- Email: [email protected]
Comparative effectiveness of robot-assisted vs. open radical cystectomy
Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Original article
Comparative effectiveness of robot-assisted vs. open radical cystectomy Nawar Hanna, M.D.a,1, Jeffrey J. Leow, M.B.B.S., M.P.H.a, Maxine Sun, M.P.H.a, David F. Friedlander, M.D.a, Thomas Seisen, M.D.a, Firas Abdollah, M.D.b,2, Stuart R. Lipsitz, Sc.D.a, Mani Menon, M.D.b, Adam S. Kibel, M.D.a, Joaquim Bellmunt, M.D.c, Toni K. Choueiri, M.D.c, Quoc-Dien Trinh, M.D.a,⁎ a
Division of Urological Surgery and Center for Surgery and Public Health, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA b Henry Ford Hospital, Vattikuti Institute of Urology, Center for Outcomes Research, Analytics and Evaluation, Detroit, MI c Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Brigham and Women’s Hospital, Boston, MA Received 18 March 2017; received in revised form 8 June 2017; accepted 18 September 2017
Abstract Objectives: Over the past decade, robot-assisted radical cystectomy (RARC) has gained traction as an alternative to the conventional open approach open radical cystectomy (ORC). However, the benefits of RARC over ORC remain unclear. Our objective was to conduct a comparative effectiveness analysis between RARC and ORC using data from the National Cancer Data Base. Materials and methods: Within the National Cancer Data Base, we identified patients with localized muscle-invasive bladder cancer who underwent RC between 2010 and 2013. Patients were stratified according to surgical approach: ORC vs. RARC. Intraoperative endpoints included: the presence of positive surgical margins, the performance of a pelvic lymph node dissection, and number of lymph nodes (LN) removed. Postoperative endpoints included: length of stay (LOS), 30- and 90-day postoperative mortality (POM) rates, 30-day readmission rate, and overall survival (OS). To minimize selection bias, observed differences in baseline characteristics between RARC vs. ORC patients were controlled for using weighted propensity scores. Binary endpoints and OS were assessed using propensity score-adjusted logistic and Cox regression analyses, respectively. POM was assessed using propensity score weighted Kaplan-Meier survival estimates at 30 and 90 days after RC. Results: Of 9,561 patients who underwent RC, 2,048 (21.4%) and 7,513 (78.6%) underwent RARC and ORC, respectively. The use of RARC increased over time, from 16.7% in 2010 to 25.3% in 2013. With regard to intraoperative outcomes, RARC was associated with equivalent rates of positive surgical margins (9.3% vs. 10.7%, odds ratio [OR] ¼ 0.86, 95% CI: 0.72–1.03; P ¼ 0.10), higher rates of pelvic lymph node dissection (96.4% vs. 92.0%, OR ¼ 2.30, 95% CI: 1.67–3.16; P o 0.001), higher median LN count (17 vs. 12, P o 0.001), higher rates of LN count above the median (56.8% vs. 40.4%, OR ¼ 1.94, 95% CI: 1.55–2.42, P o 0.001). With regard to postoperative outcomes, receipt of RARC was associated with a shorter median LOS (7 vs. 8, P o 0.001), and lower rates of pLOS (45.0% vs. 54.8%, OR ¼ 0.68, 95% CI: 0.58–0.79; P o 0.001). The 30- and 90-day POM rates were 2.8%, 6.7% for ORC, and 1.4%, 4.8% for RARC, respectively (hazard ratio [HR] ¼ 0.48, 95% CI: 0.29–0.80, P ¼ 0.005 and HR ¼ 0.71, 95% CI: 0.54–0.93; P ¼ 0.014). Finally, with a mean follow-up of 26.9 months, on IPTW-adjusted Cox regression analysis, RARC vs. ORC was associated with a benefit in OS (HR ¼ 0.79, 95% CI: 0.71–0.88; P o 0.001). Conclusions: Our large contemporary study found an increased adoption of RARC between 2010 and 2013, with more than 1 out of 4 patients undergoing RARC by the end of the study period. We found that RARC was associated with higher LN counts, shorter LOS, and The data used in the study are derived from a deidentified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed, or the conclusions drawn from these data by the investigator. Quoc-Dien Trinh is supported by an unrestricted educational Grant from the Vattikuti Urology Institute, a Clay Hamlin Young Investigator Award from the Prostate Cancer Foundation, and a Genentech Bio-Oncology Career Development Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology. 1
Supported by a Young Investigator Award from the Quebec Urological Association.
2
Consultant/advisor of GenomeDx Biosciences. Corresponding author: Tel.: þ1-617-525-7350; fax: þ1-617-525-6348. E-mail address: [email protected] (Q.-D. Trinh). ⁎
http://dx.doi.org/10.1016/j.urolonc.2017.09.018 1078-1439/r 2017 Elsevier Inc. All rights reserved.
2
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
lower POM. Our results allude to potential benefits of RARC while we wait for more definitive answers from randomized trials. r 2017 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Radical Cystectomy; Robotic Surgery
1. Introduction Bladder cancer is the sixth most common cancer in the United States with an estimated 76,960 new cases in 2016 [1]. Of those, 20%–30% will present with muscle-invasive disease. For these patients, radical cystectomy (RC) with extended pelvic lymph node dissection is considered the gold standard treatment [2]. However, this complex procedure is associated with high rates of perioperative morbidity and mortality, which in part may be attributable to the older age and multitude of comorbid diseases that typically characterize this patient population [3–5]. Since its original description, open RC (ORC) has remained the most frequently used approach for bladder extirpation. However, beginning in 2003, robot-assisted RC (RARC) has emerged as a minimally invasive approach to RC [6]. Over the past decade, RARC has slowly gained acceptance in the urology community, growing from 0.6% of cases in 2004 to 12.8% in 2010 [7]. This is in stark contrast to robot-assisted radical prostatectomy, which has seen a dramatic rise from 1.8% in 2003 to 85% in 2013 [8]. The rapid adoption of robotassisted radical prostatectomy over open prostatectomy has been attributed to a combination of direct-to-patient advertisement, market competition, and debatably superior outcomes. That said, the benefits of RARC over ORC remain controversial [9–12]. Bochner et al. [10] recently reported results from a prospective trial of 118 patients randomized to either ORC (n ¼ 58) or RARC (n ¼ 60) at a highvolume tertiary referral center, which found similar rates of complications, positive surgical margins (PSMs), lymph node (LN) yields, length of stay (LOS), and quality of life at 3 and 6-months postoperatively. In that study, the only benefit of RARC was lower intraoperative blood loss, but with significantly longer operative time and higher costs. When examining the comparative effectiveness of RARC vs. ORC at the population level, a large study of 36,773 patients did not find any differences in terms of postoperative major complications and mortality [7]. Additionally, comparable short-term oncological and health-related quality of life outcomes have been described for both procedures [13,14]. Regardless, the most evidence regarding the comparative effectiveness of RARC vs. ORC relies on reports from academic/high-volume centers with limited numbers of participating surgeons. As such, we use the National Cancer Data Base (NCDB) to perform a contemporary comparative effectiveness analysis of RARC vs. ORC in a large sample of patients of all ages comprising 70% of all cancer cases in the United States. Our hypothesis
was that RARC in increasingly used and associated with some perioperative benefits. 2. Materials and methods 2.1. Data source The NCDB is a joint initiative of the American College of Surgeons, Commission on Cancer, and American Cancer Society. Established in 1989, it serves as a comprehensive clinical surveillance resource for cancer care in the United States [15]. The NCDB compiles data from 41,500 commission-accredited cancer programs in the United States and Puerto Rico. 2.2. Patient population We identified all patients who underwent RC for bladder cancer between 2010 and 2013 using International Classification of Disease for Oncology, third edition site codes (C67.0–C67.9). Treatment modality was recorded using the Facility Oncology Registry Standards manual. Cystectomy patients were identified using the surgery of the primary site codes 50, 60, 70, and 80 (total cystectomy, radical cystectomy, pelvic exenteration, and cystectomy not otherwise specified). Since 2010, the NCDB has recorded surgical approach in order to monitor patterns and trends in the adoption and usage of minimally invasive surgical techniques. Accordingly, patients were stratified according to surgical approach: ORC or RARC. Patients who underwent laparoscopic (nonrobot-assisted) and RARC converted to open approach were excluded. Patients with metastatic (American Join Commission on Cancer [AJCC] cMþ) or clinical positive node (AJCC cNþ) and those who received any form of radiation therapy were excluded. Finally, to fully evaluate the impact of surgical approach on postoperative outcomes, we excluded patients who received neoadjuvant chemotherapy. Following exclusions, 9,561 patients remained for further analyses (Fig. 1). 2.3. Covariates Patient covariates included age, sex, race (White, Black, and Other), residence location (metropolitan, urban, and rural), and insurance status (private, Medicare/medicaid, and uninsured). Socioeconomic status variables were defined according to census tract median household income (adjusted according to 2012 inflation) and the percentage of
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
3
Fig. 1. Flowchart diagram depicting patient selection within the National Cancer Data Base between 2010 and 2013. (Color version of figure is available online.)
patients without a high school diploma. Moreover, distance from patient’s residence to treating facility is provided and was dichotomized as ≤50 miles or 450 miles. Comorbidity status was defined using the Charlson comorbidity index (CCI), and categorized to CCI 0, 1, ≥2. Tumor characteristics included tumor grade and tumor stage based on the AJCC Stage Manual seventh edition. Provider covariates included facility type, calculated according to the number of patients diagnosed with bladder cancer within 1 year, and composed of community center (100–500 cases), comprehensive community cancer center (4500 cases), and academic (4500 cases with available graduate medical training). Location of facility was also available and reported (New England, Atlantic, East Central, West Central, and West Pacific). Individual annual hospital volume was calculated as the annual caseload of patients within our cohort treated at a given institution in a given year. Receipt of chemotherapy was defined according to SEER*RX (http://seer.cancer.gov/tools/seerrx/) criteria, which is recorded as single or multiagent chemotherapy in the NCDB, as previously reported [16]. Patients who received multiagent chemotherapy after surgery were considered as having received adjuvant chemotherapy. 2.4. Outcomes 2.4.1. Intraoperative endpoints These endpoints included the presence of PSMs, the performance of a pelvic lymph node dissection, and number of LNs removed. Pathologic information on margin status
and number of LNs removed are provided within the NCDB as it appears in the pathology report. Margins reported as residual tumor not otherwise specified, microscopic or macroscopic residual tumor was defined as positive. LN count was reported as a continuous variable as well as stratified according to the median of 14 (414 vs. ≤14). 2.4.2. Postoperative endpoints Perioperative outcomes measured included LOS, 30-day and 90-day postoperative mortality (POM), 30-day readmission following surgery, and overall survival (OS). Prolonged length of stay (pLOS) was defined as a LOS greater than the median within our cohort (7 d). Death from any cause following bladder cancer diagnosis was abstracted. Follow-up time was recorded from the time of surgery until death or loss to follow-up. 2.5. Statistical analyses In the first part of our analyses, we compared baseline characteristics of patients undergoing RARC vs. ORC. The Kruskal-Wallis and chi-square test were used to assess any statistical difference between the 2 treatment groups for continuously and categorical variables, respectively. Second, a multivariable logistic regression model was fitted using all preoperative available covariates to identify patient, tumor, and provider characteristics associated with the receipt of RARC. The proportion of patients undergoing RARC was reported for each study year and temporal trend analysis was performed using the estimated annual percentage change mixed linear regression methodology [17].
