Satoru Kawakami md phd, Department of Urology, Graduate School, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8519, Japan. Email: email@example.com
Objectives: Most of the previously reported patient selection criteria for nerve-sparing radical prostatectomy were based on conventional sextant biopsy, and those based on extended biopsy have been scarcely investigated. In the current study, we developed patient selection criteria for nerve-sparing RP based on the three-dimensional 26-core (3D26) biopsy-derived variables.
Methods: We evaluated 202 non-nerve-spared sides in 109 consecutive patients in whom prostate cancer was diagnosed by the 3D26 biopsy and who underwent RP without neoadjuvant treatment. Associations of clinical and pathological variables with side-specific posterolateral extraprostatic extension (SS-PL-EPE) were analyzed. Subgroup analysis using transperineal 14-core (TP14) and transrectal 12-core (TR12) biopsies as representative subsets of the 3D26 biopsy was also performed.
Results: Maximum cancer length in positive cores ≥5 mm and biopsy Gleason score ≥4 + 3 were independent and were significant risk factors of SS-PL-EPE in the 3D26 cohort at multivariate analysis. In the prostatic side with none, one, and two risk factors, the incidences of SS-PL-EPE were 0, 14 and 52% in the 3D26 cohort, 3.4,15 and 57% in the TP14 cohort and 2.6, 20 and 61% in the TR12 cohort, respectively.
Conclusions: We developed simple patient selection criteria for nerve-sparing RP. According to our criteria, the nerve-sparing side can be selected in the majority of patients who undergo the 3D26, TP14 or TR12 biopsy with a less-than-4% risk of SS-PL-EPE.
Circumventing postoperative deterioration of quality of life, especially in erectile function and urinary continence is one of the most important issues in radical prostatectomy (RP), the mainstay of definitive local therapy for clinically localized prostate cancer. Anatomical nerve-sparing RP, developed by Walsh,1 has been performed worldwide to secure erectile function while maintaining adequate cancer control. In addition, several investigators recently demonstrated that nerve-sparing RP has a significant beneficial impact on urinary continence after surgery.2,3 Accurate prediction of posterolateral extraprostatic extension (EPE) prior to surgery is essential for adequate cancer control in nerve-sparing RP. There have been, however, a few studies proposing practical patient selection criteria for nerve-sparing RP.4–7 Although these studies provide useful information to select candidates for nerve-sparing RP, they are essentially based on the results of conventional sextant biopsies. There has been accumulating evidence supporting the theory that conventional biopsy protocols are underpowered in either cancer detectability or cancer grade estimation.8–13 Extended biopsy has been reported to improve accuracy in predicting pathological features on prostatectomy specimens compared to the sextant biopsy.14–16 Therefore, improved predictability can be expected to relate to patient selection criteria for nerve-sparing RP based on extended biopsy.
We have already reported that the three-dimensional 26-core (3D26) systematic prostate biopsy improves cancer detection compared to the transperineal 14-core (TP14) and transrectal 12-core (TR12) biopsies.9–11 We have shown that grade predicting accuracy is also improved by the 3D26 biopsy.12 In the current study, we analyzed whether clinical and pathological variables provided by the 3D26 biopsy can predict side-specific EPE in the region of the neurovascular bundle (NVB) accurately. We also performed subgroup analysis using TP14 and TR12 biopsies as representative subsets of the 3D26 biopsy.
We studied 117 consecutive patients with clinically organ-confined (cT1/2N0M0) prostate cancer diagnosed by the 3D26 biopsy and treated with RP without neoadjuvant therapy between 2002 and 2007 at our institutions. Three patients were excluded from the study because of the absence of prostate-specific antigen (PSA) parameters. Since EPE status on the nerve-spared side cannot be evaluated precisely, five patients who underwent bilateral nerve-sparing RP and 16 nerve-spared sides in 16 patients who underwent unilateral nerve-sparing RP were excluded from the analysis. A total of 202 non-nerve-spared sides in 109 patients were evaluated in this study and defined as a 3D26 cohort.
