Robotic oesophago-gastric cancer surgery

Abstract

A postoperative complications rate of nearly 50% has compelled oesophago-gastric practice to adopt minimally invasive techniques such as robotic surgery

Oesophago-gastric surgery has long been considered demanding, challenging and high-risk for both surgeon and patient alike. With improving mortality rates during the past few decades, the focus has changed to reducing complications while maintaining the quality of oncological resections. A paradigm shift in oesophago-gastric cancer surgery in recent years has been the centralisation of specialist services.¹ By concentrating care into fewer regional units, it was envisaged that outcomes – both cancer-related and otherwise – would improve, with such a set-up fostering better research, audit and innovation. Despite strong evidence for improving care, postoperative complications remain at 40–50%.^1–4 This has been a major factor in driving minimally invasive techniques into oesophago-gastric practice.

Ivor Lewis presented his seminal Hunterian Lecture in 1946, outlining a novel transthoracic two-staged approach to performing oesophagectomy for malignancy.⁵ Even at that time, the importance of negative resection margins and an adequate lymphadenectomy were highlighted to be of paramount importance. For gastric cancers, a similar adage – based on the Japanese experience – has been instilled in contemporary surgery. Thus, the variations and improvement in practice have largely been to ensure that these two integral factors are achieved in every operation. Indeed, this has been the only necessary standard of care in oesophago-gastric surgery, with the manner in which this is achieved adopting a secondary role. For example, as yet there is no prospective randomised trial comparing an open Ivor Lewis and transhiatal oesophagectomy.⁶ Studies assessing the laparoscopic approach are limited to a few dedicated centres and, despite promising results, uptake has been well below that seen in other surgical specialties. At present, it is therefore correct to say that the nature of the operation is determined by the experience of the institution and its surgeons. It is within this context that robotic surgery has entered the scene as an alternative technique.

Robotic gastrectomy

Gastric cancer is the fifth most common malignancy worldwide.⁷ Its incidence is far greater in the Far East compared with Europe and North America. As a result, survival is also dramatically different: a 5-year survival rate of 24% in Europe compared with 60% in Japan (although the latter has a structured screening programme).^8,9 As a consequence, much of the surgical research regarding stomach cancer is based on Far Eastern experience; in recent years such focus has been in minimally invasive technique. There are two main operation types for gastric cancer: total gastrectomy for a proximal stomach or body malignancy and subtotal gastrectomy for an antral or pyloric tumour. Reconstruction is with an oesophago-jejunal anastomosis in the former and a gastro-jejunal anastomosis for the latter. Although the precise degree of lymphadenectomy remains debated between the continents, the consensus is that a minimum yield of 15 lymph nodes is mandatory.¹⁰

Laparoscopic gastrectomy was not universally adopted as a reasonable alternative to conventional open surgery. This mostly related to the technical challenges of this approach, especially for large, locally advanced tumours and for total gastrectomy. In particular, given the paramount importance of appropriate lymphadenectomy, D2 procedures were especially challenging owing to dissection along precarious anatomical structures, along with constructing an intracorporeal anastomosis.^11,12 Although representing a few specialist units, the outcomes were largely equivocal when compared with open surgery, with the exception of hospital stay and postoperative pain.^13,14 However, many of these studies were selective for early malignancies, and most authors conceded these procedures as being particularly challenging, associated with lengthy operating times and an especially steep learning curve (Table 1).¹⁵ From this background, emerged the first reported robotic gastrectomy in 2003.¹⁶

Table 1.

Larger case series of laparoscopic gastrectomy. *TG: Total gastrectomy; SG: Subtotal gastrectomy

Study	n	Surgery *TG/SG	Operation time (mins)	Nodal yield	Blood loss (ml)	Complications (%)	Hospital stay (days)
Noshiro17 (2014)	160	0/160	315	40	115	10	13
Kim18 (2016)	288	0/288	230	34.1	–	9	7.4
Shen19 (2016)	330	75/255	226	31.3	212.5	10	10.6
Wang20 (2016)	145	53/92	192.4	30	152.8	10.3	6.4

