- Email: [email protected]
Robotic versus laparoscopic distal pancreatectomy – The first meta-analysis
HPB
http://dx.doi.org/10.1016/j.hpb.2016.04.008
REVIEW ARTICLE
Robotic versus laparoscopic distal pancreatectomy – The first meta-analysis Paschalis Gavriilidis1, Chetana Lim1, Benjamin Menahem1, Eylon Lahat1, Chady Salloum1 & Daniel Azoulay1,2 1
Department of Hepato-Pancreato-Biliary Surgery and Liver Transplantation, Henri Mondor Hospital, and 2INSERM U 955, Créteil, France
Abstract Background: Minimally invasive pancreaticoduodenectomy is considered hazardous for the majority of authors and minimally distal pancreatectomy is still a debated topic. The aim of this study was to compare robotic distal pancreatectomy (RDP) versus laparoscopic distal pancreatectomy (LDP) using meta-analysis. Method: EMBASE, Medline and PubMed were searched systematically to identify full-text articles comparing robotic and laparoscopic distal pancreatectomies. The meta-analysis was performed by using Review Manager 5.3. Results: Nine studies fulfilled the inclusion criteria and included 637 patients (246 robotic and 391 laparoscopic). RDP had a shorter hospital length of stay by 1 day (P = 0.01). On the other hand, LDP had shorter operative time by 30 min, although this was statistically nonsignificant (P = 0.12). RDP showed a significantly increased readmission rate (P = 0.04). There was no difference in the conversion rate, incidence of postoperative pancreatic fistula, International Study Group of Pancreatic Fistula grade B–C rate, major morbidity, spleen preservation rate and perioperative mortality. All surgical specimens of RDP reported R0 negative margins, whereas 7 specimens in the LDP group had affected margins. Conclusions: In terms of feasibility, safety and oncological adequacy, there is no essential difference between the two techniques so far. The 30 min longer operative time of the RDP is due to the docking and undocking of the robot. The shorter length of stay by 1 day should be judged in combination with the increased 90-day readmission rate. Received 21 December 2015; accepted 22 April 2016
Correspondence Daniel Azoulay, Department of Hepato-Pancreato-Biliary Surgery and Liver transplantation, Henri Mondor Hospital, 51 avenue De Lattre De Tassigny, 94010 Créteil, France. Tel: +33 1 49 81 25 48. Fax: +33 1 49 81 24 32. E-mail: [email protected]
Introduction In 1994, Automated Endoscopic System of Optical Positioning (AESOP) inaugurated the era of robotic surgery. In 2000, Intuitive Surgical’s Da Vinci system was approved by the Food and Drug Administration and began the era of advanced computer-assisted telesurgery.1 In 2003, Melvin et al. performed the first robotic distal pancreatectomy (RDP).2 Shortly thereafter, Giulianotti et al. published his case series proving the feasibility of robotic pancreatectomies.3 Laparoscopic distal pancreatectomy (LDP) is the most commonly performed pancreatic resection using minimally
HPB 2016, 18, 567–574
invasive techniques due to the absence of reconstruction phase, and can be indicated for benign and malignant pancreatic lesions. Some of the technical challenges in pancreatic surgery include the vascular control and precision required for dissection of the pancreas. These are elements that may offset the laparoscopic approach whereas the robotic approach may overcome these limitations. Compared to laparoscopic approach, the potential benefits provided by the robots could be the reproduction of the mobility of the hand and fingers with seven degrees of freedom and therefore the possibility of performing the same action as in open pancreatic surgery. To the best of our knowledge, there are no randomized
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
568
controlled trials on this subject. To clarify the value of robotic for distal pancreatectomy, a systematic literature review and metaanalysis was performed of studies comparing RDP and LDP in terms of surgical and short-term outcomes. We aimed to synthesize conclusions that will serve as basis and trigger factors for the further theoretical and practical development of minimally invasive pancreatectomies.
Methods Study selection Using the search terms “robotics” or “robotic”, “laparoscopy” or “laparoscopic”, and “distal“ and “pancreatectomy”, we performed a systematic review of the literature in Medline, Embase and PubMed (including studies from the last 13 years) up to September 2015. The research was conducted independently by P.G, B.M and C.L; subsequently all the authors compared their results. References from the articles were investigated manually. Any differences were resolved by consensus. This review adhered to the guidelines outlined in the PRISMA statement.4 Data review extraction The following data were extracted: name of authors; study design; number of patients included in the robotic and laparoscopic cohort; age; sex; body mass index (BMI); American Society of Anesthesiologists (ASA) preoperative risk index; indications; operative time; estimated blood loss; number of red blood cell packed units transfused; conversion rate from the robotic and laparoscopic procedure to the open procedure; pancreatic fistula rate; International Study Group of Pancreatic Fistula (ISGPF) grade A, B, C; 90-day minor and major morbidity; 90-day perioperative mortality; length of stay; spleen preservation rate; Ro margin status; number of lymph nodes harvested; tumor size; mean follow-up; 90-day readmission rate; and cost. Inclusion criteria We included studies with more than five patients in each arm for comparison of clinical characteristics, outcomes and the cost difference between RDP and LDP. Moreover, from studies that compared robotic, laparoscopic and open distal pancreatectomies, we chose the robotic and the laparoscopic procedures. Exclusion criteria Narrative reviews, case series or studies without matched groups, studies with less than five cases, surveys and non-English language articles were excluded. Outcome definitions Apart from the study of Kang et al.,5 all the remaining reports and definitions of the pancreatic fistula rate were according to the ISGPF classification.6 Overall, minor and major complications were categorized according to Clavien–Dindo classification.7 The operative time of the robotic procedure included the
HPB 2016, 18, 567–574
docking time of the robot. Perioperative mortality is defined as death within 90 days. Quality assessment The quality of each study was assessed according to risk for bias using the QUIPS (quality in prognosis studies) tool.8 This is based on the identification of five domains of potential study bias (study participation, prognostic factor measurement, outcome measurement, confounding factor measurement, and analysis) with three to seven items per domain. Each item is given a score according to whether its quality limits potential bias: a score of 2 indicates that it does; a score of 1 indicates that it does so ‘partly’, and a score of 0 indicates that it does not. Data were extracted independently by the authors (P.G and C.L) and disagreements resolved by discussion. Statistical analysis We used Review Manager 5.3 software (Cochrane collaboration, Oxford, England) for all statistical analyses. Considering that patients were selected by different surgical teams and operated in different centers; we chose the random-effects model to assess this heterogeneity. I2 was used for heterogeneity assessment, and values of more than 50% were considered significant. Dichotomous variables were analyzed and assessed with an odds ratio (OR); a value of less than 1 favored the robotic cohort, while values of P < 0.05 and 95% confidence intervals (CIs) without the value of 1 supported the statistical significance of odds ratio (OR). Continuous variables analyzed with the weighted mean difference (WMD). The Mantel-Haenszel method was used to combine the OR for the outcomes of interest; Peto OR was used when necessary. Publication bias and sensitivity analysis were performed.
