ORIGINAL ARTICLE

Defining the Learning Curve for Robotic-assistedEsophagogastrectomy

Jonathan M. Hernandez & Francesca Dimou &

Jill Weber & Khaldoun Almhanna & Sarah Hoffe &

Ravi Shridhar & Richard Karl & Kenneth Meredith

Received: 5 October 2012 /Accepted: 29 April 2013# 2013 The Society for Surgery of the Alimentary Tract

AbstractIntroduction The expansion of robotic-assisted surgery is occurring quickly, though little is generally known about the “learningcurve” for the technology with utilization for complex esophageal procedures. The purpose of this study is to define the learningcurve for robotic-assisted esophagogastrectomy with respect to operative time, conversion rates, and patient safety.Methods We have prospectively followed all patients undergoing robotic-assisted esophagogastrectomy and comparedoperations performed at our institutions by a single surgeon in successive cohorts of 10 patients. Our measures of proficiencyincluded: operative times, conversion rates, and complications. Statistical analyses were undertaken utilizing Spearmanregression analysis and Mann–Whitney U test. Significance was accepted with 95 % confidence.Results Fifty-two patients (41 male: 11 female) of mean age 66.2±8.8 years underwent robotic-assisted esophagogastrectomiesfor malignant esophageal disease. Neoadjuvant chemoradiation was administered to 30 (61%) patients. A significant reduction inoperative times (p <0.005) following completion of 20 procedures was identified (514±106 vs. 397±71.9). No conversions toopen thoracotomy were required. Complication rates were low and not significantly different between any 10-patient cohort;however, no complications occurred in the final 10-patient cohort. There were no in-hospital mortalities.Conclusions For surgeons proficient in performing minimally-invasive esophagogastrectomies, the learning curve for arobotic-assisted procedure appears to begin near proficiency after 20 cases. Operative complications and conversions wereinfrequent and unchanged across successive 10-patient cohorts.

Keywords Robotic-assisted esophagogastrectomy .

Learning curve . Esophageal resectionIntroduction

Utilization of robotics in surgery has the potential to in-crease accuracy in dissection through improved visualiza-tion and maneuverability while minimizing blood loss andpostoperative recovery time.1 For instance, robotic-assistedprostatectomy has demonstrated a decrease in blood loss,decreased postoperative pain, and shorter hospital length ofstay compared to the open technique in a randomized pro-spective trial.2 As applied to thoracic procedures, severalsmall series have demonstrated feasibility using roboticassistance in lung resections.3–5 Transthoracic esophagealresections have only been reported in small numbers usingrobotic assistance, although a recent report has documentedfeasibility using the transhiatal technique.6 Presumedprolonged operative times associated with the learning curveof transthoracic robotic-assisted esophagogastrectomy hasperhaps been prohibitive with regards to research.

The learning curve for any operation is the number oftimes a particular procedure must be undertaken in order to

J. M. Hernandez : F. DimouDepartment of Surgery, University of South Florida, Tampa,FL, USA

J. Weber :R. Karl :K. Meredith (*)Department of Gastrointestinal Oncology, H. Lee Moffitt CancerCenter and Research Institute, Tampa, FL, USAe-mail: kenneth.meredith@moffitt.org

K. AlmhannaMedical Oncology, H. Lee Moffitt Cancer Center and ResearchInstitute, Tampa, FL, USA

S. Hoffe : R. ShridharRadiation Oncology, H. Lee Moffitt Cancer Center and ResearchInstitute, Tampa, FL, USA

J Gastrointest SurgDOI 10.1007/s11605-013-2225-2

complete it repeatedly with high accuracy and precision.7

The steep portion of the learning curve correlates with rapidlearning and ends with the surgeon attaining proficien-cy. Following proficiency, the curve rises slowly where-by modest additional improvement occurs until thecurve flattens (plateau), at which time no further im-provement is detectable irrespective of additional pro-cedures performed (Fig. 1). We recently implemented arobotics program at our institution with the intent ofperforming operationally analogous resections to that ofthe open technique. We have undertaken this study toreview our initial experience, as well as to describe thelearning curve associated with robotic-assisted transtho-racic esophagogastrectomies.

