Comparative analysis of open, laparoscopic and robotic distal pancreatic resection: The United Kingdom's first single-centre experience.

INTRODUCTION: Laparoscopic distal pancreatectomy (LDP) has potential advantages over its open equivalent open distal pancreatectomy (ODP) for pancreatic disease in the neck, body and tail. Within the United Kingdom (UK), there has been no previous experience describing the role of robotic distal pancreatectomy (RDP). This study evaluated differences between ODP, LDP and RDP.
METHODS: Patients undergoing distal pancreatectomy performed in the Department of Hepatobiliary and Pancreatic Surgery at the Freeman Hospital between September 2007 and December 2018 were included from a prospectively maintained database. The primary outcome measure was length of hospital stay, and the secondary outcome measures were complication rates graded according to the Clavien-Dindo classification.
RESULTS: Of the 125 patients, the median age was 61 years and 46% were male. Patients undergoing RDP (n = 40) had higher American Society of Anesthesiologists grading III compared to ODP (n = 38) and LDP (n = 47) (57% vs. 37% vs. 38%, P = 0.02). RDP had a slightly lower but not significant conversion rate (10% vs. 13%, P = 0.084), less blood loss (median: 0 vs. 250 ml, P < 0.001) and a higher rate of splenic preservation (30% vs. 2%, P < 0.001) and shorter operative time, once docking time excluded (284 vs. 300 min, P < 0.001) compared to LDP. RDP had a higher R0 resection rate than ODP and LDP (79% vs. 47% vs. 71%, P = 0.078) for neoplasms. RDP was associated with significantly shorter hospital stay than LDP and ODP (8 vs. 9 vs. 10 days, P = 0.001). While there was no significant different in overall complications across the groups, RDP was associated with lower rates of Grade C pancreatic fistula than ODP and LDP (2% vs. 5% vs. 6%, P = 0.194).
CONCLUSION: Minimally invasive pancreatic resection offers potential advantages over ODP, with a trend showing RDP to be marginally superior when compared to conventional LDP, but it is accepted that that this is likely to be at greater expense compared to the other current techniques.

INTRODUCTION

Distal pancreatectomy (DP) remains the treatment of choice for both benign and malignant lesions involving the neck, body or tail of the pancreas. Advances in radiological imaging have led to increased numbers of distal pancreatic resections because of more incidental findings on cross-sectional imaging.[1] Traditionally, DP usually involves en bloc resection with the spleen due to the anatomical association of the pancreatic tail with the splenic hilum. However, because of the potential for an increased risk of sepsis, immune deficiencies and the need for vaccinations combined with antibiotic prophylaxis, more efforts to achieve spleen-preserving distal pancreatectomy (SPDP) are now considered,[2] even though SPDP is more challenging particularly when performed laparoscopically.[3]

In recent years, the use of laparoscopic distal pancreatectomy (LDP) DP has slowly gained popularity, but in the UK, it is still limited to highly specialised laparoscopic centres.[4] Case series suggest that LDP is associated with less blood loss, lower analgesic requirements, shorter hospital stay and lower rates of morbidity when compared to that of the open approach.[5] Similarly, RDP has been demonstrated to be equally safe and effective and confers benefits similar to those associated with laparoscopic surgery.[5] Currently, robotic enthusiasts believe that RDP may allow for even further marginal gains over the laparoscopic approach which are becoming hard to quantify in small series. These include further reductions in blood loss, even lower rates of conversion to open surgery and allow more patients to access minimally invasive surgery.[6] Taken together, this is likely to be more attractive to patients and in the latter case to a surgeon who has not previously performed a minimally invasive DP especially if it is perceived that the operation is technically easier. Robotic platforms offer significant user advantages including three-dimensional vision and improved graphics and ergonomics when compared with that of standard laparoscopic techniques.[78] A good example of this has been robotic prostatectomy, where the numbers have risen exponentially in the last 10 years. Upfront capital and consumable costs maybe a significant disadvantage for RDP at this point in time, but with increased competition in the development of robotic platforms, technology costs will drop, allowing consumers greater access.[9]