Results Literature search result and quality assessment A total of 182 studies were published in the medical literature until September 2015. After screening of records, 12 duplicates were removed. The researchers investigated the titles and abstracts of the remaining 170 articles; subsequently, 24 studies were selected for full-text investigation. After the full-text investigation process, 9 articles fulfilling the selection criteria were selected (246 robotic and 391 laparoscopic cases) (Table 15,9–16 and Fig. 1). Fifteen articles were excluded: 6 studies without a matched group, 4 narrative reviews, 2 surveys, 2 articles with less than 5 patients and 1 non-English language paper. The median QUIPS score for the included studies was 28 (range: 17–30) of a maximum score of 50 (Supplementary Table S1). Results of the meta-analysis Operative time All studies reported operative times5,9–16; whereas three studies did not give a standard deviation (SD).4,8,11 The mean operative
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
569
Table 1 Characteristics of included studies comparing robotic vs. laparoscopic distal pancreatectomy
Author, year, country (study design)
Number of patients; RDP vs. LDP
Age, mean ± SD or range; RDP vs. LDP
Female, BMI, %; RDP mean ± SD; vs. LDP RDP vs. LDP
ASA score ‡ 3, %; RDP vs. LDP
Indications, %; RDP vs. LDP
Waters JA, 2010, USA16 (retrospective)
17 vs. 18
64 vs. 59
65 vs. 50 NA
2.8 vs. 2.9
Adenocarcinoma: 0 vs 11; MCN: 18 vs. 17; NET: 29 vs. 28; IPMN: 35 vs. 11; SPPT: 0; SCM: 6 vs. 11; Other: 12 vs. 17
Kang CM, 2011, Korea5 (retrospective)
20 vs. 25
44 ± 15.9 vs. 56.5 ± 13.9
60 vs. 56 24.2 ± 2.9 vs. 23.4 ± 2.6
NA
Adenocarcinoma: 0; MCN: 25 vs. 8.5; NET: 15 vs. 12; IPMN: 10 vs. 40; SPPT: 20 vs. 16; SCM: 20 vs. 12; Other: 10 vs. 12
Daouadi M, 2013, USA11 (retrospective)
30 vs. 94
59 ± 13 vs. 59 ± 16
67 vs. 65 27.9 ± 5.1 vs. 29.0 ± 7.1
63 vs. 55
Adenocarcinoma: 43 vs. 15; MCN: 13 vs. 31; NET: 27 vs. 22; IPMN: 17 vs. 12; SPPT: 0 vs. 6.4; SCM: 0; Other: 3 vs. 13
Butturini G, 2014, Italy9 (prospective non randomized)
22 vs. 21
54 (26–77) vs. 55 (20–71)
77 vs. 71.5
4.5 vs. 0
Adenocarcinoma: 13.6 vs. 9.5; MCN: 27.2 vs. 33.3; NET: 36.3 vs. 42.8; IPMN: 0; SPPT: 13.6 vs. 4.7; SCM: 0 vs. 9.5; Other: 9 vs. 0
Duran H, 2014, Spain12 (retrospective)
16 vs. 18
61 ± 11.6 vs. 58 ± 10.03
44 vs. 50 NA
0 vs. 16
Adenocarcinoma: 56 vs. 44; MCN: 0; NET: 25 vs. 27; IPMN: 12 vs. 0; SPPT: 0; SCM: 6 vs. 11; Other: 0 vs. 18
Ryan CE, 2014, USA15 (retrospective)
18 vs. 16
67 ± 12.5 vs. 60 ± 17.0
50 vs. 38 29 ± 7.1 vs. 25 ± 4.5
NA
Adenocarcinoma: 22.2 vs. 25; MCN: 16 vs. 12; NET: 16.6 vs. 18.75; IPMN: 22.2 vs. 12.5; SPPT: 0; SCM: 11.1 vs. 18.75; Other: 11.11 vs. 12.5
Lee SY, 2014, USA14 (retrospective)
37 vs. 131
58 vs. 58
73 vs. 28.2
NA
Adenocarcinoma: 11 vs. 15; MCN: 16 vs. 12; NET: 21 vs. 31; IPMN: 11 vs. 14; SPPT: 5 vs. 5; SCM: 0; Other:35 vs. 23
Chen S, 2015, China10 (prospective nonrandomized)
69 vs. 50
56.2 ± 13.3 vs. 56.5 ± 15.0
66.7 vs. 24.6 ± 2.8 vs. 64 24.6 ± 3.0
4.3 vs. 4
Adenocarcinoma: 21.7 vs. 18; MCN: 37.7 vs. 32; NET: 4.3 vs. 6; IPMN: 8.6 vs. 10; SPPT: 14.5 vs. 16; SCM: 0; Other: 13 vs. 18
Lai EC, 2015, Hong Kong/China13 (retrospective)
17 vs. 18
61.2 ± 10.4 vs. 63.2 ± 17.9
41 vs. 78 24.1 ± 2.3 vs. 25.7 ± 2.7
64.7 vs. 77.78
Adenocarcinoma: 23 vs. 11.1; MCN: 11.7 vs. 22.2; NET: 23.5 vs. 11.1; IPMN: 5.88 vs. 0; SPPT: 0; SCM: 35.29 vs. 33.33; Other: 0 vs. 22.22
25.33 vs. 24.19
28.7 vs. 28.2
RDP, robotic distal pancreatectomy; LDP, laparoscopic distal pancreatectomy; NA, not available; SD, standard deviation; BMI, body mass index; ASA, American society of Anesthesiologists; MCN, mucinous cystic neoplasm; NET, neuroendocrine tumors; IPMN, intraductal papillary mucinous neoplasm; SPPT, solid pseudopapillary tumor; SCM, serous cystadenoma.
time of the RDP group was 30 min higher than in the LDP, although this difference was not statistically significant (median operative time = 30 min (95% CI = 8.12–68.52; P = 0.12); Fig. 2).