The 3D26 biopsy scheme, a combination of TR12 and TP14 biopsies, is illustrated in Figure 1. After obtaining written informed consent, the 3D26 biopsy was performed under spinal, general or local anesthesia as described elsewhere.9,10
Each biopsy core was separately labeled to analyze the location of cancer positive cores.10 All prostatectomy specimens were submitted in their entirety. After fixation, the apical and the bladder neck portions of the prostate were separated from the rest of the gland, and serially sectioned sagittally. The remaining prostate was submitted for whole-mount processing with transverse 3 to 5 mm slices cut perpendicular to the rectal surface. All histological grading of biopsies and RP specimens were reevaluated by a single pathologist according to the 2005 International Society of Urological Pathology Consensus Conference on Gleason Grading, the modified Gleason grading system.17,18
Preoperative clinical variables evaluated were patient age, PSA, prostate volume determined by transrectal ultrasound, PSA density, free PSA, percent free PSA and side-specific digital rectal examination (DRE) finding. PSA density was determined as the quotient of PSA and ultrasound estimated prostate volume. Biopsy-derived side-specific variables including number of positive cores, maximum cancer length in positive cores and biopsy Gleason score (GS) were also evaluated. Cancer length was defined as the length of continuous cancer lesion without gap of benign tissue. Side-specific biopsy GS was defined as the highest score in the side of interest and divided into five categories (0, 5 or 6, 3 + 4, 4 + 3 and 8 or greater). When cancers were not detected by the 3D26 biopsy in the side of interest, side-specific GS 0 was assigned. Clinical parameters except for DRE were considered continuous variables. The cut-offs of number of positive cores, maximum cancer length in positive cores and categorized biopsy GS were determined as the values corresponding to points in the receiver operating characteristic (ROC) curves that maximize the sum of sensitivity and specificity.
EPE was defined as cancer invasion beyond the prostatic capsule into the periprostatic soft tissue or seminal vesicles invasion.18 Posterolateral EPE was defined as one in the region of the NVB. Stepwise logistic regression analyses were used to assess the possible association between variables and side-specific posterolateral EPE (SS-PL-EPE). All analyses were performed using JMP 6.0.2 (SAS Institute Inc, Cary, NC, USA). All P-values were two-sided and a P-value less than 0.05 was considered to be statistically significant.
Transperineal 14-core (TP14) and transrectal 12-core (TR12) prostate biopsies, representative protocols of extended biopsy,19 are subsets of the 3D26 biopsy. Of the 109 patients in the 3D26 cohort, TP14 biopsy missed cancer diagnosis in 11, and the remaining 98 patients with cancers detected on the TP14 sampling sites were defined as a TP14 cohort. A total of 183 non-nerve-spared sides were evaluated in the TP14 cohort. Similarly, the TR12 biopsy missed cancer diagnosis in 15, and the remaining 94 patients with cancers detected on the TR12 sampling sites were defined as the TR12 cohort. A total of 173 non-nerve-spared sides were evaluated in the TR12 cohort. The TP14 and the TR12 cohorts constitute parts of the 3D26 cohort.
Baseline characteristics of the patients are shown in Table 1. A total of 28 EPE were identified in 25 prostates, including 23 posterolateral EPE (21%) and five seminal vesicle invasions. Lymph node metastases were identified in two patients and both had EPE. EPE limited to the anterior aspect of the gland or bilateral EPE were not found in the current cohort. A positive surgical margin (PSM) was recognized in 34 prostates and 79% (27/34) of them were located in the apical region where obvious prostatic capsules were absent.
Table 1. Baseline characteristics of 109 patients
DRE, digital rectal examination; PSA, prostate-specific antigen.
Median age [year]
66 (range 48–76)
Median PSA [ng/mL]
6.5 (range 2.1–26)
Median prostate volume [mL]
28 (range 12–100)
No. clinical stage (Determined by DRE)
No. biopsy Gleason score
6 or less
3 + 4
4 + 3
8 or greater
No. pathological stage
There were 23 posterolateral EPE of 183 non-nerve-spared sides in the TP14 cohort, and 22 posterolateral EPE of 173 non-nerve-spared sides in the TR12 cohort.
Table 2 shows the association between side-specific variables and SS-PL-EPE in the 3D26 cohort. Table 3 shows the results of univariate and multivariate analyses for prediction of SS-PL-EPE. Using ROC analysis, the number of cancer positive cores of 4 or more, maximum cancer length of 5 mm or greater and biopsy GS 4 + 3 or more were determined to be cut-offs with the highest value for predicting SS-PL-EPE. In univariate analysis, prostate volume, PSA density, number of positive cores, maximum cancer length in positive cores and biopsy GS were significantly associated with SS-PL-EPE in the 3D26 cohort. Multivariate analysis demonstrated that maximum cancer length in positive cores and biopsy GS were independent and significant risk factors of SS-PL-EPE.
