Is patient outcome compromised during the initial experience with robot-assisted radical cystectomy? Results of 164 consecutive cases

Authors


Khurshid A. Guru, Department of Urologic Oncology, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, New York, USA. e-mail: khurshid.guru@roswellpark.org

Abstract

Study Type – Therapy (case series)

Level of Evidence 4

What’s known on the subject? and What does the study add?

There is suggestion in the literature that patient outcomes are linked to surgeon volume and/or experience. This has become particularly relevant in the examination of robot- assisted radical prostatectomy in the treatment of men with localized prostate cancer, where some have advocated a "minimum" number of cases necessary to become proficient.

The current study demonstrates that, in the treatment of locally advanced bladder cancer, sequential robot-assisted radical cystectomy case number was not associated with an increased incidence of complications, blood loss, positive surgical margins, or survival in a single high-volume institution.

OBJECTIVE

• Robot-assisted radical cystectomy (RARC) remains controversial in terms of oncologic outcomes, especially during the initial experience. The purpose of this study was to evaluate the impact of initial experience of robotic cystectomy programs on oncologic outcomes and overall survival.

PATIENTS AND METHODS

• Utilizing a prospectively maintained, single institution robotic cystectomy database, we identified 164 consecutive patients who underwent RARC since November 2005.

• After stratification by age group, gender, pathologic T stage, lymph node status, surgical margin status, and sequential case number; we used chi-squared analyses to correlate sequential case number to operative time, surgical blood loss, lymph node yield, and surgical margin status.

• We also addressed the relationship between complications and sequential case number. We then utilized Cox proportional hazard modeling and Kaplan-Meier survival analyses to correlate variables to overall mortality.

RESULTS

• Sequential case number was not significantly associated with increased incidence of complications, surgical blood loss, or positive surgical margins (P= 0.780, P= 0.548, P= 0.545). Case number was, however, significantly associated with shorter operative time and mean number of lymph nodes retrieved (P < 0.001, P < 0.001).

• Sequential case number was not significantly associated with survival; however, tumour stage, the presence of lymph node metastases, and positive surgical margins were significantly associated with death.

• Although being the largest of its kind, this was a small study with short follow-up when compared to open cystectomy series.

CONCLUSION

• Initial experience with RARC did not affect the incidence of positive surgical margins, operative/postoperative complications, or overall survival in a single-institution series.

INTRODUCTION

Traditionally the standard treatment for muscle-invasive bladder carcinoma has been open radical cystectomy with extended bilateral pelvic lymph node dissection. In recent years, however, many urologists have adopted minimally invasive surgical approaches in managing urologic malignancies, including invasive bladder cancer. Minimally invasive surgery (MIS) offers lower average blood loss, earlier return of bowel function, and quicker postoperative convalescence when compared to traditional open surgery [1,2]. Robot-assisted radical prostatectomy has emerged as an oncologically equivalent option in the treatment of prostatic adenocarcinoma [3,4] .

There exists a multi-factorial learning curve associated with MIS that remains difficult to define. Prior studies have tried to address this when it comes to prostatectomy and hysterectomy [5,6].. Radical cystectomy is a more technically demanding procedure than prostatectomy, which makes adoption of the minimally invasive approach even more challenging [7]. Recent evidence demonstrates the feasibility and safety of the robot-assisted approach [2,7–9]. Despite this, robot-assisted radical cystectomy (RARC) remains controversial in terms of oncologic outcomes, especially during the initial experience. When treating bladder cancer, invasive disease is often lethal and requires aggressive therapy. Five year disease-free survival following traditional open radical cystectomy ranges from 50–60%[10–12]. The purpose of this study was to evaluate the impact of a learning curve on operative and postoperative complications, oncologic outcomes, and overall survival.

PATIENTS AND METHODS

From November 2005 to July 2009, 164 consecutive patients diagnosed with muscle invasive bladder cancer underwent RARC at Roswell Park Cancer Institute (RPCI). The technique utilized for RARC was done as previously described [10]. RPCI began its robot-assisted surgical program in August 2004. After 100 robotic prostatectomies had been reached, a decision was made to start the robotic cystectomy program. All cystectomies done since have been robot-assisted and performed by a single surgeon (KAG). The prospectively maintained database was retrospectively reviewed after approval by the institutional review board. Patient demographics, including age and gender, and operative variables such as time to cystectomy, estimated blood loss (EBL), and intraoperative/postoperative complications were assessed. Complications were estimated using the Clavien classification system [13]. Pathological variables, including tumor stage, surgical margin status, and lymph node status, were correlated to overall survival.