To date, there have only been four studies comparing open distal pancreatectomy (ODP), LDP and RDP, but none were from the United Kingdom (UK).[5] The largest series from Memorial Sloan Kettering Cancer Centre included patients undergoing ODP prior to the introduction of LDP or RDP at their centre, which somewhat precludes a reliable analysis of outcomes from the same time period.[10] At our centre, we hypothesised that RDP would be non-inferior for oncological and post-operative outcomes when compared to that of the conventional laparoscopic approach. This is the first report comparing the outcomes of open, laparoscopic and robotic distal pancreatectomy (RDP) performed at a single UK tertiary centre, aiming to evaluate oncological and post-operative outcomes for open, LDP and RDP.

METHODS

Study design and study population

A retrospective review of a consecutive series of patients undergoing open, laparoscopic and RDP with or without splenectomy performed at the Department of Hepatobiliary and Pancreatic Surgery at Freeman Hospital between September 2007 and 1 December 2018 was undertaken from a prospectively maintained database. The first forty RDP patients were included. Patients were excluded if they had another major procedure alongside the operation (e.g., some have had simultaneous laparoscopic colectomy, n = 4, and were excluded).

Patient pathway

The decision regarding the type of procedure to be performed was determined in a multidisciplinary team meeting, consisting of consultant surgeons, oncologists, interventional radiologists, pathologists, specialist nurses and dieticians, based on patients' conditions and surgeons' preferences and dependent on era as LDP (September 2007–December 2013) was abandoned completely towards the end of 2013 in favour of RDP (December 2014–December 2018). In this series, ODP was performed (September 2007 and 1 December 2018) by five different surgeons (GS, JH, DMM, JJF and SAW), whereas LDP and RDP were performed predominantly by two different surgeons (SAW and JJF). The surgical techniques have been described elsewhere.[11]

Study data points

Data collected included patient demographics (age, gender, body mass index [BMI] and American Society of Anesthesiologists [ASA] grade), intraoperative data (operative time, preparation time [for robotics], estimated blood loss and spleen preservation rate), post-operative data (total hospital stay, length of ITU stay, 30-day morbidity, post-operative pancreatic fistula [POPF], wound complication, 30-day reoperation, 30-day readmission and recurrence) and oncological data (R0 resection status).

Definition

Conversion was defined as a change of operative procedure from either laparoscopic or robotic to open procedure (laparotomy). The operative time for all groups was recorded starting from the first incision (knife to skin) until the skin closure. In the robotic group, the preparation time was defined as set-up time of the robot. R0 resection was defined pathologically as a complete excision of the tumour with a minimum clearance margin of 1 mm. Complications were graded according to Clavien–Dindo classification of surgical complication.[12] Major complications were defined as Grade 3 complications and above. POPF was diagnosed and graded according to the International Study Group on Pancreatic Fistula (ISGPF) criteria, whereby Grade C POPF were clinically significant fistulas requiring intervention.[13] Reoperation was defined as any surgical operation done within 30 days after the operation to fix complications. Readmission was defined as an episode within 30 days after the operation when the patient was re-hospitalised after being discharged from the hospital.

Study outcome

The primary outcome measure was duration of hospital stay and the secondary outcome measures were complication rates (graded according to the Clavien–Dindo classification), the ability to preserve the spleen, blood loss and R0 status.

Statistical analysis

This study was deemed exempt from full ethical review by the Clinical Governance Newcastle Upon Tyne NHS Trust as this was an observational study and no patient consent was required. Data analysis was undertaken using R Foundation Statistical software (R 3.2.1), as previously reported.[5] Categorical data were reported in the form of frequency (percentage) and assessed using Chi-square test. Continuous variables were reported in the form of mean ± standard deviation or median (range). The assessment for the normality of the data was done by using Shapiro–Wilk test. Normally distributed data were analysed using the one-way analysis of variance. The Kruskal–Wallis test was used for non-normally distributed data. P < 0.05 was considered statistically significant. All patients were analysed on an intention-to-treat basis, in which the patients who were converted to open were still considered as the part of their original group (robotic or laparoscopic).

