Establishment and validation of nomogram for positive surgical margin of prostate cancer

Objective To establish a nomogram model for predicting positive resection margins after prostate cancer surgery, and to perform the corresponding verification, in order to predict the risk of positive resection margins after surgery. Methods A total of 2 215 prostate cancer patients from The First Affiliated Hospital of Naval Medical University, Hospital, Peking University First Hospital, Peking University Third Hospital, Peking University, and First Affiliated Hospital of Xi′an Jiaotong University were included in the PC-follow database from 2015 to 2018, and a simple random sampling method was used. They were divided into 1 770 patients in the modeling group and 445 patients in the verification group. In the modeling group, the age ( 70 years), PSA ( 20 ng/ml), pelvic MRI (negative, suspicious, positive), clinical stage of the tumor (T1-T2, ≥T3), percentage of positive needles (≤33%, 34%-66%, >66%), Gleason score of biopsy pathology (≤6 points, 7 points, ≥8 points). Univariate and multivariate logistic analysis were performed to screen meaningful indicators to construct a nomogram model. The model was used for validation in the validation group. Results The results of multivariate analysis showed that preoperative PSA level (OR=2.046, 95%CI 1.022 to 4.251, P=0.009), percentage of puncture positive needles (OR=1.502, 95%CI 1.136 to 1.978, P=0.002), Gleason score of puncture pathology (OR=1.568, 95%CI 1.063 to 2.313, P=0.028), pelvic MRI were correlated (OR=1.525, 95%CI 1.160 to 2.005, P=0.033). Establish a nomogram model for independent predictors of positive margin of prostate cancer. The area under the receiver operating characteristic (ROC) curve of the validation group is 0.776. The area under the ROC curve of the preoperative PSA level, percentage of puncture positive needles, puncture pathology Gleason score, pelvic MRI, postoperative pathology Gleason score were 0.554, 0.615, 0.556, 0.522, and 0.560, respectively. The difference between the nomogram model and other indicators was statistically significant (P<0.05). Conclusions The constructed nomogram model has higher diagnostic value than the preoperative PSA level, percentage of puncture positive needles, Gleason score of puncturing pathology, pelvic MRI, and postoperative pathological Gleason score in predicting positive margin. Key words: Prostatic neoplasms; Prostate cancer; Positive surgical margin; Models statistical