Timo Soeterik

89 Nerve Sparing and Risk of Positive Margins Statistical analysis Each prostate lobe was considered as a separate case. A multilevel regression model was used to evaluate the association between nerve sparing and positive surgical margins. Side-specific factors included nerve sparing, DRE, mpMRI local stage (organ confined vs EPE), highest ISUP grade found on biopsy and percentage of positive cores. Covariates available on patient level included PSAD, surgeon, hospital and age. To adjust for the consequential data clustering on patient level a random intercept was included in the model. Missing data were assumed to be missing at random, based on the missing data patterns, and were imputed using multiple imputations. 13 Analysis was performed using R Studio. 14 RESULTS Study population A total number of 2574 patients underwent RARP from 2013 to 2018 at the 4 hospitals. The baseline characteristics and surgical outcomes on prostate lobe level are presented in Table 1. Patient-level baseline characteristics and surgical outcomes are presented in Supplementary Table S1 and S2. Positive surgical margin rates were observed in 844 (33%) of 2574 cases. The positive margin rate was 23% in pT2 (353 of 1533) and 47% in pT3 or greater (491 of 1041) tumours. A total of 1774 (69%) patients underwent interfascial nerve sparing surgery (unilateral or bilateral). Nerve sparing vs Non-nerve sparing Nerve sparing status was not available in 97 patients and, thus, these patients could not be categorized. Baseline characteristics of 4954 prostate lobes of the remaining 2477 patients with known nerve sparing status are presented in Table 1. Overall, the nerve sparing group had relatively more favourable tumour characteristics compared with the non-nerve sparing group (Table 1). Missing data Of the 2574 patients who underwent RARP during the study period, data relevant for analysis were missing in 889. This was mainly attributable to the fact that 364 patients (14%) did not undergo preoperative mpMRI and 263 (10%) underwent targeted biopsies without systematic biopsies. Thus, radiological T-stage and prostate side-specific percentage of positive cores were not available in these cases. In addition, preoperative prostate volume was not determined in 86 (3%) cases. Extensive information regarding missing data is given in a patient flow chart (Supplementary Figure). 5