Journal of Robotic Surgery

, Volume 6, Issue 4, pp 311–316

Predictors of positive surgical margins at open and robot-assisted laparoscopic radical prostatectomy: a single surgeon series

  • Mahesha Weerakoon
  • Shomik Sengupta
  • Kapil Sethi
  • Joseph Ischia
  • David R. Webb
Original Article

DOI: 10.1007/s11701-011-0313-4

Cite this article as:
Weerakoon, M., Sengupta, S., Sethi, K. et al. J Robotic Surg (2012) 6: 311. doi:10.1007/s11701-011-0313-4
  • 103 Downloads

Abstract

Robot-assisted laparoscopic radical prostatectomy (RALRP), increasingly used to treat localized prostate cancer, has advantages over open radical prostatectomy (ORP) in terms of reduced bleeding and quicker convalescence. However, debate continues over whether RALRP provides superior or at least equivalent surgical outcomes. This study compares positive surgical margins (+SM), as a surrogate for long-term cancer control, at RALRP and ORP performed by a single experienced surgeon during the process of taking up RALRP. 400 consecutive patients undergoing surgery for prostate cancer under a single surgeon (DW) between November 1999 and July 2009 were studied. Prior to July 2005, all patients underwent ORP; after this date, most patients were treated by RALRP. Data were collected by retrospective chart review and analysed independently of the treating surgeon. +SM were defined as the presence of cancer at an inked surface. Overall, 23 (11.5%) of 200 patients undergoing RALRP had +SM, compared to 40 (20.0%) of 200 patients undergoing ORP (P < 0.05). On univariate logistic regression analysis, in addition to surgical approach (odds ratio [OR] = 1.92), patient age (OR = 1.05), pathologic stage (OR = 3.93) and specimen Gleason (GS) score (OR = 1.86) were significant predictors of +SM. On multivariate analysis, surgical approach, p-stage and specimen GS remained significant predictors of +SM. RALRP is associated with lower rates of +SM compared to ORP, even after adjusting for other known risk factors. Of note, the RALRP in this study were part of the surgeon’s learning curve.

Keywords

Prostatectomy, robotic Prostatectomy, open Prostate cancer Positive surgical margins Risk factors Multivariate analysis 

Introduction

The daVinci robotic system is used to perform an increasing proportion of radical prostatectomies (RP) for the treatment of early stage prostate cancer (CaP). To date, robot-assisted laparoscopic RP (RALRP) has been clearly shown to have benefits over open RP (ORP) in terms of reduced bleeding and shorter hospital stay and convalescence [1, 2]. However, the principal surgical aims of RP are the so-called “trifecta” of cancer control, urinary continence and sexual potency. Opinion remains divided on whether RALRP provides superior or at least equivalent outcomes with respect to each of these three endpoints [1, 2].

The robust assessment of prostate cancer control requires long-term follow-up with accurate documentation of biochemical or clinical recurrence and death from CaP and other causes. However, in the short term, positive surgical margins (+SM) can be utilized as a surrogate for cancer control. Although it is well recognized that many patients with +SM remain recurrence-free in the long-term, +SM are an independent risk factor for subsequent recurrence, and can be viewed as a measure of oncological adequacy of RP [3]. Published literature shows conflicting data relating to the risk of +SM at RALRP relative to ORP [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]. The aim of this study was to examine the rate of +SM during RALRP and ORP as performed by a single experienced surgeon during the process of taking up RALRP.

Patients and methods

We studied 400 consecutive patients undergoing surgery for localized CaP under a single surgeon (who had performed in excess of 400 ORPs prior to the study period) between November 1999 and July 2009. Prior to July 2005, all patients underwent ORP; after this date, most patients were treated by RALRP. ORP was carried out using a standard approach with retrograde dissection of the prostate. RALRP was carried out trans-peritoneally, with antegrade dissection of the prostate. As previously described, a posterior approach was utilized to dissect the neuro-vascular bundles off the prostate and develop the “Veil of Aphrodite” during RALRP [16].

Data pertaining to patient demographics and pathologic details were collected by retrospective chart review, and analysed independently of the treating surgeon. Staging was according to the 2007 UICC-AJCC TNM staging system and grading according to the Gleason system. Percentage of positive cores (PPC) was defined as 100 × number of biopsy cores involved by prostate cancer/total number of biopsy cores obtained. +SM were defined as the presence of cancer at an inked surface of the specimen.