Methods

From April 2010 until September 2011, all patients referredwith esophageal pathology necessitating esophagogastrectomywere operated upon with the general intent of undertakingrobotic-assisted procedures. Patients were not considered eli-gible for a robotic-assisted procedure if their preoperativeforced expiratory volume was less than one or delay in robotscheduling would have resulted in an undue delay in under-taking the resection. Operative time was defined as the timefrom incision to time of closure. All procedure-related compli-cations were recorded.We grouped procedures in cohorts of 10cases for the purposes of comparison based upon previousreports. We defined long operations by operative times, whichwere longer than the longest mean operative time for any 10-patient cohort, and short operations by operative times, whichwere shorter than the shortest mean operative time for any 10-patient cohort. Statistical analyses were undertaken utilizing

Spearman regression analysis and Mann–Whitney U test.Significance was accepted with 95 % confidence.

Surgical Technique

All surgeries were performed using the DaVinci SurgicalSystem (Intuitive Surgical, Sunnyvale, CA). All patients hadrobotic transthoracic approaches with an intra-corporal anasto-mosis. Briefly, for the robotic abdominal mobilization of thestomach, three, 8 mm robotic ports were placed in the right andleft subcostal regions (robotic port 2 (R2/A) was placedthrough a 10-mm port-in-port technique), a 5-mm assistantport was placed in the left paramedian, and a 10-mm camerawas placed in the supra-umbilical area (Fig. 2a). The liverretractor was placed through a stab incision in the subxiphoidregion. The abdomenwas insufflated with CO2 to a pressure of15 mmHg and the abdominal cavity inspected for evidence ofmetastatic disease. The gastroepiploic pedicle was identifiedand protected, and 1 cm inferior to this, omental attachmentswere taken off of the colon to gain entrance into the lesser sac.This dissection is taken over inferiorly to the origin of thegastroepiploic vessel and anteriorly along the surface of thepylorus. Posterior pancreaticogastric attachments were takendown with the Harmonic scalpel (Ethicon, San Angelo, TX).The dissection was carried along the greater curve of thestomach, up to the level of the short gastric vessels, whichwere takenwith the Harmonic scalpel to the level of left crus ofthe diaphragm and carried circumferentially over to the right.The left gastric vessel was identified and dissected down to theorigin of the celiac artery with dissection of the commonhepatic and splenic artery. The left gastric was then transectedat its origin with a vascular stapler completing the celiaclymphadenectomy. Next, the pylorus was identified andinjected in two locations with 150 units of onabotulinum toxin

Proficiency

Fig. 1 The learning curve.With the introduction andimplementation of a newtechnique, high rates ofconversion and complications(slow beginning) occur. Thesteep portion of the learningcurve correlates with rapidlearning and ends with thesurgeon attaining proficiency.After achieving proficiency,technical improvements areobserved, but on anincrementally smaller scale,until reaching a plateau.Following the plateau, nofurther improvements aredemonstrable, irrespective ofthe number of additional casesundertaken

J Gastrointest Surg

A (Botox). A gastric conduit was created with several fires ofan Echelon stapler (Ethicon, San Angelo, TX). A feedingjejunostomy tube was placed and brought out through one ofthe existing 8-mm ports.

The patient was then placed in a lateral decubitusposition and the right thorax is entered through the sixthintercostal space, where a 10-mm port was inserted for therobotic camera. A robotic port was placed in the thirdintercostal space for robotic arm number 1. A 4-cm inci-sion was made at the ninth intercostal space, which servesas the access for robotic arm number 3 and access for theassistant (Fig. 2b). CO2 insufflation is not routinely usedin the chest. The lung is retracted via fan retractor placedthrough this incision. The pleura over the azygos vein wasincised and mobilized completely with the hook cautery.The vein was transected with a vascular stapler. Theesophagus was mobilized en bloc down to the gastro-esophageal junction dissecting all visible lymph nodes inthe periesophageal area, periaortic area, inferior pulmonaryligament, and subcarinal nodal basins. The specimenwas removed and checked for margin status. Theesophagogastric anastomosis was created using a 25-mm anvil passed transorally (Orvil, Autosuture, andNorwalk, CT) and all were performed intra-corporallywith robotic assistance. The gastric conduit staple line isnot routinely oversewn. After a nasogastric tube wasbrought down proximal to the newly created anastomo-sis, the chest was filled with saline and 40 mL of airare instilled into the proximal esophagus to check foranastomotic leaks. An omental patch was securedaround the anastomosis.

