Introduction

Radical prostatectomy (RP) is a treatment option with curative intent in clinically localized prostate cancer (PC) [1]. However, urinary incontinence (UI) represents a significant adverse consequence, detrimentally impacting daily quality of life [2] and fostering surgical regret [3]. Despite a gradual improvement of urinary continence (UC) recovery during the first year after robot-assisted radical prostatectomy (RARP), the rate of UI employing a pad-free definition ranges from 4 to 31%, with a mean value of 16% at 12 months [4]. To preserve UC after RARP, several surgical techniques have been proposed [5]. In contrast to conventional RP, whether performed through open, laparoscopic or robotic techniques, these still reach the prostate anteriorly, dissecting the bladder from the anterior abdominal wall. The novel Retzius-sparing robot-assisted radical prostatectomy (RS-RARP) accesses the prostate from a posterior approach, preserving the integrity of the pubovesical complex and the endopelvic fascia, while maintaining the anterior fixation of the bladder to the abdominal wall [6, 7]. Compared to conventional surgery, RS-RARP is associated with faster UC recovery [8].

The search for predictors of UC recovery through measurable anatomic parameters on preoperative pelvic magnetic resonance imaging (MRI) is work in progress [9, 10]. Among these predictors are prostate volume (PV), membranous urethral length (MUL), membranous urethral volume (MUV), levator ani muscle thickness (LAT), inner (ILD) and outer (OLD) levator ani muscle distance [9,10,11,12,13,14]. So far, predictors for short-term UC recovery after RS-RARP include PV, pre- and post-operative MUL [15,16,17]. The impact of measurable anatomic parameters on preoperative MRI for long-term continence outcomes after RS-RARP has yet to be described.

The objectives of the present study include the comprehensive assessment of both clinical and anatomical factors to elucidate their correlation with UC as well as the development of a reliable yet straightforward predictive tool, for estimating the probability of definitive UC recovery. This endeavor requires a longitudinal evaluation of parameters that may impact urinary outcomes.

Patients and methods

Patients and functional status evaluation

This study comprises 213 consecutive patients who were evaluated with a preoperative 3.0-Tesla MRI and underwent RS-RARP, both performed at our institution between July 2017 and December 2022. The following parameters were collected prospectively: age, body mass index (BMI), Charlson comorbidity index (CCI), preoperative oncological parameters such as prostate-specific antigen (PSA), highest International Society of Urological Pathology (ISUP) grade group on biopsy, and MRI-based tumor clinical staging using Prostate Imaging-Reporting and Data System Version 2.1 (PI-RADSv2.1) [18]. All patients had localized PC and were evaluated for urinary function preoperatively and postoperatively at 3, 6 and 12 months through pad count, by answering the third question of the Expanded Prostate Cancer Index Composite-26 questionnaire (EPIC-26) [19]. Continence was defined as no pad use. Excluded were those who underwent neoadjuvant or adjuvant radiotherapy (n = 6), prior pelvic surgery (n = 6) or missing follow-up information for continence (n = 43), leaving 158 patients eligible for analysis. Data collection followed the principles outlined in the Declaration of Helsinki. All patients provided written informed consent for study inclusion, approved by the Institutional Ethics Committee (Approval 7.7.2017).

MRI protocol and measurements

MRI was performed with a 3.0-Tesla system (Ingenia, Philips Healthcare, The Netherlands), equipped with a phased array coil, parallel imaging and multi-transmission radiofrequency capabilities, in the following free-breathing imaging sequences: (a) fast sagittal TSE T2-weighted (TR/TE, 7400 ms/100 ms; FOV, 300 mm × 300 mm; section thickness 5 mm; acquisition time, 36 s) used as localizer; (b) axial high b-value diffusion-weighted (TR/TE, 4000 ms/90 ms; b-values of 0 and 1400 s/mm2; FOV, 200 mm × 200 mm; section thickness 4 mm); (c) axial, coronal and sagittal high-resolution TSE T2-weighted (TR/TE, 3000–4000 ms/85–135 ms; FOV, 180–250 mm × 180–300 mm; section thickness 3 mm) and (d) axial TSE T1-weighted (TR/TE, 580 ms/10 ms; FOV, 290 mm × 350 mm; section thickness 5 mm). These were complemented by dynamic contrast-enhanced volumetric gradient echo T1-weighted imaging (TR/TE, 6 ms/3 ms; FOV, 180 mm × 180 mm; section thickness 3 mm).