Data are expressed as median and range or mean and standard deviation, as appropriate. Comparisons between groups were undertaken using the chi-squared test for categorical variables and the Wilcoxon test for continuous variables. Logistic regression analysis was carried out, using univariate and multivariate models, to analyse the relationship between +SM and surgical approach as well as other risk variables. All statistical analyses were carried out using online calculators at http://www.statpages.org [17], and statistical significance was set at P < 0.05.

Results

The clinical and pathologic characteristics of patients treated by ORP and RALRP are shown in Table 1. The two groups were comparable in terms of age at treatment, but patients treated by RALRP had a lower level of prostate-specific antigen (PSA) and a greater likelihood of Gleason 7 disease at biopsy than those treated by ORP. The prostate size and pathologic stage (p-stage) had a trend towards being lower in the RALRP group, but this did not reach statistical significance. The specimen Gleason score (sGS) showed evidence of some upgrading from Gleason 6 to 7 in both groups, but in keeping with the distribution of biopsy Gleason scores (bGS), patients treated by RALRP continued to have a greater likelihood of Gleason 7 cancers.
Table 1

Clinical and pathologic characteristics of patients treated by open or robot-assisted laparoscopic radical prostatectomy

Variable

ORP (N = 200)

RALRP (N = 200)

P-value

Age (years), median (range)

62 (45–82)

63 (47–80)

0.49

PSA (μg/l), median (range)

6.9 (1.6–43.0)

6.1 (0.6–43.6)

<0.003

Clinical stage, N (%)

  

0.6

 T1

109 (54.5)

106 (53.0)

 

 T2a

46 (23.0)

54 (27.0)

 

 T2b

31 (15.5)

24 (12.0)

 

 T2c

13 (6.5)

16 (8.0)

 

Biopsy Gleason score, N (%)

  

<0.001

 <7

138 (69.0)

104 (52.0)

 

 3 + 4 = 7

26 (13.0)

62 (31.0)

 

 4 + 3 = 7

21 (10.5)

26 (13.0)

 

 >7

14 (7.0)

6 (3.0)

 

Percent positive cores, N (%)

  

0.96

 <37.5

93 (46.5)

101 (50.5)

 

 37.5–50

46 (23.0)

47 (23.5)

 

 >50

45 (22.5)

50 (25.0)

 

Prostate weight (g), median (range)

53.0 (19.0–198.0)

50.5 (22.0–148.0)

0.054

Specimen Gleason score, N (%)

  

<0.03

 <7

79 (39.5)

64 (32.0)

 

 3 + 4 = 7

81 (40.5)

92 (46.0)

 

 4 + 3 = 7

25 (12.5)

39 (19.5)

 

 >7

14 (7.0)

5 (2.5)

 

Pathologic stage, N (%)

  

0.09

 T2 or less

141 (70.5)

156 (78.0)

 

 T3 or more

59 (29.5)

44 (22.0)

 
Overall, +SM occurred in 40 (20%) patients undergoing ORP and 23 (11.5%) patients undergoing RALRP (P < 0.05). The rates of +SM stratified by various clinical features are shown in Fig. 1. The year of surgery had no association with the rate of +SM (data not shown). Table 2 shows the univariate relationship between clinical factors and +SM. In addition to surgical approach, patient age, PPC and bGS were significant predictors of +SM, but serum PSA and clinical stage (c-stage) were not. On multivariate analysis, the surgical approach was the only factor that remained a statistically significant predictor of +SM (Table 2).
Fig. 1

Rates of positive surgical margins, comparing open to robotic radical prostatectomy, stratified by a clinical stage, b biopsy Gleason score, c pre-operative prostate-specific antigen and d percent of positive cores

Table 2

Univariate and multivariate logistic regression analysis of clinical variables predictive of positive surgical margins during open or robot-assisted laparoscopic radical prostatectomy

Variable

Univariate

Multivariate

OR (95% CI)

P-value

OR (95% CI)

P-value

Age

1.05 (1.00–1.09)

<0.05

1.03 (0.99–1.08)

0.12

Prostate-specific antigen

1.03 (0.99–1.07)

0.11

  

Clinical stage

1.17 (0.87–1.58)

0.29

  

Percent +ve cores

1.01 (1.00–1.03)

<0.05

1.01 (1.00–1.02)

0.07

Biopsy Gleason

1.47 (1.12–1.94)

<0.01

1.33 (0.99–1.79)

0.05

Surgical approach

0.52 (0.30–0.91)

<0.05

0.50 (0.28–0.88)