Although different techniques are reported for robotic gastrectomy, most authors describe the placement of several ports in a similar position. A peri-umbilical port is used for camera placement in the supine patient. Normally, a further 4 (8–12mm) ports are placed -2 on the right and 2 on the left (Fig 1), using 3 robotic arms and a laparoscopic assistant. A thorough laparoscopy is performed and, if negative, dissection proceeds. In a total gastrectomy, an energy device is used to divide the greater omentum off the transverse colon, proceeding superiorly towards the left gastroepiploic and short gastric vessels, which too are divided. The hiatus is reached and the left crus mobilised. Attention then turns to the lesser omentum, which is divided with the right gastric vessels and the right crus mobilised, thus liberating the lower oesophagus. The remainder of the greater omen-tum is dissected off the transverse colon, and the duodenum (D1) is transected using a linear stapler. The stomach is lifted anteriorly to expose the left gastric vessels, which are either skeletonised and ligated, or divided using a vascular stapler. The field of dissection incorporates the D2 nodal stations, which are resected en bloc with the specimen, and the distal oesophagus divided. Although an intracorporeal anastomosis is possible, most authors describe a stapled extracorporeal oesophago-jejunal (Roux-en-Y) anastomosis through a mini-laparotomy made to retrieve the specimen. A subtotal gastrectomy is performed in a similar manner, aside from the short gastric vessels being spared and the hiatus not requiring mobilisation. The distal two-thirds of the stomach is transected and, perhaps reflecting a more robust anastomosis, an intracorporeal stapled or hand-sewn gastro-jejunal anastomosis fashioned (Roux-en-Y or Bilroth II). A small suprapubic or extended port-site incision is made to extract the specimen.

Figure 1.

Port placements for robot-assisted gastrectomy. The camera () is placed in the peri-umbilical port and a liver retractor is placed through the epigastric port (). Four further ports are placed: the first robotic arm in the left lateral position and robot arms 2 and 3 on the right side (). A laparoscopic assistant port is placed on the left, between the camera and the robotic arm (). An extended port site (total gastrectomy) or suprapubic (subtotal gastrectomy) incision is used to extract the specimen and may facilitate in constructing the anastomosis (---).

The advent of robotic gastrectomy has been quite different to laparoscopy because this procedure appears to minimise some of the complexities associated with the latter approach. In particular, more precise movements – coupled with tremor eradication plus better magnification with stereoscopic and 3D vision – facilitate better dissection, whereas the more ergonomic wrist movements of the robot arm make tasks like lymphadenectomy and anastomosis less exigent.¹⁵ Given that robotic systems remain expensive – and the abdomen is not as confined as, for example, the thorax – their use in gastrectomy remains limited to a few specialist centres, largely for earlier disease. Thus most reports are from the Far East where this cancer is more common and where enthusiasm for robotics has been more palpable. Short-term outcomes of more contemporary series are highlighted in Table 2. In the absence of randomised Level 1 evidence, these reports constitute the best measure of comparative practice. It is notable that in a shorter time-frame there are more reports relating to robotic gastrectomy compared with the laparoscopic approach.

Table 2.

Robotic gastrectomy series with main short-term outcome measures. *TG: Total gastrectomy; SG: Subtotal gastrectomy

Study	n	Surgery *TG/SG	Operation time (mins)	Nodal yield	Blood loss (ml)	Complications (%)	Hospital stay (days)
Kim21 (2012)	436	109/327	226	40.2	85	9.6	7.5
Liu22 (2013)	92	54/38	302/264	23.1	80.8	11.5	6.2
Park23 (2013)	200	46/154	248.8	37.9	146.1	10	8
Junfeng24 (2014)	118	26/92	234.8	34.6	118.3	5.8	7.8
Coratti25 (2015)	98	38/60	296	30.6	105.4	12.2	7
Lee26 (2015)	133	0/133	217.5	41.2	47	10.5	6.2
Kim27 (2016)	202	42/160	226	33	50	30	6
Shen19 (2016)	93	23/70	257.1	33	176.6	9.8	9.4