Patients were divided according to sequential case number into 3 groups (1–50, 51–100, 101–164). After stratification by age group, gender, pathologic T stage, lymph node status, surgical margin status, and sequential case number; chi-squared analysis was used to correlate sequential case number to operative time, surgical blood loss, lymph node yield, and surgical margin status. To evaluate the association of sequential case number to overall survival, patients were divided into an initial group of 50 patients and a subsequent group of 114 patients so that follow-up would be roughly equivalent. Cox proportional hazard modeling and Kaplan−Meier survival analyses was then utilized to correlate the aforementioned variables to overall mortality. Hazard ratios were calculated with associated P-values. Statistical significance was achieved when the 95% CI did not cross 1.0. Statistical analyses were performed with Stata, version 10.0 (College Station, TX).

RESULTS

A total of 164 patients underwent RARC at RPCI. Mean patient age was 68 years (36–90). One hundred thirty patients (79%) were male. One hundred fifty two patients (93%) had an ileal conduit urinary diversion, 11 (7%) had an orthotopic neobladder, and 1 (0.6%) had a continent cutaneous diversion. Eighty-three patients (51%) had ≤T2 disease on the cystectomy specimen, 59 (36%) had T3 disease and 22 (13%) patients had T4 disease. Mean lymph node yield was 22.0 nodes (range 2–64 nodes) while surgical margin positivity was 8.5% (14 of 164 patients). Demographics and pathologic parameters are listed in Table 1.

Table 1.  Demographics and overall survival by age group, gender, tumor T stage, nodal status, surgical margin status, and sequential case number for patients who underwent RARC for bladder cancer. Hazard ratios are listed with associated P-values. Statistically significant values are depicted in bold. Statistical significance was achieved when the 95% CI does not cross 1.0
  Alive (%)Dead (%)TotalHR (95% CI)P-value
Age Group      
 ≤6029 (75)10 (25)39Reference 1.0 
 61–7038 (79)10 (21)480.89 (0.37–2.14)0.797
 71–8043 (72)17 (28)601.11 (0.51–2.42)0.796
 ≥8110 (59)7 (41)172.36 (0.89–6.28)0.083
Gender      
 Male97 (75)33 (25)130Reference 1.0 
 Female23 (68) 11 (32)341.18 (0.59–2.35)0.634
Stage      
 ≤T274 (89)9 (11)83Reference 1.0 
 T337 (63)22 (37)594.48 (2.05–9.79)<0.001
 T49 (41)13 (59)225.95 (2.54–13.97)<0.001
Nodal status      
 Negative 101 (86)16 (14)117Reference 1.0 
 Positive19 (40)28 (60)477.10 (3.72–13.55)<0.001
Margin status      
 Negative115 (77)35 (23)150Reference 1.0 
 Positive5 (36)9 (64)143.75 (1.80–7.86)<0.001
Sequential case number      
 ≤5129 (57)21 (43)50Reference 1.0 
 51–10035 (70)15 (30)500.91 (0.46–1.79)0.792
 101+56 (86)8 (14)640.75 (0.32–1.76)0.506
Total 12044164  

Median follow-up was 8 months (range 0–47 months). Overall, 105 patients (64%) experienced some sort of intraoperative or postoperative complication (Table 3A). There was no statistically significant difference in complication rates attributable to case number (P= 0.780). The majority of patients (76%) had only minor complications requiring no secondary intervention (Clavien grade 2 or lower) (Table 3B). Grades of complication were similar between sequential case number groups, with the exception of Clavien 5 complications, 75% of which (3 out of 4) occurred in the first 50 cases (P= 0.492). There were 4 patients (2%) who died secondary to operative or postoperative complications. Two of these patients (1%) died within 30 days of surgery (not shown). There was no clinically significant difference between patients who died in the initial 50 cases versus those who died in subsequent cases. The hospital readmission rate (overall 36%) was not significantly different between sequential case group (P= 0.588) (not shown).

Table 3.  Presence of postoperative complication (A) and highest Clavien score achieved (B) by sequential case number
A
Postoperative complicationCase NumberTotal
 ≤5051–100 101+ 
No (%)15 (34)19 (38)24 (37.5)59 (36)
Yes (%)35 (68)31 (62)40 (62.5)105 (64)
Total505064164
B
Clavien classificationCase NumberTotal
  1. Chi2 = 0.780.