<0.02

The rates of +SM stratified by various pathologic features are shown in Fig. 2. On univariate logistic regression analysis, in addition to surgical approach, p-stage and sGS were significant predictors of +SM, but prostate size was not (Table 3). On multivariate analysis, p-stage had the strongest association with +SM, but even after adjusting for this and sGS, the surgical approach remained a statistically significant predictor of +SM (Table 3).
Fig. 2

Rates of positive surgical margins, comparing open to robotic radical prostatectomy, stratified by a pathologic stage, b specimen Gleason score and c prostate weight

Table 3

Univariate and multivariate logistic regression analysis of pathologic variables predictive of positive surgical margins during open or robot-assisted laparoscopic radical prostatectomy

Variable

Univariate

Multivariate

OR (95% CI)

P-value

OR (95% CI)

P-value

Age

1.05 (1.00–1.09)

<0.05

1.04 (0.99–1.08)

0.08

Prostate-specific antigen

1.03 (0.99–1.07)

0.11

  

Prostate size

0.99 (0.97–1.00)

0.11

  

Pathologic stage

3.93 (2.25–6.88)

<0.0001

2.85 (1.54–5.29)

<0.001

Specimen Gleason

1.86 (1.36–2.53)

=0.0001

1.43 (1.02–2.02)

<0.05

Surgical approach

0.52 (0.30–0.91)

<0.05

0.54 (0.30–0.96)

<0.05

Discussion

RALRP has become established as a surgical treatment for early CaP, although there is continuing debate over whether its outcomes are superior or at least equivalent compared to ORP. In this single surgeon series, we demonstrate that RALRP is associated with a lower risk of +SM compared to ORP, even after adjusting for accepted clinico-pathologic risk factors.

Interestingly, the pre-operative serum PSA and c-stage were found not to be associated with +SM, perhaps illustrating the limitations of PSA as a prognostic marker and the inaccuracy of clinical assessment of stage. Patient age at treatment was univariately but not multivariately associated with +SM, suggesting that the association was due to higher risk disease being treated in older patients. The PPC and bGS were also only univariately associated with +SM. However, the multivariate association of these two variables with +SM did approach statistical significance, suggesting that perhaps greater numbers may have provided sufficient power for this to be demonstrated. P-stage and sGS, both accepted risk factors for +SM, were strongly predictive of the risk of +SM in our study, on both univariate and multivariate analysis. Prostate size, which some studies have shown to be associated with a lower risk of +SM, had no influence on +SM in our study.

The above risk factors were found to differ somewhat between the RALRP and ORP groups in our study. Thus, the serum PSA was significantly lower among patients undergoing RALRP, and prostate size and p-stage both showed trends towards being lower. The distribution of sGS was also different, with fewer patients with Gleason 6 disease but more with Gleason 7 disease in the RALRP group. In part, these differences may be attributable to selection of patients with larger prostates or more adverse clinical features for ORP during the very early stages of taking up RALRP. However, beyond the initial 50 cases or so, patients have been considered for RALRP irrespective of these features, with ORP carried out only on the basis of patient preference. An additional factor that may also have contributed to these differences is a continuing process of stage migration, since on average the ORP patients were treated at a slightly earlier period. However, being consecutive cases, the difference in time was short.

In any event, even accounting for the differences in risk profile between the RALRP and ORP groups, on multivariate analysis the surgical approach still accounted for an almost two-fold difference in the risk of +SM in favour of RALRP in our study. Considering clinical variables, only the surgical approach remained a significant predictor of +SM, with PPC and bGS both approaching, but not achieving statistical significance. Among pathologic variables, p-stage and sGS were significant predictors of +SM, but even after adjusting for these factors, surgical approach also retained a significant association with +SM. Notably, on stratified analysis, for any particular stage or Gleason score, RALRP was associated with a lower rate of +SM compared to ORP.