MRI measurements were performed as previously described and standardized [9, 20, 21] in axial and sagittal T2 planes with simultaneous cross-coronal T2 demarcation ensuring measurements were performed in the midsagittal T2 plane as per protocol. These were recorded as follows: (i) PV (ml) = π/6 × maximum height of the prostate × maximal prostate width × maximal prostate length; (ii) maximum height of the prostate measured from base to apex (midsagittal T2-weighted image); (iii) maximal prostate width and maximal prostate length measured at the same axial level, at the level of the maximal prostate width (axial T2-weighted image); (iv) MUL (mm) = distance between the apex of the prostate to the upper border of penile bulb, at the dorsal side of the urethral lumen (sagittal T2-weighted image); (v) urethral width (mm) = maximal diameter of urethra, measured immediately below the caudal margin of prostate apex (axial T2-weighted image); (vi) MUV (ml) calculated as urethral length × π × (urethral width/2)2; (vii) ILD (mm) = the narrowest distance from inner border of levator ani muscles immediately below the prostatic apex (axial T2-weighted image); (viii) OLD (mm) = distance from outer border of levator ani muscles measured at same level as ILD (axial T2-weighted image) and (ix) LAT (mm) calculated as (ODL-IDL)/2 (Fig. 1).

Fig. 1
figure 1

a Midsagittal T2-weighted image. The yellow arrow displays maximum height of the prostate, measured from base to apex. The red arrow defines MUL, measured from the apex of the prostate to the upper border of penile bulb, at the dorsal side of the urethral lumen. b Axial T2-weighted image at maximal prostate width. The horizontal yellow double arrow shows maximal prostate width and the vertical yellow double arrow displays maximal prostate length. c Axial T2-weighted image immediately below the caudal margin of prostate apex. The horizontal red double arrow represents urethral width. The shorter and the longer yellow double arrows display the ILD and OLD, respectively. ILD inner levator ani muscle distance, MUL membranous urethral length, OLD outer levator ani muscle distance

Full size image

MRI measurements were performed by one urologist blinded to all clinical and pathological data. The urologist underwent training under the supervision of an expert radiologist (15 years of experience).

Operative technique

Surgeries were performed using the 4-arm Da Vinci Xi surgical system (Intuitive Surgical, Sunnyvale, CA). The RS-RARP technique was originally described by Galfano et al. [22]. This posterior surgical approach requires a careful dissection of detrusor fibers from the base of the prostate to prevent ureteral damage, ultimately ensuring the preservation of the bladder neck [23]. Unilateral or bilateral neurovascular excision was undertaken in patients with erectile dysfunction, high risk or locally advanced disease. The anastomosis between the urethra and the bladder was performed with a running, barbed 3/0 polyglyconate suture (V-LOC®; Covidien, Mansfield, MA). The integrity of the urethrovesical anastomosis was confirmed intraoperatively with intravesical instillation of 150 ml of sterile saline through a 16F Foley catheter.

Pathology examination

The same senior pathologist reviewed prostatic biopsies and surgical specimens. The ISUP consensus and the American Joint Committee on Cancer 8th edition schemes were followed for grading and staging.

Statistical analysis

Quantitative variables were reported as medians and interquartile range (IQR), and qualitative variables as counts or percentages. We employed univariate logistic regression analysis to evaluate possible significant associations between patients’ demographic and clinical factors, preoperative MRI parameters, intraoperative surgical variables, and UI at 3, 6 and 12 months post-surgery. We then used multivariable logistic regression analysis to assess independent significant predictors of continence at those time points. Finally, we established a comprehensive model for estimating the probability of UC recovery which only included variables that exhibited independent statistical significance in the multivariate analysis, based on a “Backward Conditional” method. The quality of the models at each time point was evaluated based on the area under the receiver operating characteristic curve (AUC).