HPB 2016, 18, 567–574
Conversion rate All studies reported the conversion rate of each technique.5,9–16 The conversion rate was 9% (19/209) in the RDP group and 20% (70/348) in the LDP group. Meta-analysis showed no statistically
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
570
Figure 1 PRISMA flow chart showing the article selection process
significant difference between the RDP and LDP for this variable (Peto OR = 0.61; 95% CI = 0.36–1.05; P = 0.08). Incidence of pancreatic fistula and ISGPF grade B–C rates Eight of nine studies reported the incidence rate of pancreatic fistula9–16; moreover, seven of them defined and scaled it according to the ISGPF classification.9–16 The rate of pancreatic fistula was 24% (55/226) in the RDP group and 24% (88/366) in the LDP group. Meta-analysis of the 8 studies demonstrated no difference in the overall pancreatic fistula rate between the two procedures (Peto OR = 0.93; 95% CI = 0.60–1.44; P = 0.74; Fig. 3). The grade B–C pancreatic fistula rate was 13% (24/191) in the RDP cohort and 12% (41/332) in the LDP cohort. Metaanalysis of 6 studies showed no difference in the subgroup of pancreatic fistulas (ISGPF B and C) between the RDP and LDP cohorts (Peto OR = 0.98; 95% CI = 0.55–1.74; P = 0.94). Major morbidity (Dindo–Clavien classification > III) and mortality All 9 studies reported major morbidity according to the Dindo–Clavien classification 5, 9–16. The rate of major morbidity was 16% (9/246) in the RDP group and 17% (67/391) in the LDP group. Meta-analysis of 9 studies showed no
HPB 2016, 18, 567–574
significant difference in the incidence of major morbidity between RDP and LDP (Peto OR = 1.19; 95% CI 0.75–1.90; P = 0.46). All studies reported 90-day perioperative mortality.5,9–16 There was 1 death in the robotic (0.4%) and 1 death in the laparoscopic group (0.3%).11,15 Meta-analysis could not be performed for this item due to the very small rate of this event. Oncologic adequacy parameters Seven of 9 studies reported the R0 margin status and the number of harvested lymph nodes.9–12,14–16 All surgical specimens of the RDP reported with R0 negative margins. However, 7/348 (2%) of the LDP specimens were diagnosed with positive margins.10,11,15 Seven of the nine studies reported data on the number of harvested lymph nodes from both procedures.9–12,14–16 Chen et al. reported the highest mean of total lymph nodes for RDP (15.5 ± 3.1) and for LDP (15.2 ± 3.3; P = 0.819).10 Lee et al., Ryan et al., Duran et al., Butturini et al. and Daouadi et al. reported a higher number of harvested lymph nodes for the robotic compared with the laparoscopic group.9,11,12,14,15 Only Waters et al. showed a lower number of harvested lymph nodes in the robotic group.16 Meta-analysis of these outcomes could not be performed as the standard deviation (SD) could not be retrieved from the reports.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
571
Figure 2 Forest plot displaying the result of the meta-analysis comparing operative time following RDP vs. LDP
Figure 3 Forest plot displaying the result of the meta-analysis comparing incidence of pancreatic fistula following RDP vs. LDP
Figure 4 Forest plot displaying the result of the meta-analysis comparing hospital length of stay following RDP vs. LDP
Total length of hospital stay Five studies reported the length of stay.5,10,11,13,15 The mean total length of the hospital stay was 8.2 days in the RDP group and 10.9 days in the LDP group. Meta-analysis of the 5 studies showed that RDP had a statistically significant shorter hospital length of stay by 1 day (WMD = 0.97 (95% CI = 1.73–0.22; P = 0.01); Fig. 4). Only three provided the SD prices5,9,16 but meta-analysis could not be performed for this item. Readmission rate Three studies reported the 90-day readmission rate. In all of them, the robotic group showed a higher rate of readmission than its counterpart.9,11,15 Meta-analysis showed a significantly increased
HPB 2016, 18, 567–574
readmission rate, 22.85% (16/70) in RDP versus 17.55% (23/131) in LDP (Peto OR = 2.32; 95% CI = 1.03–5.21; P = 0.04; Fig. 5). Sensitivity analysis and publication bias Sensitivity analysis was performed with the aim of determining the significance of the results. For the overall incidence of pancreatic fistula, ISGPF grade B–C rate, major morbidity, conversion rate and readmission rate, the combined OR was calculated using both a fixed-effect model and a random-effects model, and the results were compared. Because of the inherent nature of non-randomized trials and the selective reporting of retrospective reviews, selective reporting and non-publication bias may influence this study.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
572
Figure 5 Forest plot displaying the result of the meta-analysis comparing readmission rate following RDP vs. LDP
Discussion Summary of evidence During the past 5 years, many authors from established pancreatic centers reported studies that supported the feasibility, reproducibility, effectiveness and cost efficacy of the robotic distal pancreatectomy compared with the laparoscopic and open procedures.5,9–16 Patient demographic characteristics and range of indications were roughly similar (Table 1). Seven of nine studies were retrospective reviews of prospectively maintained databases and two were nonrandomized studies. Selection bias is the main weakness of both types of studies. Some researchers tried to minimize this effect by selecting the control laparoscopic group from a period of time when their institution only offered the laparoscopic option.11 This also included age related selection preferences; younger patients more often choose the robotic procedure than the laparoscopic or open procedure.5,17 To the best of our knowledge, this is the first meta-analysis that compares two minimally invasive techniques of distal pancreatectomy. It included 637 patients, 246 (39%) of which underwent RDP and 391 (61%) of which underwent LDP. In sum, our metaanalysis did not detect any statistically significant differences in operative time, conversion rate, incidence of postoperative pancreatic fistula, major morbidity and spleen preservation rate. RDP showed a statistically significant reduced hospital length of stay by 1 day. Although only 3 studies reported readmission rates9,11,15; we detected a significantly higher readmission rate in the RDP group than in the LDP. All RDP surgical specimens reported with negative margins and only 0.2% of cases had positive margins in the LDP group. Seven studies reported the number of retrieved lymph nodes. In all cases the number of retrieved lymph nodes was higher in the RDP group.9–12,14–16 Meta-analysis of this parameter could not be performed due to the shortage of standard deviation data. One death was reported in the RDP (0.40%) and in the LDP (0.25%) groups. The mean difference of operative time was only 30 min longer in the RDP group. It was reported that the docking time of the robot was 30 min.18 Taking into account that the docking time was included in the total operative time, there is no essential difference between the two procedures. As with any surgical innovation, the operative time of the robotic interventions not only depended on the surgeon’s skills and their stage of the
HPB 2016, 18, 567–574
learning curve, but also on institutional learning curves as well. The conversion rate to open procedure was lower in RDP (9.09%) compared with LDP (20.11%). Meta-analysis did not find any significant difference. The most common reasons for conversion were obesity, tumor proximity to major vessels, adhesions and margins assessment.11,14 The results of the conversion rate are probably affected by tertiary centers by large numbers of cancer referral cases.11,14 Recently, Daoudi et al.11 showed that RDP significantly reduces the risk of conversion to open resection compared with LDP (0% vs. 16%; P < 0.05). However, most cases of open conversion might have occurred in their early experiences with LDP, and it is possible that the authors have been able to shorten the learning curve of RDP due to large prior experience in LDP. The incidence rate of postoperative overall and major complications including the incidence of as well as the severity of pancreatic fistula was strikingly similar. Current evidence form large meta-analysis found that independent of the technique used for transection and closure of pancreatic parenchyma after distal pancreatectomy, the incidence of postoperative pancreatic fistula ranges from 0% to 47%.19 Therefore, the results from our study are in the middle of the reported ranges. The meta-analysis showed a statistically significant reduction in the hospital stay by 1 day. However, the readmission rate was significantly higher in the RDP group than in the LDP group (22.85% vs. 17.55%). The main cause of readmission included pancreatic fistula and/or abdominal collections.9,11,15 This important finding must be underlined as most of studies comparing laparoscopic and open approaches for DP do not take into account the rate of readmissions. There is a strong possibility for underreporting bias because patients admitted in another hospital may not have been reported. It was reported that the increased readmission rate of LPD limits its perceived benefit compared with the open technique.20 Yet, Rutz et al. compared the 30-day readmissions of LDP and the open technique and reported that the former is more cost-effective than the latter.21 However, data on these readmissions are available only rarely in existing studies, which does not allow drawing any solid conclusions. Regarding oncologic adequacy, the most impressive finding was that all surgical specimens had negative surgical margins in the RDP group.9–16 In the LDP group, only 7/348 (2%) had
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
573
positive margins.10,11,15 Most of the surgeons consider that they were able to achieve adequate oncological results owing to the better ergonomic conditions offered by the robot.5,9–12,14–16 All authors except Waters et al.16 reported a higher number of retrieved lymph nodes in the RDP group.10,11,16 The rate of spleen preservation depends on technical factors but also mainly from indication of pancreatectomy (spleen preservation is usually not considered for malignancies). But this criterion cannot be used for comparison in this meta-analysis, except if this rate is compared to the rate of benign diseases. However, for some indications such as neuroendocrine tumors, it is usually not possible to extract from every study the number of benign and malignant tumors, which does not allow drawing any solid conclusions. The cost of a robotic surgical system ranges between $ 1 million to $ 2.5 million; its single use appliances and maintenance are also costly.22 It is reported that the additional variable cost of the robotic intervention compared with the laparoscopic or open procedures was $ 1600, or approximately 6%. However, including the amortized cost of the investment, the additional cost rose approximately 13%, or $ 3200.22 Data on costs are available only in 3/9 studies, which does not allow drawing any firm conclusions. Waters et al. reported that thanks to the shorter hospital stay, the cost of RDP was lower ($ 10,588) compared with the laparoscopic ($ 12,986) procedure.16 Kang et al. found RDP to be 2.5 times more expensive than LDP ($ 8304 vs. $ 3861).5 Butturini et al. also found RDP to be more expensive compared with LDP (2700–3190V vs. 1434–1674V).9 One Italian team lowered the cost by 750V for each robotic intervention by using laparoscopic ultrasonic shears instead of robotic shears and for transection of the pancreatic parenchyma; they used ultrasonic shears instead of a linear stapler.
