World Journal of Urology

, Volume 33, Issue 1, pp 77–83

Safe-R: a novel score, accounting for oncological safe nerve-sparing at radical prostatectomy for localized prostate cancer


    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
    • Department of UrologyUniversity-Hospital Hamburg-Eppendorf
  • Carolina Coelius
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Meike Adam
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
    • Department of UrologyUniversity-Hospital Hamburg-Eppendorf
  • Pierre Tennstedt
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Luis Kluth
    • Department of UrologyUniversity-Hospital Hamburg-Eppendorf
  • Thomas Steuber
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Hans Heinzer
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Markus Graefen
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Thorsten Schlomm
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
  • Uwe Michl
    • Martini-ClinicProstate Cancer Center Hamburg-Eppendorf
Original Article

DOI: 10.1007/s00345-014-1273-x

Cite this article as:
Becker, A., Coelius, C., Adam, M. et al. World J Urol (2015) 33: 77. doi:10.1007/s00345-014-1273-x



The objectives of the study were to describe a novel score (safe-R), combining information on surgical margin status (SM) and extend of nerve-sparing (NS) applicable for all patients undergoing radical prostatectomy (RP), and to test the impact of our frozen-section navigated nerve-sparing approach (NeuroSAFE) on safe-R score.


We retrospectively analyzed 9,635 RPs performed at our center between 2002 and 2011. Of these, 47 % were conducted with NeuroSAFE. Proportions of NS and SM status were assessed. Subsequently, a score for oncological safe NS (safe-R) was developed; Safe-R was categorized as 3 (for negative SM and bilateral NS), 2 (for negative SM and unilateral NS), 1 (for negative SM without NS), and 0 (for patients with positive SM), respectively. The impact of NeuroSAFE on safe-R was analyzed by chi-square test and confirmed by multinomial logistic regression, adjusting for preoperative risk factors.


Applying NeuroSAFE resulted in enhanced safe-R score, indicating lower rates of positive SM and higher rates of NS, across all risk categories (all p < 0.001). For example in high-risk patients, NeuroSAFE resulted in lower proportions of safe-R 0 (27.6 vs. 33.6 %) and higher proportions of safe-R 3 (32.4 vs. 17.1 %, p < 0.001). Linkage between the NeuroSAFE approach and safe-R was confirmed after multinomial logistic adjustment for preoperative risk factors. All results were confirmed in a propensity-matched cohort (matched for preoperative risk factors and year of surgery, data not shown).


Safe-R represents a novel tool to assess and report on oncological safe nerve-sparing in RP. NeuroSAFE is associated with enhanced safe-R scores.


Prostate cancerProstatectomyMarginNerve-sparing


Radical prostatectomy (RP) represents the only treatment strategy, which has been shown to improve cancer-specific survival in patients with localized prostate cancer (PCa) compared to watchful waiting in a prospective randomized trail [1]. During the last decade, substantial advances in perioperative functional outcomes have been achieved due to refinements in RP technique [24]. Specifically, intraoperative nerve-sparing (NS) has been shown to improve not only postoperative potency [5], but also urinary continence [2].

However, even in big case series from centers of excellence, the proportion of patients who benefit from NS-RP ranges from 53 to over 90 % [6, 7], depending on clinical risk profile and surgeons’ decision. Moreover, selection of candidates for NS-RP based on preoperative findings results in significant misclassification in up to 50 % of all patients using clinical risk stratification, [6] and up to, respectively, 30 and 16 % using tree-decision- or nomogram-based prediction of extracapsular extension (ECE) [8, 9]. Thus, patients with organ-confined disease are exposed to unnecessary resection of neurovascular bundles and patients with non-organ-confined disease to positive surgical margins (SM). Moreover, optimal reporting on surgical outcome after RP should simultaneously consider enhanced preservation of anatomical structures (more NS) and enhanced oncological safety (less positive SM) on the other hand. Considerations to combine oncological and functional outcomes after RP resulted in the definition of trifecta [10] and pentafecta [11] classification. However, only patients who are preoperatively potent and continent are eligible for reporting, making the trifecta and pentafecta classification applicable for only 30–60 % of all RP patients [6, 11]. To address this void and in order to simplify reporting on this double-edged outcome, we aimed to develop a novel score (safe-R), combining information on SM and extent of NS, which is applicable for virtually all patients after RP.

