Prediction of a positive surgical margin and biochemical recurrence after robot-assisted radical prostatectomy

The positive surgical margin (PSM) and biochemical recurrence (BCR) are two main factors associated with poor oncotherapeutic outcomes after prostatectomy. This is an Asian population study based on a single-surgeon experience to deeply investigate the predictors for PSM and BCR. We retrospectively included 419 robot-assisted radical prostatectomy cases. The number of PSM cases was 126 (30.1%), stratified as 22 (12.2%) in stage T2 and 103 (43.6%) in stage T3. Preoperative prostate-specific antigen (PSA) > 10 ng/mL (p = 0.047; odds ratio [OR] 1.712), intraoperative blood loss > 200 mL (p = 0.006; OR 4.01), and postoperative pT3 stage (p < 0.001; OR 6.901) were three independent predictors for PSM while PSA > 10 ng/mL (p < 0.015; hazard ratio [HR] 1.8), pT3 stage (p = 0.012; HR 2.264), International Society of Urological Pathology (ISUP) grade > 3 (p = 0.02; HR 1.964), and PSM (p = 0.027; HR 1.725) were four significant predictors for BCR in multivariable analysis. PSMs occurred mostly in the posterolateral regions (73.8%) which were associated with nerve-sparing procedures (p = 0.012) while apical PSMs were correlated intraoperative bleeding (p < 0.001). A high ratio of pT3 stage after RARP in our Asian population-based might surpass the influence of PSM on BCR. PSM was less significant than PSA and ISUP grade for predicting PSA recurrence in pT3 disease. Among PSM cases, unifocal and multifocal positive margins had a similar ratio of the BCR rate (p = 0.172) but ISUP grade > 3 (p = 0.002; HR 2.689) was a significant BCR predictor. These results indicate that PSA and pathological status are key factors influencing PSM and BCR.


Introduction
Up to 2019, an estimated 4986 robotic systems have been installed in medical centers worldwide, of which 561 are located in Asian countries 1 . Robot-assisted radical prostatectomy (RARP) has become a standard approach for localized prostate cancer (PCa) treatment 2 . This method yields comparable oncological outcomes as previous open and laparoscopic methods, primarily with respect to positive surgical margin (PSM) and biochemical recurrence (BCR), along with enhancements in functional outcomes, including urinary continence and recovery of erectile potency [3][4][5] .
PSM detected in radical prostatectomy (RP) specimens is considered a poor oncological outcome 6 ; however, its long-term effect on mortality remains uncertain 7 . Previous studies have reported several predictors for PSM, including prostate-speci c antigen (PSA) concentration, prostate weight 8 , obesity 9 , the histopathological ndings from biopsy and RP specimens 10 , surgeon experience 11 , pathologist interpretation 12 , surgical approach 13 , and surgical method [14][15][16][17] , may potentially in uence postoperative PSM. However, data from Asian countries regarding the prediction of PSM and BCR are still lacking owing to differences in PCa phenotypes between individuals in Asian and Western countries 18 . It remains di cult for surgeons to determine the risk of PSM before surgery and the effect of PSM on the BCR rate after surgery. In this study, we used case-cohort data from an Asian medical center to further the current understanding of PSM and BCR after RARP.
Characteristics Median (IQR) Furthermore, the PSM rate was signi cantly higher at a higher grade of ISUP in RP specimens (p < 0.001). The prostate weight in prostatectomy was similar between the two groups (p = 0.141). Overall, the postoperative ndings indicated that the more advanced the disease's progression was, the higher the PSM rate was upon nal pathological examination.

The location and number of PSMs
The percentage of PSMs (Table 3) in the apex, bladder neck, and posterolateral regions was 27.7%, 13.5%, and 73.8%. Ninety-four (74.6%) RP specimens presented unifocal, while 29 (23%) presented multifocal positive margins. The PSA level, pT stage, and ISUP grade at RP were signi cantly associated with PSMs in the bladder neck, posterolateral, and unifocal and multifocal regions, respectively. The amount of intraoperative estimated blood loss was signi cantly higher in the apex of the PSM (p < 0.001), whereas an enlarged prostate median lobe was signi cantly more common in the bladder neck of PSM (9.6% vs. 3.3%; p = 0.047). Moreover, the nerve-sparing (NS) procedure was signi cantly associated with PSM in the bladder neck (p < 0.001) and posterolateral (p = 0.012) regions. Regarding the number of PSMs, BMI, previous abdominal surgery, and NS procedures were predictive factors for PSMs in multifocal regions. Prostate weight, surgical cases, and previous TURP history were not signi cantly correlated with the foci between the positive and negative margins.