interventions that can be used safely to treat patients with benign, borderline and, in selected cases, malignant neoplasms. The current evidence is insufficient to prove the superiority of one technique over another. Further studies are needed for detailed evaluation of oncologic adequacy, long-term survival, cost of postoperative complications, length of stay and patterns of readmission. Funding or grants None.
Conflict of interest None declared.
References 1. Heemskerk J, Bouvy ND, Baeten CG. (2013) The end of robot-assisted laparoscopy? A critical appraisal of scientific evidence on the use of robot-assisted laparoscopic surgery. Surg Endosc 28:1388–1398. 2. Melvin WS, Needleman BJ, Krause KR, Ellison EC. (2003) Robotic resection of pancreatic neuroendocrine tumor. J Laparoendosc Adv Surg Tech A 13:33–36. 3. Giulianotti PC, Coratti A, Angelini M, Sbrana F, Cecconi S, Balestracci T et al. (2003) Robotics in general surgery: personal experience in a large community hospital. Arch Surg 138:777–784. 4. Moher D, Liberati A, Tetzlaff J, Altman DG. (2010) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg 8:336–341. 5. Kang CM, Kim DH, Lee WJ, Chi HS. (2010) Conventional laparoscopic and robot-assisted spleen-preserving pancreatectomy: does da Vinci have clinical advantages? Surg Endosc 25:2004–2009. 6. Liang TB, Bai XL, Zheng SS. (2007) Pancreatic fistula after pancreaticoduodenectomy: diagnosed according to International Study Group Pancreatic Fistula (ISGPF) definition. Pancreatology 7:325–331. 7. Clavien PA, Barkun J, de Oliveria ML, Vauthey JN, Dindo D, Schulick RD
Limitations of the study One of the principal limitations of our study was that we are investigating interventions that are not performed at great frequency worldwide. There are no randomized controlled studies so far, therefore, selection bias was a real problem and these findings have to be interpreted with caution. Moreover, followup was very short and the readmission rate might be inconsistently defined or underreported. One of the major limitations of this systematic review is the lacking consideration of parameters of safety, such as blood loss and reoperation rate and of costs in the majority of all analyzed studies. Particularly when discussing minimally invasive pancreatectomy, these parameters are important to give valid recommendations. Finally, two selected studies from China10,13 showed excessive lengths of stay for the laparoscopic procedures (>14 days); limiting the external validity for application in the western countries. In conclusion, this systematic review and meta-analysis of RDP vs. LDP represents the first and most comprehensive analysis of available evidence within this context. This study suggests that both techniques are valuable minimally invasive
HPB 2016, 18, 567–574
et al. (2009) The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg 250:187–196. 8. Hayden JA, Cote P, Bombardier C. (2006) Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 144:427–437. 9. Butturini G, Damoli I, Crepaz L, Malleo G, Marcheguani G, Daskalaki D et al. (2015) A prospective non-randomised single-center study comparing laparoscopic and robotic distal pancreatectomy. Surg Endosc 29:3163–3170. 10. Chen S, Zhan Q, Chen JZ, Jin JB, Deng XX, Chen H et al. (2015) Robotic approach improves spleen-preserving rate and shortens postoperative hospital stay of laparoscopic distal pancreatectomy: a matched cohort study. Surg Endosc. 11. Daouadi M, Zureikat AH, Zenati MS, Choudry H, Tsung A, Bartlett DL et al. (2013) Robot-assisted minimally invasive distal pancreatectomy is superior to the laparoscopic technique. Ann Surg 257:128–132. 12. Duran H, Ielpo B, Caruso R, Ferri V, Quijano Y, Diaz E et al. (2014) Does robotic distal pancreatectomy surgery offer similar results as laparoscopic and open approach? A comparative study from a single medical center. Int J Med Robot 10:280–285. 13. Lai EC, Tang CN. (2015) Robotic distal pancreatectomy versus conventional laparoscopic distal pancreatectomy: a comparative study for short-term outcomes. Front Med 9:356–360.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
574
14. Lee SY, Allen PJ, Sadot E, D’Angelica MI, DeMatteo RP, Fong Y et al.
19. Zhou W, Lv R, Wang X, Mou Y, Cai X, Herr I. (2010) Stapler vs suture
(2015) Distal pancreatectomy: a single institution’s experience in open,
closure of pancreatic remnant after distal pancreatectomy: a meta-
laparoscopic, and robotic approaches. J Am Coll Surg 220:18–27.
analysis. Am J Surg 200:529–536.
15. Ryan CE, Ross SB, Sukharamwala PB, Sadowitz BD, Woot TW,
20. Baker MS, Bentrem DJ, Ujiki MB, Stocker S, Talamonti MS. (2011)
Rosemurgy AS. (2015) Distal pancreatectomy and splenectomy: a robotic
Adding days spent in readmission to the initial postoperative length of
or LESS approach. JSLS 19. e2014 00246.
stay limits the perceived benefit of laparoscopic distal pancreatectomy
16. Waters JA, Canal DF, Wiebke EA, Dumas RP, Beane JD, AguilarSaavedra JR et al. (2010) Robotic distal pancreatectomy: cost effective? Surgery 148:814–823.
when compared with open distal pancreatectomy. Am J Surg 201: 295–299. discussion 299–300. 21. Rutz DR, Squires MH, Maithel SK, Sarmiento JM, Etra JW, Perez SD
17. Kwon W, Jang JY, Park JW, Han IW, Kang MJ, Kim SW. (2014) Which method of pancreatic surgery do medical consumers prefer among open, laparoscopic, or robotic surgery? A survey. Ann Surg Treat Res 86:7–15.
et al. (2014) Cost comparison analysis of open versus laparoscopic distal pancreatectomy. HPB 16:907–914. 22. Barbash GI, Glied SA. (2010) New technology and health care costs– the case of robot-assisted surgery. N Engl J Med 363:701–704.
18. de’Angelis N, Alghamdi S, Renda A, Azoulay D, Brunetti F. (2015) Initial experience of robotic versus laparoscopic colectomy for transverse
Appendix A. Supplementary data
colon cancer: a matched case-control study. World J Surg Oncol 13:
Supplementary data related to this article can be found at http://dx.doi.org/
295.
10.1016/j.hpb.2016.04.008.
