Learning Curve for Robotic-Assisted Laparoscopic Retroperitoneal Lymph Node Dissection

Abstract

Introduction:

Robotic retroperitoneal lymph node dissection (R-RPLND) is a challenging procedure. We hypothesized that surgical times and operative complications would decrease as surgeons became more facile with R-RPLND.

Materials and Methods:

We retrospectively reviewed 121 consecutive R-RPLNDs performed at Naval Medical Center San Diego and Mayo Clinic Arizona by 4 fellowship trained robotic surgeons between 2008 and 2018. Linear regression was used to analyze independent predictors of setup time, operative time, and lymph node counts. Logistic regression was used to analyze open conversions, overall complications, and high-grade complications. Variables included as independent predictors were: sequential case number, surgeon, clinical stage, chemotherapy status, RPLND template, and body mass index. Univariate and multivariate analyses were conducted. Statistical significance was established at α = 0.05.

Results:

There was no change in setup time with case number (p = 0.317), but differences were noted between surgeons. Operative times decreased with increasing case number (p < 0.001) but were negatively affected by clinical stage III testis cancer (p = 0.029) and history of chemotherapy exposure (p = 0.050). Surgical times are predicted to decrease by 1 hour after 44 cases. Lymph node counts were dependent only on the surgeon. No factors were predictive of open conversions. Fewer overall complications occurred as experience was gained (p = 0.001), but high-grade complications could not be predicted.

Conclusions:

Consistent with the learning curves shown for other technologically advanced surgical techniques, experience appears to improve surgical times and lower complication rates for R-RPLND.

Introduction

Since the da Vinci^™ surgical robot (Intuitive Surgical, Sunnyvale, CA) was approved in 2000 by the United States Food and Drug Administration, robotic-assisted laparoscopic procedures have rapidly evolved and continue to reshape the conventional surgical approach. The benefits of this technology and its adoption within urology are well established.^1,2 Retroperitoneal lymph node dissection (RPLND) may represent another opportunity for infusion of robotic technology into patient care.

Robotic-assisted technology was first applied to RPLND in 2006.³ Since that time, use of this minimally invasive technique has largely been limited to experienced surgeons in high-volume centers; however, retrospective studies comparing the safety of this procedure with its open and laparoscopic alternative have been favorable.^3

–7 In a review article by Schwen and colleagues, primary robotic-retroperitoneal lymph node dissection (R-RPLND) for low-stage non-seminomatous germ cell tumor was reported to have decreased intraoperative blood loss and decreased length of hospital stay compared with the conventional open approach.⁸ In addition, both complication rates and short-term oncologic outcomes were comparable.

As robotic surgery expands, there has been greater consideration of the learning curve associated with transition to this new technology. The term “learning curve,” first described by Herman Ebbinghaus in 1885, uses graphical means to display how a piece of knowledge or a skill is obtained through repetition.⁹ Performance outcomes typically examined include operative time, estimated blood loss, perioperative complications, and oncologic efficacy such as surgical margin status, lymph node yield (LNY), and recurrence of disease. Overall, studies have demonstrated improvement of these measures with experience; however, there has been significant variation between procedures.

To date, there is no literature examining the learning curve of R-RPLND. We hypothesized that an improvement of surgical time and operative complications would be present as surgeons gained experience with this procedure. The focus of this question was directed at the immediate 30-day perioperative outcomes associated with the learning curve for this procedure, and not at its long-term effects and oncologic outcomes, which are debated elsewhere in the literature.

Materials and Methods

A retrospective review of consecutive R-RPLNDs performed at Naval Medical Center San Diego and Mayo Clinic Arizona by four fellowship trained robotic surgeons between 2008 and 2018 was conducted. The participating surgeons began their experience with R-RPLND after fellowship training. Although each had extensive laparoscopic skills commensurate with their fellowship training, performance of laparoscopic RPLND before adoption of the robotic technique was not common among the group. These data represent their initial experiences with this robotic-assisted laparoscopic procedure. This retrospective review was approved by the Institutional Review Board, including the appropriate data sharing agreement between Naval Medical Center San Diego and Mayo Clinic Arizona. R-RPLND was completed in a manner previously described by Santomauro and colleagues and Cheney and colleagues.^4,10 All cases were performed by using a supine, low-abdominal, transperitoneal approach. Data were collected for patient characteristics, operative details, pathology results, and postoperative course. Descriptive statistics were calculated. Setup time was defined as time between wheels-in-room time and surgical cut time. Surgical time was measured from cut time until completion of wound closure. Complications were graded according to the Clavien-Dindo classification system.¹¹