Results

Fifty-two patients underwent robotic-assisted esophagogastrec-tomy at our institution during an 18-month period. Forty-one(78%) were male and 11 (22%)were female with amean age of65±11 years (Table 1). Esophageal cancer (biopsy-proven) was

the operative indication in 50 (96 %) patients: 46 (88 %) patientshad adenocarcinoma, 3 patients (6 %) had squamous cell carci-noma, 1 patient (2%) had a neuroendocrine tumor, and 2 patients(4 %) had Barrett’s metaplasia with multifocal high-gradedysplasia.

Thirty-seven (71 %) were stage II or higher, with themajority of patients (n=27; 52 %) presenting as stage IIIdisease. Thirty-five (67 %) patients received neoadjuvanttherapy and 17 (33 %) were treated with primary roboticesophagectomy. All patients underwent an R0 resection.The mean and median lymph node harvest was 19±9 and20.3 (range 8–63).

The mean operative time for all patients undergoingrobotic-assisted esophagogastrectomy was 442.4±86.8 min.A significant reduction in operative times (p <0.005) wasobserved following completion of 20 procedures (514±106vs. 397±71.9). Additionally, we identified a reduction in thenumber of long procedures (defined as an operative timelonger than the longest mean operative time for any 10-

Fig. 2 a Trocar positioning. Ccamera, A 5-mm assistant port,R1 8-mm robotic port for arm 1,R2/A 10-mm port-in-porttechnique to be used forstapling and robotic arm 2, R38-mm robotic port for arm 3,and LR stab incision for liverretractor. b R1 robotic arm 1, Ccamera, R2 robotic Arm 2, Aassistant, and E extractionincision

Table 1 Demographic data for patients undergoing robotic-assistedesophagogastrectomy

Number(N)

Percent(%)

Age 65±11 years

Gender Male 41 78 %

Female 11 22 %

Race Caucasian 45 87 %

African American 7 13 %

AJCC stage 0 2 4 %

I 13 25 %

IIa 3 6 %

IIb 7 13 %

III 27 52 %

Neoadjuvant therapy Yes 35 67 %

No 17 33 %

J Gastrointest Surg

patient cohort) with completion of successive 10-patientcohorts (Table 2). Similarly, we identified an increase inthe number of short procedures (defined as operativetime shorter than the shortest mean operative time forany 10-patient cohort) with completion of successive10-patient cohorts (Fig. 3).

No patients required conversions to open thoracoto-my, although one patient required an additional incisiondue to difficulty encountered during the esophagogastricanastomosis. Complication rates were low (n=14;26.9 %), and not significantly different between any10-patient cohort (Fig. 4). The most frequent complica-tions encountered by patients undergoing robotic-assistedesophagogastrectomy were: pneumonia (9.6 %), atrial fibril-lation, (9.6 %), chylous thorax (3.8 %), anastomotic leak(1.9 %), and gastric conduit staple line leak (1.9 %). Of note,no complications occurred in the most recent 10-patient co-hort. We did not observe any in-hospital mortalities.

Discussion

Robotic-assisted surgery remains in its infancy despiteburgeoning interest and an increasing number of reportedsmall case series. Although studies demonstrating efficacyof robotic surgery exist,8–15 there are limitations to wide-spread adoption of robotic assistance in thoracic surgery. Inparticular, a steep learning curve has been described with itsuse in numerous procedures despite the advantages of 3-dimensional visualization and advanced articulation thatreplicates human wrist-like movements.3–6,16–18 Wetherefore implemented a robotics program, focusing ontransthoracic esophagogastrectomy and herein report ourexperience. For surgeons experienced with minimally-invasive laparoscopic and thoracoscopic procedures, thelearning curve for robotic-assisted esophagogastrectomy be-gins near proficiency: a significant reduction in operative timewas observed following completion of 20 cases, while com-plication rates remained low and unchanged across successive10-patient cohorts.