HPB 2016, 18, 567–574
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.hpb.2016.04.008
REVIEW ARTICLE
Robotic versus laparoscopic distal pancreatectomy – The first meta-analysis Paschalis Gavriilidis1, Chetana Lim1, Benjamin Menahem1, Eylon Lahat1, Chady Salloum1 & Daniel Azoulay1,2 1
Department of Hepato-Pancreato-Biliary Surgery and Liver Transplantation, Henri Mondor Hospital, and 2INSERM U 955, Créteil, France
Abstract Background: Minimally invasive pancreaticoduodenectomy is considered hazardous for the majority of authors and minimally distal pancreatectomy is still a debated topic. The aim of this study was to compare robotic distal pancreatectomy (RDP) versus laparoscopic distal pancreatectomy (LDP) using meta-analysis. Method: EMBASE, Medline and PubMed were searched systematically to identify full-text articles comparing robotic and laparoscopic distal pancreatectomies. The meta-analysis was performed by using Review Manager 5.3. Results: Nine studies fulfilled the inclusion criteria and included 637 patients (246 robotic and 391 laparoscopic). RDP had a shorter hospital length of stay by 1 day (P = 0.01). On the other hand, LDP had shorter operative time by 30 min, although this was statistically nonsignificant (P = 0.12). RDP showed a significantly increased readmission rate (P = 0.04). There was no difference in the conversion rate, incidence of postoperative pancreatic fistula, International Study Group of Pancreatic Fistula grade B–C rate, major morbidity, spleen preservation rate and perioperative mortality. All surgical specimens of RDP reported R0 negative margins, whereas 7 specimens in the LDP group had affected margins. Conclusions: In terms of feasibility, safety and oncological adequacy, there is no essential difference between the two techniques so far. The 30 min longer operative time of the RDP is due to the docking and undocking of the robot. The shorter length of stay by 1 day should be judged in combination with the increased 90-day readmission rate. Received 21 December 2015; accepted 22 April 2016
Correspondence Daniel Azoulay, Department of Hepato-Pancreato-Biliary Surgery and Liver transplantation, Henri Mondor Hospital, 51 avenue De Lattre De Tassigny, 94010 Créteil, France. Tel: +33 1 49 81 25 48. Fax: +33 1 49 81 24 32. E-mail: [email protected]
Introduction In 1994, Automated Endoscopic System of Optical Positioning (AESOP) inaugurated the era of robotic surgery. In 2000, Intuitive Surgical’s Da Vinci system was approved by the Food and Drug Administration and began the era of advanced computer-assisted telesurgery.1 In 2003, Melvin et al. performed the first robotic distal pancreatectomy (RDP).2 Shortly thereafter, Giulianotti et al. published his case series proving the feasibility of robotic pancreatectomies.3 Laparoscopic distal pancreatectomy (LDP) is the most commonly performed pancreatic resection using minimally
HPB 2016, 18, 567–574
invasive techniques due to the absence of reconstruction phase, and can be indicated for benign and malignant pancreatic lesions. Some of the technical challenges in pancreatic surgery include the vascular control and precision required for dissection of the pancreas. These are elements that may offset the laparoscopic approach whereas the robotic approach may overcome these limitations. Compared to laparoscopic approach, the potential benefits provided by the robots could be the reproduction of the mobility of the hand and fingers with seven degrees of freedom and therefore the possibility of performing the same action as in open pancreatic surgery. To the best of our knowledge, there are no randomized
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
568
controlled trials on this subject. To clarify the value of robotic for distal pancreatectomy, a systematic literature review and metaanalysis was performed of studies comparing RDP and LDP in terms of surgical and short-term outcomes. We aimed to synthesize conclusions that will serve as basis and trigger factors for the further theoretical and practical development of minimally invasive pancreatectomies.
Methods Study selection Using the search terms “robotics” or “robotic”, “laparoscopy” or “laparoscopic”, and “distal“ and “pancreatectomy”, we performed a systematic review of the literature in Medline, Embase and PubMed (including studies from the last 13 years) up to September 2015. The research was conducted independently by P.G, B.M and C.L; subsequently all the authors compared their results. References from the articles were investigated manually. Any differences were resolved by consensus. This review adhered to the guidelines outlined in the PRISMA statement.4 Data review extraction The following data were extracted: name of authors; study design; number of patients included in the robotic and laparoscopic cohort; age; sex; body mass index (BMI); American Society of Anesthesiologists (ASA) preoperative risk index; indications; operative time; estimated blood loss; number of red blood cell packed units transfused; conversion rate from the robotic and laparoscopic procedure to the open procedure; pancreatic fistula rate; International Study Group of Pancreatic Fistula (ISGPF) grade A, B, C; 90-day minor and major morbidity; 90-day perioperative mortality; length of stay; spleen preservation rate; Ro margin status; number of lymph nodes harvested; tumor size; mean follow-up; 90-day readmission rate; and cost. Inclusion criteria We included studies with more than five patients in each arm for comparison of clinical characteristics, outcomes and the cost difference between RDP and LDP. Moreover, from studies that compared robotic, laparoscopic and open distal pancreatectomies, we chose the robotic and the laparoscopic procedures. Exclusion criteria Narrative reviews, case series or studies without matched groups, studies with less than five cases, surveys and non-English language articles were excluded. Outcome definitions Apart from the study of Kang et al.,5 all the remaining reports and definitions of the pancreatic fistula rate were according to the ISGPF classification.6 Overall, minor and major complications were categorized according to Clavien–Dindo classification.7 The operative time of the robotic procedure included the
HPB 2016, 18, 567–574
docking time of the robot. Perioperative mortality is defined as death within 90 days. Quality assessment The quality of each study was assessed according to risk for bias using the QUIPS (quality in prognosis studies) tool.8 This is based on the identification of five domains of potential study bias (study participation, prognostic factor measurement, outcome measurement, confounding factor measurement, and analysis) with three to seven items per domain. Each item is given a score according to whether its quality limits potential bias: a score of 2 indicates that it does; a score of 1 indicates that it does so ‘partly’, and a score of 0 indicates that it does not. Data were extracted independently by the authors (P.G and C.L) and disagreements resolved by discussion. Statistical analysis We used Review Manager 5.3 software (Cochrane collaboration, Oxford, England) for all statistical analyses. Considering that patients were selected by different surgical teams and operated in different centers; we chose the random-effects model to assess this heterogeneity. I2 was used for heterogeneity assessment, and values of more than 50% were considered significant. Dichotomous variables were analyzed and assessed with an odds ratio (OR); a value of less than 1 favored the robotic cohort, while values of P < 0.05 and 95% confidence intervals (CIs) without the value of 1 supported the statistical significance of odds ratio (OR). Continuous variables analyzed with the weighted mean difference (WMD). The Mantel-Haenszel method was used to combine the OR for the outcomes of interest; Peto OR was used when necessary. Publication bias and sensitivity analysis were performed.
