Midterm oncological outcomes of laparoscopic vs open radical prostatectomy (RP)


Correspondence: Douglas M. Dahl, Department of Urology, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114, USA.

e-mail: ddahl@partners.org



  • To compare the midterm risks of biochemical recurrence (BCR) and salvage radiation therapy (SRT) after laparoscopic (LRP) and open retropubic radical prostatectomy (RRP). Strong evidence that these techniques are comparable to the ‘gold standard’ of open RRP is lacking, as most comparative studies are limited by short follow-up or rely on historical controls.

Patients and Methods

  • We studied 1000 consecutive patients concurrently treated by either LRP or RRP between 2001 and 2005.
  • LRPs were performed by a single surgeon and RRP by four surgeons.
  • Primary outcomes were BCR and SRT. Survival analysis included relevant clinical and pathological variables.


  • Of 844 included patients, 244 underwent LRP and 600 RRP.
  • Clinical and pathological characteristics were similar in both groups. Most patients had Gleason 6 tumours (68%) and pT2 disease (86%). The median follow-up was 6.1 years and median time to recurrence 3.4 years.
  • Overall, BCR occurred in 14% of patients: 13.1% after LRP and 14.7% after RRP.
  • SRT was performed in 10.7% of patients both after LRP and RRP. In uni- and multivariate Cox regression models, surgical technique was not a significant predictor of BCR or SRT.


  • Our results suggest that in high-volume centres, LRP provides equivalent oncological control to RRP.

biochemical recurrence


body mass index


Institutional Review Board


lymph node


(laparoscopic) (open retropubic) radical prostatectomy


(salvage) radiation therapy


The use of minimally invasive approaches in prostate cancer surgery has gained increasing popularity over the past decade. In 2009, ≈66% of all radical prostatectomies (RPs) in the USA were performed by minimally invasive techniques [1]. The well-documented improvements in short-term outcomes, e.g. recovery time and blood loss, are particularly appealing to patients and surgeons. However, the few available studies that compare long-term oncological outcomes of laparoscopic (LRP) and open retropubic RPs (RRP) are limited by unequal patient populations. Many studies rely on historical controls and are subject to bias, as study populations, pathological reporting and PSA assays have evolved over the years. Currently available studies of contemporaneously treated patients address these biases, but they are limited by short follow-up times (Table 1) [2-9].

Table 1. Studies comparing the outcomes of concurrent open RRP and LRP (performed with and without robotic assistance), sorted by study type and size. Most have a median or mean follow-up of <2 years and few recurrences. BCR rates were not always the primary outcome of the cited studiesThumbnail image of

Our institution has extensive experience in both open RRP and LRP. We have previously published the pathological outcomes of LRP and RRP of 1000 patients treated between 2001 and 2005, and shown that both groups are comparable in their baseline characteristics and have similar rates of positive surgical margins [10].

The goal of the present study was to evaluate the association of surgical technique with the risk of biochemical recurrence (BCR) and salvage radiation therapy (SRT) in these patients. Furthermore, we investigate whether the effect of clinical and pathological variables on tumour recurrence varies with surgical technique.

Patients and Methods

After Institutional Review Board (IRB) approval we retrospectively reviewed all RP cases that were included in an IRB-approved tumour bank between November 2001 and March 2005. As previously described [10], this tumour bank consists of tissue specimens and clinicopathological data of RP cases completed in this hospital during normal business hours (≈80% of all cases). Exclusion criteria are shown in Fig. 1. Patients were mainly of White race (95% White, 3% Black, 2% Asian).

Figure 1.

Cohort selection.

A single experienced laparoscopic surgeon (D.M.D.) performed all LRPs using a modification of the technique by Guillonneau and Vallencien [11]. Both pre- and transperitoneal approaches were used. Four experienced urologists of the same department performed the open RRPs. Pelvic lymph node (LN) dissection was performed in all RRPs and in 35% of LRPs depending on preoperative risk assessment. When performed, a median of four LNs were removed per patient both by RRP and LRP. Processing and examination of the specimens was done according to a standardised protocol by a dedicated uro-pathologist (C.L.W.) [10].

Follow-up data was extracted from the electronic medical system and/or obtained from external documents. The PSA level was measured quarterly for the first 2 years, then yearly. BCR was defined as a PSA level of ≥0.2 ng/mL, followed by a confirmatory value, or exceptionally as any detectable PSA level of <0.2 ng/mL motivating SRT and/or androgen deprivation. SRT was defined as radiation to the prostatic fossa (±LNs) in the setting of a newly detectable PSA.