RESULTS

Pre-operative data

In this study, 125 patients underwent DP, of which 40 (32%) were robotic and 47 (38%) were laparoscopic. Patients' demographics and pre-operative data across the different surgical approaches are presented in Table 1. The most common indications [Table 1] for DP were neuroendocrine tumour (n = 40), intraductal papillary mucinous neoplasms (n = 21), adenocarcinoma (n = 20), mucinous cystic neoplasm (n = 12) and chronic pancreatitis (n = 12). The three cohorts were similar in terms of age, sex, BMI, malignancy rate and pathological types. Patients receiving RDP were associated with higher ASA Grade III than those receiving ODP or LDP (57% vs. 37% vs. 38%, P = 0.02). The median lesion size in the ODP group was statistically significantly larger compared to that of LDP and RDP (35 vs. 22 vs. 24 mm, P = 0.003).

Table 1

Pre-operative data of patients undergoing distal pancreatectomy stratified by surgical approach (percentages reported in parentheses)

	Open (n=38)	Laparoscopic (n=47)	Robotic (n=40)	P
Age (years)	56 (47, 68)	67 (50, 72)	62 (55, 72)	0.236
Sex, male	17 (45)	23 (49)	17 (42)	0.828
BMI (kg/m2)	27 (22, 29)	28 (25, 30)	28 (23, 30)	0.288
ASA grade
I	3 (8)	12 (26)	2 (5)	0.02
II	21 (55)	17 (36)	14 (38)
III	14 (37)	18 (38)	21 (57)
Histopathology
Adenocarcinoma	8 (21)	7 (15)	5 (12)	0.084
Chronic pancreatitis	8 (21)	3 (6)	1 (2)
IPMN	7 (18)	8 (17)	6 (15)
MCN	1 (3)	8 (17)	3 (8)
NET	9 (24)	14 (30)	17 (42)
Others	5 (13)	7 (15)	8 (20)
Malignancy, yes	19 (50)	24 (51)	24 (60)	0.613
Size of lesion (mm)	35 (24, 50)	22 (15, 29)	24 (16, 35)	0.003

BMI: Body mass index, ASA: American Society of Anaesthesiologists, IPMN: Intraductal papillary mucinous neoplasm, MCN: Mucinous cystic neoplasm, NET: Neuroendocrine tumour

Intraoperative data

There were 10 conversions in total: 10% (n = 4) in RDP and 12% (n = 6) in LDP (P = 0.088). The causes of conversion were suspicion of bowel injury (n = 1 LDP), tumour progression and invasion into other adjacent organs (e.g., stomach/small bowel) with concerns over margin (n = 3 LDP, RDP n = 1), anomalous hepatic arterial anatomy (n = 1 LDP), tumour close to the portal vein needing venoplasty (LDP n = 1), unable to disengage linear stapler (n = 1 LDP), adhesions without reasonable progression (n = 2 RDP) and splenic vein thrombosis with sinistral varices in chronic pancreatitis (n = 1 RDP). There was no significant difference in the conversion rates after RDP between the first twenty and the second twenty (15% vs. 5%, P = 0.563) although the trend showed improvement as the series progressed. The operative time for RDP having removed docking time was significantly shorter than that of LDP but longer than that of ODP (median: 284 vs. 300 vs. 190 min, P < 0.001). This remained the same even with operative time including docking time (median: 334 vs. 300 vs. 190 min, P < 0.001). The estimated blood loss was statistically significantly lower in RDP than that of LDP and ODP (median: 0 vs. 250 vs. 462 ml, P < 0.001), indeed blood loss was minimal in the RDP group ranging from 0 to 500 ml. The spleen preservation rate in RDP was statistically significantly better than that of LDP but lower than that of ODP (30% vs. 2% vs. 47%, P < 0.001). For oncological outcomes in malignant lesions, RDP had higher rates of R0 resections than that of LDP and ODP (79% vs. 71% vs. 47%, P = 0.078) but was not statistically significant. There were no significant differences in the number of lymph nodes harvested between the groups [Table 2].