Linear regression was used to analyze independent predictors of setup time, operative time, and lymph node counts. Logistic regression was used to analyze open conversions, overall complications, and high-grade complications. Variables included as independent predictors were: sequential case number, surgeon, clinical stage, chemotherapy status, RPLND template, and body mass index (BMI). Univariate and multivariate analyses were conducted. Statistical significance was established at α = 0.05. STATA^®12 (College Station, TX) was used for the analysis.

Results

From 2008 to 2018, 121 R-RPLNDs were performed at 2 institutions, Naval Medical Center San Diego (57 cases, 47.1%) and Mayo Clinic Arizona (64 cases, 52.9%). All four surgeons were fellowship trained in programs with a strong emphasis on robotic and other minimally invasive operating techniques. The average age of the patients was 30.3 ± 10.7 years with an average BMI of 26.6 ± 4.3. With reference to clinical stage, 60 out of 121 (49.6%) cases were Stage I at the time of surgery, 30 out of 121 (24.8%) were Stage II, 18 out of 121 (14.9%) were Stage III, and 3 out of 121 (2.5%) were Stage IS. Primary malignancies that were not of testicular origin made up 8.3% (10/121) of the cohort and included three rhabdomyosarcomas, three mesotheliomas, two Leydig cell tumors, one desmoplastic round cell tumor, and one undefined mass. Ninety-nine (81.8%) of the surgeries used a bilateral template whereas the remainder were unilateral (10 right template, 12 left template). Finally, 65 (53.7%) of the cases were primary and 56 (46.3%) were post-chemotherapy. Median (interquartile range) follow-up for the cohort was 26 (3–60) months.

The mean setup time was 68 ± 27 minutes (median 61) with operative times of 341 ± 117 minutes (median 339). There were conversions to an open approach in 5.8% (7/121). Reasons for conversion included: three for brisk bleeding, two for poor exposure, one for robotic malfunction, and one for discovery of a mass encasing the duodenum that required open excision. Two of the bleeds involved the renal artery. Lymph node counts averaged 26 ± 16 (median 24). Median length of stay was 2 days. Complications were encountered in a total of 27.3% (33/121). These included 14.0% (17/121), 8.3% (10/121), 0.8% (1/121), 3.3% (4/121), and 0.8% (1/121) with Clavien I, II, IIIa, IIIb, and IVa, respectively. The IIIa complication was a lymphocele that required percutaneous aspiration. The IIIb complications were for two ureteral injuries that required repair, one nephrectomy due to injury of the right renal artery, and one fasciotomy in a patient who developed compartment syndrome. One patient with a history of prior bleomycin exposure had acute hypoxic respiratory failure (Clavien IVa).

Over the study period, room setup time did not decrease with respect to case sequence number (Table 1). In contrast, operative time decreased significantly on both univariate (p = 0.020) and multivariate analysis (p < 0.001) as surgeons gained experience (Fig. 1). Other factors were found that significantly affected operative time; specifically, prior chemotherapy exposure (p = 0.050) and malignancies of testicular origin with preoperative clinical Stage III (p = 0.029) increased operative time (Table 2). The coefficient (95% confidence interval) for case number (on multivariate analysis) was −1.369 (−1.969 to −0.770), suggesting that after 44 cases, operative times would be reduced by 1 hour. Lymph node counts did not change as surgeon experience grew (p = 0.600) (Table 3).

FIG. 1.

Robotic RPLND learning curve for operative time. A decrease in operative time was seen as more experience was gained. Surgical times are predicted to decrease by 1 hour after 44 cases. RPLND = retroperitoneal lymph node dissection. Color images are available online.

Table 1.