All statistical analyses were performed with Stata12 (College Station, TX, USA) and PASS 11 (NCSS, Kaysville, UT, USA). Variables of interest included age and body mass index (BMI) at surgery, type and date of RP, surgeon identity, Gleason score, tumour stage, prostate weight, surgical margins and their localisation, duration of follow-up, BCR and SRT. Univariate analysis relied on non-parametric Kruskal–Wallis tests, chi-squared and Fisher's exact tests, as needed. For multivariate survival analysis, a Cox regression model was built using a non-automated forward stepwise approach. Interactions between variables were tested when plausible. The proportional hazard assumption was graphically assessed and formally tested for all variables. Possible transformations of continuous variables evaluated using fractional polynomials and spline functions. All statistical tests were two-sided with P < 0.05 considered to indicate statistical significance.

In this cohort of 844 patients who were treated over a 5-year period, with a median follow-up of 6 years, a log-rank test would allow the detection of a 9% difference in BCR-free survival. This calculation considers that 70% of the patients are treated by RRP, and is based on a 6-year attrition rate of one third of the RRPs and two-thirds of the LRPs, as seen in this study.


Clinical and Pathological Characteristics

From the initial 1000 patients, 57 met exclusion criteria, and 99 could not be included due to lack of follow-up data (Fig. 1). The proportion of missing patients was similar in both treatment groups (11% missing in LRP, 14% in RRP, P = 0.22).

Of the remaining 844 patients, 244 underwent LRP and 600 RRP. Clinical and pathological characteristics were similar in both groups (Table 2). Margin rates were comparable in both groups across all Gleason scores and across both pT2 and pT3 tumour stages (pT2: 13.6% in LRP vs 14.0% in RRP, P = 0.714; pT3: 29.0% in LRP vs 34.9% in RRP, P = 0.531).

Table 2. Clinicopathological characteristics of 844 patients undergoing RP between 2001 and 2005 by open RRP (n = 600) or LRP (n = 244)
VariableTotalRP techniqueP
  1. IQR, interquartile range. *In 144 cases, BMI was not documented. In two cases of RRP, the margin location was not documented.
Patients, n (%)844600 (71)244 (29) 
Median (IQR):    
Age at surgery, years59 (55–64)59 (54–64)59 (55–63)0.200
BMI, kg/m2*26.8 (25–30)27.0 (24.8–29.8)26.3 (24.1–29.8)0.132
Preoperative PSA level, ng/mL5.0 (3.9–6.8)5.0 (3.8–6.8)5.2 (4.0–7.1)0.441
Prostate weight, g41 (33–54)40.8 (33–52)41.8 (34–57)0.119
N (%):    
Gleason score:   0.595
≤6573 (67.9)407 (67.9)166 (68.03) 
7242 (28.7)107 (28.3)72 (29.8) 
≥829 (3.4)23 (3.8)6 (2.5) 
Stage:   0.619
pT2729 (86.4)516 (86.0)213 (87.3) 
pT3115 (13.6)84 (14.0)31 (12.7) 
Positive margins:139 (16.5)101 (16.8)38 (15.6)0.655
Location   0.342
Apex35 (25.9)22 (22.2)13 (36.1) 
Peripheral84 (62.2)66 (66.7)18 (50.0) 
Bladder neck6 (4.4)4 (4.0)2 (5.6) 
Multiple10 (7.4)7 (7.1)3 (8.3) 
Median (IQR):    
Follow-up, years6.1 (3.7–7.3)6.6 (5.0–7.6)4.6 (1.9–6.0)<0.001
Time to BCR, years3.4 (1.8–5.0)3.4 (1.7–4.9)3.4 (2.4–5.7)0.639
BCR, n (%)120 (14.2)88 (14.7)32 (13.1)0.558
Median (IQR) time to SRT, years4.0 (2.5–5.7)4.0 (2.5–5.6)4.0 (2.7–5.7)0.936
SRT, n (%)90 (10.7)64 (10.7)26 (10.7)0.990

The proportion of patients treated by LRP moderately increased over the inclusion period (from 19% in 2001/2002 to 32% in 2004/2005). We therefore examined whether clinical and pathological characteristics also changed over the inclusion period and found a 3% increase in organ -confined prostate tumours (85.9% in 2001/2002 vs 88.9 in 2004/2005, P = 0.04) and a slight shift from Gleason 6 to Gleason 7 tumours (a 5% decrease of Gleason 6 tumours and a 6% increase of Gleason 7, P = 0.01). However, these changes equally affected patients treated by LRP and RRP.