Effect of PSMs on biochemical recurrence
In total, 395 (94.3%) patients were available for follow-up evaluation, and 97 (24.6%) patients developed recurrent PSA. The overall 5-year biochemical recurrence (BCR)-free survival rate was 66.7%. Figure 1 indicates that PSM's presence signi cantly decreased the 5-year BCR-free survival rate in the log-rank test (p < 0.001). Among patients with a PSM, the 1-, 3-, and 5-year BCR-free survival rates were 79.7%, 61.1%, and 41.9%; by comparison, those with an NSM were 88.9%, 80.9%, and 71.4%, respectively. Regarding the number of PSMs, the 5-year BCR-free survival curve was similar in the multifocal and unifocal groups (p = 0.172).
In univariate analysis (   Discussion RARP offers potential bene ts such as a low PSM rate compared with open RP owing to better visibility and less blood loss 19,20 . In our series, the overall PSM rate was 30.1%, which is a higher rate than that reported in high-volume RARP studies, in which the range typically was 10.8-22% 21 . This may be attributable to a much higher percentage (56.6%) of patients in the pT3 stage in our study than in other studies, where the percentage of pT3 stage patients ranged from 9.3 to 37.5% 21 .
We determined the major preoperative predictor for the PSM rate to be PSA > 10 ng/mL. The predictors of the ISUP score upon biopsy and the cT stage de ned through MRI did not exhibit a positive correlation in multivariate analysis. Liss  However, the current results demonstrated that the NS group had a lower risk of PSM in the posterolateral region than the non-NS group did (3% vs. 20%). This is probably because of the higher percentage of patients in the pT3 stage than in the pT2 stage in the non-NS group (79% vs. 21%), which may be more in uential than NS techniques.
Previous studies have been reported the apex is the most frequent region of PSM in RP specimens 31,41 .
This is attributable to the unclear prostate capsular margins, which are di cult to identify in pT2 and pT3 stages 38 . More other studies have reported that apical PSM is correlated with the surgeon's approach and skills rather than the tumor stage 21,26,41 . The surgeon in this study has extensive laparoscopic experience, explaining the relatively low rate of PSM in the apex. The estimated blood loss was signi cantly higher among patients with apical PSM, which implies potential bleeding from the dorsal vein complex upon dissection of the apical prostate 42 . Coelho et al. reported that high BMI was a predictor for apical PSM in a cohort study involving 876 RARPs 31 , and our study determined that high BMI was signi cantly associated with higher odds of PSMs at multifocal regions (p = 0.025) but not apical regions.
Koizumi et al. reported that employing the RARP approach has a higher likelihood of leading to PSM at the bladder neck than either ORP or LARP do 29 . This is possible because of the excessive preservation of bladder neck tissue to secure postoperative urinary continence 15 . Furthermore, the presence of a prominent median lobe during surgery might increase the risk of PSM over the bladder neck (p = 0.047), indicating the challenging task of identifying surgical margins between the protruding prostatic lobe and bladder neck.
Several studies have reported patients who underwent prostatectomy with a 5-year BCR-free survival rate ranging from 74-87% and a median PSA recurrence time of 2.6 years [43][44][45] . In this study, the 5-year BCRfree survival rate was 66.7%, which was lower than that reported in other studies; this is probably owing to the higher ratio (53%) of aggressive pT3 disease at the outset of the accumulation of RARP cases.
Evidence supports the characterization of PSM as a strong predictor of disease progression [46][47][48]  Exclusion criteria: patients with a history of neoadjuvant hormone therapy or any focal treatment, including radiotherapy, cryotherapy, high-intensity focused ultrasound therapy, and salvage prostatectomy, were excluded. Furthermore, patients lost to follow-up within two months of RARP, and those who received adjuvant therapy owing to adverse pathological outcomes before PSA relapse were excluded.
Preoperative PSA levels were determined immediately before the prostate biopsy, and postoperative PSA levels were determined within 1-month post-biopsy and then at 3-month intervals until PSA recurrence was con rmed. Time zero marked the date of RARP, and patients without BCR did not present PSA recurrence on the most recent follow-up evaluation before the end of 2019. An increase in serum PSA levels was identi ed twice, and other factors potentially elevating PSA levels were excluded.

RARP
All RARPs were performed by a single urologist (H.J.C) who had > 15 years' experience in laparoscopic urological surgery. For RARP, the transperitoneal approach was adopted, employing the Intuitive Surgical da Vinci® Surgical System Si or Xi with six ports. The prostatic anterior fat pad was removed to visualize the prostatic boundaries. The bladder neck was opened and separated from the prostate. By dividing the vesicoprostatic muscle, the vas deferens and seminal vesicles were exposed. Bilateral vas deferens were transected and then pulled anteriorly to facilitate the dissection of seminal vesicles. The Denonvilliers' fascia was identi ed, and the posterior plane was carefully dissected from the base to the prostate apex to preserve neurovascular bundles. The prostatic apex was dissected, thus maximally preserving the urethral stump. The apex was laterally dissected from the anteromedial components of the levator ani.
The urethra was posteriorly incised, and the prostate was removed. The vesicourethral anastomosis was conducted using a continuous suture. Finally, the incised detrusor apron was reapproximated.

Statistical analysis
The IBM SPSS® Statistics 20 United States was used for statistical analysis. Pearson's chi-square and independent samples t-test were used for assessing categorical and continuous data, respectively. The predictive factors of the PSM rate were compared through multivariable logistic regression analysis. The Kaplan-Meier method was used to estimate the 5-year BCR-free survival rate, and the log-rank test was performed to compare the correlations between each factor and BCR-free survival determined through univariate analysis. The factors in uencing BCR-free survival were analyzed using multivariate Cox proportional hazards regression analysis in strati ed pathological stage and surgical margin status, separately.

Declarations Author contributions
Ching-Wei Yang is the rst author. He is responsible for patient data collection, statistical analysis, and manuscript writing. Hsiao-Hsien Wang, Mohamed Fayez Hassouna, and Manish Chand are co-authors.
They review the manuscript and provide academic reinforces. Hsiao-Jen Chung is the corresponding author. He is responsible for manuscript proofreading.
Competing interests