You have accessJournal of UrologyPenile & Testicular Cancer: Penile & Testicular Cancer III (MP76)1 Apr 2020MP76-05 VALIDATION OF THE BEST MODELS TO PREDICT PATHOHISTOLOGY IN GERM CELL TUMOR PATIENTS UNDERGOING POSTCHEMOTHERAPY RETROPERITONEAL LYMPH NODE DISSECTION Tim Nestler*, Pia Paffenholz, Bettina Baeßler, Martin Hellmich, Andreas Hiester, Alessandro Nini, David Pfister, Peter Albers, and Axel Heidenreich Tim Nestler*Tim Nestler* More articles by this author , Pia PaffenholzPia Paffenholz More articles by this author , Bettina BaeßlerBettina Baeßler More articles by this author , Martin HellmichMartin Hellmich More articles by this author , Andreas HiesterAndreas Hiester More articles by this author , Alessandro NiniAlessandro Nini More articles by this author , David PfisterDavid Pfister More articles by this author , Peter AlbersPeter Albers More articles by this author , and Axel HeidenreichAxel Heidenreich More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000000962.05AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: The aim of our study was to validate and evaluate the two currently best performing prediction models (Vergouwe and Leao) for final pathohistology in NSGCT patients undergoing postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) and we introduce a new radiomics approach. METHODS: We performed a retrospective analysis including 496 patients who underwent a PC-RPLND between 2008 and 2018 to validate the two prediction models using published formulas and thresholds. ROC were plotted and AUC was calculated. We determined the optimal cut point and used bootstrapping (1,000 replications) to estimate its variability. For radiomics, lymph nodes of 80 patients were identified on CT images, semiautomatically segmented with 93 radiographic features (pyRadiomics package). A linear support vector machine algorithm was applied to analyze reproducible radiomics features. A continuous reduction of features analyzed was performed using Random Forest algorithms and ROC analysis. RESULTS: In our validation cohort, the Vergouwe model had a significantly better AUC compared to Leao model (0.749 [CI 0.706-0.792] vs. 0.689 [0.642-0.736], p = 0.004) to predict benign histology. At a threshold of > 70% for the probability of benign disease, the Leao model would have avoided PC-RPLND in 8.6% with benign disease with an error rate of 5.6% for viable tumor. The Vergouwe model would avoid PC-RPLND in 23.4% with benign disease with an error rate of 12.7% for viable tumor/teratoma. Of the 93 radiomic features analyzed, 51 features were reproducible. Applying the trained algorithm on the training dataset resulted in an accuracy of 0.96 (93% sensitivity, 100% specificity, 100% PPV), on an independent validation cohort the accuracy was 0.81 (88% sensitivity, 72% specificity, 78% PPV). CONCLUSIONS: According to our data, the discriminatory accuracy of both models is not sufficient to safely select patients for surveillance strategy instead of PC-RPLND. The radiomics model is promising but needs prospective validation. Further studies including new biomarkers are needed to optimize the accuracy of potential prediction models. Source of Funding: none © 2020 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 203Issue Supplement 4April 2020Page: e1153-e1153 Advertisement Copyright & Permissions© 2020 by American Urological Association Education and Research, Inc.MetricsAuthor Information Tim Nestler* More articles by this author Pia Paffenholz More articles by this author Bettina Baeßler More articles by this author Martin Hellmich More articles by this author Andreas Hiester More articles by this author Alessandro Nini More articles by this author David Pfister More articles by this author Peter Albers More articles by this author Axel Heidenreich More articles by this author Expand All Advertisement PDF downloadLoading ...
Read full abstract