Follow-Up and Survival Analysis

The median time to BCR was 3.4 years in both groups. The median follow-up of patients in remission was 6.1 years. Follow-up was significantly longer in patients treated by RRP than LRP (6.6 vs 4.6 years, P < 0.001). This difference in duration of follow-up was evenly distributed over the study period.

Overall, cumulative BCR was 14.2% (120/844), including 13.1% (32/244) of patients treated by LRP and 14.7% (88/600) of patients treated by RRP (P = 0.56). BCR rates did not differ by surgeon in patients undergoing RRP (P = 0.33). Kaplan–Meier estimates of survival at 6 years were 83% for LRP cases and 86% for RRP cases. The Kaplan–Meier plots showed overlapping survival curves with no association between type of surgery and probability of BCR (Fig. 2, log-rank, P = 0.177).

Figure 2.

Kaplan–Meier curves of BCR-free survival in patients treated either by LRP or open RRP. At 6 years, the Kaplan–Meier estimates of survival were 83% and 86% for patients treated by LRP and RRP, respectively (log-rank P = 0.18).

Univariate survival analysis of established prognostic factors; however, revealed significant differences in survival (Fig. 3 and Tables 3 and 4). Stratified analyses showed LRP and RRP achieved equivalent results across different risk groups and patient characteristics such as obesity (data not shown).

Figure 3.

Kaplan–Meier curves representing BCR-free survival by Gleason score (a), pathological stage (b), and margin status (c).

Table 3. Association of clinical and pathological factors with BCR in univariate and multivariate Cox regression analyses
 Univariate analysisMultivariate analysis
HR (95% CI)PHR (95% CI)P
PSA1.1 (1.10–1.16)<0.0011.1 (1.04–1.12)<0.001
pT2Reference Reference 
pT35.0 (3.44–7.16)<0.0011.9 (1.21–2.89)0.005
Gleason sum    
≤6Reference Reference 
76.6 (4.3–10.0)<0.0014.3 (2.63–6.90)<0.001
≥816.4 (9.0–29.8)<0.0019.3 (4.78–18.17)<0.001
NegativeReference Reference 
Positive3.0 (2.30–3.83)<0.0012.0 (1.37–3.01)<0.001
Prostate weight1.0 (0.97–0.999)0.032 
BMI1.1 (1.05–1.14)<0.0011.1 (1.05–1.15)<0.001
Age at RP1.0 (0.98–1.04)0.513 
Date of RP1.0 (1.00–1.00)0.640 
RRPReference Reference 
LRP1.3 (0.88–1.99)0.1780.9 (0.61–1.43)0.745
Table 4. Association of clinical and pathological factors with SRT in univariate and multivariate Cox regression analyses
 Univariate analysisMultivariate analysis
HR (95% CI)PHR (95% CI)P
PSA1.1 (1.08–1.16)<0.0011.1 (1.01–1.11)0.010
pT2Reference Reference 
pT34.7 (3.07–7.15)<0.0012.0 (1.21–3.19)0.007
Gleason sum    
≤6Reference Reference 
75.7 (3.5–9.1)<0.0013.8 (2.20–6.42)<0.001
≥812.2 (6.1–24.4)<0.0017.1 (3.31–15.06)<0.001
NegativeReference Reference 
Positive3.6 (2.32–5.43)<0.0012.4 (1.52–3.77)<0.001
Prostate weight9.8 (0.97–0.998)0.025 
BMI1.1 (1.04–1.14)<0.0011.08 (1.02–1.13)0.002
Age at RP1.0 (0.95–1.01)0.197 
Date of RP1.0 (1.00–1.00)0.883 
RRPReference Reference 
LRP1.6 (0.98–2.46)0.061.0 (0.61–1.60)0.963

In a multivariate Cox regression model, surgical technique had no significant impact on recurrence (hazard ratio 0.9, P = 0.75), whereas Gleason score, pathological stage, preoperative PSA level, margin status and BMI were significant prognostic factors (Table 3). No interactions between the different variables were detected.

Of the 120 patients presenting with BCR, 90 (75%) underwent SRT. Therefore, the use of SRT as the endpoint of survival analysis yielded similar results as the use of BCR: univariate analysis showed no difference in SRT-free survival (Fig. 4), and in a multivariate Cox regression model, surgical technique did not predict SRT, whereas Gleason score, stage, margin status, PSA level and BMI were significant predictors (Table 4).