All statistical analyses were performed using IBM® SPSS® Statistics version 27 for Windows, the level of significance was set at p < 0.05 and a CI of 95% was assumed.

Results

The median patient age was 60 years, median BMI 27 kg/m2 and 86% of the patients had a CCI ≤ 2. Of note, median PV was 32.7 ml (IQR 25.1−43.3), median MUL 15.1 mm (IQR 12.1−17.4) and median MUV 1.2 ml (IQR 0.9−1.6). Remaining patient and tumor characteristics as well as MRI measurements are presented in Table 1.

Table 1 Patient and tumor characteristics, and MRI anatomic parameters
Full size table

Pathological assessment revealed locally advanced disease in 31% of the patients, and the rate of positive surgical margins (over 3 mm in length) was 12%. A total of 59% of the patients underwent bilateral nerve-sparing, while bladder neck preservation was performed in 97%. Additional surgical variables and tumor pathological staging are presented in Table 2.

Table 2 Surgical variables and tumor pathological staging
Full size table

Concerning UC, it improved with time, 63.3%, 77.9% and 89.1% of the patients were pad-free at 3, 6 and 12 months post-surgery, respectively. Univariate analysis demonstrated that age, PV, MUL, MUV and CCI were significant predictors for UI at 3 months; PV, MUL and MUV were significant predictors for UI at 6 months; MUL was the only significant predictor of UI at 12 months (Table 3). Multivariate analysis revealed that except for CCI, the same variables remained significant at 3 months. However, by 6 months, age and MUL had lost significance (Table 4). Regarding the performance of the multivariate models at 3, 6 and 12 months, the AUC was 0.782 (CI 0.707–0.856), 0.783 (CI 0.700–0.865) and 0.690 (CI 0.529−0.850), respectively. Our prediction model for UC recovery at 3 months after RS-RARP includes age, PV, MUL and MUV. At 6 months, the model includes PV and MUV, while at 12 months, MUL stands as the sole predictor. The prediction model and tool are available in the supplementary material. At 12 months, the risk of UI with MUL < 10 mm was 25%. Conversely, the risk of UI with MUL > 15 mm was 5% (Fig. 2).

Table 3 Univariate logistic regression of factors predicting post-operative urinary incontinence at 3, 6 and 12 months
Full size table
Table 4 Multivariate logistic regression of factors predicting post-operative urinary incontinence at 3, 6 and 12 months
Full size table
Fig. 2
figure 2

The blue line represents the probability of a patient being urinary continent at 12 months after Retzius-sparing robot-assisted radical prostatectomy, based on the logistic regression model. The orange dots represent the observed proportion of urinary continent patients, calculated according to 2.5 mm MUL intervals

Full size image

Discussion

The advantageous impact of RS-RARP over conventional RARP on early UC recovery has been established in a landmark study in which at 3 and 12 months post-surgery, 75.7% (95% CI 64.6%−86.9%) and 95.8% (95% CI 90.4%−100%) of patients were pad free, accordingly [8]. Early on, Chang et al. established through post-operative cystourethrography, that RS-RARP was associated with less bladder neck descent and improved early continence outcome [24]. Several randomized controlled trials have established similar outcomes [25,26,27], alike our dataset, where the rate of pad-free patients was 63.3% and 89.1% at 3 and 12 months post-surgery, respectively. The rate of UC recovery consistently slows after the first three months. Consequently, the development of a predictive tool for UC recovery within the 3-month timeframe holds significance, as well as the definitive long-term status by the 12-month mark [28].