Room Setup Time

Variable	Univariate			Multivariate
Variable	Coefficient	95% CI	p	Coefficient	95% CI	p
Sequential case No.	−0.168	−0.306 to −0.029	0.018	−0.076	−0.226 to 0.074	0.317
Surgeon^a
Surgeon 2	−14.850	−30.088 to 0.389	0.056	−16.823	−32.608 to −1.039	0.037
Surgeon 3
Surgeon 4	−27.568	−36.944 to −18.192	<0.001	−29.35	−40.265 to −18.435	<0.001
Clinical stage^b
Stage II testis cancer	−6.474	−18.588 to 5.640	0.292	−3.379	−21.823 to 15.066	0.717
Stage III testis cancer	5.856	−8.650 to 20.361	0.425	8.746	−11.325 to 28.818	0.389
Stage IS testis cancer	−26.439	−64.201 to 11.324	0.168	−26.655	−61.009 to 7.699	0.127
Non-germ cell tumor	−8.771	−27.600 to 10.056	0.358	7.336	−10.798 to 25.470	0.424
Not available	−22.439	−75.388 to 30.511	0.403	−9.834	−59.610 to 39.942	0.696
Prior chemotherapy	−2.445	−12.368 to 7.477	0.626	7.661	−9.732 to 25.055	0.384
Template^c
Left	9.650	−12.963 to 32.263	0.400	5.697	−14.999 to 26.393	0.586
Bilateral	1.552	−16.033 to 19.137	0.861	2.673	−14.428 to 19.774	0.757
BMI (kg/m²)	0.269	−0.883 to 1.421	0.645	0.355	−0.733 to 1.442	0.519

Room setup time varied by case number and surgeon on univariate analysis; however, on multivariate analysis, only differences between surgeons were notable.

Relative to surgeon 1.

Relative to clinical stage I testis cancer.

Relative to right-sided template.

BMI = body mass index; CI = confidence interval.

Table 2.

Operative Time

Variable	Univariate			Multivariate
Variable	Coefficient	95% CI	p	Coefficient	95% CI	p
Sequential case No.	−0.708	−1.303 to −0.112	0.020	−1.369	−1.969 to −0.770	<0.001
Surgeon^a
Surgeon 2	−22.418	−90.858 to 46.021	0.518	5.494	−54.341 to 65.328	0.856
Surgeon 3	216.049	−13.505 to 445.603	0.065	133.010	−66.682 to 332.701	0.189
Surgeon 4	−41.048	−86.702 to 4.606	0.078	−54.418	−98.958 to −9.878	0.017
Clinical stage^b
Stage II testis cancer	35.155	−13.378 to 83.687	0.154	16.004	−59.064 to 91.074	0.673
Stage III testis cancer	135.350	77.680 to 193.020	<0.001	91.394	9.502 to 173.285	0.029
Stage IS testis cancer	115.517	−38.731 to 269.764	0.141	95.949	−44.249 to 236.148	0.178
Non-germ cell tumor	7.683	−69.025 to 84.391	0.843	52.226	−21.507 to 125.959	0.163
Not available	120.017	−96.356 to 336.389	0.274	140.021	−63.104 to 343.146	0.175
Prior chemotherapy	72.724	32.119 to 113.329	0.001	70.272	0.088 to 140.455	0.050
Template^c
Left	−18.650	−116.884 to 79.584	0.708	−44.291	−128.719 to 40.137	0.301
Bilateral	41.755	−34.444 to 117.953	0.280	38.28	−31.418 to 107.978	0.279
BMI (kg/m²)	2.726	−2.252 to 7.703	0.280	3.111	−1.270 to 7.492	0.162

Operative time was related to case number, clinical stage III testis cancer, and a history of prior exposure to chemotherapy, on both univariate and multivariate analysis.

Relative to surgeon 1.

Relative to clinical stage I testis cancer.

Relative to right-sided template.

Table 3.

