Early oncological outcomes of robot-assisted radical prostatectomy for high-grade prostate cancer


Chris Wambi, Vattikuti Urology Institute, Henry Ford Hospital, 2799 West Grand Boulevard, Detroit, MI 48202, USA.
e-mail: cow2101@columbia.edu


Study Type – Therapy (case series)
Level of Evidence 4


To evaluate the oncological outcomes of patients with specimen Gleason 8 and 9 prostate cancers and to determine factors that predict biochemical recurrence-free survival (BCRFS) after robot-assisted radical prostatectomy (RARP).


Of 4156 patients who underwent RARP from January 2001 to 2009, we identified 368 men with Gleason 8 or 9 tumours who met the inclusion criteria. BCR was defined as a PSA level of ≥0.2 ng/mL with a second rising value. The Kaplan–Meier method and log-rank test were used to compare BCRFS while factors that predict BCRFS were determined by Cox proportional hazards modelling.


The median age and PSA level were 62 years and 6.4 ng/mL for men with Gleason 8, and 63 years and 6.7 ng/mL for Gleason 9 cancers. The median (interquartile range, IQR) overall follow-up was 23 (10–46) months and 19 (7–37) months for Gleason 8 and 9 tumours, respectively. At 60 months the mean (se) overall BCRFS was 36 (5)% and for Gleason 8 it was 47 (6)% and for Gleason 9 it was 21 (7)% (P < 0.001). At 5 years, extraprostatic extension (pT3a) resulted in BCRFS of 52 (9)% for Gleason 8 tumours and 21 (11)% for Gleason 9 (P= 0.012). On multivariable analysis, lymph node invasion, specimen Gleason score, pathological stage and tumour volume predicted BCRFS.


Early results suggest RARP monotherapy performs comparably to RP for BCRFS in men with high-grade prostate cancer. There are significant oncological differences between Gleason 8 and 9 tumours.


biochemical recurrence


biochemichal recurrence-free survival


radical retropubic prostatectomy


robot-assisted radical prostatectomy


tumour volume


lymph node (invasion)


interquartile range


hazard ratio.


Robot-assisted radical prostatectomy (RARP) is now a common surgical approach in the treatment of prostate cancer in the USA. Although several centres have detailed functional results with RARP, reports of oncological outcomes remain sparse [1–4]. Longitudinal studies of patients treated conservatively for prostate cancer show that most cancer-specific deaths occur in men with Gleason 8–10 disease [5,6]. Furthermore, patients with Gleason 8–10 cancers are most likely to have disease progression and cancer-specific mortality after RP or radiation therapy and therefore stand to benefit the most from adjuvant or salvage therapy [7–9]. Long-term biochemical recurrence-free survival (BCRFS) is nonetheless feasible in select patients with Gleason 8–10 tumours after RP alone [10–12]. The oncological outcomes of RARP in men with high-grade prostate cancer have not been reported. There is a need to determine the oncological efficacy of RARP [2,4] particularly in men with high-risk tumours given the theoretical concern that this surgical approach may compromise cancer control due to a lack of tactile feedback.

PSA screening has resulted in a downward stage migration as well as significantly decreased median tumour volume (TV) in all clinical stages and specimen Gleason grades in men with prostate cancer thereby resulting in potentially improved oncological outcomes [13–15]. There has been renewed interest in surgical treatment for patients with high-grade prostate cancer [16]. Surgery is generally advocated as part of a multimodal treatment regimen for patients with Gleason 8–10 prostate cancer. Identification of patients with high-grade tumours who are least likely to have disease progression after surgery would enhance the judicious recommendation of adjuvant treatment. We sought to determine oncological outcomes in patients with high-grade prostate cancer treated with RARP alone and to identify factors that determine disease progression after RARP in this heterogeneous group.


Data were collected prospectively using a protocol approved by the Institutional Review Board at Henry Ford Hospital and in compliance with the provisions of the Health Insurance Portability and Accountability Act. Between January 2001 and January 2009, 4156 patients underwent RARP [17]. Of these men, 398 had Gleason 8–10 cancers on final pathological specimens. We excluded 23 patients treated with neoadjuvant androgen deprivation, chemotherapy, or radiation therapy. The remaining 375 patients were treated with RARP as monotherapy. Most patients were clinically staged with CT of the abdomen/pelvis and bone scans whereas endo-rectal MRI was not routinely performed. Our general approach was to perform a conventional nerve-sparing procedure in potent men with ≥25% biopsy TV, Gleason 7, a PSA level of <20 ng/mL, and clinical stage ≤T2a. In men with a PSA level of ≥20 ng/mL, clinical stage ≥T2b or biopsy Gleason ≥8 we commonly performed a wide resection on the side with high-grade or palpable disease to include the neurovascular bundle, lateral prostatic fascia and perirectal fat. Preservation of the lateral prostatic fascia (‘Veil of Aphrodite’) is undertaken in men with <25% biopsy TV Gleason ≤7, cT1c and a PSA level of <10 ng/mL. Our earlier lymph node (LN) dissection template included the obturator as well as external iliac LN packets. In the last 2 years, we have included the hypogastric LN packet.

