Mp10-15 Using Artificial Intelligence to Characterize Body Composition in Patients Undergoing Cytoreductive Nephrectomy

You have accessJournal of UrologyKidney Cancer: Advanced (including Drug Therapy) I (MP10)1 May 2024MP10-15 USING ARTIFICIAL INTELLIGENCE TO CHARACTERIZE BODY COMPOSITION IN PATIENTS UNDERGOING CYTOREDUCTIVE NEPHRECTOMY Grant M. Henning, Madeline Dorr, Ekamjit S. Deol, Spyridon P. Basourakos, Daniel D. Shapiro, E. Jason Abel, Reza Nabavizadeh, R. Houston Thompson, Stephen A. Boorjian, Bradley C. Leibovich, and Vidit Sharma Grant M. HenningGrant M. Henning , Madeline DorrMadeline Dorr , Ekamjit S. DeolEkamjit S. Deol , Spyridon P. BasourakosSpyridon P. Basourakos , Daniel D. ShapiroDaniel D. Shapiro , E. Jason AbelE. Jason Abel , Reza NabavizadehReza Nabavizadeh , R. Houston ThompsonR. Houston Thompson , Stephen A. BoorjianStephen A. Boorjian , Bradley C. LeibovichBradley C. Leibovich , and Vidit SharmaVidit Sharma View All Author Informationhttps://doi.org/10.1097/01.JU.0001008588.39303.c9.15AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Appropriate patient selection for cytoreductive nephrectomy (CN) in patients with metastatic renal cell carcinoma (RCC) is paramount. Prior work has shown associations between skeletal muscle mass and adiposity with survival for patients with RCC, yet manual segmentation is labor intensive and poorly reproducible. We evaluated the efficiency of an automated artificial intelligence (AI) algorithm to characterize body composition for patients undergoing CN. METHODS: Our institutional database was retrospectively queried for patients undergoing CN between 2005-2021. Clinicopathologic features were reviewed. Preoperative CT scans were analyzed using our deep convolutional neural network AI algorithm for automated segmentation of skeletal muscle and adipose areas at the L3 vertebral body (Figure 1). Sarcopenia was defined according to gender-specific consensus definitions, sarcopenic obesity was defined according to relative fat mass to fat-free mass ratios, and visceral-subcutaneous fat ratios were calculated. Associations with body composition indices and major complications and overall survival (OS) were evaluated using logistic and Cox regression analyses. RESULTS: 324 patients were identified, of whom 185 had preoperative CT imaging satisfactory for automated segmentation. Median age at surgery was 60 years (IQR 54-67), 48 (25.9%) patients received systemic therapy prior to surgery, and the rate of major surgical complications was 15.1% (N=28). Using our AI algorithm, we identified 66 (36%) patients with radiographic sarcopenia, 25 (14%) with sarcopenic obesity, and 64 (35%) with a visceral to subcutaneous fat ratio of at least 1. Sarcopenia was not found to be independently associated with OS following surgery. On regression analysis, neither sarcopenia (OR 0.55, p=0.21), sarcopenic obesity (OR 0.45, p=0.30), nor elevated visceral to subcutaneous fat ratio (OR 0.90, p=0.82) were statistically associated with major complications. CONCLUSIONS: Our automated AI algorithm identified that a high proportion of patients undergoing CN have radiographic sarcopenia and unfavorable adipose distribution. This preliminary work warrants further investigation to determine if AI identified body composition metrics can be incorporated into existing models to improve patient selection for CN. Download PPT Source of Funding: None © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e145 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Grant M. Henning More articles by this author Madeline Dorr More articles by this author Ekamjit S. Deol More articles by this author Spyridon P. Basourakos More articles by this author Daniel D. Shapiro More articles by this author E. Jason Abel More articles by this author Reza Nabavizadeh More articles by this author R. Houston Thompson More articles by this author Stephen A. Boorjian More articles by this author Bradley C. Leibovich More articles by this author Vidit Sharma More articles by this author Expand All Advertisement PDF downloadLoading ...