Timo Soeterik

122 CHAPTER 7 METHODS Patient population and study data After receiving institutional review board approval, data from 1871 consecutive patients diagnosed with prostate cancer that underwent robot-assisted radical prostatectomy (RARP) at three teaching hospitals were extracted from prospectively maintained databases. Of these, one patient was excluded due to prior treatment with androgen deprivation therapy. The cohort of patients that underwent RARP from 2014-2018 at the Canisius Wilhelmina Hospital (CWH) Nijmegen, was used for nomogram development. This cohort was selected for model derivation due to the population size and its multi- centre nature. Since 2013, regional prostate cancer surgery has been centralised and patients from two other hospitals (Catharina Hospital Eindhoven and Radboud University Medical Centre Nijmgen) all undergo RARP at CWH. The cohorts of patients undergoing RARP at the Hospital Group Twente in Almelo-Hengelo (validation cohort 1) and St. Antonius Hospital Nieuwegein-Utrecht (validation cohort 2) from 2015 – 2018 were used for external validation. Predictor selection We used a clinically driven, evidence-based approach for prediction selection. First, a very recent literature review was used to identify significant predictors for side-specific EPE. 13 Second, three consensus meetings were organized with clinical experts including urologists (HvM, JW, SS, and JPvB), an expert uro-radiologist (IS) and uro-pathologist (HKV). Predictors were selected based on relevance, availability and usefulness. Patient- based (prostate-specific antigen [PSA] density [PSAD]) and side-specific covariates (digital rectal examination [DRE] local staging, mpMRI-based local staging, highest International Society of Urological Pathology (ISUP) biopsy grade and percentage of positive systematic cores) were included. MRI protocol MRI was performed using 3 Tesla scanners and a body coil. Gadolinium (1 mg/kg) was intravenously administrated. Radiological reporting was done by dedicated radiologists with at least two years of experience with prostate MRI reading. MRI reporting in 2013 and 2014 was done according to the European Society of Urogenital Radiology (ESUR) guidelines. 14 From 2015 and onwards, the principles of PI-RADS v2 were followed. 15 Imaging-based T-stage was defined according to the American Joint Committee on Cancer TNM classification. 16