Preoperative predictive model of recovery of urinary continence after radical prostatectomy

Abstract

Objective

● To build a predictive model of urinary continence recovery following radical prostatectomy that incorporates magnetic resonance imaging parameters and clinical data.

Patients and Methods

● We conducted a retrospective review of data from 2,849 patients who underwent pelvic staging magnetic resonance imaging prior to radical prostatectomy from November 2001 to June 2010.

● We used logistic regression to evaluate the association between each MRI variable and continence at 6 or 12 months, adjusting for age, body mass index (BMI), and American Society of Anesthesiologists (ASA) score and then used multivariable logistic regression to create our model.

● A nomogram was constructed using the multivariable logistic regression models.

Results

● In total, 68% (n=1,742/2,559) and 82% (n=2,205/2,689) regained function at 6 and 12 months, respectively.

● In the base model, age, BMI, and ASA score were significant predictors of continence at 6 or 12 months on univariate analysis (p <0.005).

● Among the preoperative magnetic resonance imaging measurements, membranous urethral length, which showed great significance, was incorporated into the base model to create the full model.

● For continence recovery at 6 months, the addition of membranous urethral length increased the AUC to 0.664 for the validation set, an increase of 0.064 over the base model. For continence recovery at 12 months, the AUC was 0.674, an increase of 0.085 over the base model.

Conclusions

● Using our model, the likelihood of continence recovery increases with membranous urethral length and decreases with age, body mass index, and ASA score.

● This model could be used for patient counseling and for the identification of patients at high risk for urinary incontinence in whom to study changes in operative technique that improve urinary function after radical prostatectomy.