R.S. Lance, Department of Surgery, Urology Service, Madigan Army Medical Center, Tacoma, Washington, 98431-1100, USA.
Objective To review and compare the outcome of patients undergoing radical retropubic prostatectomy (RRP) or radical perineal prostatectomy (RPP) for clinically localized prostate cancer.
Patients and methods From 1988 to 1997, 1382 men who were treated by RRP and 316 by RPP were identified from databases of the Uniformed Services Urology Research Group. The following variables were assessed; age, race, prostate-specific antigen (PSA) level before surgery, clinical stage, biopsy Gleason sum, estimated blood loss (EBL), margin-positive rate, pathological stage, biochemical recurrence rate, short and long-term complication rates, impotence and incontinence rates. To eliminate selection bias, the analysis was concentrated on pairs of patients matched by race, preoperative PSA level, clinical stage and biopsy Gleason sum.
Results In the 190 matched patients there were no significant differences between the RRP and RPP groups in either organ-confined (57% vs 55%), margin-positive (39% vs 43%), or biochemical recurrence rates (12.9% vs 17.6% at a mean follow-up of 47.1 vs 42.9 months), respectively. The mean EBL was 1575 mL in the RRP group and 802 mL in the RPP group (P < 0.001). The only significant difference in complication rates was a higher incidence of rectal injury in the RPP group (4.9%) than in the RRP group (none, P < 0.05).
Conclusions In similar populations of patients, RPP offers equivalent organ-confined, margin-positive and biochemical recurrence rates to RRP, while causing significantly less blood loss.
Since the early 1980s, with the anatomical studies of Walsh and Donker , radical retropubic prostatectomy (RRP) has become the ‘gold standard’ for the curative treatment of clinically localized prostate cancer. The ability to harvest pelvic lymph nodes and spare cavernosal nerves were key advantages leading to the acceptance of this technique as the current standard against which all others are compared. The major disadvantage is blood loss from the dorsal vein complex, with a mean estimated blood loss (EBL) in large studies of 579–2266 mL, requiring blood transfusions in most patients [2,3].
Recently, Partin et al., reviewing the incidence of pelvic lymph node metastasis in patients with clinically localized prostate cancer, developed a nomogram which provides predictive factors to exclude many patients from mandatory pelvic lymph node dissection (PLND). Such studies, coupled with the development of laparoscopic techniques, have led to a resurgence in the use of radical perineal prostatectomy (RPP). Billroth carried out the first planned perineal prostatectomy in 1867  and Young popularized the classic supra-sphincteric RPP in 1905 . The advantages of RPP over RRP include decreased blood loss by avoiding the dorsal vein complex, and early mobilization and discharge from the hospital after surgery. The major disadvantages of RPP are that PLND cannot be conducted through the same incision, there is a higher incidence of rectal injury and there are technical limitations of gland size. With the development of the cavernosal nerve-sparing RPP technique described by Weldon and Tavel , there is no clear advantage of anatomical RRP over RPP for preserving potency.
A recent study  compared patients undergoing RRP or RPP; there were no significant differences between the groups in margin-positive, organ-confined or PSA failure rates. The remaining questions are whether RPP provides comparable pathological specimen quality, biochemical recurrence rates, and short and long-term morbidity compared with RRP.
Thus the objective of the present study was to answer these questions by expanding the study population to include patients treated by urologists using both techniques, and to eliminate selection bias by using matched comparisons. We used the newly organized tri-service Uniformed Services Urology Research Group (USURG), a collaborative group of military academic medical centres. The congressionally mandated Center for Prostate Disease Research (CPDR) database was used to provide data on patients at participating USURG medical centres.
Patients and methods
The prostate cancer databases were queried for the years 1988–97 at the following USURG institutions: Madigan Army Medical Center, Tacoma, Washington; Naval Medical Center, San Diego, California; Wilford Hall Air Force Medical Center and Brooke Army Medical Center, both in San Antonio, Texas; and Walter Reed Army Medical Center, Washington, DC; 1698 men were identified who had undergone radical prostatectomy (mean age 63.3 years, range 39–76).