  2. Chi2 = 0.492.

 ≤5051–100 101+ 
016192459
11371131
29 111535
39121435
40000
53104
Total complications505064164

Sequential case number was not significantly associated with mean estimated intraoperative blood loss (P= 0.548) or positive surgical margins (P= 0.545). Case number was, however, significantly associated with shorter operative time (P < 0.001) and the mean number of lymph nodes retrieved (Table 2). Mean cystectomy operative time decreased from 180 min in the initial cases to 136 min in the last cohort, and lymph node yield increased from an average of 16 nodes in the first 50 cases to 26 nodes in the latest patients.

Table 2.  Outcomes of the first 164 single-institution robotic assisted radical cystectomies. Statistical analysis is by chi2 comparison
  Mean time for cystectomyMean EBLMedian No. LNPositive margins
Case number     
   1–50180566164
  51–100165 631236
 101–164136 521244
P-value <0.0010.548<0.0010.545

Tumor stage, the presence of lymph node metastasis, and positive surgical margins were significantly associated with death (P < 0.001 in all groups, respectively) (Table 1, Fig. 1). Patient age group and gender were not significantly associated with overall survival. Overall survival between patients in the initial 50 cases and the subsequent 114 cases did not differ significantly (P= 0.613) (Fig. 2). A total of 13 patients (8%) received neoadjuvant chemotherapy, and 41 patients (25%) received adjuvant chemotherapy (not shown). Adjuvant chemotherapy was offered to all patients with node-positive disease and/or pathological T3 or higher stage. There was no significant difference between case groups in terms of receipt of neoadjuvant or adjuvant chemotherapy (P= 0.208 and P= 0.181, respectively). The percentage of patients receiving chemotherapy was evenly distributed among the groups and did not significantly affect the survival analyses.

Figure 1.

Overall survival stratified by age group, pathologic T stage, lymph node involvement, and surgical margin status.

Figure 2.

Overall survival by sequential case number.

DISCUSSION

With the advent of the da VinciTM Surgical System (Intuitive Surgical, Sunnyvale, CA), urologists have yet another tool in their oncologic armamentarium. As with any new surgical technique, however; there comes an associated learning curve. In oncologic surgery, the surgeon must quickly become proficient without compromising cancer control. With this in mind, we sought to evaluate the impact of a learning curve associated with RARC on various outcome parameters, including surgical margin status, operative time, complication rates, and ultimately survival.

Several studies have addressed the relationship between sequential volume and outcomes in various open surgical procedures. McCabe et al. confirmed a significant inverse relationship between surgeon caseload and mortality in 6308 patients undergoing radical cystectomy for invasive bladder cancer in England [14]. They identified that a caseload of at least eight operations per year was associated with the lowest mortality rate. In our series, survival was driven largely by tumor characteristics, including stage and nodal status. While overall volume may play a role, sequential case number did not affect overall survival.

Learning curves for minimally invasive urologic procedures have been examined with a variety of metrics to assess outcome. Vickers et al. compared learning curves between open and laparoscopic approaches for radical prostatectomy and found that surgical outcomes seemed to improve with more cases in each technique, albeit at a slower rate in laparoscopic procedures [15,16]. The authors alluded to the fact that laparoscopy is inherently more difficult to learn, as it belies the loss of three-dimensions and a majority of haptic feedback. Pruthi et al. reported on a learning curve for robot-assisted laparoscopic radical cystectomy using their initial 50 consecutive patients with clinically localized disease (≤T2) [17]. Despite the high operative times and blood loss, there were no positive margins and lymph node yields remained stable throughout their experience in this small population. In our group, only 51% of patients had ≤pT2 disease. Our positive margin rates were not driven by sequential case number, but instead by tumor pathologic characteristics. Lymph node yields, however, did differ significantly between the initial and latter cases. This is, in part, explained by an increasing familiarity and level of experience with complex peri-vascular dissection. However, this is also explained by the fact that, of the lymphadenectomies done in the initial 50 cases, 20% were a standard bilateral obturator node dissection only, while 100% of the lymphadenectomies from cases 100 onward were extended to the aortic bifurcation (not shown). One must also keep in mind that the learning curves in both of these series, however, reflect that of experienced robotic surgeons.