Results Literature search result and quality assessment A total of 182 studies were published in the medical literature until September 2015. After screening of records, 12 duplicates were removed. The researchers investigated the titles and abstracts of the remaining 170 articles; subsequently, 24 studies were selected for full-text investigation. After the full-text investigation process, 9 articles fulfilling the selection criteria were selected (246 robotic and 391 laparoscopic cases) (Table 15,9–16 and Fig. 1). Fifteen articles were excluded: 6 studies without a matched group, 4 narrative reviews, 2 surveys, 2 articles with less than 5 patients and 1 non-English language paper. The median QUIPS score for the included studies was 28 (range: 17–30) of a maximum score of 50 (Supplementary Table S1). Results of the meta-analysis Operative time All studies reported operative times5,9–16; whereas three studies did not give a standard deviation (SD).4,8,11 The mean operative
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
569
Table 1 Characteristics of included studies comparing robotic vs. laparoscopic distal pancreatectomy
Author, year, country (study design)
Number of patients; RDP vs. LDP
Age, mean ± SD or range; RDP vs. LDP
Female, BMI, %; RDP mean ± SD; vs. LDP RDP vs. LDP
ASA score ‡ 3, %; RDP vs. LDP
Indications, %; RDP vs. LDP
Waters JA, 2010, USA16 (retrospective)
17 vs. 18
64 vs. 59
65 vs. 50 NA
2.8 vs. 2.9
Adenocarcinoma: 0 vs 11; MCN: 18 vs. 17; NET: 29 vs. 28; IPMN: 35 vs. 11; SPPT: 0; SCM: 6 vs. 11; Other: 12 vs. 17
Kang CM, 2011, Korea5 (retrospective)
20 vs. 25
44 ± 15.9 vs. 56.5 ± 13.9
60 vs. 56 24.2 ± 2.9 vs. 23.4 ± 2.6
NA
Adenocarcinoma: 0; MCN: 25 vs. 8.5; NET: 15 vs. 12; IPMN: 10 vs. 40; SPPT: 20 vs. 16; SCM: 20 vs. 12; Other: 10 vs. 12
Daouadi M, 2013, USA11 (retrospective)
30 vs. 94
59 ± 13 vs. 59 ± 16
67 vs. 65 27.9 ± 5.1 vs. 29.0 ± 7.1
63 vs. 55
Adenocarcinoma: 43 vs. 15; MCN: 13 vs. 31; NET: 27 vs. 22; IPMN: 17 vs. 12; SPPT: 0 vs. 6.4; SCM: 0; Other: 3 vs. 13
Butturini G, 2014, Italy9 (prospective non randomized)
22 vs. 21
54 (26–77) vs. 55 (20–71)
77 vs. 71.5
4.5 vs. 0
Adenocarcinoma: 13.6 vs. 9.5; MCN: 27.2 vs. 33.3; NET: 36.3 vs. 42.8; IPMN: 0; SPPT: 13.6 vs. 4.7; SCM: 0 vs. 9.5; Other: 9 vs. 0
Duran H, 2014, Spain12 (retrospective)
16 vs. 18
61 ± 11.6 vs. 58 ± 10.03
44 vs. 50 NA
0 vs. 16
Adenocarcinoma: 56 vs. 44; MCN: 0; NET: 25 vs. 27; IPMN: 12 vs. 0; SPPT: 0; SCM: 6 vs. 11; Other: 0 vs. 18
Ryan CE, 2014, USA15 (retrospective)
18 vs. 16
67 ± 12.5 vs. 60 ± 17.0
50 vs. 38 29 ± 7.1 vs. 25 ± 4.5
NA
Adenocarcinoma: 22.2 vs. 25; MCN: 16 vs. 12; NET: 16.6 vs. 18.75; IPMN: 22.2 vs. 12.5; SPPT: 0; SCM: 11.1 vs. 18.75; Other: 11.11 vs. 12.5
Lee SY, 2014, USA14 (retrospective)
37 vs. 131
58 vs. 58
73 vs. 28.2
NA
Adenocarcinoma: 11 vs. 15; MCN: 16 vs. 12; NET: 21 vs. 31; IPMN: 11 vs. 14; SPPT: 5 vs. 5; SCM: 0; Other:35 vs. 23
Chen S, 2015, China10 (prospective nonrandomized)
69 vs. 50
56.2 ± 13.3 vs. 56.5 ± 15.0
66.7 vs. 24.6 ± 2.8 vs. 64 24.6 ± 3.0
4.3 vs. 4
Adenocarcinoma: 21.7 vs. 18; MCN: 37.7 vs. 32; NET: 4.3 vs. 6; IPMN: 8.6 vs. 10; SPPT: 14.5 vs. 16; SCM: 0; Other: 13 vs. 18
Lai EC, 2015, Hong Kong/China13 (retrospective)
17 vs. 18
61.2 ± 10.4 vs. 63.2 ± 17.9
41 vs. 78 24.1 ± 2.3 vs. 25.7 ± 2.7
64.7 vs. 77.78
Adenocarcinoma: 23 vs. 11.1; MCN: 11.7 vs. 22.2; NET: 23.5 vs. 11.1; IPMN: 5.88 vs. 0; SPPT: 0; SCM: 35.29 vs. 33.33; Other: 0 vs. 22.22
25.33 vs. 24.19
28.7 vs. 28.2
RDP, robotic distal pancreatectomy; LDP, laparoscopic distal pancreatectomy; NA, not available; SD, standard deviation; BMI, body mass index; ASA, American society of Anesthesiologists; MCN, mucinous cystic neoplasm; NET, neuroendocrine tumors; IPMN, intraductal papillary mucinous neoplasm; SPPT, solid pseudopapillary tumor; SCM, serous cystadenoma.
time of the RDP group was 30 min higher than in the LDP, although this difference was not statistically significant (median operative time = 30 min (95% CI = 8.12–68.52; P = 0.12); Fig. 2).
