Robot-assisted laparoscopic prostatectomy is not associated with early postoperative radiation therapy

Authors


Junzo Chino, DUMC 3085, Durham, NC, USA.
e-mail: junzo.chino@duke.edu

Abstract

OBJECTIVE

To compare open radical prostatectomy (RP) and robot-assisted laparoscopic prostatectomy (RALP), and to determine whether RALP is associated with a higher risk of features that determine recommendations for postoperative radiation therapy (RT).

PATIENTS AND METHODS

Patients undergoing RP from 2003 to 2007 were stratified into two groups: open RP and RALP. Preoperative (PSA level, T stage and Gleason score), pathological factors (T stage, Gleason score, extracapsular extension [ECE] and the status of surgical margins and seminal vesicle invasion [SVI]) and early treatment with RT or referral for RT within 6 months were compared between the groups. Multivariate analysis was used to control for selection bias in the RALP group.

RESULTS

In all, 904 patients were identified; 368 underwent RALP and 536 underwent open RP (retropubic or perineal). Patients undergoing open RP had a higher pathological stage with ECE present in 24.8% vs 19.3% in RALP (P = 0.05) and SVI in 10.3% vs 3.8% (P < 0.001). In the RALP vs open RP group, there were positive surgical margins in 31.5% vs 31.9% (P = 0.9) and there were postoperative PSA levels of 3 0.2 ng/mL in 5.7% vs 6.3% (P = 0.7), respectively. On multivariate analysis to control for selection bias, RALP was not associated with indication for RT (odds ratio (OR) 1.10, P = 0.55), or referral for RT (OR 1.04, P = 0.86).

CONCLUSION

RALP was not associated with an increase in either indication or referral for early postoperative RT.

Abbreviations
RALP

robot-assisted laparoscopic prostatectomy

RP

radical prostatectomy

RT

radiation therapy

ECE

extracapsular extension

SVI

seminal vesicle involvement

PSM

positive surgical margin

OR

odds ratio.

INTRODUCTION

After USA Food and Drug Administration approval of the daVinci Surgical System (Intuitive Surgical, CA, USA) in 2000, enormous interest has been directed towards developing the technique of robot-assisted laparoscopic prostatectomy (RALP) [1–3]. The use of RALP has increased dramatically in the last 5 years [4]. Benefits can include reduced hospital stay and reduced blood loss [5,6]. However, as with the introduction of any complex procedure the learning curve for RALP appears to be steep and the benefits may be overshadowed by the risks [7,8]. One recent report of >2500 men reported an association between minimally invasive radical prostatectomy (RP), including RALP, and a higher risk of salvage therapy [4]. The purpose of the present analysis, from a large academic hospital, was to determine if RALP is associated with a higher risk of features that determine recommendations for postoperative radiation therapy (RT).

PATIENTS AND METHODS

In this retrospective, Institutional Review Board approved study, all patients undergoing RP from 2003 until mid-2007 at Duke University Medical Center were identified using an approved in-house database. These dates were selected as they coincided with the start of RALP utilization in 2003, and the need for a minimum of 6 months follow-up for patterns of referral [9].

The clinical records were reviewed to identify pretreatment clinical characteristics: age at diagnosis, race, PSA level at diagnosis, biopsy Gleason score, percentage of positive cores and clinical T stage. The date and surgical technique (RALP, perineal or retropubic RP) were recorded. Pathology reports were then individually reviewed for Gleason score and the presence of known postoperative risk factors for recurrence: extracapsular extension (ECE), seminal vesicle involvement (SVI), and positive surgical margins (PSMs). Additionally, PSMs were further categorized according to magnitude (focal ≤5 mm or non-focal) and location (apical, peripheral, or bladder neck). Alhough the presence of ECE and SVI would not be expected to be affected by the surgical approach; these data were collected as they may influence the decision for postoperative RT.

Postoperative serum PSA levels were also analysed. The first postoperative PSA was excluded if performed within 1 month of surgery, to exclude spurious high levels from inadequate clearance of circulating antigen. The PSA level and follow-up notes were then reviewed for a minimum of 6 months to determine if RT referral was planned. RT referrals were scored if the urologist clearly stated that referral was to be pursued, regardless of whether there was evidence that an RT consult was performed or if RT was received. If RT referral was clearly planned but delayed due to improving urinary continence during this period, referral was scored. Patients with <6 months follow-up (n = 187) were excluded from this part of the study.