The variables examined included: (before surgery) age at surgery, race, PSA level, biopsy Gleason sum and clinical stage; (during surgery) EBL, organ-confined rate, margin-positive rate, location of margins, pathological stage, Gleason sum and transfusion rate; (after surgery) short- and long-term complication rates, and biochemical recurrence rates.
The EBL was defined as the volume of blood loss recorded by the anaesthetist during surgery. The clinical and pathological staging was based on the 1992 American Joint Committee for Cancer Staging TNM classification . Organ-confined disease was defined as a pathological stage of < T2c, margin-positive as inked surgical margin(s) involved with cancer, and biochemical recurrence as a PSA level of > 0.2 ng/mL (Tandem, Hybritech, San Diego, CA). Incontinence after surgery was defined as the involuntary loss of any urine with or without pad usage, and impotence as the inability to gain an erection sufficient for satisfactory sexual intercourse.
To provide the most accurate comparison of patients undergoing the two types of radical prostatectomy the analysis used pairs of patients from the two surgical groups, matched by race, preoperative PSA level, clinical stage and biopsy Gleason sum. The PSA value was log-transformed and thereafter treated as a normally distributed continuous variable for analysis. The log PSA values were transformed back into linear units for descriptive purposes. The Gleason sum, clinical and pathological stage were treated as ordinal variables. In the unmatched dataset, chi-square, Mann–Whitney, and independent t-tests were used to compare the groups for categorical, ordinal and continuous data, respectively. Methods appropriate for paired data (McNemar's, Wilcoxon sign rank, Wilkes-Shapiro test and paired t-tests) were used with the matched dataset. All statistical tests were two-sided. In the matched analysis patients with one or both data fields missing were excluded from the analysis for that variable. Statistical significance was defined as P < 0.05.
In all, 1698 men underwent radical prostatectomy at the participating USURG medical centres between 7 January 1988 and 30 December 1997. Of these, 1382 men were treated by RRP and 316 by RPP. Men in the RRP group had a higher preoperative PSA level than their counterparts in the RPP group; the mean and median PSA in the RRP group were 12.43 and 7.40 ng/mL, compared with 7.36 and 5.35 ng/mL in the RPP group, respectively (P < 0.01). Although the differences were not statistically significant there was a trend toward a higher mean clinical stage and biopsy Gleason sum in the RRP group. This reflects a selection bias in most of the participating institutions toward RRP and PLND for those patients with higher clinical stage, biopsy Gleason sum and preoperative PSA level. Therefore, to eliminate selection bias, the subsequent analysis used patients matched by race, preoperative PSA level, clinical stage and Gleason sum.
In each group, 190 patients were identified who met the matching criteria; 120 pairs were perfectly matched for the Gleason sum, there was a difference of one grade in 67 pairs (higher in the RRP mate in 42 pairs and higher in the RPP mate in 25) and two grades different in three pairs (higher in the RRP mate in two pairs and the RPP mate in one). Table 1 summarizes the matched patient characteristics.
Table 1. The characteristics, outcome and location of positive margins of the matched pairs of patients
Mean age, years
Mean PSA level, ng/mL
Mean Gleason sum
Outcome [no. of pairs assessed]
EBL, mL 
Organ-confined, % 
Margin-+ve, % 
Seminal vesicle +ve 
Biochemical recurrence, % 
(mean follow-up, months
Mean blood transfusion, units:
Wound infection 
Wound separation 
Rectal injury 
(mean follow-up, months
Impotence, % 
(mean follow-up, months
Bladder neck contracture 
Location of positive margins, n
> two sites
Overall, there were no significant differences in either organ-confined, margin-positive, or biochemical recurrence rates (Fig. 1) in this matched group analysis (Table 1). The location of positive margins was assessed and classified as unknown, apical, lateral, anterior, posterior or bladder neck. There were 75 patients in the RRP group with positive margins and 82 in the RPP group. Table 1 also lists the analysis of margin location in the two groups; the location of the positive margin could not be determined from the pathology report in 11 patients in the RRP group and 17 in the RPP group. There was a higher incidence of posterior margin positivity in the RRP group, while the incidence of margin involvement at the bladder neck and the anterior prostate was higher in the RPP subgroup.