Table 2

Intraoperative data of patients undergoing distal pancreatectomy stratified by surgical approach (percentages reported in parentheses)

	Open (n=38)	Laparoscopic (n=47)	Robotic (n=40)	P
Conversion	N/A	6 (13)	4 (10)	0.083
Operative time (min)	190 (90, 475)	300 (120, 540)	334 (200, 529)	<0.001
Operative time (min)*	190 (90, 475)	300 (120, 540)	284 (168, 529)	<0.001
Estimated blood loss (ml)	462 (0, 1055)	250 (0, 465)	0 (0, 500)	<0.001
Spleen preservation	18 (47)	1 (2)	12 (30)	<0.001
Oncological outcome	n=19	n=24	n=24
R1 resection	10 (53)	7 (29)	5 (21)	0.078
Lymph node harvested	10 (3, 19)	12 (8, 19)	9 (6, 14)	0.611

*Operating time after docking time of the robot has been removed. N/A: Not available

Post-operative outcomes

The median length of follow-up in all cohorts was 12 months (range: 4–69 months). The post-operative outcomes are presented in Table 3. Patients undergoing RDP had a statistically significantly shorter length of hospital stay than that of LDP and ODP (median 8 vs. 9 vs. 10 days, P = 0.001). There were no significant differences in the rates of overall complications between patients undergoing RDP, LDP and ODP (55% vs. 47% vs. 37%, P = 0.274). Patients undergoing RDP had statistically significantly lower rates of major complications (Clavien–Dindo ≥ Grade III) than that of LDP and ODP (10% vs. 28% vs. 11%, P = 0.042). There were no statistically significant differences in the rates of clinically significant Grade C pancreatic fistula, which were lower in patients undergoing RDP compared to that of LDP and ODP (2% vs. 6% vs. 5%, P = 0.194). Patients undergoing RDP had statistically significantly lower rates of post-operative diabetes mellitus than those undergoing ODP and LDP (2% vs. 18% vs. 28%, P = 0.007). In terms of discharge analgesia, all patients in the RDP group were given analgesia compared to those in the LDP and ODP groups (100% vs. 87% vs. 82%, P = 0.023). Patients in the RDP group had a trend for lower rates of weak opioid use than those in the LDP group (28% vs. 45%, P = 0.241), but there was no statistically significant difference in non-opioid usage (98% vs. 87%, P = 0.076).

Table 3

Post-operative data of patients undergoing distal pancreatectomy stratified by surgical approach (percentages reported in parentheses)

	Open (n=38)	Laparoscopic (n=47)	Robotic (n=40)	P
ITU/HDU stay (days)	1 (0, 20)	1 (0, 9)	1 (0, 3)	0.165
Length of stay (days)	10 (6, 77)	9 (4, 46)	8 (4, 17)	0.001
Overall complications	14 (37)	22 (47)	22 (55)	0.274
Major complications	4 (11)	13 (28)	4 (10)	0.042
Pancreatic fistula grade
None	30 (79)	33 (70)	21 (53)	0.194
Grade A	4 (11)	6 (13)	10 (25)
Grade B	2 (5)	5 (11)	8 (20)
Grade C	2 (5)	3 (6)	1 (2)
Readmission (90 days)	4 (11)	3 (6)	6 (15)	0.423
Analgesia at discharge	31 (82)	41 (87)	40 (100)	0.023
Non-opioids	31 (82)	41 (87)	39 (98)	0.076
Weak opioids	13 (34)	21 (45)	11 (28)	0.241
Strong opioids	7 (18)	3 (6)	12 (30)	0.015