Clinical and pathological staging were in accordance with the 2002 American Joint Cancer Committee TNM system. Prostatectomy specimens were processed according to the Stanford protocol until 2005 and by whole-mount specimen analysis thereafter. The TV was assessed by visual estimate of percentage carcinoma in each processed section and calculation of the mean [18].

After RARP PSA tests were recommended at 1.5, 4, 8, 12, 18, 24 months and at least on a yearly basis thereafter. Biochemical failure was defined as a PSA value of ≥0.2 ng/mL and a second rising value [19]. Patients with biochemical persistence at 6 weeks after RARP were assumed to have biochemical failure at the time of surgery (67 patients). Patients who underwent immediate adjuvant treatment before the first postoperative PSA level measurement were excluded (seven patients). There was only one patient with Gleason 10 cancer and this patient was excluded from subsequent analysis. The Kaplan–Meier method was used to estimate BCRFS and survival curves were compared using the log-rank test. Univariate and multivariable Cox proportional hazards models were used to estimate the hazard ratio (HR) and 95% CI of factors that predict BCRFS. Covariates consisted of age, PSA level, biopsy Gleason score, clinical stage, race, specimen Gleason score, pathological stage, LN invasion (LNI), specimen TV (%), surgical margin status and specimen weight (g). Statistically significant variables on univariate analysis were subsequently entered in the multivariable models in a stepwise fashion. Continuous variable means were compared using the Mann–Whitney test or the Student t-test with Welch correction. Proportions were compared using the chi-square test. All tests were two-sided with significance level set at P < 0.05.


The clinical and pathological characteristics of the men with high-grade prostate cancer are shown in Table 1. The median (interquartile range, IQR) overall follow-up was 23 (10–46) months and 19 (7–37) months for Gleason 8 and 9 tumours, respectively (P= 0.07). The median (IQR) PSA follow-up was 16 (5–40) months for Gleason 8 and 14 (4–34) months for Gleason 9 patients (P= 0.14; Table 1). There were no differences in age, PSA level, clinical stage and ethnicity between Gleason 8 and Gleason 9 patients (Table 1). By comparison, there were significant differences in biopsy Gleason score, pathological stage, specimen TV, positive surgical margin rate, and BCR rate (all P < 0.001) between both high Gleason groups (Table 1). The surgical positive margin rate was closely related to pathological stage in non-LNI invasive tumours (pTN0 or pTNx) with 19.8% (18/91) in pT2, 56% (79/141) in pT3a and 67.7% (65/96) in pT3b. The median (IQR) overall LN dissection yield was 5 (3–7). Since 2008 the median LN dissection yield increased to 9 (5–14).

Table 1.  Clinical and pathological characteristics of patients with Gleason score 8 and 9 tumours
CharacteristicGleason scoreP
  • *


  • Mann–Whitney;

  • ‡chi-square.

No. patients214154 
Median (IQR):   
 Age, years62 (58–67)63 (58–67)0.49*
 PSA level, ng/mL6.4 (4–9.6)6.7 (4.9–11.3)0.28
Biopsy, n (%):7 (7–8)8 (7–9)<0.001
 637 (17.3)9 (5.8)0.001
 3 + 430 (14.0) 11 (7.1)0.053
 4 + 348 (22.4)31 (20.1)0.69
 879 (36.9)44 (28.6)0.12
 918 (8.4)56 (36.4)<0.001
 1003 (1.9)0.15
 Unknown2 (0.009) 0
Clinical stage, n (%):  0.22
 ≤T1c103 (40.5)66 (42.8) 
 T2a57 (26.6)39 (25.3) 
 ≥T2b32 (12.6)36 (23.4)0.01
 Unknown22 (10.3)13 (8.4) 
Ethnicity, n (%):  0.69
 African-American33 (15.4)22 (14.3) 
 Caucasian136 (63.6)93 (60.4) 
 Other 11 (5.1)7 (4.5) 
 Unknown34 (15.9)32 (20.8) 
Pathological stage, n (%) or n/N (%): <0.001 
 pT272 (33.6)20 (13.0)<0.001
 pT3a98 (45.8)58 (37.7)0.15
 pT3b43 (20.1)73 (47.4)<0.001
 pT41 (0.005)3 (1.9)0.16
 pTN+18/193 (9.3)19/147 (12.9)0.38
Median (IQR) TV, %25 (12.5–35)30 (20–50)<0.001
Positive margins, n (%)87 (40.6)98 (63.6)<0.001