Regardless of technique, the most crucial goal of the procedure remains the optimization of oncologic outcomes – long-term survival and prevention of local or metastatic recurrence. Evidence demonstrates that patients with positive surgical margins have poor prognoses; and positive surgical margins have been found to be independently associated with disease specific death [18]. In the Southwest Oncology Group 8710 study of 268 patients, Herr et al. reported a 9.3% positive surgical margin rate [19]. In Dotan et al. ’s study of 1589 patients treated by radical cystectomy, the incidence of positive soft tissue surgical margins was 4.2%. Risk factors for positive soft tissue surgical margins included female gender, locally advanced cancer, the presence of vascular invasion, and mixed tumor histology. Patients with organ-confined disease had no incidence of a positive surgical margin, while 9% of patients with extravesical disease had a positive surgical margin. Soft tissue surgical margin rates in these open series are comparable to the current study of RARC, with a surgical margin positivity rate of 8.5%. Again, we found that sequential case number did not affect margin status; instead margin status was reflective of tumor stage. The dissections done in our series included all perivesical tissue posteriorly to the rectum and laterally to the pelvic sidewall. Yuh et al. stressed the importance of dissecting these perivesical spaces in order to attempt to decrease the surgical margin positivity rate in patients with bulky tumors [20].

In addition to achieving a negative surgical margin, performing an adequate lymph node dissection is an integral part of bladder cancer surgery in terms of prognosis, and some argue, therapy [7,9,12]. Evidence from several open series has demonstrated the clinical significance of lymph node metastases on overall survival [7,21,22]. Five year survival in contemporary series ranges from 31% to 38% in patients with nodal metastases. Similarly, we found that the presence of positive lymph nodes was the most highly correlative variable to overall survival; patients with nodal metastases were 7.1 times more likely to die over the follow-up period than those without metastases. Stein et al. evaluated their long term surgical experience and clinical outcome of 1054 patients who had undergone radical cystectomy and bilateral extended pelvic lymphadenectomy [9]. Median follow-up in this study was 10.2 years, and 246 patients (24%) had lymph node involvement. Nearly 31% of these node-positive patients were alive at 5 years, and 23% were alive at 10 years. Since the inception of RARC at our institution in 2005, 43% of patients with nodal involvement are currently alive, and short-term survival rates appear comparable to open series.

In conjunction with oncologic outcomes and survival, complication rates are an important consideration in the analysis of any new surgical technique. In our series, we found that 105 patients (64%) had some sort of measurable complication during or after their RARC. This is essentially equivalent to the 64% 90-day complication rate seen by Shabsigh et al. when they retrospectively reviewed 1142 consecutive open cystectomies from Memorial Sloan Kettering Cancer Center [23]. We also found that the majority of patients (76%) did not require any subsequent procedural intervention (i.e. Clavien grade 2 complication or less), and that this rate remained relatively constant throughout the sequential case number groups (76% in the first 50 cases, 74% in cases 51–100, and 78% in cases 101 and greater). The only difference came with Clavien grade 5 complications (patient death), in which 3 patients died in the initial 50 cases, compared to 1 in the next 50 cases, and 0 in the remaining 64 cases.

To our knowledge, this is the first study to assess the impact of a learning curve and initial experience with RARC on patient outcomes and overall survival. This study demonstrates not only the feasibility and safety of the procedure, but similar oncologic, complication, and survival data compared to open series. The current study population was diverse with 51% of patients having organ-confined disease (pT≤2) and 49% with locally advanced disease (pT3-4). Furthermore, our series is unique in that it is devoid of selection bias, as all patients with an indication for radical cystectomy were treated with RARC. No open cystectomies have been performed at our institution since the initiation of the robotic program. This study demonstrates that sequential case number is not significantly associated with an increased incidence of complications, positive surgical margins, or overall survival.

Limitations to our study include its short overall follow-up and, despite being the largest single-institution series of RARC, an overall small number of patients compared to open series. Other limitations include the retrospective nature of the database query and, as with any database analysis, the possibility of data entry miscoding. This potential error, however, should be nonselective over the cohort analyzed, and in effect, cancel out any overt bias.

CONCLUSION

Initial experience with RARC did not affect the incidence of positive surgical margins, operative/postoperative complications, or overall survival in a single-institution series. Although larger series with longer follow-up are necessary, these results suggest that RARC can meet the standards set by open radical cystectomy.

CONFLICT OF INTEREST

Khurshid Guru is a paid speaker for Intuitive Surgical Systems and is a partner in Simulated Surgical Systems.