Table 2. Association between side-specific variables and side-specific posterolateral extraprostatic extension in the three-dimensional 26-core cohort
DRE, digital rectal examination; PSA, prostate-specific antigen.
Prostate volume [mL]
Free PSA [ng/mL]
Percent free PSA
Side-specific number of cancer positive cores
Side-specific maximum cancer length in positive cores [mm]
Side-specific biopsy Gleason score
6 or less
3 + 4
4 + 3
8 or greater
Table 3. Multivariate logistic regression analysis for prediction of side-specific posterolateral extraprostatic extension in the three-dimensional 26-core cohort
P-value (full model)
P-value (reduced model)
Odds ratio (95% CI)
CI, confidence interval; DRE, digital rectal examination; PSA, prostate-specific antigen.
Percent free PSA
Side-specific Abnormal DRE
Side-specific number of cancer positive cores (≥4)
Side-specific maximum cancer length in positive cores (≥5 mm)
Side-specific biopsy Gleason score (≥4 + 3)
We developed a simple criteria to predict SS-PL-EPE (Fig. 2), stratifying patients into three risk groups according to the number of the aforementioned risk factors: the low- (0), intermediate- (1) and high-risk (2) groups. Observed incidences of SS-PL-EPE in the 3D26 cohort are shown in Figure 2. The area under the ROC curves (AUC) of the criteria was 0.899. We applied the criteria to the TP14 and the TR12 cohorts. Observed incidences of SS-PL-EPE according to the risk groups in the TP14 and the TR12 cohorts are shown in Figure 2. AUC of the criteria in the TP14 and the TR12 cohorts was 0.818 and 0.848, respectively. Predictive accuracy of the criteria is shown in Table 4. Positive predictive values for the absence of SS-PL-EPE were more than 95% in the low-risk group in all three cohorts.
Table 4. Predictive accuracy of the patient selection criteria for nerve-sparing radical prostatectomy
The purpose of our study was to identify a patient population in which preservation of the NVB can be performed without deteriorating cancer control. When nerve-sparing procedure was performed on a side with EPE in the NVB region, the risk of PSM and consequently the risk of compromising cancer control would increase. Therefore, preoperative precise prediction of SS-PL-EPE is essential for securing both postoperative quality of life and cancer control.
Previously, we did not have clear criteria for nerve-sparing RP, and we have performed it by the individual surgeon's discretion. Consequently, most of the nerve-sparing procedures have been performed at the sides without cancer on biopsy in our institutions. Considering the current results, we could afford more patients the opportunity to undergo nerve-sparing RP.
The current analysis revealed that maximum cancer length in positive cores and biopsy GS are two independent risk factors of SS-PL-EPE. Using the current criteria, the nerve-sparing procedure can be performed with a substantially low risk of PSM on the side of interest. The criteria provide urologists the advantage of deciding on which side the NVB should be spared without compromising cancer control. It may also help patients by reducing their ambivalence between a desire to remain potent and a desire to remain free of cancer.
Several models predicting side-specific EPE have been reported (Table 5). Ohori et al. constructed and internally validated nomograms to predict the side-specific probability of EPE based on sextant-based biopsy.4 AUC of their full model nomogram incorporating PSA, clinical stage, biopsy GS, percent positive cores and percent cancer in cores was 0.806. Although they reported 65% of the EPE was found in the NVB region (posterolateral EPE) in their cohort, their model cannot predict EPE in the NVB region. Naya et al. demonstrated a simple risk stratification of SS-PL-EPE using two risk factors, a maximum cancer length of 7 mm or greater in any core and a positive basal core,5 and validated it recently.7 According to their criteria, the nerve-sparing side could be selected in 26% and 60% of the patients in their cohort with 4% and 10% incidences of SS-PL-EPE, respectively. Tsuzuki et al. reported a model predicting SS-PL-EPE using PSA, DRE finding, biopsy GS, average percent of prostatic tissue involved by tumor and percent positive core.6 The model was devised to identify the nerve-sparing side with a less-than-10% incidence of SS-PL-EPE. Taken together, these previously reported models can reduce the risk of SS-PL-EPE down to approximately 10%.
Table 5. Comparison of patient selection models for nerve-sparing radical prostatectomy
No. sampling cores in PBx
Incidence of target
AUC, area under the receiver operating characteristics curve; DRE, digital rectal finding; EPE, extraprostatic extension; GS, Gleason score; PBx, prostate biopsy; PSA, prostate-specific antigen; SS-PL-EPE, side-specific posterolateral EPE.