Furthermore, we hypothesized that using neurovascular structure-adjacent frozen-section examination (NeuroSAFE) to control for positive SM during NS-RP [12] significantly enhances oncological safe NS as measured by safe-R.

Patients and methods


From January 2002 to January 2011, 9,635 consecutive patients underwent open or robotic RP for clinically localized PCa at our center. Clinical characteristics (PSA, clinical stage, biopsy Gleason grade) were recorded, and patients were stratified using the D`Amico risk groups for disease progression [13]. Collection of data was approved by our internal reviewer board, and all patients provided written informed consent. Frozen-section navigated nerve-sparing approach was performed in 4,509 (47 %) of all patients using a technique previously described (NeuroSAFE) [12].

Indication and performance of the NeuroSAFE technique

The NeuroSAFE technique was used to assist intraoperative selection of potential candidates for NS. Between 2002 and 2007, we relied on Tree-decision- (2002–2006) or nomogram-based models (after 2006), predicting ECE [8, 9] to identify NS candidates. The NeuroSAFE procedure was performed only in cases with suspect intraoperative findings or in patients with a nomogram-predicted higher risk for ECE (>20 %), but who were strongly requesting NS. Since 2007, NeuroSAFE was routinely used in virtually every patient to assure negative SM during NS-RP, resulting in a NeuroSAFE frequency of 91 % (94, 95, and 77 % for low-, intermediate-, and high-risk patients, respectively) of all cases in 2011. The NeuroSAFE procedure is elaborately described in our previous publication [12], and crucial steps of the procedure are sketched below; the whole dorsolateral aspect of the prostatic surface, containing the complete neurovascular structure-adjacent tissue, is dissected after the removal of the prostate. By inking the inner and outer SM with different colors, assignment of the correct anatomical location by the pathologist is assured. According to the Stanford protocol, serial step sections at 3 mm [14] are performed and histopathological assessment is done on two 6-μm-thick cryosections of each block by a trained uropathologist. In case of a positive NeuroSAFE result, the surgeon is informed and typically performs an ipsilateral secondary neurovascular resection, including the rectolateral part of the Denonvieller’s fascia. The secondarily resected tissue is submitted separately for permanent section analysis. The definitive SM status is determined on the whole RP specimen including the NeuroSAFE sections [15]. Resection margins are considered positive if at least one cancer cell extended to the inked surface [14].

Pathological outcome was assessed using the AJCC 2002 staging system, and tumour grading was classified using the Gleason Grading system (1992–2005) and revised 2005 Gleason grading system afterward.

Statistical assessment

Pre-, intra-, and postoperative characteristics were stratified for NeuroSAFE or non-NeuroSAFE approach. The frequency and proportion and means, medians, and interquartile ranges were computed for categorical and continuous variables, and compared using chi-square and Mann–Whitney U tests, respectively.

A score for oncological safe NS (safe-R) was developed. Safe-R was categorized as 3 (for negative SM and bilateral NS), 2 (for negative SM and unilateral NS), 1 (for negative SM without NS), and 0 (for patients with positive SM), respectively (Fig. 1). The impact of NeuroSAFE on safe-R was analyzed by chi-square test and confirmed with a multinomial logistic regression, controlling for preoperative risk factors (PSA, biopsy Gleason score, and clinical stage). All analyses were stratified for patients of low-, intermediate-, or high-risk according to the D’Amico classification. In subanalyses, a propensity-matched analysis was performed in a regression model adjusting for preoperative risk factors (PSA, clinical stage, and biopsy Gleason score) and year of surgery. The R packages non-random was used as previously described [16].
Fig. 1

Depicting higher safe-R scores in patients undergoing non-NeuroSAFE RP versus NeuroSAFE RP, stratified for risk groups, according to the D’Amico classification


Baseline characteristics

Overall, 9,635 patients with a median age of 64.1 years, interquartile rang {IQR}: 59.2–68.0 underwent RP (Table 1). Of those, respectively, 4,093 (42.5 %), 4,228 (43.9 %), and 1,314 (13.6 %) were classified as D`Amico low risk, intermediate risk, and high risk, according to preoperative risk factors (PSA level, clinical stage, and biopsy Gleason score). A safe-R score of 0 was assigned to 1,647 patients (17.1 %) who showed positive surgical margins after RP. Safe-R scores of 3, 2, and 1 were assigned to patients with negative surgical margins after bilateral (5,635, 59.0 %), unilateral (1,925, 20.0 %), and none nerve-sparing RP (380, 3.9 %), respectively (Table 2).
Table 1