ITU/HDU: High-dependency unit

DISCUSSION

Minimally access techniques reduce the physiological impact of surgery for lesions of the distal pancreas, potentially permitting shorter hospital stays with enhanced recovery compared to that of traditional open procedures.[14] In the UK, very few centres have adopted laparoscopic pancreatic surgery and until now, none have reported their experience of robotic techniques for DP. While several centres outside of the UK have published their experience of RDP, these are mainly from centres in North America, Europe (France and Italy) and Asia,[5] and the current meta-analyses are based on these data.[5] This is the first report from the UK. We previously implemented a laparoscopic programme in 2007 which ceased in November 2013 after introducing a robotic programme. This tertiary centre experience of distal pancreatic resections, for both benign and malignant indications, demonstrates that minimally invasive surgery, whether robotic or laparoscopic, is safe and feasible with comparable rates of morbidity and mortality when compared to that of the open technique. No cost analysis was performed between techniques as this can only be done accurately if the data are collected prospectively, specifically itemising all aspects of the procedure accurately along with complications and interventions and further readmissions.[9]

This study found no significant differences in the operative time between RDP (docking time excluded) and LDP, although the operative time was statistically significantly longer compared to that of the ODP group (284 vs. 300 vs. 190 min, P < 0.001). These findings are not surprising and have been reported in the majority of previously published studies,[1516] while only three have demonstrated shorter operating time with RDP.[17] A number of previously published meta-analyses in robotic surgery including prostatectomy,[18] gastrectomy[19] and hysterectomy[19] have all suggested that the operative time for robotic procedures was significantly extended compared to that of laparoscopic surgery. This variability might be explained by two main factors: (i) the robotic set-up time, which includes the time needed to prepare the robotic cart and for trocars to be connected to the robot – we accept this adds time to the procedure especially if there is additional training needed for new surgical practitioners, and (ii) learning curve; surgeons' experience, skill and their position in the learning curve influences the operating time (with experience will improve).

In the present study, blood loss in patients undergoing RDP was significantly lower than that of LDP. Only five studies from previously published series have demonstrated similar findings to that of our study.[515] Zhang et al. and Chen et al. have shown that blood loss during RDP was significantly lower than that of LDP, similar to our study.[16] Only one study has demonstrated that RDP was associated with significantly lower transfusion rates than that of LDP (3% vs. 14%, P = 0.05).[17] The majority of other studies did not demonstrate any difference in transfusion rates between RDP and LDP.[520]

In terms of conversion rates, RDP was associated with slightly lower rates compared to that of LDP (10% vs. 12%), but this was not a statistically significant finding. It is difficult to interpret whether the lower conversion rate in the RDP group resulted from the technical advantages of the robotic system, which plays an important role in facilitating haemostasis or whether it was just due to surgeons becoming better at minimally invasive pancreatectomy.[21] Only two studies, by Raoof et al.[22] (10% vs. 27%, P < 0.001) and Xourafas et al.[23] (8% vs. 16%, P = 0.007), have demonstrated statistically significantly lower conversion rates after RDP when compared to that of LDP, while others have not shown any significant difference.[5]

Spleen preservation during minimally invasive DP is difficult and therefore could be used as a guide for progression. This not only has immunological advantages for the patient but also has reduced prescription costs. This series demonstrates that the splenic preservation rate in the RDP was similar to that of ODP, but significantly better compared to that of LDP. Although some studies have similar findings (96% vs. 39%, P < 0.001),[17] the majority of previously published studies do not.[5] Importantly, the presence of malignant distal pancreatic lesions has to be taken into consideration because spleen preservation is not always recommended in these cases.[5] In our series, there were no significant differences in the rates of malignant lesions between the groups, yet splenic preservation remains higher in the RDP group. Nevertheless, we accept that the splenic preservation rate was low in the LDP group because from the outset, our intention was not to perform splenic preservation. We now realise that our learning curve may have not been reached for this particular operation. However, with increasing experience of minimally invasive techniques, there has been an immediate realisation that splenic preservation is much easier with the robotic platform.