4
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Third, to account for selection bias, observed differences in baseline characteristics between patients who received RARC vs. ORC were controlled for using a weighted propensity score analysis. After fitting a propensity score model including all available preoperative covariates (as listed in Table 1), each patient was weighted by the inverse probability of being in the RARC vs. ORC, with the goal of balancing out observable characteristics between the 2 treatment groups; this approach is known as the inverse probability of treatment weighting (IPTW). Balance between covariates in weighted groups was also assessed by using the standardized differences approach and by comparing their distribution with unadjusted data. Using weighted data, rates for each endpoint were compared using chi-square and an extension of the Wilcoxon test with complex survey data [18] for categorical and continuous endpoints, respectively. Intraoperative and postoperative categorical endpoints were assessed using IPTW-adjusted logistic regression. POM and OS analyses were limited to 7,099 patients with follow-up and vital status information. IPTW-adjusted Kaplan-Meier curves were used to depict OS rates between the 2 groups. POM was assed using 30- and 90-day estimates derived from propensity score weighted KaplanMeier curves, with hazard ratios estimated at 30 and 90 days using a time-varying Cox regression model allowing for different hazards for each time point. An IPTWadjusted Cox regression model was used to evaluate the OS of RARC vs. ORC, after adjusting for preoperative covariates included in the propensity score model. All models included random hospital effects to account for clustering of patients within hospitals. All statistical analyses were performed using Stata 13.1 with a 2-sided significance level set at P o 0.05. An institutional review board waiver was obtained before conducting this study, in accordance with institutional regulation when dealing with deidentified administrative data.
3. Results 3.1. Baseline characteristics Overall, our cohort consisted of 9,561 patients who underwent RC between 2010 and 2013. Of those, 2,048 (21.4%) and 7,513 (78.6%) underwent RARC and ORC, respectively (Table 1). In our unweighted cohort, patients treated with RARC were more often male (78.8% vs. 74.1%, P o 0.001). With regard to sociodemographic characteristics, RARC patients were more often treated at an academic facility (66.8% vs. 54.1%, P o 0.001), living in a metropolitan county (80.8% vs. 76.8%, P ¼ 0.002), held higher education (29.5% vs. 24.3% with o7% of population without a high school degree, P o 0.001), and income (36.3% vs. 29.3% with 463,000$/y, P o 0.001).
Additionally, patients treated with RARC were more often treated at a high-volume hospital (median hospital annual volume of 9 vs. 7, P o 0.001). Finally, regarding tumor characteristics, RARC patients had lower clinical primary tumor stage (26.8% vs. 23.5% with ≤cT1, P ¼ 0.017). Following IPTW adjustment, all covariates were well balanced with an absolute standardized difference of less than 10% for all variables. The effect of IPTW adjustment on baseline characteristics distribution is depicted in Fig. 2. 3.1.1. Use of RARC The use of RARC has increased over time, from 16.7% in 2010 to 25.3% in 2013 (estimated annual percentage change þ13.4, 95% CI: þ3.3–24.5, P ¼ 0.03). 3.2. Predictors of the receipt of RARC On multivariable logistic regression, male sex (P o 0.001), academic center (P ¼ 0.001 vs. community center), and lower tumor stage (P ¼ 0.029 cT1 vs. cT2) were associated with receipt of RARC (Table 2). 3.3. Endpoints With regard to intraoperative outcomes, IPTW-adjusted analyses showed that RARC was associated with equivalent rates of PSMs (9.4% vs. 10.7%, odds ratio [OR] ¼ 0.86, 95% CI: 0.72–1.04; P ¼ 0.12), higher rates of lymphadenectomy (96.4% vs. 92.0%, OR ¼ 2.30, 95% CI: 1.67– 3.16; P o 0.001), higher median LN count (17 vs. 12, P o 0.001) and higher rates of LN count above the median (56.8% vs. 40.4%, OR ¼ 1.94, 95% CI: 1.55–2.42; P o 0.001) (Tables 3 and 4). With regard to postoperative outcomes, receipt of RARC was associated with a shorter median LOS (7 vs. 8, P o 0.001), lower rates of pLOS (45.1% vs. 54.8%, OR = 0.68, 95% CI: 0.58–0.79; P o 0.001), and similar 30-day readmission rates (10.2% vs. 10.2%, OR = 1.00, 95% CI: 0.83–1.22; P = 0.983) (Tables 3 and 4). IPTW-adjusted Kaplan-Meier curves revealed 30- and 90-day POM rates of 2.8%, 6.7% for ORC and 1.4%, 4.8% for RARC, respectively (hazard ratio [HR] = 0.48, 95% CI: 0.29–0.80; P = 0.005 and HR = 0.71, 95% CI: 0.54–0.93, P = 0.014) favoring RARC. With a mean follow-up of 26.9 months, propensity score-adjusted OS rates at 2 years were 70.2% and 62.5% for patients treated with RARC and ORC, respectively (P o 0.001) (Fig. 3). On IPTW-adjusted Cox regression analysis, RARC vs. ORC was associated with a benefit in OS (HR ¼ 0.79, 95% CI: 0.71–0.88; P o 0.001). 4. Discussion The adoption of RARC has not followed the same trajectory as robot-assisted radical prostatectomy, as the
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
5
Table 1 Baseline characteristics of 9,561 patients with bladder cancer who underwent radical cystectomy according to surgical approach in the National Cancer Data Base between years 2010 and 2013 Overall RARC ORC P valuea Standardized mean Standardized mean (n ¼ 9,561, 100%) (n ¼ 2,048, 21.4%) (n ¼ 7,513, 78.6%) difference before difference after IPTW (%)c IPTW (%)c Age at diagnosis, median (IQR) Age o50 50–59 60–69 70–79 ≥80 Sex Male Female Race White non-Hispanic Black non-Hispanic Other CCI 0 1 2 Insurance status Private Government No insurance Unknown Facility location New England Atlantic East Central West Central West Unknown Facility type Academic Community Comprehensive community Integrated cancer center Unknown Patient location Metro county Urban county Rural county Unknown % Population without high school degree 421 13–20.9 7–12.9 o7 Unknown Median income quartiles o38,000$ 38–47,999$ 48–62,999$ 463,000$ Unknown Distance to facility ≤50 450 Unknown
70 (62–77) 364 1,413 2,898 3,424 1,462
(3.8) (14.8) (30.3) (35.8) (15.3)
69 (62-76) 62 319 666 695 306
(3.0) (15.6) (32.5) (33.9) (14.9)
70 (62–77) 302 1,094 2,232 2,729 1,156
1,614 (78.8) 434 (21.2)
5,568 (74.1) 1,945 (25.9)
8,432 (88.2) 581 (6.1) 548 (5.7)
1,793 (87.6) 113 (5.5) 142 (6.9)
6,639 (88.4) 468 (6.2) 406 (5.4)
6,349 (66.4) 2,442 (25.5) 770 (8.1)
1,371 (66.9) 534 (26.1) 143 (7.0)
4,978 (66.3) 1,908 (25.4) 627 (8.4)
2,703 6,495 265 98
(28.3) (67.9) (2.8) (1.0)
611 1,370 49 18
(29.8) (66.9) (2.4) (0.9)
2,092 5,125 216 80
(27.9) (68.2) (2.9) (1.1)
497 3,362 2,563 1,623 1,462 54
(5.2) (35.2) (26.8) (17.0) (15.3) (0.6)
151 713 566 237 374 7
(7.4) (34.8) (27.6) (11.6) (18.3) (0.3)
346 2,694 1,997 1,386 1,088 47
(4.6) (35.3) (26.6) (18.5) (14.5) (0.6)
5,391 426 3,148 493 103
(56.4) (4.5) (32.9) (5.2) (1.1)
1,327 48 528 93 52
(64.8) (2.3) (25.8) (4.5) (2.5)
4,064 378 2,620 400 51
(54.1) (5.0) (34.9) (5.3) (0.7)
7,422 1,622 236 281
(77.6) (17.0) (2.5) (2.9)
1,654 298 41 55
(80.8) (14.6) (2.0) (2.7)
5,768 1,324 195 226
(76.8) (17.6) (2.6) (3.0)
1,323 2,472 3,265 2,428 73
(13.8) (25.9) (34.2) (25.4) (0.8)
268 454 707 604 15
(13.1) (22.2) (34.5) (29.5) (0.7)
1,055 2,018 2,558 1,824 58
(14.0) (26.9) (34.1) (24.3) (0.8)
1,484 2,451 2,603 2,945 78
(15.5) (25.6) (27.2) (30.8) (0.8)
285 467 538 743 15
(13.9) (22.8) (26.3) (36.3) (0.7)
1,199 1,984 2,065 2,202 63
(16.0) (26.4) (27.5) (29.3) (0.8)
1,551 (75.7) 482 (23.5) 15 (0.7)
−0.7 0.1
o0.001 −11.1
0.3
(4.0) (14.6) (29.7) (36.3) (15.4)
7,182 (75.1) 2,379 (24.9)
7,089 (74.1) 2,401 (25.1) 71 (0.7)
−1.5 −6.2
0.207b 0.016
5,538 (73.7) 1,919 (25.5) 56 (0.8)
0.018
4.5
−0.7
0.128
−3.2
−0.2
0.207
−5.2
−0.3
o0.001
−4.4
2.2
o0.001 −14.3
4.8
0.002
−8.0
0.4
o0.001
11.6
o0.1
o0.001
13.4
−0.3
0.176
−4.5
o0.1
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
6 Table 1 Continued
Overall RARC ORC P valuea Standardized mean Standardized mean (n ¼ 9,561, 100%) (n ¼ 2,048, 21.4%) (n ¼ 7,513, 78.6%) difference before difference after IPTW (%)c IPTW (%)c Hospital annual volume (cases/y) Median (IQR) Grade Well/moderately differentiated Poorly/undifferentiated Unknown AJCC clinical T stage ≤cT1 cT2 cT3 cT4 cTx AJCC pathologic T stage pT0 ≤pT1 pT2 pT3 pT4 pTx AJCC pathologic nodal status pN0 pN1 pNx Adjuvant chemotherapy
8 (4-20)
9 (5-17)
7 (3-22)
601 (6.3) 7,508 (78.5) 1,452 (15.2)
107 (5.2) 1,576 (77.0) 365 (17.8)
494 (6.6) 5,932 (79.0) 1,087 (14.5)
2,316 3,871 481 371 2,522
(24.2) (40.5) (5.0) (3.9) (26.4)
549 797 90 71 541
(26.8) (38.9) (4.4) (3.5) (26.4)
1,767 3,074 391 300 1,981
(23.5) (40.9) (5.2) (4.0) (26.4)
322 1,781 2,707 3,153 1,206 392
(3.4) (18.6) (28.3) (33.0) (12.6) (4.1)
91 455 568 616 226 92
(4.4) (22.2) (27.7) (30.1) (11.0) (4.5)
231 1,326 2,139 2,537 980 300
(3.1) (17.7) (28.5) (33.8) (13.0) (4.0)
6,928 1,818 815 1,461
(72.5) (19.0) (8.5) (15.3)
1,539 374 135 324
(75.2) (18.3) (6.6) (15.8)
5,389 1,444 680 1,137
(71.7) (19.2) (9.1) (15.1)
o0.001b
−7.8
4.7
o0.001
10.3
0.3
0.017
−2.5
0.1
o0.001
–
–
0.001
–
–
0.444
–
–
IQR ¼ interquartile range. a Chi-square test. b Kruskal-Wallis test. c Reported for variables used in propensity score calculation (preoperative variables).