Univariate analysis examined the association of RALP with high-risk features (ECE, SVI, PSMs), detectable PSA level (≥0.2 ng/mL at >4 weeks after surgery) and referral for RT within 6 months of surgery using the chi-square test [10]. Patient selection bias in the RALP group was tested using the Mann–Whitney U-test for differences in age, preoperative PSA level, Gleason score, percentage of positive biopsy cores and clinical T stage [11].

Multivariate models were generated via logistic regression to determine the influence of preoperative clinical characteristics (clinical T stage, Gleason score, PSA level, year of surgery, race, age at surgery) and the use of RALP upon PSMs, PSM subcategories, detectable PSA level, and RT referral [12]. All P-values provided are for two-tailed tests.

RESULTS

In all, 904 patients met the entry criteria. Preoperative patient and disease characteristics are listed in Table 1. Important differences in disease characteristics were evident. Patients in the RALP cohort were more likely to have a lower pretreatment PSA level (P = 0.004), a lower Gleason score (P = 0.013), and a lower clinical T stage (P = 0.002).

Table 1.  Preoperative patient characteristics
VariableOpen RPRALPP
  • *

    Biopsy quantification available in 414 cases.

N536 (retropubic 428, perineal 108)368NA
Median (range)
 Age, years 60 (40–78) 59 (42–75) 0.075
N (%)
 Year surgery:  <0.001
  2003 90 12 
  2004103 79 
  2005150 71 
  2006148127 
  2007(first half) 45 79 
 Race   0.032
  White375 (70)276 (75) 
  Black101 (19) 60 (16) 
  Other 60 (11) 32 (9) 
 PSA level, ng/mL   0.004
  <10415 (85)319 (91) 
  ≥10, <20 52 (11) 28 (8) 
  ≥20 21 (4)  3 (1) 
 Clinical T stage   0.002
  T1c353 (73)281 (81) 
  T2a 94 (20) 55 (16) 
  T2b 16 (3)  5 (1) 
  T2c 12 (2)  5 (1) 
  T3a  9 (2)  0 
  T3b  1 (0.2)  0 
 Gleason score   0.013
  ≤6302 (61)245 (68) 
  3 + 4107 (22) 80 (22) 
  4 + 3 46 (9) 25 (7) 
  ≥8 38 (8)  11 (6) 
 >50% core biopsy involvement* 49/250 (20) 28/164 (17) 0.52

The postoperative disease characteristics are listed in Table 2. On univariate analysis, there were no differences in the incidence of PSMs (odds ratio (OR) 0.88, P = 0.90) or incidence of detectable PSA levels (OR 0.90, P = 0.74). Patients treated with RALP had less incidence of SVI (OR 0.34, P < 0.001) and a borderline trend towards less ECE (OR 0.72, P = 0.051).

Table 2.  Postoperative characteristics
VariableOpen RPRALPUnivariate model OR for postop. characteristic RALP vs open RPMultivariate model (adjusting for preop. characteristics) OR for postop. characteristic RALP vs open RP
  • *

    Any tertiary Gleason 5 included in ≥8 catagory;

  • Focal margins defined at ≤5 mm in extent;

  • First postoperative PSA taken ≥1 month after RP;

  • §

    RT indications include ECE, SVI, positive margins, or PSA level of ≥0.2 ng/mL;

  • Patients with 6 months follow-up after surgery; preop., preoperative; postop., postoperative.