Characteristics of 9,635 patients who underwent RP for clinically localized PCa between 2002 and 2011









Age (years)


 Mean (median)

63.2 (64.1)

63.0 (64.0)

63.4 (64.3)


 Interquartile range





PSA (ng/ml)



 Mean (median)

8.72 (6.48)

8.76 (6.70)

8.68 (6.20)


 Interquartile range





D’Amico group



 Low risk

4,093 (42.5)

1,287 (28.5)

2,806 (54.7)


 Intermediate risk

4,228 (43.9)

2,556 (56.7)

1,672 (32.6)


 High risk

1,314 (13.6)

666 (14.8)

648 (12.6)


Pathological stage (pT)




6,899 (71.6)

3,042 (67.5)

3,857 (75.2)



1,886 (19.6)

1,039 (23.0)

847 (16.5)



850 (8.8)

428 (9.5)

422 (8.3)


Gleason score



 ≤3 + 3

2,743 (28.5)

681 (15.1)

2,062 (40.2)


 3 + 4

5,303 (55.0)

3,001 (66.6)

2,302 (44.9)


 4 + 3

1,243 (12.9)

680 (15.1)

563 (11.0)


 ≥4 + 4

346 (3.6)

147 (3.3)

199 (3.9)






6,579 (68.3)

2,868 (63.6)

3,711 (72.4)



2,430 (25.2)

1,522 (33.8)

908 (17.7)



626 (6.5)

119 (2.6)

507 (9.9)


Surgical margins




1,647 (17.1)

726 (16.1)

921 (18.0)



7,988 (82.9)

3,783 (83.9)

4,205 (82.0)


Safe-R score




1,647 (17.1)

726 (16.1)

921 (18.0)



380 (3.9)

75 (1.7)

305 (6.0)



1,925 (20.0)

1,192 (26.4)

733 (14.3)



5,635 (59.0)

2,516 (55.8)

3,167 (61.8)


RP radical prostatectomy, PCa prostate cancer, PSA prostate-specific antigen

Table 2

Categorizing safe-R score as: 3 (for negative surgical margin and bilateral nerve-sparing), 2 (for negative surgical margin and unilateral nerve-sparing), 1 (for negative surgical margin without nerve-sparing), and 0 (for all patients with positive surgical margins), respectively


Safe-R 0

Safe-R 1

Safe-R 2

Safe-R 3

Overall patients

1,647 (17 %)

380 (4 %)

1,925 (20 %)

5,683 (59 %)

Surgical margin





Extent of nerve-sparing





Bivariate associations

A total of 4,509 patients (47 %) underwent NeuroSAFE-navigated RP and were compared to 5,126 patients (53 %) who underwent standard RP. Pre- and postoperative tumour characteristics were significantly worse in the NeuroSAFE group (Table 1). However, NeuroSAFE was associated with significantly higher safe-R scores (all p < 0.001). Within the D`Amico low-risk group, a safe-R 0 (positive SM) was significantly less prevalent (10.6 vs. 13.3 %) and safe-R scores of 2 or 3 (uni- or bilateral NS and negative SM) were more prevalent (88.1 vs. 85.7 %, p < 0.001) in patients undergoing NeuroSAFE, compared to non-NeuroSAFE patients (Fig. 1). Similarly, in intermediate- and high-risk patients, NeuroSAFE resulted in lower proportions of safe-R score 0 (15.9 vs. 19.8 % and 27.6 vs. 33.6 %) and higher proportions of safe-R score 3 (53.6 vs. 43.7 % and 32.4 vs. 17.1 %, respectively, all p < 0.001, Fig. 1).

Multinomial logistic adjustment

Linkage between the NeuroSAFE approach and safe-R was confirmed after multinomial logistic adjustment for preoperative risk factors. Specifically, patients who underwent NeuroSAFE-navigated RP had a lower risk of safe-R 1 [odds ration (OR) 0.24, 95 % confidence interval (CI) 0.18–0.31, p < 0.001) and higher probabilities of safe-R 2 (OR 1.59, 95 % CI 1.38–1.83, p < 0.001) and safe-R 3 (OR 1.39, 95 % CI 1.2–1.54, p < 0.001) compared to non-NeuroSAFE patients.