most cases continue to be performed in a conventional open approach. Indeed, 3 randomized trials comparing the 2 approaches found opposing results [10–12]. These conflicting findings highlight the difficulties of interpreting surgical trials, as they may be influenced by the technique of the surgeons participating in the trial. From a broader payer and societal perspective, observational studies based on results from the community may provide a more accurate depiction of the benefits of a surgical tool, especially when there are significant cost differences between approaches [7,14,19]. Against this backdrop, we used the NCDB to compare the perioperative and oncological outcomes of RARC vs. ORC. To our knowledge, this study represents the largest comparative analysis of RARC vs. ORC controlling for patient, hospital, and tumor characteristics. Our contemporary analysis provides compelling evidence of the increasing adoption of robotic surgery for the treatment of bladder cancer in the United States and it is pressing to better define the role of RARC. In our study, patients treated within academic institutions were more likely to undergo RARC, as previously suggested [7,20]. We also found that there were only small differences in tumor characteristics between both treatment groups. Our findings suggest that in the most contemporary period,
robotic surgeons are willing to undertake equally complex cases as their open counterparts. Interestingly, many patient and hospital characteristics including race and CCI as well as insurance status and patient/hospital location were not significant predictors of RARC receipt. These results suggest that the diffusion of RARC is not associated with disparities usually seen with the adoption of novel technologies [21]. Our finding of superior perioperative outcomes within the RARC cohort is most significant. Specifically, we found that RARC was associated with a 52% and 29% lower odds of 30-day and 90-day POM, respectively. Although the randomized trial by Bochner et al. did not demonstrate a reduction in POM with the use of RARC, we do corroborate the findings of Musch et al. [22], which showed in a single center retrospective study a 3% reduction in POM after RARC vs. ORC. Our results are also in line with a report from the National Inpatient Sample, a population-based study from the United States [23]. Similarly, we found that patients undergoing RARC were 32% less likely to experience a pLOS, which is consistent with existent evidence based on institutional [24] and population-based [14] data. Although we acknowledge the limitations of our observational study design, our findings are reassuring and
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎ Facility_Type
7
Table 2 Multivariable logistic regression predicting receipt of robotic-assisted radical cystectomy adjusting for multiple patient and hospital characteristics in patient with bladder cancer underdoing radical cystectomy in the NCDB between years 2010 and 2013 (n ¼ 9,561)
Median_Income Hospital_Annual_Volume Education_Level Sex Grade
Factors
Patient_Location Insurance_Status
Odds ratio
95% CI
P value
Race Distance_to_Facility Facility_Location CCI AJCC_Clinical_T_Stage Age –.2
–.1 0 .1 Standardardised difference Before Adjustment
.2
After Adjustment
Fig. 2. The effect of inverse probability of treatment weighting adjustment on standardised differences distribution in baseline characteristics between ORC and RARC patients. (Color version of figure is available online.)
allude to potential benefits for the robotic approach, which may (or not) offset the costs of robot acquisition and disposable supplies. Historically, investigators have questioned the ability to complete extended LN dissections with the robotic technique. In our analysis of data including 70% of all cancer cases in the United States, we found that patients undergoing the robotic approach were more likely to receive a LN dissection, and when performed, with a higher median LN count than their open counterparts. These results may be attributed to the centralization of robotic procedures to highvolume or academic centers, with potentially associated differences in pathologic sampling methodology, which may ultimately affect LN count [25]; our findings suggest that an adequate LN dissection is possible with the robotic approach, as reported by Nix et al. [12] in their randomized trial. Of note, median LN counts for RARC within our cohort (17, interquartile range: 10–25) is comparable to other series from highvolume institutions (16, interquartile range: 10–24) [26]. In a similar fashion, there have been questions about how the lack of tactile feedback would impact surgical margin rates. However, in our cohort of patients who did not receive neoadjuvant chemotherapy, we found that the PSM rates were equivalent between the 2 surgical approaches, thus corroborating data from randomized trials [9,11,19]. Finally, on relatively short-term mean follow-up (26.9 mo), we found that RARC was associated with a 7.7% higher OS rate at 2 years. Few studies have reported comparison of survival between RARC and ORC. Our results are in line with a small single center nonrandomized comparison, which showed 10% higher disease-specific survival at a mean follow-up of 38 months [27]. However, a recent SEER-Medicare analysis with a longer mean follow-up (49 mo) did not demonstrate this benefit after propensity matching adjustments [14]. Caution should be
Race White non-Hispanic 1.0 (ref) Black non-Hispanic 0.94 0.73–1.22 0.649 Other 1.20 0.89–1.60 0.226 Age at diagnosis 0.98 0.91–1.05 0.505 (per 10 y difference) Sex Male 1.0 (ref) Female 0.77 0.68–0.87 o0.001 CCI 0 1.0 (ref) 1 1.05 0.92–1.19 0.466 ≥2 0.89 0.72–1.09 0.258 Insurance status Medicare/medicaid 1.0 (ref) Private 1.07 0.94–1.22 0.284 No insurance 0.83 0.58–1.18 0.292 Unknown 0.78 0.47–1.27 0.313 Facility location Atlantic 1.0 (ref) New England 1.51 0.68–3.34 0.314 East Central 1.10 0.70–1.73 0.677 West Central 0.69 0.44–1.08 0.105 West 1.35 0.83–2.20 0.230 Unknown 0.01 0.01–0.04 o0.001 Facility type Academic 1.0 (ref) Community 0.28 0.16–0.51 o0.001 Comprehensive community 0.47 0.32–0.69 o0.001 Integrated network program 0.52 0.27–0.99 0.046 Unknown 28.9 20.38–41.07 o0.001 Patient location Metro county 1.0 (ref) Urban county 0.95 0.77–1.17 0.629 Rural county 0.97 0.59–1.59 0.911 Unknown 0.82 0.55–1.23 0.343 Median income quartiles o38,000$ 1.0 (ref) 38–47,999$ 1.01 0.82–1.24 0.922 48–62,999$ 0.98 0.79–1.23 0.885 463,000$ 1.08 0.78–1.50 0.655 Unknown 3.47e-07 1.44e-07–8.40e-07 o0.001 Percentage population without high school degree in patient’s area of residence 421% 1.0 (ref) 13%–20.9% 0.91 0.74–1.14 0.419 7%–12.9% 1.11 0.84–1.47 0.476 o7% 1.22 0.88–1.67 0.231 Unknown 1.59 0.53–4.75 0.405 Distance to facility ≤50 1.0 (ref) 450 0.97 0.75–1.26 0.836 Unknown 2,571,467 6,49,236.3– o0.001 1.02e07 Hospital annual volume 0.98 0.97–1.00 0.032 (cases/y)
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Table 2 Continued Odds ratio
Grade Well/moderately differentiated Poorly/undifferentiated Unknown AJCC clinical T stage T1 T2 T3 T4 Unknown
95% CI
P value
IPTW-adjusted OR RARC vs. ORC
1.0 (ref) 0.116 0.020
1.0 (ref) 0.87 0.77 0.80 0.92
0.77–0.98 0.58–1.01 0.61–1.05 0.79–1.08
0.027 0.058 0.111 0.321
taken in interpreting oncologic outcomes with shorter follow-up and potential bias of patient selection. Our study is not without limitations. First, the analyses are retrospective in nature; which comes with an unavoidable selection bias, inherent to all nonprospective, nonrandomized studies. We attempted to address such bias by performing propensity score based analyses that may approximate randomization of observed confounding. However, we recognize the limitations of propensity adjustment. Second, this represents an analysis of only commission on cancer-accredited hospitals and may not be generalizable to the US population at large. Third, the use of any large administrative dataset is prone to coding errors and missing observations. The latter were excluded and this may also have introduced a selection bias. Fourth, the NCDB does Table 3 Postoperative outcomes in 9,561 patients within the NCDB who underwent radical cystectomy between 2010 and 2013 according to surgical approach after propensity score weighting
10.7 89.3
96.4 3.6
92.0 8.0
17 (10–25) 43.2 56.8
12 (7–20) 59.6 40.4
10.2 89.8
10.2 89.8
o0.001 o0.001
0.983
5. Conclusion Our contemporary study shows the increased adoption of RARC between 2010 and 2013 in the United States, with
8 (6–11) 45.2 54.8
IQR ¼ interquartile range. a All P values based on Pearson test with weighted data.