Lymph Node Count

Variable	Univariate			Multivariate
Variable	Coefficient	95% CI	p	Coefficient	95% CI	p
Sequential case No.	0.253	−0.059 to 0.110	0.553	0.046	−0.043 to 0.135	0.305
Surgeon^a
Surgeon 2	−5.991	−14.390 to 2.407	0.160	−7.283	−16.098 to 1.531	0.104
Surgeon 3	62.875	34.792 to 90.958	<0.001	55.277	25.817 to 84.738	<0.001
Surgeon 4	−10.371	−16.014 to −4.727	<0.001	−14.382	−20.961 to −7.803	<0.001
Clinical stage^b
Stage II testis cancer	−1.431	−8.718 to 5.856	0.698	−0.364	−11.408 to 10.681	0.948
Stage III testis cancer	5.798	−2.950 to 14.546	0.192	1.009	−11.119 to 13.137	0.869
Stage IS testis cancer	7.700	−11.136 to 26.536	0.420	8.466	−9.342 to 26.275	0.348
Non-germ cell tumor	−1.967	−13.950 to 10.017	0.746	1.579	−9.934 to 13.091	0.786
Not available	−4.967	−37.069 to 27.135	0.760	−3.662	−33.582 to 26.258	0.809
Prior chemotherapy	1.765	−4.116 to 7.647	0.553	4.539	−5.755 to 14.834	0.384
Template^c
Left	−7.500	−20.989 to 5.989	0.273	−11.04	−23.493 to 1.413	0.082
Bilateral	0.316	−10.158 to 10.789	0.952	−3.735	−14.019 to 6.550	0.473
BMI (kg/m²)	0.092	−0.599 to 0.784	0.792	−0.009	−0.661 to 0.642	0.978

Lymph node counts varied only in association with surgeon on both univariate and multivariate analysis.

Relative to surgeon 1.

Relative to clinical stage I testis cancer.

Relative to right-sided template.

Conversions and complications were analyzed. No factors associated with conversion were significant on univariate analysis, so the multivariate analysis was not performed. Specifically, case sequence number (p = 0.269), surgeon (p = 0.137), clinical stage (p = 0.614), prior exposure to chemotherapy (p = 0.556), template (p = 0.776), and BMI (p = 0.358) failed to reach significance. Overall complication rates decreased with surgeon experience on univariate and multivariate analysis (p = 0.003 and p = 0.001, respectively), although the number of complications classified as high grade (Clavien-Dindo Grade III, IV, or V) did not change with experience (Table 4).

Table 4.

Risk of High-Grade Complications

Variable	Univariate			Multivariate
Variable	Odds ratio	95% CI	p	Odds ratio	95% CI	p
Sequential case No.	1.007	0.983 to 1.031	0.575	1.008	0.978 to 1.039	0.600
Surgeon^a
Surgeon 2	1.392	0.117 to 16.581	0.793	0.896	0.067 to 12.037	0.934
Surgeon 3	Empty			Empty
Surgeon 4	0.959	0.153 to 6.001	0.964	0.887	0.083 to 9.492	0.921
Clinical stage^b
Stage II testis cancer	0.483	0.052 to 4.520	0.523	0.098	0.004 to 2.175	0.142
Stage III testis cancer	0.824	0.086 to 7.872	0.866	0.187	0.009 to 3.979	0.282
Stage IS testis cancer	Empty			Empty
Non-germ cell tumor	Empty			Empty
Not available	Empty			Empty
Prior chemotherapy	1.170	0.227 to 6.041	0.851	4.864	0.272 to 87.130	0.283
Template^c
Left	1.709	0.183 to 15.990	0.639	2.200	0.173 to 27.923	0.543
Bilateral	Empty
BMI (kg/m²)	1.019	0.844 to 1.229	0.846	1.023	0.840 to 1.248	0.818

None of the evaluated factors were significantly associated with the risk of high-grade complications.

Relative to surgeon 1.

Relative to clinical stage I testis cancer.

Relative to right-sided template.

There was some variation between operating surgeons with respect to setup time (Table 1), operative time (Table 2), lymph node count (Table 3), and total complication number. This surgeon effects on setup time and lymph node count were identified on both univariate and multivariate analysis. However, there was no difference in the number of high-grade complications between surgeons (Table 4).