HPB 2016, 18, 567–574
Conversion rate All studies reported the conversion rate of each technique.5,9–16 The conversion rate was 9% (19/209) in the RDP group and 20% (70/348) in the LDP group. Meta-analysis showed no statistically
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
570
Figure 1 PRISMA flow chart showing the article selection process
significant difference between the RDP and LDP for this variable (Peto OR = 0.61; 95% CI = 0.36–1.05; P = 0.08). Incidence of pancreatic fistula and ISGPF grade B–C rates Eight of nine studies reported the incidence rate of pancreatic fistula9–16; moreover, seven of them defined and scaled it according to the ISGPF classification.9–16 The rate of pancreatic fistula was 24% (55/226) in the RDP group and 24% (88/366) in the LDP group. Meta-analysis of the 8 studies demonstrated no difference in the overall pancreatic fistula rate between the two procedures (Peto OR = 0.93; 95% CI = 0.60–1.44; P = 0.74; Fig. 3). The grade B–C pancreatic fistula rate was 13% (24/191) in the RDP cohort and 12% (41/332) in the LDP cohort. Metaanalysis of 6 studies showed no difference in the subgroup of pancreatic fistulas (ISGPF B and C) between the RDP and LDP cohorts (Peto OR = 0.98; 95% CI = 0.55–1.74; P = 0.94). Major morbidity (Dindo–Clavien classification > III) and mortality All 9 studies reported major morbidity according to the Dindo–Clavien classification 5, 9–16. The rate of major morbidity was 16% (9/246) in the RDP group and 17% (67/391) in the LDP group. Meta-analysis of 9 studies showed no
HPB 2016, 18, 567–574
significant difference in the incidence of major morbidity between RDP and LDP (Peto OR = 1.19; 95% CI 0.75–1.90; P = 0.46). All studies reported 90-day perioperative mortality.5,9–16 There was 1 death in the robotic (0.4%) and 1 death in the laparoscopic group (0.3%).11,15 Meta-analysis could not be performed for this item due to the very small rate of this event. Oncologic adequacy parameters Seven of 9 studies reported the R0 margin status and the number of harvested lymph nodes.9–12,14–16 All surgical specimens of the RDP reported with R0 negative margins. However, 7/348 (2%) of the LDP specimens were diagnosed with positive margins.10,11,15 Seven of the nine studies reported data on the number of harvested lymph nodes from both procedures.9–12,14–16 Chen et al. reported the highest mean of total lymph nodes for RDP (15.5 ± 3.1) and for LDP (15.2 ± 3.3; P = 0.819).10 Lee et al., Ryan et al., Duran et al., Butturini et al. and Daouadi et al. reported a higher number of harvested lymph nodes for the robotic compared with the laparoscopic group.9,11,12,14,15 Only Waters et al. showed a lower number of harvested lymph nodes in the robotic group.16 Meta-analysis of these outcomes could not be performed as the standard deviation (SD) could not be retrieved from the reports.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
571
Figure 2 Forest plot displaying the result of the meta-analysis comparing operative time following RDP vs. LDP
Figure 3 Forest plot displaying the result of the meta-analysis comparing incidence of pancreatic fistula following RDP vs. LDP
Figure 4 Forest plot displaying the result of the meta-analysis comparing hospital length of stay following RDP vs. LDP
Total length of hospital stay Five studies reported the length of stay.5,10,11,13,15 The mean total length of the hospital stay was 8.2 days in the RDP group and 10.9 days in the LDP group. Meta-analysis of the 5 studies showed that RDP had a statistically significant shorter hospital length of stay by 1 day (WMD = 0.97 (95% CI = 1.73–0.22; P = 0.01); Fig. 4). Only three provided the SD prices5,9,16 but meta-analysis could not be performed for this item. Readmission rate Three studies reported the 90-day readmission rate. In all of them, the robotic group showed a higher rate of readmission than its counterpart.9,11,15 Meta-analysis showed a significantly increased
HPB 2016, 18, 567–574
readmission rate, 22.85% (16/70) in RDP versus 17.55% (23/131) in LDP (Peto OR = 2.32; 95% CI = 1.03–5.21; P = 0.04; Fig. 5). Sensitivity analysis and publication bias Sensitivity analysis was performed with the aim of determining the significance of the results. For the overall incidence of pancreatic fistula, ISGPF grade B–C rate, major morbidity, conversion rate and readmission rate, the combined OR was calculated using both a fixed-effect model and a random-effects model, and the results were compared. Because of the inherent nature of non-randomized trials and the selective reporting of retrospective reviews, selective reporting and non-publication bias may influence this study.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
572
Figure 5 Forest plot displaying the result of the meta-analysis comparing readmission rate following RDP vs. LDP
Discussion Summary of evidence During the past 5 years, many authors from established pancreatic centers reported studies that supported the feasibility, reproducibility, effectiveness and cost efficacy of the robotic distal pancreatectomy compared with the laparoscopic and open procedures.5,9–16 Patient demographic characteristics and range of indications were roughly similar (Table 1). Seven of nine studies were retrospective reviews of prospectively maintained databases and two were nonrandomized studies. Selection bias is the main weakness of both types of studies. Some researchers tried to minimize this effect by selecting the control laparoscopic group from a period of time when their institution only offered the laparoscopic option.11 This also included age related selection preferences; younger patients more often choose the robotic procedure than the laparoscopic or open procedure.5,17 To the best of our knowledge, this is the first meta-analysis that compares two minimally invasive techniques of distal pancreatectomy. It included 637 patients, 246 (39%) of which underwent RDP and 391 (61%) of which underwent LDP. In sum, our metaanalysis did not detect any statistically significant differences in operative time, conversion rate, incidence of postoperative pancreatic fistula, major morbidity and spleen preservation rate. RDP showed a statistically significant reduced hospital length of stay by 1 day. Although only 3 studies reported readmission rates9,11,15; we detected a significantly higher readmission rate in the RDP group than in the LDP. All RDP surgical specimens reported with negative margins and only 0.2% of cases had positive margins in the LDP group. Seven studies reported the number of retrieved lymph nodes. In all cases the number of retrieved lymph nodes was higher in the RDP group.9–12,14–16 Meta-analysis of this parameter could not be performed due to the shortage of standard deviation data. One death was reported in the RDP (0.40%) and in the LDP (0.25%) groups. The mean difference of operative time was only 30 min longer in the RDP group. It was reported that the docking time of the robot was 30 min.18 Taking into account that the docking time was included in the total operative time, there is no essential difference between the two procedures. As with any surgical innovation, the operative time of the robotic interventions not only depended on the surgeon’s skills and their stage of the
HPB 2016, 18, 567–574
learning curve, but also on institutional learning curves as well. The conversion rate to open procedure was lower in RDP (9.09%) compared with LDP (20.11%). Meta-analysis did not find any significant difference. The most common reasons for conversion were obesity, tumor proximity to major vessels, adhesions and margins assessment.11,14 The results of the conversion rate are probably affected by tertiary centers by large numbers of cancer referral cases.11,14 Recently, Daoudi et al.11 showed that RDP significantly reduces the risk of conversion to open resection compared with LDP (0% vs. 16%; P < 0.05). However, most cases of open conversion might have occurred in their early experiences with LDP, and it is possible that the authors have been able to shorten the learning curve of RDP due to large prior experience in LDP. The incidence rate of postoperative overall and major complications including the incidence of as well as the severity of pancreatic fistula was strikingly similar. Current evidence form large meta-analysis found that independent of the technique used for transection and closure of pancreatic parenchyma after distal pancreatectomy, the incidence of postoperative pancreatic fistula ranges from 0% to 47%.19 Therefore, the results from our study are in the middle of the reported ranges. The meta-analysis showed a statistically significant reduction in the hospital stay by 1 day. However, the readmission rate was significantly higher in the RDP group than in the LDP group (22.85% vs. 17.55%). The main cause of readmission included pancreatic fistula and/or abdominal collections.9,11,15 This important finding must be underlined as most of studies comparing laparoscopic and open approaches for DP do not take into account the rate of readmissions. There is a strong possibility for underreporting bias because patients admitted in another hospital may not have been reported. It was reported that the increased readmission rate of LPD limits its perceived benefit compared with the open technique.20 Yet, Rutz et al. compared the 30-day readmissions of LDP and the open technique and reported that the former is more cost-effective than the latter.21 However, data on these readmissions are available only rarely in existing studies, which does not allow drawing any solid conclusions. Regarding oncologic adequacy, the most impressive finding was that all surgical specimens had negative surgical margins in the RDP group.9–16 In the LDP group, only 7/348 (2%) had
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
573
positive margins.10,11,15 Most of the surgeons consider that they were able to achieve adequate oncological results owing to the better ergonomic conditions offered by the robot.5,9–12,14–16 All authors except Waters et al.16 reported a higher number of retrieved lymph nodes in the RDP group.10,11,16 The rate of spleen preservation depends on technical factors but also mainly from indication of pancreatectomy (spleen preservation is usually not considered for malignancies). But this criterion cannot be used for comparison in this meta-analysis, except if this rate is compared to the rate of benign diseases. However, for some indications such as neuroendocrine tumors, it is usually not possible to extract from every study the number of benign and malignant tumors, which does not allow drawing any solid conclusions. The cost of a robotic surgical system ranges between $ 1 million to $ 2.5 million; its single use appliances and maintenance are also costly.22 It is reported that the additional variable cost of the robotic intervention compared with the laparoscopic or open procedures was $ 1600, or approximately 6%. However, including the amortized cost of the investment, the additional cost rose approximately 13%, or $ 3200.22 Data on costs are available only in 3/9 studies, which does not allow drawing any firm conclusions. Waters et al. reported that thanks to the shorter hospital stay, the cost of RDP was lower ($ 10,588) compared with the laparoscopic ($ 12,986) procedure.16 Kang et al. found RDP to be 2.5 times more expensive than LDP ($ 8304 vs. $ 3861).5 Butturini et al. also found RDP to be more expensive compared with LDP (2700–3190V vs. 1434–1674V).9 One Italian team lowered the cost by 750V for each robotic intervention by using laparoscopic ultrasonic shears instead of robotic shears and for transection of the pancreatic parenchyma; they used ultrasonic shears instead of a linear stapler.