Kaplan Meier survival estimates
0
o0.001 7 (6–10) 55.0 45.0
0.71–1.03 0.100 1.67–3.16 o0.001 1.55–2.42 o0.001 0.83–1.22 0.983 0.58–0.79 o0.001
not contain information on important patient characteristics such as body mass index, smoking status, American Society of Anesthesiologists class, and diversion type, all of which could significantly impact our measured outcomes. Fifth, it was not possible to determine if extracorporeal vs. intracorporeal diversion was performed in patients undergoing RARC. Additionally, most of the patients did not have information on diversion type; hence we could not adjust for this variable. Furthermore, the database does not provide information on operative time, blood loss, postoperative complications, analgesic use, return of bowel movements, use of epidural anesthesia, or participation in any perioperative recovery pathways such as enhanced recovery After surgery protocols [28]. Moreover, no information is provided on the extent of lymphadenectomy and there was no central pathology review. Finally, cause of POM could not be determined.
0.100 9.3 90.7
0.86 2.30 1.94 1.00 0.68
Based on logistic regression adjusted for IPTW using propensity score.
1.00
Margins Positive 10.0 Negative 90.0 Lymph node dissection Yes 94.2 No 5.8 LN count Median (IQR) 14 (8–23) ≤14 51.2 414 48.8 Readmission within 30 d Yes 10.2 No 89.8 LOS Median (IQR) 7 (6–10) ≤7 d 50.1 47 d 49.9
Open radical P valuea Roboticassisted radical cystectomy (%) cystectomy (%)
a
overall survival (%)
Overall (%)
Positive margins Lymph node dissection performed LN count (414 nodes) Readmission within 30 d LOS (48 d)
0.75
0.95–1.53 1.06–1.99
P valuea
0.50
1.21 1.45
95% CI
0.25
Factors
Table 4 Propensity score-adjusted logistic regressions predicting postoperative outcome according to surgical approach in 9,561 patients with bladder cancer who underwent radical cystectomy in the NCDB between 2010 and 2013
0.00
8
10
20
30
40
50
60
months after diagnosis RARC
ORC
Fig. 3. Propensity adjusted Kaplan-Meier curves depicting overall survival following radical cystectomy between ORC (red) and RARC (blue). (Color version of figure is available online.)
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
currently more than 1 out of 4 patients undergoing RARC. Compared to ORC, RARC was associated with higher LN counts, shorter LOS, lower POM, and higher OS. Our results allude to potential benefits for RARC over ORC while we wait for more definitive answers from randomized trials. References [1] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30, http://dx.doi.org/10.3322/caac.21332. [2] Witjes JA, Comperat E, Cowan NC, de Santis M, Gakis G, Lebret T, et al. EAU guidelines on muscle-invasive and metastatic bladder cancer: summary of the 2013 guidelines. Eur Urol 2014;65:778–92, http://dx.doi.org/10.1016/j.eururo.2013.11.046. [3] Konety BR, Allareddy V, Herr H. Complications after radical cystectomy: analysis of population-based data. Urology 2006;68:58–64, http://dx.doi.org/10.1016/j.urology.2006.01.051. [4] Hautmann RE, de Petriconi RC, Volkmer BG. Lessons learned from 1,000 neobladders: the 90-day complication rate [quiz 1235]. J Urol 2010;184:990–4, http://dx.doi.org/10.1016/j.juro.2010.05.037. [5] Lowrance WT, Rumohr JA, Chang SS, Clark PE, Smith JAJ, Cookson MS. Contemporary open radical cystectomy: analysis of perioperative outcomes. J Urol 2008;179:1313–8. [discussion 1318]. http://dx.doi.org/10.1016/j.juro.2007.11.084. [6] Menon M, Hemal AK, Tewari A, Shrivastava A, Shoma AM, El-Tabey NA, et al. Nerve-sparing robot-assisted radical cystoprostatectomy and urinary diversion. BJU Int 2003;92:232–6. [7] Leow JJ, Reese SW, Jiang W, Lipsitz SR, Bellmunt J, Trinh Q-D, et al. Propensity-matched comparison of morbidity and costs of open and robot-assisted radical cystectomies: a contemporary populationbased analysis in the United States. Eur Urol 2014;66:569–76, http://dx.doi.org/10.1016/j.eururo.2014.01.029. [8] Leow JJ, Chang SL, Meyer CP, Wang Y, Hanske J, Sammon JD, et al. Robot-assisted versus open radical prostatectomy: a contemporary analysis of an all-payer discharge database. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo.2016.01.044. [9] Khan MS, Gan C, Ahmed K, Ismail AF, Watkins J, Summers JA, et al. A single-centre early phase randomised controlled three-arm trial of open, robotic, and laparoscopic radical cystectomy (CORAL). Eur Urol 2015. http://dx.doi.org/10.1016/j.eururo.2015.07.038. [10] Bochner BH, Sjoberg DD, Laudone VP. A randomized trial of robot-assisted laparoscopic radical cystectomy. N Engl J Med 2014; 371:389–90, http://dx.doi.org/10.1056/NEJMc1405213. [11] Parekh DJ, Messer J, Fitzgerald J, Ercole B, Svatek R. Perioperative outcomes and oncologic efficacy from a pilot prospective randomized clinical trial of open versus robotic assisted radical cystectomy. J Urol 2013;189:474–9, http://dx.doi.org/10.1016/j.juro.2012.09.077. [12] Nix J, Smith A, Kurpad R, Nielsen ME, Wallen EM, Pruthi RS. Prospective randomized controlled trial of robotic versus open radical cystectomy for bladder cancer: perioperative and pathologic results. Eur Urol 2010;57:196–201, http://dx.doi.org/10.1016/j.eururo.2009.10.024. [13] Messer JC, Punnen S, Fitzgerald J, Svatek R, Parekh DJ. Healthrelated quality of life from a prospective randomised clinical trial of robot-assisted laparoscopic vs open radical cystectomy. BJU Int 2014;114:896–902, http://dx.doi.org/10.1111/bju.12818. [14] Hu JC, Chughtai B, O’Malley P, Halpern JA, Mao J, Scherr DS, et al. Perioperative outcomes, health care costs, and survival after roboticassisted versus open radical cystectomy: a national comparative effectiveness study. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo. 2016.03.028.
9
[15] Winchester DP, Stewart AK, Bura C, Jones RS. The National Cancer Data Base: a clinical surveillance and quality improvement tool. J Surg Oncol 2004;85:1–3, http://dx.doi.org/10.1002/jso.10320. [16] Gray PJ, Fedewa SA, Shipley WU, Efstathiou JA, Lin CC, Zietman AL, et al. Use of potentially curative therapies for muscle-invasive bladder cancer in the United States: results from the National Cancer Data Base. Eur Urol 2013;63:823–9, http://dx.doi.org/10.1016/j. eururo.2012.11.015. [17] Roghmann F, Antczak C, McKay RR, Choueiri T, Hu JC, Kibel AS, et al. The burden of skeletal-related events in patients with prostate cancer and bone metastasis. Urol Oncol 2015;33:17.e9–e18. http://dx. doi.org/10.1016/j.urolonc.2014.09.010. [18] Natarajan S, Lipsitz SR, Fitzmaurice GM, Sinha D, Ibrahim JG, Haas J, et al. An extension of the Wilcoxon rank-sum test for complex sample survey data. J R Stat Soc Ser C Appl Stat 2012;61:653–64, http://dx.doi.org/10.1111/j.1467-9876.2011.01028.x. [19] Bochner BH, Dalbagni G, Sjoberg DD, Silberstein J, Keren Paz GE, Donat SM, et al. Comparing open radical cystectomy and robot-assisted laparoscopic radical cystectomy: a randomized clinical trial. Eur Urol 2015;67:1042–50, http://dx.doi.org/10.1016/j.eururo.2014.11.043. [20] Monn MF, Cary KC, Kaimakliotis HZ, Flack CK, Koch MO. National trends in the utilization of robotic-assisted radical cystectomy: an analysis using the nationwide inpatient sample. Urol Oncol 2014;32:785–90, http://dx.doi.org/10.1016/j.urolonc.2014.04.007. [21] Trinh Q-D, Sammon J, Sun M, Ravi P, Ghani KR, Bianchi M, et al. Perioperative outcomes of robot-assisted radical prostatectomy compared with open radical prostatectomy: results from the nationwide inpatient sample. Eur Urol 2012;61:679–85, http://dx.doi.org/ 10.1016/j.eururo.2011.12.027. [22] Musch M, Janowski M, Steves A, Roggenbuck U, Boergers A, Davoudi Y, et al. Comparison of early postoperative morbidity after robot-assisted and open radical cystectomy: results of a prospective observational study. BJU Int 2014;113:458–67, http://dx.doi.org/ 10.1111/bju.12374. [23] Yu H-Y, Hevelone ND, Lipsitz SR, Kowalczyk KJ, Nguyen PL, Choueiri TK, et al. Comparative analysis of outcomes and costs following open radical cystectomy versus robot-assisted laparoscopic radical cystectomy: results from the US nationwide inpatient sample. Eur Urol 2012;61:1239–44, http://dx.doi.org/10.1016/j.eururo.2012. 03.032. [24] Novara G, Catto JWF, Wilson T, Annerstedt M, Chan K, Murphy DG, et al. Systematic review and cumulative analysis of perioperative outcomes and complications after robot-assisted radical cystectomy. Eur Urol 2015;67:376–401, http://dx.doi.org/10.1016/j.eururo. 2014.12.007. [25] Zehnder P, Moltzahn F, Mitra AP, Cai J, Miranda G, Skinner EC, et al. Radical cystectomy with super-extended lymphadenectomy: impact of separate vs en bloc lymph node submission on analysis and outcomes. BJU Int 2016;117:253–9, http://dx.doi.org/10.1111/ bju.12956. [26] Raza SJ, Wilson T, Peabody JO, Wiklund P, Scherr DS, Al-Daghmin A, et al. Long-term oncologic outcomes following robot-assisted radical cystectomy: results from the International Robotic Cystectomy Consortium. Eur Urol 2015. http://dx.doi.org/10.1016/j.eururo.2015.04.021. [27] Khan MS, Challacombe B, Elhage O, Rimington P, Coker B, Murphy D, et al. A dual-centre, cohort comparison of open, laparoscopic and robotic-assisted radical cystectomy. Int J Clin Pract 2012;66:656–62, http://dx.doi.org/10.1111/j.1742-1241.2011.02888.x. [28] Azhar RA, Bochner B, Catto J, Goh AC, Kelly J, Patel HD, et al. Enhanced recovery after urological surgery: a contemporary systematic review of outcomes, key elements, and research needs. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo.2016.02.051.