Discussion

The landscape of surgical technology is rapidly evolving. The goal of any innovative surgical technique should be its reproducibility and safety. R-RPLND was developed with the goal of matching the oncologic efficacy of open-retroperitoneal lymph node dissection (O-RPLND) while providing the benefits of a minimally invasive approach. Despite reports of comparable perioperative and intermediate-term oncologic outcomes to the conventional open approach, R-RPLND remains limited to a small number of surgeons and thus, a small number of patients. R-RPLND continues to be in the early stages of the diffusion of the innovation cycle (innovators to early adopters). Understanding the learning curve of R-RPLND could benefit experienced robotic surgeons in preparation for the critical surgical steps, which will ensure patient safety and oncologic efficacy.

Our study is the first evaluation of the learning curve with respect to operative time and surgical complications for R-RPLND in the primary and post-chemotherapy testis cancer setting. We found that there was a learning curve associated with operative time but not case setup time. A decrease in operative time as a surgeon becomes more familiar with a new procedure is a phenomenon that has been demonstrated multiple times throughout the recent literature of robotic urologic procedures. Prostatectomy was the first and remains the most commonly performed procedure by a robotic approach. In a meta-analysis of more than 17,000 prostatectomies, Davis and coworkers found surgical time to decrease with surgeon experience and noted that it continued to decrease after a surgeon's 100th case.¹² A similar decrease in operative time with increased experience has been found for robotic nephrectomy with learning curves as short as five cases in a study by Lavery and colleagues.¹³ Finally, in a multi-institutional study of 21 surgeons performing robotic radical cystectomy, Hayn and associates found a difference of 115 minutes of operative time (p < 0.0001) between surgeons who performed <30 robotic procedures and surgeons who performed >50.¹⁴ Our model would predict surgical times to decrease by 60 minutes by the 45th case.

Another aspect of the learning curve demonstrated in this study was the decrease of overall complication rate as experience grew. This finding is consistent throughout the research for transition from open to robotic techniques in urology. Davis and coworkers found a decrease in complication rate from 11.75% to 8.93% for 67 surgeons between their first case and their 100th case of robotic prostatectomy.¹² A study by Larcher and colleagues examined perioperative outcomes for robotic partial nephrectomy of two highly experienced surgeons. This study found that the complication rate continued to decrease without a plateau after 300 cases.¹⁵ Importantly in this study, although overall complication rate decreased continually, the rate of complications graded Clavien-Dindo Grade III, IV, or V was stable throughout the study period. This may suggest that although the surgeons are able to better refine their technique as they become more practiced in this operation, the safety of the patients is not compromised during a surgeon's early case sequence numbers for this procedure (in the hands of experienced robotic surgeons).

Historically, LNY has been an important performance measure in oncologic surgery. In primary RPLND for testis cancer, the total number of lymph nodes resected has been shown to have a higher probability of finding a positive lymph node. Total lymph node count has also been shown to be an independent predictor of recurrence in the post-chemotherapy RPLND setting, even in patients with favorable pathology (i.e., only teratoma and fibrosis). However, LNY is often believed to be subjective, dependent on center, processing, and pathologist. In addition, controversies regarding the boundaries of RPLND for testicular cancer persist today as the result of the development of the modified template and nerve-sparing technique. Recent findings demonstrate that mapping studies supporting the modified templates underestimated the incidence of retroperitoneal metastases due to lack of follow-up and liberal use of postoperative chemotherapy in these patients.^16,17 In our study, we demonstrate LNY variation between surgeons on both univariate and multivariate analysis. Despite this variation between surgeons, LNY did not change over time. Others have reported no learning curve associated with LNY in robotic-assisted laparoscopic para-aortic lymphadenectomy combined with oncologic pelvic surgery.¹⁸ This suggests that LNY is probably not impacted in the early adoption of this procedure by an experienced robotic surgeon.