The overall median BCRFS was 36 months. The mean (se) rate of BCRFS at 60 months was 36 (5)% for all patients. For Gleason 8 patients median BCRFS was 60 months compared with 13 months for men with Gleason 9 cancer (P < 0.001, Fig. 1A). At 60 months BCRFS was was 47 (6)% and 21 (7)% for Gleason 8 and 9, respectively (P < 0.001, Fig. 1A). In Gleason 8 cancers, median BCRFS was not reached for pT2 and pT3a tumours while it was 6 and 0 months, respectively, for pT3b and pT4 tumours (P < 0.001, Fig. 1B). In Gleason 9 cancers median BCRFS was not reached for pT2 or was 13, 6, and 5 months for pT3a, pT3b and pT4 tumours, respectively (P= 0.001, Fig. 1C). Stratification by specimen TV (<10, 10–40 and >40%) also resulted in BCRFS differences (P < 0.001, Fig. 1D). Negative surgical margin status was associated with BCRFS of 50 (7)% at 60 months, whereas positive margin status resulted in 20 (7)% BCRFS at 5 years (P < 0.001, Fig. 2A). Margin status did not impact BCRFS in Gleason 8 cancer (P= 0.11, Fig. 2B) whereas it was associated with a profound difference in BCRFS in Gleason 9 cancer (P < 0.001, Fig. 2C). Patients with margin negative and no LNI tumours (N0R0) had a median BCRFS of 88 months while those with margin positive or LNI tumours (N1R1) had a median survival of 10 months (P < 0.001, Fig. 2D).

Figure 1.

The BCRFS according to: A, specimen Gleason score (GS); B, pathological stage for GS 8 tumours; C, pathological stage for GS 9 tumours; and D, specimen TV (%). Median survival not reached (NR)

Figure 2.

The BCRFS according to: A, margin status; B, margin status for Gleason score (GS) 8 tumours; C, margin status for GS 9 tumours; and D, margin-negative and LN-negative tumours (N0R0) compared with with margin-positive or LN-positive tumours (N1R1). Median survival not reached (NR).

Univariate Cox proportional hazard analysis showed that PSA level, biopsy Gleason score, clinical stage, pathological Gleason score, pathological stage, LNI status, TV and surgical margin status were predictors of BCRFS for all patients. On multivariable analysis only pathological Gleason score, pathological stage, LNI and TV remained significant predictors of BCRFS for all patients (Table 2).

Table 2.  Cox proportional hazards analysis of factors predicting time to BCR for Gleason score 8 and 9 cancers
VariableUnivariateMultivariable (stepwise)
HR (95% CI)PHR (95% CI)P
Age, years0.98 (0.96–1.01)0.13  
PSA level, ng/mL1.03 (1.02–1.05)<0.001  
Biopsy: 0.013  
 60.39 (0.20–0.77)0.007  
 3 + 40.82 (0.47–1.43)0.48  
 4 + 30.96 (0.60–1.54)0.86  
 91.35 (0.86–2.11)0.19  
 101.55 (0.21–11.1)0.67  
Clinical stage: 0.01  
 T2a1.25 (0.81–1.93)0.31  
 ≥T2b2.07 (1.34–3.21)0.001  
African-American vs Caucasian1.18 (0.76–1.84)0.47  
Specimen weight, g1.00 (0.99–1.01)0.77  
Pathological Gleason score (9 vs 8)2.18 (1.54–3.08)<0.0011.5 (1.04–2.16)0.03
Pathological stage: <0.001  
 pT20.38 (0.22–0.66)<0.0010.46 (0.26–0.82)0.008
 pT3b2.14 (1.47–3.10)<0.0011.45 (0.98–2.14)0.06
 pT43.09 (1.12–8.54)0.03  
 pTN1 vs pTN03.64 (2.33–5.69)<0.0012.51 (1.58–3.99)<0.001
TV, %1.03 (1.02–1.03)<0.0011.02 (1.01–1.03)0.002
Positive margins2.14 (1.50–3.03)<0.001  