Original article
Comparative effectiveness of robot-assisted vs. open radical cystectomy Nawar Hanna, M.D.a,1, Jeffrey J. Leow, M.B.B.S., M.P.H.a, Maxine Sun, M.P.H.a, David F. Friedlander, M.D.a, Thomas Seisen, M.D.a, Firas Abdollah, M.D.b,2, Stuart R. Lipsitz, Sc.D.a, Mani Menon, M.D.b, Adam S. Kibel, M.D.a, Joaquim Bellmunt, M.D.c, Toni K. Choueiri, M.D.c, Quoc-Dien Trinh, M.D.a,⁎ a
Division of Urological Surgery and Center for Surgery and Public Health, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA b Henry Ford Hospital, Vattikuti Institute of Urology, Center for Outcomes Research, Analytics and Evaluation, Detroit, MI c Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Brigham and Women’s Hospital, Boston, MA Received 18 March 2017; received in revised form 8 June 2017; accepted 18 September 2017
Abstract Objectives: Over the past decade, robot-assisted radical cystectomy (RARC) has gained traction as an alternative to the conventional open approach open radical cystectomy (ORC). However, the benefits of RARC over ORC remain unclear. Our objective was to conduct a comparative effectiveness analysis between RARC and ORC using data from the National Cancer Data Base. Materials and methods: Within the National Cancer Data Base, we identified patients with localized muscle-invasive bladder cancer who underwent RC between 2010 and 2013. Patients were stratified according to surgical approach: ORC vs. RARC. Intraoperative endpoints included: the presence of positive surgical margins, the performance of a pelvic lymph node dissection, and number of lymph nodes (LN) removed. Postoperative endpoints included: length of stay (LOS), 30- and 90-day postoperative mortality (POM) rates, 30-day readmission rate, and overall survival (OS). To minimize selection bias, observed differences in baseline characteristics between RARC vs. ORC patients were controlled for using weighted propensity scores. Binary endpoints and OS were assessed using propensity score-adjusted logistic and Cox regression analyses, respectively. POM was assessed using propensity score weighted Kaplan-Meier survival estimates at 30 and 90 days after RC. Results: Of 9,561 patients who underwent RC, 2,048 (21.4%) and 7,513 (78.6%) underwent RARC and ORC, respectively. The use of RARC increased over time, from 16.7% in 2010 to 25.3% in 2013. With regard to intraoperative outcomes, RARC was associated with equivalent rates of positive surgical margins (9.3% vs. 10.7%, odds ratio [OR] ¼ 0.86, 95% CI: 0.72–1.03; P ¼ 0.10), higher rates of pelvic lymph node dissection (96.4% vs. 92.0%, OR ¼ 2.30, 95% CI: 1.67–3.16; P o 0.001), higher median LN count (17 vs. 12, P o 0.001), higher rates of LN count above the median (56.8% vs. 40.4%, OR ¼ 1.94, 95% CI: 1.55–2.42, P o 0.001). With regard to postoperative outcomes, receipt of RARC was associated with a shorter median LOS (7 vs. 8, P o 0.001), and lower rates of pLOS (45.0% vs. 54.8%, OR ¼ 0.68, 95% CI: 0.58–0.79; P o 0.001). The 30- and 90-day POM rates were 2.8%, 6.7% for ORC, and 1.4%, 4.8% for RARC, respectively (hazard ratio [HR] ¼ 0.48, 95% CI: 0.29–0.80, P ¼ 0.005 and HR ¼ 0.71, 95% CI: 0.54–0.93; P ¼ 0.014). Finally, with a mean follow-up of 26.9 months, on IPTW-adjusted Cox regression analysis, RARC vs. ORC was associated with a benefit in OS (HR ¼ 0.79, 95% CI: 0.71–0.88; P o 0.001). Conclusions: Our large contemporary study found an increased adoption of RARC between 2010 and 2013, with more than 1 out of 4 patients undergoing RARC by the end of the study period. We found that RARC was associated with higher LN counts, shorter LOS, and The data used in the study are derived from a deidentified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed, or the conclusions drawn from these data by the investigator. Quoc-Dien Trinh is supported by an unrestricted educational Grant from the Vattikuti Urology Institute, a Clay Hamlin Young Investigator Award from the Prostate Cancer Foundation, and a Genentech Bio-Oncology Career Development Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology. 1
Supported by a Young Investigator Award from the Quebec Urological Association.
2
Consultant/advisor of GenomeDx Biosciences. Corresponding author: Tel.: þ1-617-525-7350; fax: þ1-617-525-6348. E-mail address: [email protected] (Q.-D. Trinh). ⁎
http://dx.doi.org/10.1016/j.urolonc.2017.09.018 1078-1439/r 2017 Elsevier Inc. All rights reserved.
2
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
lower POM. Our results allude to potential benefits of RARC while we wait for more definitive answers from randomized trials. r 2017 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Radical Cystectomy; Robotic Surgery
1. Introduction Bladder cancer is the sixth most common cancer in the United States with an estimated 76,960 new cases in 2016 [1]. Of those, 20%–30% will present with muscle-invasive disease. For these patients, radical cystectomy (RC) with extended pelvic lymph node dissection is considered the gold standard treatment [2]. However, this complex procedure is associated with high rates of perioperative morbidity and mortality, which in part may be attributable to the older age and multitude of comorbid diseases that typically characterize this patient population [3–5]. Since its original description, open RC (ORC) has remained the most frequently used approach for bladder extirpation. However, beginning in 2003, robot-assisted RC (RARC) has emerged as a minimally invasive approach to RC [6]. Over the past decade, RARC has slowly gained acceptance in the urology community, growing from 0.6% of cases in 2004 to 12.8% in 2010 [7]. This is in stark contrast to robot-assisted radical prostatectomy, which has seen a dramatic rise from 1.8% in 2003 to 85% in 2013 [8]. The rapid adoption of robotassisted radical prostatectomy over open prostatectomy has been attributed to a combination of direct-to-patient advertisement, market competition, and debatably superior outcomes. That said, the benefits of RARC over ORC remain controversial [9–12]. Bochner et al. [10] recently reported results from a prospective trial of 118 patients randomized to either ORC (n ¼ 58) or RARC (n ¼ 60) at a highvolume tertiary referral center, which found similar rates of complications, positive surgical margins (PSMs), lymph node (LN) yields, length of stay (LOS), and quality of life at 3 and 6-months postoperatively. In that study, the only benefit of RARC was lower intraoperative blood loss, but with significantly longer operative time and higher costs. When examining the comparative effectiveness of RARC vs. ORC at the population level, a large study of 36,773 patients did not find any differences in terms of postoperative major complications and mortality [7]. Additionally, comparable short-term oncological and health-related quality of life outcomes have been described for both procedures [13,14]. Regardless, the most evidence regarding the comparative effectiveness of RARC vs. ORC relies on reports from academic/high-volume centers with limited numbers of participating surgeons. As such, we use the National Cancer Data Base (NCDB) to perform a contemporary comparative effectiveness analysis of RARC vs. ORC in a large sample of patients of all ages comprising 70% of all cancer cases in the United States. Our hypothesis
was that RARC in increasingly used and associated with some perioperative benefits. 2. Materials and methods 2.1. Data source The NCDB is a joint initiative of the American College of Surgeons, Commission on Cancer, and American Cancer Society. Established in 1989, it serves as a comprehensive clinical surveillance resource for cancer care in the United States [15]. The NCDB compiles data from 41,500 commission-accredited cancer programs in the United States and Puerto Rico. 2.2. Patient population We identified all patients who underwent RC for bladder cancer between 2010 and 2013 using International Classification of Disease for Oncology, third edition site codes (C67.0–C67.9). Treatment modality was recorded using the Facility Oncology Registry Standards manual. Cystectomy patients were identified using the surgery of the primary site codes 50, 60, 70, and 80 (total cystectomy, radical cystectomy, pelvic exenteration, and cystectomy not otherwise specified). Since 2010, the NCDB has recorded surgical approach in order to monitor patterns and trends in the adoption and usage of minimally invasive surgical techniques. Accordingly, patients were stratified according to surgical approach: ORC or RARC. Patients who underwent laparoscopic (nonrobot-assisted) and RARC converted to open approach were excluded. Patients with metastatic (American Join Commission on Cancer [AJCC] cMþ) or clinical positive node (AJCC cNþ) and those who received any form of radiation therapy were excluded. Finally, to fully evaluate the impact of surgical approach on postoperative outcomes, we excluded patients who received neoadjuvant chemotherapy. Following exclusions, 9,561 patients remained for further analyses (Fig. 1). 2.3. Covariates Patient covariates included age, sex, race (White, Black, and Other), residence location (metropolitan, urban, and rural), and insurance status (private, Medicare/medicaid, and uninsured). Socioeconomic status variables were defined according to census tract median household income (adjusted according to 2012 inflation) and the percentage of
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
3
Fig. 1. Flowchart diagram depicting patient selection within the National Cancer Data Base between 2010 and 2013. (Color version of figure is available online.)
patients without a high school diploma. Moreover, distance from patient’s residence to treating facility is provided and was dichotomized as ≤50 miles or 450 miles. Comorbidity status was defined using the Charlson comorbidity index (CCI), and categorized to CCI 0, 1, ≥2. Tumor characteristics included tumor grade and tumor stage based on the AJCC Stage Manual seventh edition. Provider covariates included facility type, calculated according to the number of patients diagnosed with bladder cancer within 1 year, and composed of community center (100–500 cases), comprehensive community cancer center (4500 cases), and academic (4500 cases with available graduate medical training). Location of facility was also available and reported (New England, Atlantic, East Central, West Central, and West Pacific). Individual annual hospital volume was calculated as the annual caseload of patients within our cohort treated at a given institution in a given year. Receipt of chemotherapy was defined according to SEER*RX (http://seer.cancer.gov/tools/seerrx/) criteria, which is recorded as single or multiagent chemotherapy in the NCDB, as previously reported [16]. Patients who received multiagent chemotherapy after surgery were considered as having received adjuvant chemotherapy. 2.4. Outcomes 2.4.1. Intraoperative endpoints These endpoints included the presence of PSMs, the performance of a pelvic lymph node dissection, and number of LNs removed. Pathologic information on margin status
and number of LNs removed are provided within the NCDB as it appears in the pathology report. Margins reported as residual tumor not otherwise specified, microscopic or macroscopic residual tumor was defined as positive. LN count was reported as a continuous variable as well as stratified according to the median of 14 (414 vs. ≤14). 2.4.2. Postoperative endpoints Perioperative outcomes measured included LOS, 30-day and 90-day postoperative mortality (POM), 30-day readmission following surgery, and overall survival (OS). Prolonged length of stay (pLOS) was defined as a LOS greater than the median within our cohort (7 d). Death from any cause following bladder cancer diagnosis was abstracted. Follow-up time was recorded from the time of surgery until death or loss to follow-up. 2.5. Statistical analyses In the first part of our analyses, we compared baseline characteristics of patients undergoing RARC vs. ORC. The Kruskal-Wallis and chi-square test were used to assess any statistical difference between the 2 treatment groups for continuously and categorical variables, respectively. Second, a multivariable logistic regression model was fitted using all preoperative available covariates to identify patient, tumor, and provider characteristics associated with the receipt of RARC. The proportion of patients undergoing RARC was reported for each study year and temporal trend analysis was performed using the estimated annual percentage change mixed linear regression methodology [17].
