Urol J. 2023 Dec 13. doi: 10.22037/uj.v20i.7758. Online ahead of print.
ABSTRACT
PURPOSE: To analyze the risk factors for the relapse of prostate cancer (PC) after radical prostatectomy (RP) and build a nomogram as a predictive model. Materials andMethods: The patients who underwent PR from March 2019 to February 2022 were retrospectively enrolled in our hospital’s case system. During the follow-up process, two consecutive prostate-specific antigens (PSA) ≥0.2 μg/L were performed. And needle biopsy was performed to further determine whether the patient had prostate cancer recurrence. According to the follow-up results, the patients were divided into non-relapsed and relapsed groups.The related parameters of the two groups were collected. Independent risk factors for postoperative recurrence were determined using a Cox proportional hazards regression model. Statistical software, R, was used to build nomograms. R software was used to construct a nomogram, and the prediction effect of the nomogram was evaluated by the calibration curve and the area under the ROC curve (AUC).
RESULTS: Among the 367 patients who underwent RP, 112 (30.52%) had, and 255 (69.48%) did not have relapses after surgery. Cox multivariableregression analysis revealed that preoperative Gleason score, preoperative PSA, pathological staging, positive margin, and seminal vesicle invasion, were the risk factors for postoperative recurrence after RP (all P < 0.05). Verification of the predictive model by ROC curve demonstrated that the AUC of the ROC curves for patients’ relapses 3 and 5 years after RP was 0.986 (95%CI0.975-0.998) and 0.974 (95%CI0.961-0.987), respectively. This model validation showed that the results of the predictive model were basically consistent with the actual results, suggesting that the nomogram was able to accurately predict a patient’s relapse.
CONCLUSION: The nomogram of this study was a good predictor of postoperative recurrence of PC after RP, which will help doctors provide personalized treatment and follow-up strategies for patients.
PMID:38088088 | DOI:10.22037/uj.v20i.7758