N536 (retropubic 428, perineal 108)368  
N (%)
Pathological T stage  Overall for trend P = 0.090Overall for trend P = 0.47
 pTx  2 (0.4)  0  
 pT2a 77 (14) 52 (14)  
 pT2b  8 (2)  9 (2)  
 pT2c307 (57)230 (62)  
 pT3a 85 (16) 62 (17)  
 pT3b 53 (10) 14 (4)  
 pT4  4 (1)  1 (0.3)  
Pathological Gleason score  Overall for trend P = 0.10Overall for trend P = 0.74
 ≤6211 (39)158 (43)  
 3 + 4191 (36)131 (36)  
 4 + 3 54 (10) 51 (14)  
 ≥8* 79 (15) 28 (8)  
ECE present133 (25) 71 (19)0.72, P = 0.0510.92 P = 0.67
SVI present 55 (10) 14 (4)0.34, P < 0.0010.38, P = 0.006
Margin status
 Positive171 (32)116 (31)0.99, P = 0.901.16, P = 0.38
 Focally positive 64 (12) 55 (15)1.30, P = 0.191.33, P = 0.21
PSM location
 Apex 65 (12) 57 (16)1.33, P = 0.151.67, P = 0.03
 Periphery128 (24) 73 (20)0.79, P = 0.150.90, P = 0.59
 Bladder neck 28 (5) 17 (5)0.88, P = 0.681.05, P = 0.89
PSA level ≥0.2 ng/mL 31 (6) 19 (6)0.90, P = 0.741.39, P = 0.36
Number of RT indications§  Overall for trend P = 0.23Overall for trend P = 0.69
 1137 (26)105 (28)  
 2 65 (12) 44 (12)  
 3 29 (5)  9 (2)  
 4  9 (2)  0  
Any RT indication present240 (45)158 (43)0.92, P = 0.581.10, P = 0.55
RT referral referred 62 (13) 44 (13)1.03, P = 0.851.04, P = 0.86

On multivariate analysis (Table 2), adjusting for preoperative disease characteristics, there was a higher rate of apical PSMs in the RALP cohort (OR 1.67, P = 0.03), although no change in the overall rate of PSMs. There were no differences in incidence of RT indications (OR 1.10, P = 0.55) or RT referral (OR 1.04, P = 0.86).

PSMs were more prevalent with higher clinical stage, Gleason score, and initial PSA, and when more than half of the biopsy cores were positive (Table 3). There was no interaction of surgical approach and higher-risk category, except for a lower OR for PSMs with the use of RALP with a Gleason score ≥8 (OR 0.2, P = 0.04), although there were only 11 patients undergoing RALP in this subgroup.

Table 3.  PSMs by surgical approach and preoperative characteristics
VariableUnivariate association of PSM with characteristic, Pn/N (%) PSMsOR (95%CI) for PSM, P RALP vs open RP
Open RPRALP
Clinical T stage 0.015   
 cT1c 100/353 (28)91/281 (32)1.21 (0.9–1.7), 0.27
 cT2a  38/94 (40)14/55 (26)0.50 (0.2–1.0), 0.06
 cT2b   8/16 3/51.50 (0.2–11.5), 0.70
 ≥cT2c   11/22 (50) 2/50.67 (0.1–4.8), 0.69
Clinical Gleason score<0.001   
 ≤6  76/305 (25)63/245 (26)1.0 (0.7–1.5), 0.83
 3 + 4  43/107 (40)36/80 (46)1.2 (0.7–2.2), 0.51
 4 + 3  18/46 (39)10/25 (40)1.0 (0.4–2.8), 0.94
 ≥8  24/38 (63) 3/110.2 (0.05–0.96), 0.04
%+’ve biopsy cores 0.002   
 <50%  57/201 (28)40/136 (29)1.0 (0.6–1.7), 0.83
 ≥50%  26/49 (53)10/28 (36)0.5 (0.2–1.3), 0.14
Initial PSA level, ng/mL 0.006   
 ≤10 122/415 (29)99/319 (31)1.1 (0.8–1.5), 0.63
 10–20  22/52 (42)10/28 (36)0.7 (0.3–2.0), 0.57
 >20   11/21 (52) 2/31.8 (0.1–23.2), 0.64

The multivariate model for postoperative high-risk features is shown in Table 4. Preoperative PSA level and Gleason score had significant association with high-risk features (P < 0.001 for all comparisons). There was no association between the use of RALP and high-risk features. The analysis was repeated with the percentage of positive core biopsies included in the preoperative characteristics, with no effect on the overall results.

Table 4.  Multivariate analysis for the presence of at least one indication for early postoperative RT (PSM, PSA level of ≥0.2 ng/mL, ECE, or SVI) and for RT referral within 6 months of surgery
VariableRT indicationRT referral
OR (95% CI)P*OR (95% CI)P*
  • *

    P-values given with comparison to reference category;

  • OR per 1 year increase;

  • OR per 1 ng/mL increase.