interventions that can be used safely to treat patients with benign, borderline and, in selected cases, malignant neoplasms. The current evidence is insufficient to prove the superiority of one technique over another. Further studies are needed for detailed evaluation of oncologic adequacy, long-term survival, cost of postoperative complications, length of stay and patterns of readmission. Funding or grants None.
Conflict of interest None declared.
References 1. Heemskerk J, Bouvy ND, Baeten CG. (2013) The end of robot-assisted laparoscopy? A critical appraisal of scientific evidence on the use of robot-assisted laparoscopic surgery. Surg Endosc 28:1388–1398. 2. Melvin WS, Needleman BJ, Krause KR, Ellison EC. (2003) Robotic resection of pancreatic neuroendocrine tumor. J Laparoendosc Adv Surg Tech A 13:33–36. 3. Giulianotti PC, Coratti A, Angelini M, Sbrana F, Cecconi S, Balestracci T et al. (2003) Robotics in general surgery: personal experience in a large community hospital. Arch Surg 138:777–784. 4. Moher D, Liberati A, Tetzlaff J, Altman DG. (2010) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg 8:336–341. 5. Kang CM, Kim DH, Lee WJ, Chi HS. (2010) Conventional laparoscopic and robot-assisted spleen-preserving pancreatectomy: does da Vinci have clinical advantages? Surg Endosc 25:2004–2009. 6. Liang TB, Bai XL, Zheng SS. (2007) Pancreatic fistula after pancreaticoduodenectomy: diagnosed according to International Study Group Pancreatic Fistula (ISGPF) definition. Pancreatology 7:325–331. 7. Clavien PA, Barkun J, de Oliveria ML, Vauthey JN, Dindo D, Schulick RD
Limitations of the study One of the principal limitations of our study was that we are investigating interventions that are not performed at great frequency worldwide. There are no randomized controlled studies so far, therefore, selection bias was a real problem and these findings have to be interpreted with caution. Moreover, followup was very short and the readmission rate might be inconsistently defined or underreported. One of the major limitations of this systematic review is the lacking consideration of parameters of safety, such as blood loss and reoperation rate and of costs in the majority of all analyzed studies. Particularly when discussing minimally invasive pancreatectomy, these parameters are important to give valid recommendations. Finally, two selected studies from China10,13 showed excessive lengths of stay for the laparoscopic procedures (>14 days); limiting the external validity for application in the western countries. In conclusion, this systematic review and meta-analysis of RDP vs. LDP represents the first and most comprehensive analysis of available evidence within this context. This study suggests that both techniques are valuable minimally invasive
HPB 2016, 18, 567–574
et al. (2009) The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg 250:187–196. 8. Hayden JA, Cote P, Bombardier C. (2006) Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 144:427–437. 9. Butturini G, Damoli I, Crepaz L, Malleo G, Marcheguani G, Daskalaki D et al. (2015) A prospective non-randomised single-center study comparing laparoscopic and robotic distal pancreatectomy. Surg Endosc 29:3163–3170. 10. Chen S, Zhan Q, Chen JZ, Jin JB, Deng XX, Chen H et al. (2015) Robotic approach improves spleen-preserving rate and shortens postoperative hospital stay of laparoscopic distal pancreatectomy: a matched cohort study. Surg Endosc. 11. Daouadi M, Zureikat AH, Zenati MS, Choudry H, Tsung A, Bartlett DL et al. (2013) Robot-assisted minimally invasive distal pancreatectomy is superior to the laparoscopic technique. Ann Surg 257:128–132. 12. Duran H, Ielpo B, Caruso R, Ferri V, Quijano Y, Diaz E et al. (2014) Does robotic distal pancreatectomy surgery offer similar results as laparoscopic and open approach? A comparative study from a single medical center. Int J Med Robot 10:280–285. 13. Lai EC, Tang CN. (2015) Robotic distal pancreatectomy versus conventional laparoscopic distal pancreatectomy: a comparative study for short-term outcomes. Front Med 9:356–360.
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.
HPB
574
14. Lee SY, Allen PJ, Sadot E, D’Angelica MI, DeMatteo RP, Fong Y et al.
19. Zhou W, Lv R, Wang X, Mou Y, Cai X, Herr I. (2010) Stapler vs suture
(2015) Distal pancreatectomy: a single institution’s experience in open,
closure of pancreatic remnant after distal pancreatectomy: a meta-
laparoscopic, and robotic approaches. J Am Coll Surg 220:18–27.
analysis. Am J Surg 200:529–536.
15. Ryan CE, Ross SB, Sukharamwala PB, Sadowitz BD, Woot TW,
20. Baker MS, Bentrem DJ, Ujiki MB, Stocker S, Talamonti MS. (2011)
Rosemurgy AS. (2015) Distal pancreatectomy and splenectomy: a robotic
Adding days spent in readmission to the initial postoperative length of
or LESS approach. JSLS 19. e2014 00246.
stay limits the perceived benefit of laparoscopic distal pancreatectomy
16. Waters JA, Canal DF, Wiebke EA, Dumas RP, Beane JD, AguilarSaavedra JR et al. (2010) Robotic distal pancreatectomy: cost effective? Surgery 148:814–823.
when compared with open distal pancreatectomy. Am J Surg 201: 295–299. discussion 299–300. 21. Rutz DR, Squires MH, Maithel SK, Sarmiento JM, Etra JW, Perez SD
17. Kwon W, Jang JY, Park JW, Han IW, Kang MJ, Kim SW. (2014) Which method of pancreatic surgery do medical consumers prefer among open, laparoscopic, or robotic surgery? A survey. Ann Surg Treat Res 86:7–15.
et al. (2014) Cost comparison analysis of open versus laparoscopic distal pancreatectomy. HPB 16:907–914. 22. Barbash GI, Glied SA. (2010) New technology and health care costs– the case of robot-assisted surgery. N Engl J Med 363:701–704.
18. de’Angelis N, Alghamdi S, Renda A, Azoulay D, Brunetti F. (2015) Initial experience of robotic versus laparoscopic colectomy for transverse
Appendix A. Supplementary data
colon cancer: a matched case-control study. World J Surg Oncol 13:
Supplementary data related to this article can be found at http://dx.doi.org/
295.
10.1016/j.hpb.2016.04.008.
HPB 2016, 18, 567–574
© 2016 International Hepato-Pancreato-Biliary Association Inc. Published by Elsevier Ltd. All rights reserved.