Risk assessment is imperative in the treatment of all men diagnosed with prostate cancer. Despite the historic view that RP for patients with Gleason 8–10 disease is not curative, several series show that a significant number of men with high-grade prostate cancer may be cured with RP as monotherapy in the PSA era [10–12,16,20]. Oncological outcomes of high Gleason grade patients treated with RARP have not been reported, despite a theoretical concern that the lack of tactile feedback in RARP may compromise cancer control. We determined BCRFS in men with Gleason 8 and Gleason 9 tumours treated with RARP as monotherapy and identified factors that predict BCR.

The median age and PSA level of patients in the present study were lower than those reported in earlier large single-institution series, which is consistent with increased PSA screening [10–12,20]. Furthermore, most men in the present cohort had clinically localized prostate cancer. The 60 month BCRFS rate of 36 (5)% in the present study is comparable with previously reported rates for similar patient populations undergoing RP without adjuvant therapy (38, 38, 40 and 71%) [10–12,20]. It is worth noting that there were different definitions of BCR used in these high Gleason grade RP series. Moreover, there was a variable incidence of biochemical persistence in those studies. Two studies used a BCR definition of a PSA level of ≥0.4 ng/mL [11,20] and another study used a PSA level of ≥0.2 ng/mL [10]. While the study that reported the highest 60 month BCRFS survival defined BCR as a PSA level of ≥0.1 ng/mL, only five of 188 (2.6%) patients in that series had biochemical persistence after RP [12]. The rate of biochemical persistence in the present study was 20.4% (67/328), which may indicate differences in patient population despite the common finding of high-grade disease at the time of RARP. Nonetheless, we report for the first time important oncological differences between patients with Gleason 8 and Gleason 9 cancers not previously addressed in RP series that, if confirmed by other centres, may improve the selection of patients for adjuvant therapy.

Despite comparable median ages, PSA levels, and clinical stages patients with Gleason 8 cancers were 1.7-fold more likely to have prostate-confined (pT2) tumours compared with seminal vesicle invasion (pT3b), patients with Gleason 9 disease were 3.6-fold more likely to harbour pT3b tumours compared with pT2. In Gleason 8–10 prostate cancer, patients with pT2 disease were previously shown to have the highest BCRFS [11,12]. Similarly, in the present study men with pT2 tumours had the highest BCRFS. Furthermore, in the present study pT3b cancers had a median BCRFS of only 6 months regardless of Gleason grade. In contrast, in the present study BCRFS in tumours exhibiting extraprostatic extension (pT3a) differed significantly between Gleason 8 and 9 cancers. The median BCRFS in pT3a tumours was not reached in Gleason 8 cancers while it was 13 months in Gleason 9 cancers (P= 0.007). If these findings are confirmed by other centres, it would be reasonable to recommend adjuvant therapy and enrolment in clinical trials for patients with Gleason 9 and pT3a tumours. All men with Gleason 8 or 9 tumours and seminal vesicle invasion would be recommended to undergo adjuvant treatment and strongly considered for enrolment in clinical trials. Conversely, patients with pT2 tumours are probably better managed expectantly. The differences in pathological stage distribution as well as potentially different biological behaviour of pT3a tumours may in part explain the different BCRFS rates between men with Gleason 8 and 9 tumours. Another possible factor in the better BCRFS of Gleason 8 compared with Gleason 9 tumours is TV. Gleason 9 tumours had significantly greater specimen TV compared with Gleason 8 (P < 0.001).

We evaluated the impact of specimen TV on BCRFS, as this is a point of controversy after RP. Other investigators have identified specimen TV as a major determinant of BCRFS on multivariable analysis [18,21–26]. Epstein et al. [27] and others arrived at the opposite conclusion surmising that TV added no additional prognostic information on multivariable analysis once the Gleason score, margin status, or pathological stage are known and therefore TV needs not be routinely reported [15,28]. In the present study, specimen TV was strongly predictive of BCRFS on univariate and multivariable analysis for patients with either Gleason 8 or 9 tumours. The present study is the first to exclusively examine the impact of this tumour feature in men with high Gleason grade disease without the potential confounding effect of well-differentiated tumours. Merrill et al. [29] recently showed that the utility of TV in predicting BCRFS on univariate or multivariable analysis depends on the pathological Gleason grade. Specifically, in their study, TV was not beneficial in prognosticating BCRFS in Gleason 6 tumours, which is in agreement with the results of Kikuchi et al. [15]. There are well-acknowledged considerations in the technical aspects of TV determination at the time of surgery. The present findings using visual estimation [18,24] suggests that in high Gleason grade tumours TV should be reported as it is a primary determinant of BCRFS. Interestingly, Chun et al. [22] have reported that TV, which essentially represents tumour bulk, is the most accurate predictor of positive surgical margin status on multivariable logistic regression and second most accurate predictor of BCR second only to the percentage of Gleason grade 4/5 tumour in the specimen [21,22,25].