Twenty patients in the RRP group received a non-homologous transfusion, while 30 did so in the RPP group (P = 0.13). The only significant differences between the matched groups were a higher EBL in the RRP group (P < 0.001) and a higher rectal injury rate in the RPP group (P = 0.03). There were no differences between the groups for incontinence, impotence, bladder neck contracture or short-term complication rates (Table 1). No data were available to analyse differences in those undergoing a nerve-sparing approach or not, and the hospital stay for patients in the two surgical groups was not available for comparative analysis.
The analysis of matched patients showed no significant difference between RRP and RPP in margin-positive, organ-confined or biochemical recurrence rates. Previous unmatched comparison trials have likewise reported margin-positive and biochemical failure rates that are equivalent between the approaches [10,11]. A recent article challenges these findings; Boccon-Gibod et al. compared the incidence of positive surgical margins in patients who underwent RRP and RPP for clinically localized prostate cancer. In this retrospective trial, 94 patients underwent RPP (48) or RRP (46) for stage T1 and T2 prostate cancer. The authors differentiated between pT3a (extracapsular extension with no positive margins) and pT3b (extracapsular extension with positive margins), and intraprostatic carcinoma with or without incisions into benign tissue (iatrogenic positive margins). The overall incidence of positive margins was 56% in the perineal and 61% in the retropubic group, and biochemical failure-free survival, with a mean follow-up 25 months, was identical at 67% in each. However, the incidence of positive surgical margins in pT2 tumours was higher in the perineal than the retropubic group (43% vs 29%, P < 0.05). More surprising was the incidence of capsular incisions exposing benign glandular tissue, which was significantly higher in the perineal than in the retropubic group (90% vs 37%, P < 0.05), irrespective of positive surgical margins. The authors then concluded that even though the two procedures were equivalent in the incidence of margin-positive and biochemical recurrence rates, the radical retropubic approach was better because there was less likelihood of capsular incision.
Therefore, the authors found no data that would support their supposition. The most impressive result from their study may be that the biochemical recurrence rates were identical between the procedures, despite an incidence of iatrogenic capsular incisions of 90% in the perineal group. A more likely explanation for the high iatrogenic margin-positive rates reported in the perineal group may be lack of experience with the perineal technique in the hands of the authors. In larger series of radical perineal prostatectomies, in the hands of more experienced perineal surgeons [13,14], the long-term survival, margin-positive and biochemical recurrence rates are similar to those reported in RRP series . Furthermore, in the opinion of one of the present authors (J.B.T), the ability to obtain a wide dissection using the perineal approach is equivalent to RRP in properly selected patients, and the apical dissection, a region susceptible to margin involvement , is much easier.
The present series represents the largest contemporary comparison in which patients were matched by clinical stage, Gleason sum, race and preoperative PSA level. Potential bias is possible in comparative trials because the perineal group includes patients with a lower preoperative PSA level and Gleason sum, to prevent the need for PLND. We attempted to eliminate this bias by comparing the two procedures in matched patients. Additionally, academic centres were chosen where surgeons performing the procedures were trained in both RPP and RRP, to eliminate differences in technical expertise as much as possible. The series shows clearly that RPP offers equivalent organ-confined and margin-positive rates to RRP when the procedures are undertaken in similar patient populations. A potential weakness of the study is the relatively short mean follow-up used to assess biochemical recurrence. Ultimately, a longer follow-up is needed to conclude that RPP offers equivalent cancer control to RRP.
The present study, with other comparative series, shows a clear advantage of the perineal approach in having lower EBL and homologous transfusion rates. Furthermore, we confirm the relatively low morbidity of both approaches, except for the higher rectal injury rate in the perineal group.