4
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Third, to account for selection bias, observed differences in baseline characteristics between patients who received RARC vs. ORC were controlled for using a weighted propensity score analysis. After fitting a propensity score model including all available preoperative covariates (as listed in Table 1), each patient was weighted by the inverse probability of being in the RARC vs. ORC, with the goal of balancing out observable characteristics between the 2 treatment groups; this approach is known as the inverse probability of treatment weighting (IPTW). Balance between covariates in weighted groups was also assessed by using the standardized differences approach and by comparing their distribution with unadjusted data. Using weighted data, rates for each endpoint were compared using chi-square and an extension of the Wilcoxon test with complex survey data [18] for categorical and continuous endpoints, respectively. Intraoperative and postoperative categorical endpoints were assessed using IPTW-adjusted logistic regression. POM and OS analyses were limited to 7,099 patients with follow-up and vital status information. IPTW-adjusted Kaplan-Meier curves were used to depict OS rates between the 2 groups. POM was assed using 30- and 90-day estimates derived from propensity score weighted KaplanMeier curves, with hazard ratios estimated at 30 and 90 days using a time-varying Cox regression model allowing for different hazards for each time point. An IPTWadjusted Cox regression model was used to evaluate the OS of RARC vs. ORC, after adjusting for preoperative covariates included in the propensity score model. All models included random hospital effects to account for clustering of patients within hospitals. All statistical analyses were performed using Stata 13.1 with a 2-sided significance level set at P o 0.05. An institutional review board waiver was obtained before conducting this study, in accordance with institutional regulation when dealing with deidentified administrative data.
3. Results 3.1. Baseline characteristics Overall, our cohort consisted of 9,561 patients who underwent RC between 2010 and 2013. Of those, 2,048 (21.4%) and 7,513 (78.6%) underwent RARC and ORC, respectively (Table 1). In our unweighted cohort, patients treated with RARC were more often male (78.8% vs. 74.1%, P o 0.001). With regard to sociodemographic characteristics, RARC patients were more often treated at an academic facility (66.8% vs. 54.1%, P o 0.001), living in a metropolitan county (80.8% vs. 76.8%, P ¼ 0.002), held higher education (29.5% vs. 24.3% with o7% of population without a high school degree, P o 0.001), and income (36.3% vs. 29.3% with 463,000$/y, P o 0.001).
Additionally, patients treated with RARC were more often treated at a high-volume hospital (median hospital annual volume of 9 vs. 7, P o 0.001). Finally, regarding tumor characteristics, RARC patients had lower clinical primary tumor stage (26.8% vs. 23.5% with ≤cT1, P ¼ 0.017). Following IPTW adjustment, all covariates were well balanced with an absolute standardized difference of less than 10% for all variables. The effect of IPTW adjustment on baseline characteristics distribution is depicted in Fig. 2. 3.1.1. Use of RARC The use of RARC has increased over time, from 16.7% in 2010 to 25.3% in 2013 (estimated annual percentage change þ13.4, 95% CI: þ3.3–24.5, P ¼ 0.03). 3.2. Predictors of the receipt of RARC On multivariable logistic regression, male sex (P o 0.001), academic center (P ¼ 0.001 vs. community center), and lower tumor stage (P ¼ 0.029 cT1 vs. cT2) were associated with receipt of RARC (Table 2). 3.3. Endpoints With regard to intraoperative outcomes, IPTW-adjusted analyses showed that RARC was associated with equivalent rates of PSMs (9.4% vs. 10.7%, odds ratio [OR] ¼ 0.86, 95% CI: 0.72–1.04; P ¼ 0.12), higher rates of lymphadenectomy (96.4% vs. 92.0%, OR ¼ 2.30, 95% CI: 1.67– 3.16; P o 0.001), higher median LN count (17 vs. 12, P o 0.001) and higher rates of LN count above the median (56.8% vs. 40.4%, OR ¼ 1.94, 95% CI: 1.55–2.42; P o 0.001) (Tables 3 and 4). With regard to postoperative outcomes, receipt of RARC was associated with a shorter median LOS (7 vs. 8, P o 0.001), lower rates of pLOS (45.1% vs. 54.8%, OR = 0.68, 95% CI: 0.58–0.79; P o 0.001), and similar 30-day readmission rates (10.2% vs. 10.2%, OR = 1.00, 95% CI: 0.83–1.22; P = 0.983) (Tables 3 and 4). IPTW-adjusted Kaplan-Meier curves revealed 30- and 90-day POM rates of 2.8%, 6.7% for ORC and 1.4%, 4.8% for RARC, respectively (hazard ratio [HR] = 0.48, 95% CI: 0.29–0.80; P = 0.005 and HR = 0.71, 95% CI: 0.54–0.93, P = 0.014) favoring RARC. With a mean follow-up of 26.9 months, propensity score-adjusted OS rates at 2 years were 70.2% and 62.5% for patients treated with RARC and ORC, respectively (P o 0.001) (Fig. 3). On IPTW-adjusted Cox regression analysis, RARC vs. ORC was associated with a benefit in OS (HR ¼ 0.79, 95% CI: 0.71–0.88; P o 0.001). 4. Discussion The adoption of RARC has not followed the same trajectory as robot-assisted radical prostatectomy, as the
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
5
Table 1 Baseline characteristics of 9,561 patients with bladder cancer who underwent radical cystectomy according to surgical approach in the National Cancer Data Base between years 2010 and 2013 Overall RARC ORC P valuea Standardized mean Standardized mean (n ¼ 9,561, 100%) (n ¼ 2,048, 21.4%) (n ¼ 7,513, 78.6%) difference before difference after IPTW (%)c IPTW (%)c Age at diagnosis, median (IQR) Age o50 50–59 60–69 70–79 ≥80 Sex Male Female Race White non-Hispanic Black non-Hispanic Other CCI 0 1 2 Insurance status Private Government No insurance Unknown Facility location New England Atlantic East Central West Central West Unknown Facility type Academic Community Comprehensive community Integrated cancer center Unknown Patient location Metro county Urban county Rural county Unknown % Population without high school degree 421 13–20.9 7–12.9 o7 Unknown Median income quartiles o38,000$ 38–47,999$ 48–62,999$ 463,000$ Unknown Distance to facility ≤50 450 Unknown
70 (62–77) 364 1,413 2,898 3,424 1,462
(3.8) (14.8) (30.3) (35.8) (15.3)
69 (62-76) 62 319 666 695 306
(3.0) (15.6) (32.5) (33.9) (14.9)
70 (62–77) 302 1,094 2,232 2,729 1,156
1,614 (78.8) 434 (21.2)
5,568 (74.1) 1,945 (25.9)
8,432 (88.2) 581 (6.1) 548 (5.7)
1,793 (87.6) 113 (5.5) 142 (6.9)
6,639 (88.4) 468 (6.2) 406 (5.4)
6,349 (66.4) 2,442 (25.5) 770 (8.1)
1,371 (66.9) 534 (26.1) 143 (7.0)
4,978 (66.3) 1,908 (25.4) 627 (8.4)
2,703 6,495 265 98
(28.3) (67.9) (2.8) (1.0)
611 1,370 49 18
(29.8) (66.9) (2.4) (0.9)
2,092 5,125 216 80
(27.9) (68.2) (2.9) (1.1)
497 3,362 2,563 1,623 1,462 54
(5.2) (35.2) (26.8) (17.0) (15.3) (0.6)
151 713 566 237 374 7
(7.4) (34.8) (27.6) (11.6) (18.3) (0.3)
346 2,694 1,997 1,386 1,088 47
(4.6) (35.3) (26.6) (18.5) (14.5) (0.6)
5,391 426 3,148 493 103
(56.4) (4.5) (32.9) (5.2) (1.1)
1,327 48 528 93 52
(64.8) (2.3) (25.8) (4.5) (2.5)
4,064 378 2,620 400 51
(54.1) (5.0) (34.9) (5.3) (0.7)
7,422 1,622 236 281
(77.6) (17.0) (2.5) (2.9)
1,654 298 41 55
(80.8) (14.6) (2.0) (2.7)
5,768 1,324 195 226
(76.8) (17.6) (2.6) (3.0)
1,323 2,472 3,265 2,428 73
(13.8) (25.9) (34.2) (25.4) (0.8)
268 454 707 604 15
(13.1) (22.2) (34.5) (29.5) (0.7)
1,055 2,018 2,558 1,824 58
(14.0) (26.9) (34.1) (24.3) (0.8)
1,484 2,451 2,603 2,945 78
(15.5) (25.6) (27.2) (30.8) (0.8)
285 467 538 743 15
(13.9) (22.8) (26.3) (36.3) (0.7)
1,199 1,984 2,065 2,202 63
(16.0) (26.4) (27.5) (29.3) (0.8)
1,551 (75.7) 482 (23.5) 15 (0.7)
−0.7 0.1
o0.001 −11.1
0.3
(4.0) (14.6) (29.7) (36.3) (15.4)
7,182 (75.1) 2,379 (24.9)
7,089 (74.1) 2,401 (25.1) 71 (0.7)
−1.5 −6.2
0.207b 0.016
5,538 (73.7) 1,919 (25.5) 56 (0.8)
0.018
4.5
−0.7
0.128
−3.2
−0.2
0.207
−5.2
−0.3
o0.001
−4.4
2.2
o0.001 −14.3
4.8
0.002
−8.0
0.4
o0.001
11.6
o0.1
o0.001
13.4
−0.3
0.176
−4.5
o0.1
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
6 Table 1 Continued
Overall RARC ORC P valuea Standardized mean Standardized mean (n ¼ 9,561, 100%) (n ¼ 2,048, 21.4%) (n ¼ 7,513, 78.6%) difference before difference after IPTW (%)c IPTW (%)c Hospital annual volume (cases/y) Median (IQR) Grade Well/moderately differentiated Poorly/undifferentiated Unknown AJCC clinical T stage ≤cT1 cT2 cT3 cT4 cTx AJCC pathologic T stage pT0 ≤pT1 pT2 pT3 pT4 pTx AJCC pathologic nodal status pN0 pN1 pNx Adjuvant chemotherapy
8 (4-20)
9 (5-17)
7 (3-22)
601 (6.3) 7,508 (78.5) 1,452 (15.2)
107 (5.2) 1,576 (77.0) 365 (17.8)
494 (6.6) 5,932 (79.0) 1,087 (14.5)
2,316 3,871 481 371 2,522
(24.2) (40.5) (5.0) (3.9) (26.4)
549 797 90 71 541
(26.8) (38.9) (4.4) (3.5) (26.4)
1,767 3,074 391 300 1,981
(23.5) (40.9) (5.2) (4.0) (26.4)
322 1,781 2,707 3,153 1,206 392
(3.4) (18.6) (28.3) (33.0) (12.6) (4.1)
91 455 568 616 226 92
(4.4) (22.2) (27.7) (30.1) (11.0) (4.5)
231 1,326 2,139 2,537 980 300
(3.1) (17.7) (28.5) (33.8) (13.0) (4.0)
6,928 1,818 815 1,461
(72.5) (19.0) (8.5) (15.3)
1,539 374 135 324
(75.2) (18.3) (6.6) (15.8)
5,389 1,444 680 1,137
(71.7) (19.2) (9.1) (15.1)
o0.001b
−7.8
4.7
o0.001
10.3
0.3
0.017
−2.5
0.1
o0.001
–
–
0.001
–
–
0.444
–
–
IQR ¼ interquartile range. a Chi-square test. b Kruskal-Wallis test. c Reported for variables used in propensity score calculation (preoperative variables).