There remains debate regarding the oncologic efficacy of robotic approaches for oncologic procedures when compared with the standard open technique. This topic has been brought to the forefront in the wake of data from gynecologic oncology, suggesting worsened survival for robotic-assisted hysterectomies performed for cervical cancer. Specifically, Ramirez and coworkers performed a randomized control trial comparing minimally invasive (robotic or laparoscopic) vs open hysterectomy for Stage I cervical cancer and found both a decrease in disease-free survival rate (86.0% vs 96.5%) and decreased overall survival (93.8% vs 99.0%) with the minimally invasive approaches.¹⁹ The authors point to increased manipulation and carbon dioxide insufflation as possible sources of increasing the spread of malignant cells at the time of operation. Despite this evidence, the literature does not suggest generalizability of these data to other oncologic operations. Retrospective reviews of robotic approaches for rectal cancer, lung cancer, and endometrial cancer all show similar survival and recurrence rates when compared with the traditional open or laparoscopic approach.^16,20

–23 In addition, Hussein and colleagues analyzed major oncologic endpoints at 10 years from a total of 446 patients who underwent robotic-assisted robotic cystectomy.²⁴ Recurrence-free survival, disease-specific survival, and overall survival were similar to those reported historically for open cystectomy.

There have been a few studies looking at early oncologic outcomes for R-RPLND. The three largest retrospective studies examining 2 years recurrence-free survival of patients undergoing R-RPLND show rates of 91% to 100%, and all recurrences cited in these studies occurred outside of the original operative field.^5
–7 These rates are similar to the quoted recurrence rate of 92.5% for O-RPLND from a large systematic review performed by Rassweiler and colleagues Although these data are encouraging, a multi-institutional or prospective study would be valuable in further evaluating the oncologic efficacy of R-RPLND.²⁵

Limitations of this study include its retrospective study design and small case number. Since this procedure is infrequently performed at most institutions, additional multi-institutional studies would help to corroborate our findings. In addition, this study involved four experienced surgeons with many years of practice. To further identify how these learning curves affect training requirements for residents and fellows, a study examining more junior staff would be necessary to further answer these questions. In addition, the surgeons may have shown a tendency to tackle more challenging cases with this approach as their comfort level with the procedure grew with experience, representing a different patient selection criterion for the procedure at the end of the series compared with the beginning. This limitation, however, would only strengthen the notion that there is a learning curve. This study may have also benefited from additional longitudinal follow-up, principally for oncologic outcomes. Overall, this article is intended to focus on the perioperative safety data and technical learning curve and so additional follow-up was not felt critical for this study, but is being tracked by both institutions. Further, in this study, operative time continued to decrease in a linear fashion throughout the entire study period, but a longer study may be able to better identify whether there is point at which the curve plateaus, thus concluding the operative time-learning curve.

Conclusions

Consistent with the learning curves shown for other technologically advanced surgical techniques, experience appears to improve surgical times and lower complication rates. Attention to variations in individual surgeons' techniques for R-RPLND may provide opportunities to improve setup time and LNY, improving the overall quality of care for these patients. Disease characteristics play a role in operative times, but they do not predict the likelihood of a perioperative complication. These findings suggest that this procedure can be successfully adopted, safely performed, and continually improved by experienced robotic surgeons.

Footnotes

Acknowledgments

The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of the Navy, Department of Defense, or the U.S. Government.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this project.

Abbreviations Used

References

Fantus

, Cohen

, Riedinger

, et al. Facility-level analysis of robot utilization across disciplines in the National Cancer Database. J Robot Surg, 2019; 13:293–299.

, Tam

. Current status of robot-assisted surgery. Hong Kong Med J, 2014; 20:241–250.

Davol

, Sumfest

, Rukstalis

. Robotic-assisted laparoscopic retroperitoneal lymph node dissection. Urology, 2006; 67:199.

Cheney

, Andrews

, Leibovich

, Castle

. Robot-assisted retroperitoneal lymph node dissection: Technique and initial case series of 18 patients. BJU Int, 2015; 115:114–120.

Pearce

, Golan

, Gorin

, et al. Safety and early oncologic effectiveness of primary robotic retroperitoneal lymph node dissection for nonseminomatous germ cell testicular cancer. Eur Urol, 2017; 71:476–482.

Rocco

, Stroup

, Abdul-Muhsin

, et al. Primary robotic RLPND for nonseminomatous germ cell testicular cancer: A two-center analysis of intermediate oncologic and safety outcomes. World J Urol, 2020; 38:859–867.

Stepanian

, Patel

, Porter

. Robot-assisted laparoscopic retroperitoneal lymph node dissection for testicular cancer: Evolution of the technique. Eur Urol, 2016; 70:661–667.