Age1.02 (0.99–1.04)0.161.00 (0.96–1.03)0.84
Year of surgery1.06 (0.92–1.21)0.441.25 (1.02–1.55)0.04
Race
 White1.00 reference 1.00 reference 
 Black1.27 (0.86–1.88)0.230.94 (0.53–1.66)0.93
 Other0.98 (0.52–1.85)0.950.55 (0.20–1.52)0.55
Preoperative PSA level1.09 (1.05–1.13)<0.0011.08 (1.04–1.12)<0.001
T stage
 T1c1.00 reference 1.00 reference 
 T2a1.35 (0.90–2.01)0.141.03 (0.57–1.86)0.92
 T2b1.88 (0.70–5.06)0.211.60 (0.47–5.47)0.45
 ≥T2c2.09 (0.77–5.66)0.151.45 (0.47–4.48)0.52
Biopsy Gleason score
 ≤61.00 reference 1.00 reference 
 3 + 42.45 (1.70–3.55)<0.0011.76 (0.99–3.11)0.053
 4 + 32.99 (1.71–5.25)<0.0013.55 (1.79–7.04)<0.001
 ≥85.76 (2.73–12.2)<0.0015.30 (2.50–11.3)<0.001
Use of RALP1.10 (0.80–1.52)0.551.04 (0.64–1.72)0.86

The multivariate model for early RT referral is shown in Table 4. Age, race, clinical T stage, and use of RALP had no association with referral, while Gleason score (7 and ≥8 vs ≤6, P < 0.001), PSA level (OR 1.08 per ng/mL, P < 0.001), and year of surgery (OR 1.25, P = 0.04) were significant.

DISCUSSION

RALP is increasingly used in the treatment of prostate cancer. The enthusiasm for this technique is high, although the relative benefits and risks of this technique compared with the open RP remain controversial. Of particular concern is the oncological soundness of minimally invasive procedures and any change in the need for salvage or adjuvant therapies is of keen interest. The results of the present study suggest that, in the hands of an experienced robotic surgeon, there is no increased indication or utilization of further postoperative treatment incurred with this technique.

In multivariate analysis controlling for preoperative risk characteristics, there was no evidence for an association between RALP and higher overall margin positivity or indication for RT after surgery. The risk of apical PSMs was higher in the RALP cohort; however, this is in the context of the fact that overall rates of PSMs were no different between these groups, and not seen in the univariate comparison.

Differences in ECE and SVI would clearly not be expected to be dependent on surgical approach. The differences in univariate analysis can be explained by selection bias towards lower-risk patients in the RALP group (Table 1). The differences in ECE rates resolved with the multivariate model controlling for preoperative risk factors; however, a higher rate of SVI remains significant in the open RP group. This probably represents an incomplete correction for preoperative risk in these high-risk patients.

Hu et al. [4] examined the use and outcomes of minimally invasive RP in a random sample of Medicare beneficiaries treated between 2003 and 2005. In that analysis, RALP was associated with an increase in salvage therapies within 6 months of surgery (OR 3.7, 95% CI 2.81–4.81). This result is at variance with the present analysis but several important differences between the studies might explain the apparent lack of consistency.

First, full preclinical and pathological data was available in the present analysis, which allowed for proper accounting for differences in risk factors, as well as the ability to determine the incidence of indications for RT in both cohorts. Secondly, salvage therapy was defined as either RT or hormone therapy in the Medicare study whereas the present report only considered RT. Thirdly, year of surgery is not adjusted for by Hu et al.[4], which was associated with postoperative RT referral in the present study. This is probably related to the publication of several randomized studies indicating a benefit for postoperative RT [13–15]. Finally, the present analysis presents results from a single high-volume institution. Therefore, RALPs performed during the early learning curve at our institution have a limited impact on the present analysis, whereas the Medicare results probably include a higher proportion of RALPs performed early in the learning curve [16,17].

The present analysis is limited by the retrospective nature of the data. There are many other risk stratification variables that are unknown, such as PSA history before diagnosis and patient body mass index. There is some selection bias in the patients who returned for follow-up within 6 months, as often patients may be followed by urologists outside of the institution if they have an uncomplicated course postoperatively. These caveats stated, we think the absence of associations in the present data concerning RALP and postoperative high-risk features are a defence of the oncological soundness of the technique in skilled hands.

In conclusion, in this large single-institution series, RALP was not associated with an increase in indications or referrals for early postoperative RT.

CONFLICT OF INTEREST

None declared.

Ancillary