most cases continue to be performed in a conventional open approach. Indeed, 3 randomized trials comparing the 2 approaches found opposing results [10–12]. These conflicting findings highlight the difficulties of interpreting surgical trials, as they may be influenced by the technique of the surgeons participating in the trial. From a broader payer and societal perspective, observational studies based on results from the community may provide a more accurate depiction of the benefits of a surgical tool, especially when there are significant cost differences between approaches [7,14,19]. Against this backdrop, we used the NCDB to compare the perioperative and oncological outcomes of RARC vs. ORC. To our knowledge, this study represents the largest comparative analysis of RARC vs. ORC controlling for patient, hospital, and tumor characteristics. Our contemporary analysis provides compelling evidence of the increasing adoption of robotic surgery for the treatment of bladder cancer in the United States and it is pressing to better define the role of RARC. In our study, patients treated within academic institutions were more likely to undergo RARC, as previously suggested [7,20]. We also found that there were only small differences in tumor characteristics between both treatment groups. Our findings suggest that in the most contemporary period,
robotic surgeons are willing to undertake equally complex cases as their open counterparts. Interestingly, many patient and hospital characteristics including race and CCI as well as insurance status and patient/hospital location were not significant predictors of RARC receipt. These results suggest that the diffusion of RARC is not associated with disparities usually seen with the adoption of novel technologies [21]. Our finding of superior perioperative outcomes within the RARC cohort is most significant. Specifically, we found that RARC was associated with a 52% and 29% lower odds of 30-day and 90-day POM, respectively. Although the randomized trial by Bochner et al. did not demonstrate a reduction in POM with the use of RARC, we do corroborate the findings of Musch et al. [22], which showed in a single center retrospective study a 3% reduction in POM after RARC vs. ORC. Our results are also in line with a report from the National Inpatient Sample, a population-based study from the United States [23]. Similarly, we found that patients undergoing RARC were 32% less likely to experience a pLOS, which is consistent with existent evidence based on institutional [24] and population-based [14] data. Although we acknowledge the limitations of our observational study design, our findings are reassuring and
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎ Facility_Type
7
Table 2 Multivariable logistic regression predicting receipt of robotic-assisted radical cystectomy adjusting for multiple patient and hospital characteristics in patient with bladder cancer underdoing radical cystectomy in the NCDB between years 2010 and 2013 (n ¼ 9,561)
Median_Income Hospital_Annual_Volume Education_Level Sex Grade
Factors
Patient_Location Insurance_Status
Odds ratio
95% CI
P value
Race Distance_to_Facility Facility_Location CCI AJCC_Clinical_T_Stage Age –.2
–.1 0 .1 Standardardised difference Before Adjustment
.2
After Adjustment
Fig. 2. The effect of inverse probability of treatment weighting adjustment on standardised differences distribution in baseline characteristics between ORC and RARC patients. (Color version of figure is available online.)
allude to potential benefits for the robotic approach, which may (or not) offset the costs of robot acquisition and disposable supplies. Historically, investigators have questioned the ability to complete extended LN dissections with the robotic technique. In our analysis of data including 70% of all cancer cases in the United States, we found that patients undergoing the robotic approach were more likely to receive a LN dissection, and when performed, with a higher median LN count than their open counterparts. These results may be attributed to the centralization of robotic procedures to highvolume or academic centers, with potentially associated differences in pathologic sampling methodology, which may ultimately affect LN count [25]; our findings suggest that an adequate LN dissection is possible with the robotic approach, as reported by Nix et al. [12] in their randomized trial. Of note, median LN counts for RARC within our cohort (17, interquartile range: 10–25) is comparable to other series from highvolume institutions (16, interquartile range: 10–24) [26]. In a similar fashion, there have been questions about how the lack of tactile feedback would impact surgical margin rates. However, in our cohort of patients who did not receive neoadjuvant chemotherapy, we found that the PSM rates were equivalent between the 2 surgical approaches, thus corroborating data from randomized trials [9,11,19]. Finally, on relatively short-term mean follow-up (26.9 mo), we found that RARC was associated with a 7.7% higher OS rate at 2 years. Few studies have reported comparison of survival between RARC and ORC. Our results are in line with a small single center nonrandomized comparison, which showed 10% higher disease-specific survival at a mean follow-up of 38 months [27]. However, a recent SEER-Medicare analysis with a longer mean follow-up (49 mo) did not demonstrate this benefit after propensity matching adjustments [14]. Caution should be
Race White non-Hispanic 1.0 (ref) Black non-Hispanic 0.94 0.73–1.22 0.649 Other 1.20 0.89–1.60 0.226 Age at diagnosis 0.98 0.91–1.05 0.505 (per 10 y difference) Sex Male 1.0 (ref) Female 0.77 0.68–0.87 o0.001 CCI 0 1.0 (ref) 1 1.05 0.92–1.19 0.466 ≥2 0.89 0.72–1.09 0.258 Insurance status Medicare/medicaid 1.0 (ref) Private 1.07 0.94–1.22 0.284 No insurance 0.83 0.58–1.18 0.292 Unknown 0.78 0.47–1.27 0.313 Facility location Atlantic 1.0 (ref) New England 1.51 0.68–3.34 0.314 East Central 1.10 0.70–1.73 0.677 West Central 0.69 0.44–1.08 0.105 West 1.35 0.83–2.20 0.230 Unknown 0.01 0.01–0.04 o0.001 Facility type Academic 1.0 (ref) Community 0.28 0.16–0.51 o0.001 Comprehensive community 0.47 0.32–0.69 o0.001 Integrated network program 0.52 0.27–0.99 0.046 Unknown 28.9 20.38–41.07 o0.001 Patient location Metro county 1.0 (ref) Urban county 0.95 0.77–1.17 0.629 Rural county 0.97 0.59–1.59 0.911 Unknown 0.82 0.55–1.23 0.343 Median income quartiles o38,000$ 1.0 (ref) 38–47,999$ 1.01 0.82–1.24 0.922 48–62,999$ 0.98 0.79–1.23 0.885 463,000$ 1.08 0.78–1.50 0.655 Unknown 3.47e-07 1.44e-07–8.40e-07 o0.001 Percentage population without high school degree in patient’s area of residence 421% 1.0 (ref) 13%–20.9% 0.91 0.74–1.14 0.419 7%–12.9% 1.11 0.84–1.47 0.476 o7% 1.22 0.88–1.67 0.231 Unknown 1.59 0.53–4.75 0.405 Distance to facility ≤50 1.0 (ref) 450 0.97 0.75–1.26 0.836 Unknown 2,571,467 6,49,236.3– o0.001 1.02e07 Hospital annual volume 0.98 0.97–1.00 0.032 (cases/y)
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Table 2 Continued Odds ratio
Grade Well/moderately differentiated Poorly/undifferentiated Unknown AJCC clinical T stage T1 T2 T3 T4 Unknown
95% CI
P value
IPTW-adjusted OR RARC vs. ORC
1.0 (ref) 0.116 0.020
1.0 (ref) 0.87 0.77 0.80 0.92
0.77–0.98 0.58–1.01 0.61–1.05 0.79–1.08
0.027 0.058 0.111 0.321
taken in interpreting oncologic outcomes with shorter follow-up and potential bias of patient selection. Our study is not without limitations. First, the analyses are retrospective in nature; which comes with an unavoidable selection bias, inherent to all nonprospective, nonrandomized studies. We attempted to address such bias by performing propensity score based analyses that may approximate randomization of observed confounding. However, we recognize the limitations of propensity adjustment. Second, this represents an analysis of only commission on cancer-accredited hospitals and may not be generalizable to the US population at large. Third, the use of any large administrative dataset is prone to coding errors and missing observations. The latter were excluded and this may also have introduced a selection bias. Fourth, the NCDB does Table 3 Postoperative outcomes in 9,561 patients within the NCDB who underwent radical cystectomy between 2010 and 2013 according to surgical approach after propensity score weighting
10.7 89.3
96.4 3.6
92.0 8.0
17 (10–25) 43.2 56.8
12 (7–20) 59.6 40.4
10.2 89.8
10.2 89.8
o0.001 o0.001
0.983
5. Conclusion Our contemporary study shows the increased adoption of RARC between 2010 and 2013 in the United States, with
8 (6–11) 45.2 54.8
IQR ¼ interquartile range. a All P values based on Pearson test with weighted data.