Schwen

, Gupta

, Pierorazio

. A review of outcomes and technique for the robotic-assisted laparoscopic retroperitoneal lymph node dissection for testicular cancer. Adv Urol, 2018; 2018:2146080.

Ebbinghaus

. Memory: A Contribution to Experimental Psychology. New York: University Microfilms, 1913.

10.

Santomauro

, Stroup

, L'Esperance

, Masterson

, Derweesh

, Auge

, Crain

, L'Esperance

. Supine robotic-assisted retroperitoneal lymph node dissection for testicular cancer. CRSLS, 2014; 1–8. DOI: 10.4293/CRSLS.2014.00326.

11.

Dindo

, Demartines

, Clavien

. Classification of surgical complications: A new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg, 2004; 240:205–213.

12.

Davis

, Kreaden

, Gabbert

, Thomas

. Learning curve assessment of robot-assisted radical prostatectomy compared with open-surgery controls from the premier perspective database. J Endourol, 2014; 28:560–566.

13.

Lavery

, Small

, Samadi

, Palese

. Transition from laparoscopic to robotic partial nephrectomy: The learning curve for an experienced laparoscopic surgeon. JSLS, 2011; 15:291–297.

14.

Hayn

, Hussain

, Mansour

, et al. The learning curve of robot-assisted radical cystectomy: Results from the International Robotic Cystectomy Consortium. Eur Urol, 2010; 58:197–202.

15.

Larcher

, Muttin

, Peyronnet

, et al. The learning curve for robot-assisted partial nephrectomy: Impact of surgical experience on perioperative outcomes. Eur Urol, 2019; 75:253–256.

16.

Carver

, Shayegan

, Eggener

, et al. Incidence of metastatic nonseminomatous germ cell tumor outside the boundaries of a modified postchemotherapy retroperitoneal lymph node dissection. J Clin Oncol, 2007; 25:4365–4369.

17.

Eggener

, Carver

, Sharp

, Motzer

, Bosl

, Sheinfeld

. Incidence of disease outside modified retroperitoneal lymph node dissection templates in clinical stage I or IIA nonseminomatous germ cell testicular cancer. J Urol, 2007; 177:937–942; discussion 42–43.

18.

Ponce

, Barahona

, Pla

, et al. Robotic transperitoneal infrarenal para-aortic lymphadenectomy with double docking: Technique, learning curve, and perioperative outcomes. J Minim Invasive Gynecol, 2016; 23:622–627.

19.

Ramirez

, Frumovitz

, Pareja

, et al. Minimally invasive versus abdominal radical hysterectomy for cervical cancer. N Engl J Med, 2018; 379:1895–1904.

20.

Spaggiari

, Sedda

, Maisonneuve

, et al. A brief report on survival after robotic lobectomy for early-stage lung cancer. J Thorac Oncol, 2019; 14:2176–2180.

21.

Kneuertz

, D'Souza

, Richardson

, Abdel-Rasoul

, Moffatt-Bruce

, Merritt

. Long-term oncologic outcomes after robotic lobectomy for early-stage non-small-cell lung cancer versus video-assisted thoracoscopic and open thoracotomy approach. Clin Lung Cancer, 2020; 21:214.e2–224.e2.

22.

Jayne

, Pigazzi

, Marshall

, et al. Efficacy and mechanism evaluation. Robotic-assisted surgery compared with laparoscopic resection surgery for rectal cancer: The ROLARR RCT. Southampton, UK: NHIR Journals Library, 2019.

23.

Jorgensen

, Mogensen

, Wu

, Korsholm

, Lund

, Jensen

. Survival after a nationwide introduction of robotic surgery in women with early-stage endometrial cancer: A population-based prospective cohort study. Eur J Cancer, 2019; 109:1–11.

24.

Hussein

, Elsayed

, Aldhaam

, et al. Comparison of long-term oncologic outcomes among historical open and minimally invasive retrospective studies. J Urol, 2019; 202:927–935.

25.

Rassweiler

, Scheitlin

, Heidenreich

, Laguna

, Janetschek

. Laparoscopic retroperitoneal lymph node dissection: Does it still have a role in the management of clinical stage I nonseminomatous testis cancer? A European perspective. Eur Urol, 2008; 54:1004–1015.