The patients described in this series are somewhathomogenous in that the vast majority underwent theprocedure for malignant disease, which is the mostcommon indication for esophageal resections in theUnited States.19–23 Nevertheless, about half of our patientsreceived preoperative chemoradiation, which despite the po-tential added difficulty, did not appreciably complicate theesophageal dissection. We utilized loose selection criteria inoffering robotic-assisted procedures to our patients. In otherwords, our cohort does not represent a highly selected patientpopulation based upon anticipated ease. We therefore believeour results have significant external validity, and are repro-ducible by surgeons with experience in minimally-invasivethoracic techniques.

The mean operative time for the first 10-patient co-hort was greater than 8 h. Following completion of 20procedures, we observed a statistically significant de-crease in the mean operative times to less than 7 h.We attribute this decrease in operative time to bothsurgeon-specific and team-related factors. We believethat the operative team should consist of a core groupof personnel to establish routine, and streamline partic-ulars related to the robot such as set-up, docking, andinstrument exchange. Conversions fromminimally-invasivetechniques to open procedures typically occur because offailure to progress.5,21,24,25 Interestingly, no patients inour study were converted to open procedures, althoughwe may have realized a statistically significant reductionin mean operative time early had we adopted a moreliberal approach to thoracotomy.

Postoperative complications were infrequent and not sig-nificantly different across 10-patient cohorts. Pneumonia wasthe most frequent complication encountered by patients un-dergoing robotic-assisted procedures. At present, it is unclearif our rate of pneumonia is related to anesthesia time orsingle-lung ventilation time. Gastroesophageal anastomotic

Table 2 Operative times, including long and short cases, for patientsundergoing robotic-assisted esophagogastrectomy in successive 10-patient cohorts

Patientcohort

Mean operativetime (min)

Standarddeviation

Range(min)

Longcases

Shortcases

1–10 514 106 360–694 4 1

11–20 459 35 390–503 0 1

21–30 414 90 322–570 2 6

31-40 424 83 315–586 1 6

41–52 410 58 310–501 0 5

Fig. 3 Operative times for patients undergoing robotic-assistedesophagogastrectomy depicted in successive 10-patient cohorts

J Gastrointest Surg

leak is a serious complication associated with transthoracicesophagectomy.26,27 A single patient in our series experi-enced an anastomotic leak. We believe that utilization of therobot will not increase the leak rate associated withesophagectomies, over standard open or other minimally-invasive approaches.

With the advent of any new technique, it is impera-tive that operative standards remain uncompromised.With the implementation of progressive minimally-invasive technology, the identical procedure must beundertaken to that of the open technique, particularlywith respect to the treatment of malignancies. Robotic-assisted transthoracic esophagogastrectomy has not beenmeaningfully evaluated in the management of esophage-al cancers given it is a relatively nascent technology with alimited number of centers employing its use. Minimally-invasive transthoracic esophagectomy has, however, beenextensively studied, and has been demonstrated to be equiva-lent to open procedures in standard measures of oncologicresection including margin status, lymph nodes sampled,etc.6,20,21,28–30 Nevertheless, it is essential to document thetechnical execution of the oncologic aspects of the robotic-assisted esophagogastrectomy going forward. Although it wasnot the intent of the study to document oncologic feasibility,we did complete R0 resections in all patients undergoingresection with adequate lymph node harvest.

The senior author has extensive experience with minimally-invasive esophagectomy via the laparoscopic abdominal andthoracoscopic approaches. Prior to robotic approach, this wasthe author’s preferred approach. However after the initial learn-ing curve, the author found robotic to be the preferred tech-nique. The most challenging initial obstacles the authorsovercame were operating room staffing and patient set-up.

However, after several in-services, these were quickly rectified.Most improvements were seen in the chest, however after the 3-port technique was perfected, the chest phase could routinely beperformed in times comparable or less then the thoracoscopicapproach. In the abdomen, the third arm allowed for easierretraction and the surgeon control of the camera improvedvisualization.

Cost is certainly another obstacle to increased implementa-tion of robotic-assisted procedures. The “up-front” costs asso-ciated with using the robot is undeniably higher than thoseassociated with laparoscopic/thoracoscopic techniques or opensurgery;31 although any direct comparison is beyond the scopeof this report. It is noteworthy, however, that our median lengthof stay for patients undergoing robotic-assisted resections isless than patients undergoing open resections at our institution(data not shown). Nonetheless, for surgeons not currentlyand/or unwilling to employ laparoscopic/thoracoscopic tech-niques, robotic assistance may impart cost savings in terms ofdecreased ICU and hospital length of stay.