PSA ≥10, abnormal DRE, GS ≥7, average percent of prostatic tissue involved by tumor >20, percent positive core >33.3
Less than 10% in cases with 1 or fewer of the risk factors
12 and 14
Maximum cancer length ≥5 mm, GS ≥4 + 3
Approximately 3% in cases with no risk factors
Maximum cancer length ≥5 mm, GS ≥4 + 3
Nearly 0% in cases with no risk factors
Using our criteria, the nerve-sparing side could be selected in 85% (93/109) of the 3D26 cohort, 89% (87/98) of the TP14 cohort and 91% (86/94) of the TR12 cohort with 0%, 3.4% and 2.6% incidences of SS-PL-EPE, respectively. The rate of patients with SS-PL-EPE in our cohort was 21% which is almost equivalent to those of Tsuzuki's (21%), Ohori's (20%) and Naya's (18%) cohorts. Compared with the previously reported models, our model can provide patients with higher safety and a less-than-4% risk of SS-PL-EPE in the majority of patients. Higher predictability of SS-PL-EPE by our criteria would be due to higher sampling density of the gland around the NVB by the 3D26 biopsy.
In the presence of stronger variables provided by the 3D26 biopsy, all the clinical variables evaluated did not contribute to the patient selection criteria. Tsuzuki's and Ohori's models greatly relied on PSA for prediction of side-specific EPE. Our criteria are not influenced by clinical variables, such as PSA and DRE findings. We assume that the evaluation of cancer characteristics by the 3D26 biopsy is more accurate than by sextant biopsy, and therefore the impact of clinical variables are diminished.
Biopsy GS is one of the two significant risk factors of SS-PL-EPE in the current study. The impact of biopsy GS on the prediction of SS-PL-EPE in the current criteria is much larger than those in the previously reported models. We consider that accurate grade estimation by the 3D26 biopsy enhances the importance of biopsy GS for predicting SS-PL-EPE. Regarding categorization of biopsy GS, between 3 + 4 and 4 + 3 was the most discriminable in the current study. Our finding is in accordance with those reported by other investigators who have demonstrated that the prediction of pathological stage or biochemical recurrence would be more accurate by discriminating GS 3 + 4 from GS 4 + 3 disease than GS 3 + 3 from GS 3 + 4.20,21 We included GS 0 (no cancer in the side of interest) in the current analysis because GS 0 was not considered to be completely safe for the nerve-sparing procedure, even in the extended sampling. In fact, one SS-PL-EPE was found in the side of GS 0 in the TR12 cohort. There was no EPE in the side of GS 0 in the 3D26 cohort, suggesting that the absence of EPE might be predicted only by extensive sampling such as the 3D26 biopsy.
Maximum cancer length in positive cores, along with the number of positive cores, is a representative variable reflecting tumor volume.22,23 Naya et al. also reported that maximum cancer length in positive cores is a significant risk factor of SS-PL-EPE independent of biopsy GS.5,7 Variables reflecting tumor volume were also included in Tsuzuki's and Ohori's models. Taken together, information on tumor volume obtained by biopsy is closely associated with SS-PL-EPE.
In the TP14 and TR 12 cohorts, significant risk factors of SS-PL-EPE were identical to those in the 3D26 cohort. Although AUC of the criteria in the TP14 and the TR12 cohort was lower than that in the 3D26 biopsy, these criteria are applicable to patients diagnosed by TP14 and TR12 biopsies.
Several limitations of our study should be acknowledged. Although clinical variables including PSA parameters and volume were not significant predictors for predicting SS-PL-EPE in multivariate analysis, these impacts should be investigated in a larger cohort. Also a validation study with a larger cohort would be relevant to confirm the performance of the criteria. Two risk factors with different odds ratios were handled equally in the current study. The final goal of this study was to make simple criteria for predicting EPE in a small number of patients, so we did not perform an analysis matched to the value of the odds ratio of each variable. However, in a larger cohort more detailed risk evaluation for predicting SS-PL-EPE, such as a nomogram using continuous variables, should be investigated.
The 3D26 biopsy-based patient selection criteria for nerve-sparing RP using maximum cancer length in positive cores and biopsy GS is simple and sufficiently accurate. According to our criteria, the nerve-sparing side could be selected in many of the patients who underwent the 3D26, TP14 and TR12 biopsy with less than 4% incidence of SS-PL-EPE.