Establishment Of A Prediction Model For Biochemical Recurrence Following Radical Prostatectomy Based On Postoperative Short-term PSA Changes

Background:Radical prostatectomy is one of the most important treatment modes for early prostate cancer（PCa）,but approximately 15%–45%of patients experience biochemical recurrence（BCR）following the operation.Accurately predicting postoperative BCR time is important for guiding the subsequent treatment.In 1999,based on preoperative PSA levels and postoperative Gleason scores（GS）,extracapsular extension（ECE）,seminal vesicle invasion（SVI）,lymph node（LN）,and positive surgical margins（PSM）,Kattan first established a nomogram that could predict the BCR of patients with PCa following radical prostatectomy,which was later verified at multiple centers in the United States.Subsequently,a series of similar prediction models based on the Kattan postoperative nomogram were proposed.However,the predictors included in the previous prediction models were limited to the clinical or pathological information obtained during the perioperative period and could not be combined with the information obtained during the postoperative follow-up.PSA as a specific tumor marker for prostate cancer,was a serum indicator that required regular detected after radical prostatectomy.Theoretically,PSA in patients at 6^th weeks after radical prostatectomy could be reduced to undetectable level（less than 0.02ng/ml）,while PSA level in some patients was still higher than 0.02ng/ml at 6^th weeks after radical prostatectomy,which might indicate that there were still tumor residues in patients and they were more possibilities to biochemical recurrence.Therefore,this study attempted to build a new nomogram prediction model based on the PSA level at the 6^th week after surgery and combined with other pathological information of patients,in an attempt to further improve the predictive efficacy.Objective:We aimed to establish a prediction model for biochemical recurrence in patients with prostate specific antigen at the 6^th weeks after radical prostatectomy and combined it with pathological stage,Gleason scores and positive surgical margins.Then we compared the new model with the classic Kattan postoperative nomogram.Methods:We screened 2220 patients from the PC-follow database who underwent radical prostatectomy between February 2004 and February 2018 and adopted 304 cases for modeling and another 102 cases for validation.Using the COX proportional hazards model,we identified independent predictors associated with biochemical recurrence and then introduced the 6th-week postoperative PSA levels,pathological stage,Gleason scores and positive surgical margins for modeling.The R software was used to plot the nomogram of the prediction model for biochemical recurrence following radical prostatectomy,whose consistency index（C-index）was then calculated to obtain the calibration curve between the predicted value from the nomogram and the actual value.Internal verification was accomplished via repeated sampling（1,000 times）using the Bootstrap method,while external verification was conducted using data from the validation group.Finally,integrated discrimination improvement（IDI）was adopted to evaluate the prediction performances of the new nomogram model and the Kattan postoperative nomogram.Results:The C-index of the new model was 0.905（95%CI:0.876-0.934）.The calibration curve of the new model demonstrated superior consistency between the predicted and actual value.The C-index of the validation group was 0.875（95%CI:0.833-0.917）.External verification validated the applicability of the new model,confirming that it could accurately predict 3-year biochemical-recurrence-free probability.The IDI demonstrated a 31.92%improvement in prediction performance over that of the Kattan nomogram（P<0.001）.Patients were divided into high risk groups and low risk group based on the cut-off value.The BCR-free rate was 88.6%in the low risk group,however,was 0 in the high risk group.Almost 69.41%unnecessary follow-up could be reduced based on the new model.Conclusion:The new nomogram model could predict biochemical recurrence probability based on the 6^th weeks postoperative PSA levels,pathological stage,Gleason scores and positive surgical margins with a higher accuracy than that of the classic Kattan postoperative nomogram.The model improved convenience in clinical applications and provided a basis for developing individualized follow-up plans.