In conclusion, surgeons familiar with minimally-invasivetechniques for esophageal resection can successfully imple-ment robotic-assisted programs without adversely effectingpatient safety. We have shown that operative times can bereduced following the completion of 20 procedures, suggestingthat the learning curve for robotic-assisted esophagectomybegins near proficiency.

References

1. D’Amico TA. Robotics in thoracic surgery: Applications and out-comes. The Journal of Thoracic and Cardiovascular Surgery2006;131(1):19–20.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1-10 11-20 21-30 31-40 41-52

Strata

Nu

mb

er o

f C

om

plic

atio

ns

PneumoniaA FibPELeakAnycomp

Fig. 4 Complications forpatients undergoing robotic-assisted esophagogastrectomy,stratified by successive10-patient cohorts

J Gastrointest Surg

2. Menon M, Tewari A, Baize B, Guillonneau B, Vallancien G.Prospective comparison of radical retropubic prostatectomy androbot-assisted anatomic prostatectomy: The Vattikuti UrologyInstitute experience. Urology 2002;60(5):864–68.

3. Bodner J, Wykypiel H, Wetscher G, Schmid T. First experienceswith the da Vinci™ operating robot in thoracic surgery. EuropeanJournal of Cardio-Thoracic Surgery 2004;25(5):844–51.

4. Flores RM, Alam N. Video-Assisted Thoracic Surgery Lobectomy(VATS), Open Thoracotomy, and the Robot for Lung Cancer. TheAnnals of Thoracic Surgery 2008;85(2):S710-S15.

5. Morgan JA, Ginsburg ME, Sonett JR, Morales DLS, Kohmoto T,Gorenstein LA, et al. Advanced thoracoscopic procedures arefacilitated by computer-aided robotic technology. EuropeanJournal of Cardio-Thoracic Surgery 2003;23(6):883–87.

6. Dunn DH, Johnson EM, Morphew JA, Dilworth HP, KruegerJL, Banerji N. Robot-assisted transhiatal esophagectomy: a 3-year single-center experience. Diseases of the Esophagus2012:no-no.

7. Teplitz CJ. The Learning Curve Deskbook: A Reference Guide toTheory, Calculations, and Applications New York: Quorum Books,1991.

8. Galvani C, Gorodner M, Moser F, Jacobsen G, Chretien C, Espat N,et al. Robotically assisted laparoscopic transhiatal esophagectomy.Surgical Endoscopy 2008;22(1):188–95.

9. Giulianotti P CAAM, et al. Robotics in general surgery: Personalexperience in a large community hospital. Archives of Surgery2003;138(7):777–84.

10. Horgan S, Berger RA, Elli EF, Espat NJ. Robotic-assisted mini-mally invasive transhiatal esophagectomy. The American surgeon2003;69(7):624–26.

11. Landry CS, Grubbs EG, Stephen Morris G, Turner NS,Christopher Holsinger F, Lee JE, et al. Robot assisted transaxillarysurgery (RATS) for the removal of thyroid and parathyroid glands.Surgery 2011;149(4):549–55.

12. Puntambekar SP, Rayate N, Joshi S, Agarwal G. Robotic transthoracicesophagectomy in the prone position: Experience with 32 patients withesophageal cancer. J Thorac Cardiovasc Surg 2011;142(5):1283–84.

13. Ruurda JP, Gooszen HG, Broeders IAMJ. Early experience inrobot-assisted laparoscopic Heller myotomy. ScandinavianJournal of Gastroenterology 2004;39(241):4–8.

14. van Hillegersberg R, Boone J, Draaisma W, Broeders I, GiezemanM, Rinkes I. First experience with robot-assisted thoracoscopicesophagolymphadenectomy for esophageal cancer. SurgicalEndoscopy 2006;20(9):1435–39.

15. Weksler B, Sharma P, Moudgill N, Chojnacki KA, Rosato EL.Robot-assisted minimally invasive esophagectomy is equivalent tothoracoscopic minimally invasive esophagectomy. Diseases of theEsophagus 2012;25(5):403–09.