As previously reported in high Gleason grade tumours, in the present study surgical margin status was not a significant predictor of BCRFS on multivariable analysis despite the powerful significance of this tumour feature on univariate analysis [11,20]. Cautiously interpreted, this finding does not indicate that margin status is unimportant, rather, margin status is probably highly correlated with more statistically important pathological features such as Gleason grade, pathological stage and TV. Nonetheless, positive surgical margin status was associated with significantly worse BCRFS in Gleason 9 and not Gleason 8 cancers. This may suggest a different mechanism of BCR between patients with Gleason 8 and Gleason 9 disease in the current era. While micrometastatic disease probably plays a role in BCR for both Gleason 8 and 9 tumours, locally advanced disease may also have a prominent effect in Gleason 9 cancers. Patients with positive margins in the setting of Gleason 9 cancer should be strongly considered for adjuvant therapy regardless of pathological stage.

Given the variable accuracy of preoperative identification of patients at highest risk of disease progression after RP, we limited the present study to patients with postoperative Gleason 8 and 9 cancers. Clearly, the present patient cohort represents a population at high risk of disease progression. That 20.4% patients who did not undergo neoadjuvant therapy had biochemical persistence after RARP is consistent with previous observations in patients with pathological Gleason 8–10 cancers [10,11]. Yossepowitch et al. [30] evaluated eight commonly used preoperative definitions of high-risk patients and found considerable variability in their ability to predict adverse pathological features and BCR. In their review, Yossepowitch et al. [30] found that the rate of organ-confined prostate cancer varied from 22 to 63% depending on the high-risk definition used. Likewise, the 5-year BCRFS rates in organ-confined prostate cancer also varied from 49 to 80% depending on the definition used [30]. Using D’Amico’s preoperative criteria [31], Kane et al. [14] found that currently the majority of patients are characterized as high risk based upon Gleason grade alone. In the contemporary era 34–55% of biopsy Gleason 8–10 cancers are downgraded to Gleason ≤7 at the time of surgery [10,32]. Therefore, the present study and those previously reporting outcomes in patients with specimen Gleason 8–10 prostate cancer probably represent populations at high risk of disease progression by all measures or definitions. Nonetheless, the present data show reasonable early biochemical outcomes after RARP in a subset of patients with high Gleason grade cancers, particularly Gleason 8 tumours or localized Gleason 9 cancers. Although we have not reported cancer-specific or overall survival in the present study, several large series have recently shown that RP may result in the greatest disease-specific survival for patients with Gleason 8–10 cancers compared with radiation therapy, androgen deprivation or conservative management [7,33–35].

There are important limitations in the present study beyond its retrospective design. The median PSA level follow-up was 14–16 months, although this is congruent with most laparoscopic RP and RARP series [4]. In their systematic review, Wilt et al. [4] reported that in 21 non-randomized clinical trials and case series involving 2301 and 1757 patients, respectively, treated with laparoscopic RP or RARP, the median follow-up was 8 months. Lastly, despite its importance for adjuvant or salvage therapy initiation, BCRFS is not perfectly predictive of cancer-specific survival. Nonetheless, the results presented in the present study are compelling because we performed an intention-to-treat analysis and included patients with LNI as well as men with biochemical persistence (49% of all biochemical failures in the present series). Despite inclusion of these very high-risk patients, oncological outcomes between Gleason 8 and Gleason 9 cancers remained significantly different.

In conclusion, early oncological outcomes for patients with Gleason 8 and 9 cancers treated by RARP are equivalent to reported results obtained by RP in the PSA era. There appear to be important oncological differences between Gleason 8 and Gleason 9 tumours. LNI, pathological Gleason score, pathological stage and TV independently predict BCRFS on multivariable analysis. Further studies of this population are required to better tailor adjuvant treatment to patients at greatest risk for recurrence and expectant management or salvage therapy to those who may have prolonged BCRFS.


None declared.