Kaplan Meier survival estimates
0
o0.001 7 (6–10) 55.0 45.0
0.71–1.03 0.100 1.67–3.16 o0.001 1.55–2.42 o0.001 0.83–1.22 0.983 0.58–0.79 o0.001
not contain information on important patient characteristics such as body mass index, smoking status, American Society of Anesthesiologists class, and diversion type, all of which could significantly impact our measured outcomes. Fifth, it was not possible to determine if extracorporeal vs. intracorporeal diversion was performed in patients undergoing RARC. Additionally, most of the patients did not have information on diversion type; hence we could not adjust for this variable. Furthermore, the database does not provide information on operative time, blood loss, postoperative complications, analgesic use, return of bowel movements, use of epidural anesthesia, or participation in any perioperative recovery pathways such as enhanced recovery After surgery protocols [28]. Moreover, no information is provided on the extent of lymphadenectomy and there was no central pathology review. Finally, cause of POM could not be determined.
0.100 9.3 90.7
0.86 2.30 1.94 1.00 0.68
Based on logistic regression adjusted for IPTW using propensity score.
1.00
Margins Positive 10.0 Negative 90.0 Lymph node dissection Yes 94.2 No 5.8 LN count Median (IQR) 14 (8–23) ≤14 51.2 414 48.8 Readmission within 30 d Yes 10.2 No 89.8 LOS Median (IQR) 7 (6–10) ≤7 d 50.1 47 d 49.9
Open radical P valuea Roboticassisted radical cystectomy (%) cystectomy (%)
a
overall survival (%)
Overall (%)
Positive margins Lymph node dissection performed LN count (414 nodes) Readmission within 30 d LOS (48 d)
0.75
0.95–1.53 1.06–1.99
P valuea
0.50
1.21 1.45
95% CI
0.25
Factors
Table 4 Propensity score-adjusted logistic regressions predicting postoperative outcome according to surgical approach in 9,561 patients with bladder cancer who underwent radical cystectomy in the NCDB between 2010 and 2013
0.00
8
10
20
30
40
50
60
months after diagnosis RARC
ORC
Fig. 3. Propensity adjusted Kaplan-Meier curves depicting overall survival following radical cystectomy between ORC (red) and RARC (blue). (Color version of figure is available online.)
N. Hanna et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
currently more than 1 out of 4 patients undergoing RARC. Compared to ORC, RARC was associated with higher LN counts, shorter LOS, lower POM, and higher OS. Our results allude to potential benefits for RARC over ORC while we wait for more definitive answers from randomized trials. References [1] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30, http://dx.doi.org/10.3322/caac.21332. [2] Witjes JA, Comperat E, Cowan NC, de Santis M, Gakis G, Lebret T, et al. EAU guidelines on muscle-invasive and metastatic bladder cancer: summary of the 2013 guidelines. Eur Urol 2014;65:778–92, http://dx.doi.org/10.1016/j.eururo.2013.11.046. [3] Konety BR, Allareddy V, Herr H. Complications after radical cystectomy: analysis of population-based data. Urology 2006;68:58–64, http://dx.doi.org/10.1016/j.urology.2006.01.051. [4] Hautmann RE, de Petriconi RC, Volkmer BG. Lessons learned from 1,000 neobladders: the 90-day complication rate [quiz 1235]. J Urol 2010;184:990–4, http://dx.doi.org/10.1016/j.juro.2010.05.037. [5] Lowrance WT, Rumohr JA, Chang SS, Clark PE, Smith JAJ, Cookson MS. Contemporary open radical cystectomy: analysis of perioperative outcomes. J Urol 2008;179:1313–8. [discussion 1318]. http://dx.doi.org/10.1016/j.juro.2007.11.084. [6] Menon M, Hemal AK, Tewari A, Shrivastava A, Shoma AM, El-Tabey NA, et al. Nerve-sparing robot-assisted radical cystoprostatectomy and urinary diversion. BJU Int 2003;92:232–6. [7] Leow JJ, Reese SW, Jiang W, Lipsitz SR, Bellmunt J, Trinh Q-D, et al. Propensity-matched comparison of morbidity and costs of open and robot-assisted radical cystectomies: a contemporary populationbased analysis in the United States. Eur Urol 2014;66:569–76, http://dx.doi.org/10.1016/j.eururo.2014.01.029. [8] Leow JJ, Chang SL, Meyer CP, Wang Y, Hanske J, Sammon JD, et al. Robot-assisted versus open radical prostatectomy: a contemporary analysis of an all-payer discharge database. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo.2016.01.044. [9] Khan MS, Gan C, Ahmed K, Ismail AF, Watkins J, Summers JA, et al. A single-centre early phase randomised controlled three-arm trial of open, robotic, and laparoscopic radical cystectomy (CORAL). Eur Urol 2015. http://dx.doi.org/10.1016/j.eururo.2015.07.038. [10] Bochner BH, Sjoberg DD, Laudone VP. A randomized trial of robot-assisted laparoscopic radical cystectomy. N Engl J Med 2014; 371:389–90, http://dx.doi.org/10.1056/NEJMc1405213. [11] Parekh DJ, Messer J, Fitzgerald J, Ercole B, Svatek R. Perioperative outcomes and oncologic efficacy from a pilot prospective randomized clinical trial of open versus robotic assisted radical cystectomy. J Urol 2013;189:474–9, http://dx.doi.org/10.1016/j.juro.2012.09.077. [12] Nix J, Smith A, Kurpad R, Nielsen ME, Wallen EM, Pruthi RS. Prospective randomized controlled trial of robotic versus open radical cystectomy for bladder cancer: perioperative and pathologic results. Eur Urol 2010;57:196–201, http://dx.doi.org/10.1016/j.eururo.2009.10.024. [13] Messer JC, Punnen S, Fitzgerald J, Svatek R, Parekh DJ. Healthrelated quality of life from a prospective randomised clinical trial of robot-assisted laparoscopic vs open radical cystectomy. BJU Int 2014;114:896–902, http://dx.doi.org/10.1111/bju.12818. [14] Hu JC, Chughtai B, O’Malley P, Halpern JA, Mao J, Scherr DS, et al. Perioperative outcomes, health care costs, and survival after roboticassisted versus open radical cystectomy: a national comparative effectiveness study. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo. 2016.03.028.
9
[15] Winchester DP, Stewart AK, Bura C, Jones RS. The National Cancer Data Base: a clinical surveillance and quality improvement tool. J Surg Oncol 2004;85:1–3, http://dx.doi.org/10.1002/jso.10320. [16] Gray PJ, Fedewa SA, Shipley WU, Efstathiou JA, Lin CC, Zietman AL, et al. Use of potentially curative therapies for muscle-invasive bladder cancer in the United States: results from the National Cancer Data Base. Eur Urol 2013;63:823–9, http://dx.doi.org/10.1016/j. eururo.2012.11.015. [17] Roghmann F, Antczak C, McKay RR, Choueiri T, Hu JC, Kibel AS, et al. The burden of skeletal-related events in patients with prostate cancer and bone metastasis. Urol Oncol 2015;33:17.e9–e18. http://dx. doi.org/10.1016/j.urolonc.2014.09.010. [18] Natarajan S, Lipsitz SR, Fitzmaurice GM, Sinha D, Ibrahim JG, Haas J, et al. An extension of the Wilcoxon rank-sum test for complex sample survey data. J R Stat Soc Ser C Appl Stat 2012;61:653–64, http://dx.doi.org/10.1111/j.1467-9876.2011.01028.x. [19] Bochner BH, Dalbagni G, Sjoberg DD, Silberstein J, Keren Paz GE, Donat SM, et al. Comparing open radical cystectomy and robot-assisted laparoscopic radical cystectomy: a randomized clinical trial. Eur Urol 2015;67:1042–50, http://dx.doi.org/10.1016/j.eururo.2014.11.043. [20] Monn MF, Cary KC, Kaimakliotis HZ, Flack CK, Koch MO. National trends in the utilization of robotic-assisted radical cystectomy: an analysis using the nationwide inpatient sample. Urol Oncol 2014;32:785–90, http://dx.doi.org/10.1016/j.urolonc.2014.04.007. [21] Trinh Q-D, Sammon J, Sun M, Ravi P, Ghani KR, Bianchi M, et al. Perioperative outcomes of robot-assisted radical prostatectomy compared with open radical prostatectomy: results from the nationwide inpatient sample. Eur Urol 2012;61:679–85, http://dx.doi.org/ 10.1016/j.eururo.2011.12.027. [22] Musch M, Janowski M, Steves A, Roggenbuck U, Boergers A, Davoudi Y, et al. Comparison of early postoperative morbidity after robot-assisted and open radical cystectomy: results of a prospective observational study. BJU Int 2014;113:458–67, http://dx.doi.org/ 10.1111/bju.12374. [23] Yu H-Y, Hevelone ND, Lipsitz SR, Kowalczyk KJ, Nguyen PL, Choueiri TK, et al. Comparative analysis of outcomes and costs following open radical cystectomy versus robot-assisted laparoscopic radical cystectomy: results from the US nationwide inpatient sample. Eur Urol 2012;61:1239–44, http://dx.doi.org/10.1016/j.eururo.2012. 03.032. [24] Novara G, Catto JWF, Wilson T, Annerstedt M, Chan K, Murphy DG, et al. Systematic review and cumulative analysis of perioperative outcomes and complications after robot-assisted radical cystectomy. Eur Urol 2015;67:376–401, http://dx.doi.org/10.1016/j.eururo. 2014.12.007. [25] Zehnder P, Moltzahn F, Mitra AP, Cai J, Miranda G, Skinner EC, et al. Radical cystectomy with super-extended lymphadenectomy: impact of separate vs en bloc lymph node submission on analysis and outcomes. BJU Int 2016;117:253–9, http://dx.doi.org/10.1111/ bju.12956. [26] Raza SJ, Wilson T, Peabody JO, Wiklund P, Scherr DS, Al-Daghmin A, et al. Long-term oncologic outcomes following robot-assisted radical cystectomy: results from the International Robotic Cystectomy Consortium. Eur Urol 2015. http://dx.doi.org/10.1016/j.eururo.2015.04.021. [27] Khan MS, Challacombe B, Elhage O, Rimington P, Coker B, Murphy D, et al. A dual-centre, cohort comparison of open, laparoscopic and robotic-assisted radical cystectomy. Int J Clin Pract 2012;66:656–62, http://dx.doi.org/10.1111/j.1742-1241.2011.02888.x. [28] Azhar RA, Bochner B, Catto J, Goh AC, Kelly J, Patel HD, et al. Enhanced recovery after urological surgery: a contemporary systematic review of outcomes, key elements, and research needs. Eur Urol 2016. http://dx.doi.org/10.1016/j.eururo.2016.02.051.