However, other studies comparing RALRP and ORP in terms of +SM have shown conflicting findings, as summarized in Table 4. Unfortunately, these results are difficult to compare, since studies vary greatly in terms of the disease spectrum, type of institution, and number and experience of treating surgeons. Most studies are retrospective and only some undertake multivariate or matched analysis to account for the confounding effects of differences in risk factors such as stage and Gleason score between ORP and RALRP patients. Notwithstanding the limitations of collective data analysis, one review of 5,472 patients treated by RALRP and 26,691 patients treated by ORP from multiple series found the overall +SM rates to be comparable to those in our study, being 12.5 and 23.5% respectively [18], although the ORP group included a higher proportion of T3 tumours. Conversely, in a meta-analysis comparing minimally invasive RP (including laparoscopic RP as well as RALRP) to ORP, there was no difference found in rates of +SM (risk ratio 0.88, 95% CI 0.74–1.06) [19].
Table 4

Published studies comparing positive surgical margins at open radical prostatectomy (ORP) and robot-assisted laparoscopic radical prostatectomy (RALRP)

Center

ORP

RALRP

P-value

MVA

N

+SM (%)

N

+SM (%)

OR (95% CI)

Factors adjusted for

St. Vincents Sydney [4]

502

84 (17)

212

45 (21)

0.18

N/A

Matched analysis

Padua [5]

105

21 (30)

103

35 (34)

0.97

N/A

 

Mayo Clinic [6]

588

100 (17)

294

46 (16)

0.61

N/A

Matched analysis

U Wis [7]

98

12 (14)

94

11 (13)

NS

N/A

 

Johns Hopkins [8]

522

75 (14.4)

522

102 (19.5)

0.01

1.6 (1.1–2.5)

Age, race, biopsy Gleason score, PSA, c-stage, number of cases performed

Epworth Richmond [9]

102

27 (26.4)

102

14 (13.7)

<0.05

N/A

 

Milan [10]

240

61 (25)

120

26 (22)

NS

0.92 (0.53–1.60)

Age, PSA, p-stage

Duke [11]

435

122 (28)

362

106 (29)

0.7

1.38 (0.95–2.02)

Age, PSA, year of surgery, BMI, c-stage, biopsy Gleason score

Vanderbilt [12]

200

71 (35.5)

200

30 (15)

<0.001

N/A

Matched analysis

Vattikuti [13]

100

(23)

200

(9)

<0.05

N/A

Matched analysis

Metro Health Michigan [14]

50

18 (36)

50

11 (22)

0.007

N/A

 

Harvard [15]

346

30 (8.7)

524

80 (15.3)

<0.05

1.9 (1.2–3.1)

Age, PSA, year of surgery, c-stage, biopsy Gleason score, BMI, nerve sparing

MVA multivariate analysis, N/A not available, PSA prostate-specific antigen, BMI body mass index

Clearly, RALRP can be associated with lower +SM compared to ORP, but not necessarily so. RALRP affords the surgeon technologic advances including magnified binocular vision and wristed instrumentation, which are cited as factors important in reducing +SM. We believe that, in addition to these technologic advances, our approach during RALRP to the posterior dissection of the prostate and development of the plane between the prostatic capsule and neurovascular bundle is an important factor in reducing +SM in our series [16]. We find that the antegrade dissection of the prostate, as well as the optical and physical features of the daVinci robot, allow this portion of RP to be carried out under vision and to be tailored appropriately to the risk features of the patient’s disease. Thus, we routinely carry out nerve-sparing during RALRP for all patients, but with the extent of nerve sparing guided by a pre-operative assessment of the patient’s risk features.

As every practicing surgeon would realize, the daVinci robotic system is but a surgical instrument, and like any other requires appropriate surgical expertise to obtain optimum results. Studies have clearly shown the relationship of surgeon experience and volume to outcomes at RP, thus highlighting the importance of the surgeon in such analyses. It is our belief that, at least in part, the conflicting literature on +SM with various approaches to RP is a result of comparing outcomes from surgeons with disparate levels of expertise. The most appropriate comparison therefore is between the results obtained by a single surgeon using different approaches. It is to be noted that one of the strengths of our study is that it compares results from cases performed by a single surgeon. In particular, while the RALRP cases included his learning curve, the ORPs were part of a mature series from an experienced surgeon who had already performed hundreds of cases. The fact that consecutive cases were studied, and the most recent ORPs were undertaken in parallel with the RALRPs, hopefully minimizes any impact that temporal trends may have on disease characteristics, surgical technique or pathologic analysis.

Conclusions

RALRP is associated with a lower rate of +SM compared to ORP, even after adjusting for known clinical and pathologic risk factors.

Conflict of interest

None.

Copyright information

© Springer-Verlag London Ltd 2011

Authors and Affiliations

  • Mahesha Weerakoon
    • 1
  • Shomik Sengupta
    • 1
    • 2
  • Kapil Sethi
    • 1
  • Joseph Ischia
    • 1
  • David R. Webb
    • 1
  1. 1.Department of UrologyAustin HealthHeidelbergAustralia
  2. 2.East MelbourneAustralia

Personalised recommendations