were used as intraoperative surgical guidance(control group).The basic profile and perioperative data of the patients were statistically analyzed.3.2 Laparoscopic Partial Nephrectomy(LPN):A total of 179 RCC patients who underwent LPN from December 2015 to October 2020 were collected.Among them,3D reconstruction of 101 patients were completed preoperatively to guide the renal surgery(experimental group),while 78 patients without performing 3D reconstruction and CT images were used as intraoperative surgical guidance(control group).The basic profile and perioperative data of the patients were statistically analyzed.Result1.A total of 170 RCC patients were collected.Gender,renal artery diameters,target-opposite renal artery diameter ratio(TOr),and kidney length were screened as independent predictors of ARa.Gender、renal artery diameters、Tor、renal length were included in the predictive model and nomogram by logistic regression analysis and the area under the receiver operating characteristic curve(AUC)of the predictive mode was 0.791 for diagnosing ARas.When using the renal length measured in 2D square,the AUC was 0.770,and in the validating group using data form TCIA database,the AUC was 0.878.2.Tumor size was screened as an independent predictor of multiple TFSAs by logistic regression analysis,and the AUROC value of the predictive model was 0.804.When using the tumor size measured in 2D square,the AUROC was 0.789,and in the validating group using data form TCIA database,the AUROC of was 0.928.3.The basic profile and tumor characteristics of patients underwent LPN between the two groups has not statistically difference.In terms of surgical results of the experimental group and control group,the surgical time was 117.5±51.3 vs.154.713±44.7 min(p=0.033);the renal warm ischemia time was 27.12±7.413 vs 30.589±7.547 min(p=0.011).4.The basic profile and tumor characteristics of patients underwent LRN between the two groups has not statistically difference.The surgical time in the experimental group was shorter than that in the control group(117.5±51.3 vs 154.713±44.7 min,p=0.033),and the rest indexes were not statistically significant.Conclusion1.A predictive model of ARas and nomogram were developed based on gender,renal artery diameter,Tor and renal length.The predictive model could visually and accurately predict the presence or absence of ARas in RCC patients.2.A predictive model of multiply TFSAs and nomogram were developed based on tumor size,which could visually and accurately predict the presence or absence of multiply TFSAs in RCC patients.3.preoperative 3D restrictions help surgeons to have preoperative knowledges of the renal vasculature and precise separation of targeted vessels,which improve surgical efficiency and safety and reduce renal warm ischemia in LPN.