Stratified for D’Amico risk classification, low-risk patients who underwent NeuroSAFE RP had a higher probability for safe-R score 2 (OR 6.6, 95 % CI 4.8–9.1, p < 0.001) compared to non-NeuroSAFE patients. The probability for safe-R 1 (OR 1.62, 95 % CI 0.86–3.04, p = 0.1) and SAFE-R 3 (OR 1.12, 95 % CI 0.91–1.38, p = 0.3), respectively, was not statistically significantly different for patients undergoing NeuroSAFE or non-NeuroSAFE RP. Intermediate-risk patients undergoing NeuroSAFE RP were less likely to show safe-R 1 (OR 0.25, 95 % CI 0.17–0.37, p < 0.001), whereas the probability of safe-R 3 (OR 1.57, 95 % CI 1.31–1.89, p < 0.001) was statistically significantly higher compared to non-NeuroSAFE. The probability of safe-R 2 after RP was similar in both groups (OR 1.05, 95 % CI 0.86–1.27, p = 0.6). High-risk patients undergoing NeuroSAFE RP were statistically significant less likely to show safe-R 1 (OR 0.08, 95 % CI 0.04–0.15, p < 0.001) and more likely to show safe-R 2 (OR 1.49, 95 % CI 1.11–2.01, p = 0.008) and safe-R 3 (OR 2.92, 95 % CI 2.06–4.12, p < 0.001) compared to non-NeuroSAFE patients.

In subanalyses, a propensity score matching for preoperative PSA, clinical stage, biopsy Gleason score, and year of surgery was performed. Also, within the matched cohort, the NeuroSAFE approach was associated with statistically significantly higher safe-R scores, in low-, intermediate-, and high- risk patients (data not shown).


Quality of life after RP is strongly related to postoperative urinary continence and erectile function. [17] Both have been shown to be associated with the extent of intraoperative preservation of neurovascular bundles. [2, 5] Previous publications have shown that frozen-section analysis represents an oncological safe approach to ensure NS, even in high-risk patients [12, 18]. However, wide resection of the anatomical structures is a widely used strategy to avoid positive SM [19], a histopathological finding that is reported to increase the risk of disease recurrence [20], and palpable tumors were historically considered as contraindication for nerve-sparing procedure [21, 22]. Accordingly, intraoperative NS was considered as a trade-off between functional and oncological results [23], and functional and oncological outcomes have been show to vary, not only according to patients’ disease characteristics [6], but also according to surgical experience [24]. According to these considerations, preservation of ideally both neurovascular bundles and simultaneously maximal oncologic safety would represent optimal surgical outcome after RP. To simplify reporting on outcomes after RP by combining information on oncological safety (i.e., SM status) and preservation of functional tissue (i.e., extent of NS) in a single score, which is applicable for virtually all patients, we developed safe-R. Furthermore, we hypothesized that applying intraoperative neurovascular structure-adjacent frozen-section examination (NeuroSAFE)-guided nerve-sparing approach significantly enhances oncological safe NS as measured by safe-R.

First, our score represents a simple, intuitive measure of surgical outcomes after RP, assigning low values to patients with adverse oncological findings (safe-R 0 in all patients with positive SM) and gradually higher values to patients with negative SM and favorable functional outcomes (safe-R 1, 2, and 3 to patients with no, uni-, or bilateral NS, respectively, Table 2). It simplifies reporting and comparison of surgical series outcomes and is applicable for virtually every patient undergoing RP. Moreover, in contrast to the trifecta [10] or pentafecta [11] system that do not weight outcomes, the priority of oncological safety over preservation of functional tissue is emphasized by assigning a score of 0 to all patients with positive surgical margins, non-regarding of extent of NS. As RP represents, before all, a cancer-directed treatment, complete resection of the tumor should remain the most important aim of RP. Moreover, patients with positive surgical margins are at high risk to undergo adjuvant or salvage radiotherapy, which has been shown to impair urinary function. [25]

Second, our results demonstrate that using the NeuroSAFE approach resulted in significantly enhanced safe-R scores. The benefit of the NeuroSAFE approach was most pronounced in intermediate- and high-risk patients, resulting in lower proportions of safe-R score 0 (positive surgical margins) and higher proportions of safe-R score 3. For example, high-risk PCa patients according to the D’Amico classification who underwent NeuroSAFE RP showed a, respectively, 1.5- and 3-fold higher probability for negative SM and uni- (safe-R 2) or bilateral (safe-R 3) preservation of neurovascular bundles, compared to non-NeuroSAFE patients, after multivariable adjustment for tumor characteristics. In the group of low-risk patients, the effect of NeuroSAFE was less pronounced and selection of candidates for nerve-sparing using preoperative clinical data might provide similar rates of nerve-sparing. However, in low-risk patients, NeuroSAFE was associated with a lower rate of positive surgical margins (safe-R 0).