Predictive models are also useful for postoperative treatment planning. Conservative treatment, including pelvic floor stimulation and biofeedback, can be offered earlier for patients at high risk of developing long-lasting UI. Such a proactive strategy may help patients who have the potential for recovery [29]. We developed a straightforward prediction tool for continence after RS-RARP, incorporating clinical factors and anatomic parameters. We omitted predictors related to surgeon’s experience based on our prior research, which highlighted the robustness of RS-RARP for UC recovery regardless of surgeon experience [30]. In our data set, age, PV, MUL and MUV are significant predictors for continence at 3 months. The AUC of the model stands at 78.2%, indicating a moderate level of accuracy. Several studies and meta-analyses [10, 31, 32] have also demonstrated that MUL is the most studied and reliable urethral parameter to integrate a prediction tool [21]. Our investigation similarly demonstrated the significant contribution of a longer MUL to UC recovery. Most importantly, we have been able to find a threshold that enables pre-operative patient counselling regarding definitive UC recovery.

When considering several surgical approaches (open, laparoscopic and robotic prostatectomy), multivariate analysis confirmed that age and MUL were significant independent predictors for early UC recovery in a patient population with a median age of 71 years old [33]. Wang et al. also established that patients with a longer MUL had better UC recovery at one and 3 months post-RARP [34]. In another study, at 12 months post-surgery, unlike our findings, age has been described to impact continence upon univariate analysis [11]. Nonetheless, the latter study involved patients subjected to conventional RARP, once again potentially accounting for the observed discrepancy. However, in alignment with our data, both MUL and surgical experience, rather than patient age, have been described as predictors of UI at 12 months following RARP [35]. Surgical experience is a recognized predictive factor for UI following RARP [4], yet its predictive ability does not extend to RS-RARP [30]. Even though PV has been correlated with age and a protruded median lobe (p < 0.001) [36], the latter has also been described to impact early postoperative UC recovery [37]. In our study, PV is a significant predictor of early UC recovery independently of age, and we highlight its usefulness as already integrated in the PI-RADSv2.1 protocol [18]. Lastly, our lack of association between BMI and comorbidity is not unexpected given that these are also absent following RP [38].

Within our dataset, the parameters of LAT, ILD, and OLD measured in the axial T2-weighted sequence did not emerge as statistically significant predictors of UC recovery at any time point. Different measurement standards and surgical techniques may assign varying degrees of importance to identical predictors for UC recovery [39]. For example, LAT measured on an axial T2-weighted sequence was described as an independent predictor of immediate UC recovery following laparoscopic radical prostatectomy [40]. When measured on a coronal T2-weighted sequence it was found to also behave as an independent predictor of UC recovery three months after RARP [12]. However, in line with our study, a recent systematic review failed to identify any connection between LAT and long-term UC recovery [41].

Bladder neck and neurovascular bundle (NVB) preservation were not significantly associated with post-operative UC and indeed they have not consistently been reported as predictors of UC recovery [38, 42, 43]. During RS-RARP the detrusor is carefully peeled away on the posterior bladder neck to avoid ureteral lesions and the bladder neck is almost always preserved [36]. In our series, it was preserved in 97% of patients, preventing conclusions as regards the impact of bladder resection on postoperative UI.

We acknowledge that MRI measurement might have moderate interrater agreement which is a bias for the generalization of our results based on a single rater [39, 44], an issue which may soon be solved by automated MRI measurements [45]. We attempted to mitigate this limitation by following a standardized protocol for midsagittal MUL measurements [20, 21] and as such we believe it does not affect our conclusions [39]. Other limitations include a pre-operative lack of assessment for the presence of lower urinary tract symptoms which could affect UC recovery, a single institution perspective and a limited number of patients. Finally, our findings require additional external validation.

Conclusions

To the best of our knowledge, we have developed the first tool to predict individual postoperative UI rates following RS-RARP, using preoperative clinical and measurable anatomical features on preoperative MRI. Besides providing valuable guidance for counseling, it also draws the attention of the surgeon toward anatomic parameters which are not routinely reported on the PI-RADSv2.1 protocol [18] but are included in the prediction model. Our prediction tool is user-friendly, has acceptable accuracy, and is intended to help urologists estimate the individual probability of UC recovery, defined as pad-free status, at 3 months post-surgery. Furthermore, we highlight the importance of MUL as the single independent predictor of continence at 12 months. By establishing MUL thresholds for UC recovery, we enable precise patient counseling, provide realistic expectations and allow informed choices that enhance quality of life.