Oncological outcomes are also important in predicting the risk of locoregional recurrence and long-term survival. First, R0 resections were achieved in the majority of LDP and RDP compared to that of ODP (79% vs. 71% vs. 47%). Ryan et al. demonstrated that rates of R0 resections are higher after RDP compared to that of LDP (100% vs. 94%), but with no statistically significant difference.[24] No previously published studies have demonstrated higher R0 resection rates with RDP over LDP.[522] However, it is well understood that R0 resections rates can be heavily influenced by meticulous pathological analysis, case selection and surgical experience, for example easier cases are preferentially selected for minimally invasive techniques.[25] Second, lymph nodes examined, which is an important predictor of long-term survival of pancreatic cancer patients, were similar between all the three groups in this study, similar to that of previously published studies.[1526]

The post-operative outcomes in our study demonstrated that overall and major complications were similar across all groups. Previous studies have reported overall and major complications rates ranging between 10%–88% and 0%–45% for patients undergoing RDP, respectively.[1720] Our data reflect that we are consistent with previously reported studies. Furthermore, the incidence of pancreatic fistula in our case series was highest in the RDP group (despite trying different techniques to ameliorate this problem), but all of these were low-grade fistulas (Grade A and B) based on ISGPF classification. There was one Grade C fistula in the RDP group. None of the previously published studies demonstrated significant differences in high-grade POPF between RDP and LDP, although a trend favoured RDP.[2127]

Our study was the first to evaluate post-operative pain. We found that the use of weak opioids, such as codeine and tramadol, was lower in the RDP group, suggesting less post-operative pain in this particular group. We also converted the analgesia into morphine-equivalent daily dose, but, again, the result showed that the morphine-equivalent daily doses were similar between the three groups. However, use of opioids has decreased over time and may not reflect the decrease in pain in these patients but rather a change in culture.

The total cost of DP procedures was not calculated in our study as the billing of the operative procedure, length of hospital stay and any other charges related to this is complicated to accurately evaluate in the UK. It also depends on tariffs and QUALYS. There are no accurate studies which have looked at cost; all have flaws in their methodology.[9] The calculation will only be more accurate if a study is conducted in a prospective manner specifically looking at cost. Limited studies have reported this type of data. Kang et al.[28] found that RDP ($8304) was more expensive than LDP ($3861). A study by Butturini et al.[29] also reported the same finding (RADP = €2700–3190 vs. LDP = €1431–1674). In contrast, Waters et al.[10] reported that the cost of RDP was lower than that of LDP ($10588 vs. $12986), owing to the lower hospital stay in the RDP group. From this standpoint, there is an expectation that the cost of RDP can be subsided in the future, reducing overall cost particularly when combined with the implementation of enhanced recovery techniques. Nevertheless, it is also possible that this will not happen as technology advances with more useful adjuncts being applied to robotic platforms such as remote operating, better audio–visuals, diagnostic imaging and anatomical mapping, which may all in fact increase the overall cost.

This study has important limitations to be discussed. First, this was a retrospective study subject to bias associated with the nature of the study. Second, there could be a potential bias with comparing LDP with RDP as the former was started during the learning curve of surgeons and RDP was adopted when surgeons were confident with minimally invasive techniques. However, comparing RDP with ODP, the former appears to have some oncological benefits. Finally, the selection bias between ODP, LDP and RDP may affect the outcomes, but as demonstrated in Table 1, there appears to be no significant differences in patient characteristics except tumour size, suggesting limited selection bias overall.

CONCLUSION

This report is the first UK single-centre experience of minimally invasive pancreatic resection which specifically includes robotic techniques. It shows that RDP is safe and feasible with comparable oncological and surgical outcomes when compared to that of LDP and ODP. Future randomised controlled trials for this operation will be needed to demonstrate whether robotic pancreatic surgery is genuinely better for both patients and surgeons. There is also an urgent need to evaluate its true cost compared to that of LDP.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.