Figure 4.

Kaplan–Meier curves of SRT-free survival after LRP and open RRP (log-rank P = 0.058).


In the present study of 844 patients concurrently operated by either LRP or open RRP, we found no difference in the rates of BCR and SRT. In a multivariate analysis, BCR and SRT were significantly associated with preoperative PSA level, pathological stage, margin status, and patient BMI, but not with surgical approach. Both techniques achieved equivalent cancer control across different risk groups regardless of patient characteristics, e.g. BMI.

Compared with other series of concurrent LRP and open RRPs, the present patient population had a low-to-moderate oncological risk profile (Table 1). Similar proportions of stage pT3 tumours, positive surgical margins and well-to-moderately differentiated tumours are reported by Tewari et al. [9], Drouin et al. [5] and Krambeck et al. [7]. In these series, analyses of BCR-free survival did not show any difference between open and minimally invasive RP, but were likely to be underpowered due to short follow-up (<2 years in [7, 9]) and small sample sizes (<100 patients per group in [5]). With 3- and 5-year BCR-free survival rates of 95 and 91% for pT2 tumours and 76 and 60% for pT3 tumours, the present results lie well within the range of non-comparative series of open and minimally invasive RPs with a long follow-up (studies discussed in [12]).

The present study is unique in several aspects. First of all, it minimises sources of potential biases inherent to cohort studies. Patients were treated in the same department, thereby reducing biases due to differences in patient recruitment, treatment choice, perioperative care and pathological analysis. Furthermore, due to the long follow-up, this analysis permits assessment of mid- to long-term oncological outcomes, including BCR and use of SRT. While cancer-specific survival would have been a preferable outcome, both BCR and SRT are relevant events after RP and function as surrogate oncological endpoints for disease status, and additionally have considerable impact on patients' quality of life. Finally, due to its large sample size, the present cohort was sufficiently powered to detect a 9% difference in cumulative BCR-free survival.

All participating surgeons were well into their learning curve at the beginning of the present study. All LRPs were performed by a single surgeon, but we did not detect surgeon-specific differences in pathological outcomes in the RRP group. We therefore think that the present results can be generalised to other centres practicing LRPs with comparable patient volumes.

With few pure laparoscopic surgeons still performing LRP and the general trend toward use of robotic assistance, the question arises whether similar results will be realised with other forms of minimally invasive RP. Robotic assistance appears to shorten the learning curve of LRP and to provide greater technical ease [13, 14]. One might therefore hypothesise, given equal surgical proficiency, that robot-assisted LRP will offer the same cancer control as open RRP. This remains to be shown in well-balanced studies.

As in the present study, there is increasing evidence that minimally invasive techniques for RP achieve similar oncological and functional outcomes compared with RRP. The perceived benefits of minimally invasive techniques include short-term parameters; perioperative pain, blood loss, duration of hospital stay and operative costs. Additionally, there may be benefits in surgical training, visualisation and shortened learning curves. Given the cost of equipment for minimally invasive surgery, further evaluation to determine if these benefits are truly being realised will be at the centre of many future studies.

The present study has limitations due to its retrospective design. LN dissection was systematically performed in RRPs but only in a third of LRPs, depending on the preoperative risk assessment of positive LNs, which reflects current practice. Few patients present positive LNs at RP in contemporary series, and the small fraction of missed patients with positive LNs would, at worst, lead us to underestimate the tumour control provided by LRP. Differences in the duration of follow-up were carefully analysed and no biases or interactions with pathological characteristics were found. Furthermore, 10% of the present patient records lacked postoperative PSA values, and could not be included in the analysis. These patients tended to be older, suggesting the presence of competing diseases. Nevertheless, established prognostic factors were identical in included and excluded patients. Therefore, we do not expect this lack of data to bias the present results.

In conclusion, in the present cohort, the use of laparoscopy for the treatment of localised prostate cancer offered the same cancer control as open surgery. In the absence of randomised-controlled trials, this well-balanced study adds strong evidence to the topic. Additional studies on functional outcomes, surgical training and costs will provide more knowledge on the true value of minimally invasive prostate surgery.


This work was supported in part through a research grant to G.J.W. (Swiss Cancer League grant no. KLS-02593-02-2010). The authors thank Kevin McLaughlin for the logistical support.

Conflict of Interest

None declared.