16. Bokhari M, Patel C, Ramos-Valadez D, Ragupathi M, Haas E.Learning curve for robotic-assisted laparoscopic colorectal surgery.Surgical Endoscopy 2011;25(3):855–60.

17. Boone J, Schipper MEI, Moojen WA, Borel Rinkes IHM,Cromheecke GJE, van Hillegersberg R. Robot-assisted thoracoscopic

oesophagectomy for cancer. British Journal of Surgery 2009;96(8):878–86.

18. Hayn MH, Hussain A, Mansour AM, Andrews PE, Carpentier P,Castle E, et al. The Learning Curve of Robot-Assisted RadicalCystectomy: Results from the International Robotic CystectomyConsortium. European Urology 2010;58(2):197–202.

19. AA Santillan JF, KL Meredith, NR Shah, ST Kelley. MinimallyInvasive Surgery for Esophageal Cancer. Journal of the NationalComprehensive Cancer Network 2008;6(9):879–84.

20. Luketich JD, Alvelo-Rivera M, Buenaventura PO, Christie NA,McCaughan JS, Litle VR, et al. Minimally Invasive Esophagectomy:Outcomes in 222 Patients. Annals of Surgery 2003;238(4):486–95.

21. Luketich JD, Pennathur A, Awais O, Levy RM, Keeley S, ShendeM,et al. Outcomes After Minimally Invasive Esophagectomy: Reviewof Over 1000 Patients. Annals of Surgery 2012;256(1):95–103.doi:10.1097/SLA.0b013e3182590603.

22. Melis M, Weber J, McLoughlin J, Siegel E, Hoffe S, Shridhar R, et al.An Elevated Body Mass Index Does Not Reduce Survival AfterEsophagectomy for Cancer. Annals of Surgical Oncology 2011;18(3):824–31.

23. Meredith K, Weber J, Turaga K, Siegel E, McLoughlin J, Hoffe S,et al. Pathologic Response after Neoadjuvant Therapy is the MajorDeterminant of Survival in Patients with Esophageal Cancer.Annals of Surgical Oncology 2010;17(4):1159–67.

24. R.J.J. Verhage EJH, J. Boone, R. Van Hillegersberg. MinimallyInvasive Surgery Compared to Open Procedures in Esophagectomyfor Cancer: a Systematic Review of the Literature. MinervaChirugica 2009;64(2):135–46.

25. Safranek PM, Cubitt J, Booth MI, Dehn TCB. Review of open andminimal access approaches to oesophagectomy for cancer. BritishJournal of Surgery 2010;97(12):1845–53.

26. Yamamoto S, Kawahara K, Maekawa T, Shiraishi T, Shirakusa T.Minimally invasive esophagectomy for stage I and II esophagealcancer. Ann Thorac Surg 2005;80(6):2070–5.

27. RC C, BC R, LA N, DA. B. Comparing outcomes after transtho-racic and transhiatal esophagectomy: a 5-year prospective cohortof 17,395 patients. J Am Coll Surg 2007;6(205):735–40.

28. Biere SSAY, van Berge Henegouwen MI, Maas KW, Bonavina L,Rosman C, Garcia JR, et al. Minimally invasive versus openoesophagectomy for patients with oesophageal cancer: amulticentre, open-label, randomised controlled trial. The Lancet2012;379(9829):1887–92.

29. Kim DJ, Hyung WJ, Lee CY, Lee J-G, Haam SJ, Park I-K, etal. Thoracoscopic esophagectomy for esophageal cancer:Feasibility and safety of robotic assistance in the prone posi-tion. The Journal of Thoracic and Cardiovascular Surgery2010;139(1):53–59.e1.

30. Nagpal K, AhmedK, Vats A, Yakoub D, James D, Ashrafian H, et al. Isminimally invasive surgery beneficial in the management of esophagealcancer? A meta-analysis. Surgical Endoscopy 2010;24(7):1621–29.

31. Gabriel I. Barbash MD, M.P.H, Sherry A. Glied PD. NewTechnology and Health Care Costs — The Case of Robot-Assisted Surgery. N Engl J Med 2010;363:701–04.

J Gastrointest Surg