The D’ Amico classification system [13], although developed to predict PSA-specific outcome, is widely used to stratify reporting on perioperative outcomes [6, 26]. However, using preoperative data like PSA value, clinical stage, and biopsy findings is associated with a significant risk of misclassification. [27] For example, non-organ-confined tumors are found in up to 18 % of all “low-risk” patients, whereas pT2 tumors are found in up to 50 % of “high-risk” patients [6, 28]. Tree-structured and nomogram-based prediction of the risk of extracapsular extension using preoperative findings represent further developments of prediction tools and are widely accepted for the selection of candidates for NS-RP, [8, 9] but still potentially detain patients with organ-confined disease from nerve-sparing RP showing an accuracy of 70 and 84 %, respectively [9]. Moreover, nomograms can only be applied in patients with complete and detailed preoperative biopsy information, including location and percentage of positive biopsy cores and percentage of tumor in positive cores. Additionally, the validity of all available nomograms in patients who underwent targeted prostate biopsies [29] has yet to be confirmed. As applied biopsy schemes and elaborateness of histopathological reporting vary substantially in patients diagnosed in the community, complete data for site-specific calculation of ECE according to the Steuber nomogram were available for only roughly 30 % of all patients who underwent RP in our cohort. In patients with complete preoperative data, individuals with a side-specific risk <20 % for ECE typically (according to the 95th percentile of the predicted ECE risk in our cohort) received NS-RP when no NeuroSAFE was applied [3]. In the NeuroSAFE cohort, sufficient preoperative data for the side-specific Steuber nomogram was available for 3,020 prostatic lobes. Applying this “cutoff” (risk for ECE < 20 %) to these patients, NS would have been performed in only 2,134 of 3,020 lobes (71 %). However, applying NeuroSAFE, NS was performed in 2,607 of 3,020 lobes (87 %). Accordingly, NeuroSAFE resulted in additional NS in 1 of 6 prostatic lobes compared to patients receiving NS only relying on the nomogram. Simultaneously, positive SM (safe-R 0) was reduced by 3, 4, and 6 % in low-, intermediate-, and high-risk patients undergoing NeuroSAFE RP compared to non-NeuroSAFE patients. These numbers strikingly demonstrate that applying the NeuroSAFE approach enhances the proportion of NS and reduces positive SM, compared to elaborate selection of candidates for NS according to preoperative data.

Several limitations apply to our study, most of them due to its retrospective nature, applying for all studies investigating this topic. [12, 30, 31] The decision whether to perform NeuroSAFE or not largely depended on preoperative data and admittedly was influenced on rather subjective intraoperative decisions by the individual surgeon. Accordingly, a selection bias might have influenced the findings of our study. However, the distribution of pre- and postoperative disease characteristics, which was statistically significantly favorable for the patients of the non-NeuroSAFE group (more low risk according to the D’Amico classification, more organ-confined tumors with low-grade Gleason score in final pathologic report), suggests that benefits of NeuroSAFE approach, (i.e., higher values for safe-R) are rather underestimated due to this selection bias. In order to account for this potential selection bias, we performed subanalyses using a propensity score-matched cohort of patients. Essentially, our results showing that the NeuroSAFE approach was associated with higher safe-r scores were confirmed across all risk categories (data not shown). Second, the validity of safe-R and linkage to functional and cancer control outcomes should be examined in the future studies.


Safe-R represents a novel tool to assess and report on oncological safe nerve-sparing in RP, combining information on oncological safety and preservation of functional tissue in a single score, which is applicable for virtually all patients. Applying intraoperative neurovascular structure-adjacent frozen-section examination (NeuroSAFE) guided nerve-sparing approach significantly enhances oncological safe NS as measured by safe-R, even in high-risk PCa patients.

Conflict of interests

The authors declare that they have no conflict of interest.

Ethical standard

The authors confirm that the study has been performed in accordance with the ethical standards laid down in the Declaration of Helsinki 1964 and its later amendments. All patients gave their informed consent prior to their inclusion in the study.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014