The majority of reports are limited to circa 200 patients, and most form comparative studies of a robotic vs a laparoscopic or open approach. One of the largest series was by Kim et al, comprising 436 patients.²¹ Of these, more than 327 underwent subtotal gastrectomy, but this early study revealed comparable or better outcomes than the other approaches. The median operating time was 226 minutes with a median blood loss of only 85ml. Lymph node harvest – a surrogate marker of oncological adequacy – was more than 40. Complication rates were below 10% and early discharge was achieved – both lower than open surgery, albeit not significantly. Given the large series, the authors did grant an improvement in outcome over time and with improving experience. One of the main consistent findings in most studies has been the longer operating time with a robotic approach. This is related to the additional preparation and docking time, and incorporates the learning curve.²⁸ Indeed, it has been suggested this curve plateaus at 20 cases, with a sequential improvement in operating time thereafter.²¹ Most studies (Table 2) report an operating time of more than 200 minutes, which – although longer than expected for open surgery – is equivocal or shorter than laparoscopy. The other commonly cited benefit of minimally invasive surgery is reduced intraoperative blood loss. Some studies report a loss as low as 44mls. Although difficult to ascertain, this is significantly lower than both open and laparoscopic surgery. Indeed, many authors relate this finding to better views and control of instruments, and an apparent easier ability to control bleeding with the robotic arms.^15,29 This is especially important during a D2 lymphadenectomy. The lower blood loss is considered an important factor given the postulated risk of dissemination of cancer cells with increased intraoperative bleeding.³⁰

The complication rate – and, therefore, hospital stay – is lower for robotic gastrectomy compared with open surgery, but remains similar to the laparoscopic approach. The latter likely relates to the number and similar size of incisions used, and some authors report a specific improvement in intra-abdominal sepsis and wound infections.^24,26,27 The lower morbidity is an important benefit, given the long-term impact this can have on physical, emotional and nutritional aspects of patient recovery.^3,4 However, most studies do not classify morbidity according to a standardised system, such as the Clavien-Dindo classification – thus results should be treated with caution. Indeed, readmission rates are often not cited – thus making earlier discharge more difficult to interpret. However, it makes sense that minimally invasive approaches reduce morbidity and promote earlier discharge; this has been true for many other robotic procedures. This appears especially true in specific groups, such as elderly patients and those with a high BMI.^15,26 This can have a long-term positive fiscal benefit too.¹⁵

With the robotic operation, lymph node yields are significantly higher than laparoscopy, and comparable with most open series.^19,21–27 Indeed, several subgroup analyses have confirmed better lymph node yield, specifically in the difficult supra-pancreatic and peri-splenic regions.³¹ The seven degrees of freedom and more precise movements, especially in tight spaces, are commonly cited reasons for this advantage with the robotic approach. It largely remains to be seen whether this correlates with comparable long-term survival, but initial studies do suggest a stage-for-stage equivalence when compared with open surgery.^25,27,28 In future studies, it may be found that – when analysed in conjunction with the impact of lower complications – overall survival rates are improved.

In summary, there is little doubt that the robotic approach is technically and oncologically safe to perform, and is likely more beneficial in terms of morbidity and recovery than other approaches. However, this is offset by the lengthier operating time, and the considerable learning curve. It can be safely assumed that the apprehensions relating to the laparoscopic approach have not been found with the robotic method; it appears that several of the obstacles inherent with the former have been overcome. Among the advocates of robotic gastrectomy, there is plenty of additional supportive anecdotal evidence. For example, dissection around the pancreas is more delicate with the robotic arm, resulting in lower rates of postoperative pancreatitis and pancreatic fistula.³² Similarly, intracorporeal suturing is reportedly easier.^15,20,22,29 However, given that robotic gastrectomy remains limited to a few specialist units, it is difficult to generalise these findings. It is clear that this is a positive surgical adjunct, but until robotic systems are more affordable and widely used, it is difficult to predict whether some excellent outcomes are more generally replicable.

Robotic oesophagectomy

Oesophageal cancer is increasing in incidence and represents the eighth most common malignancy worldwide, and the sixth commonest cause of cancer-related mortality.⁷ Survival generally remains poor and approximately 50% of patients have unresectable disease at presentation. Despite this, surgery remains the mainstay of curative treatment.

The enthusiasm for robotics has been far more evident for oesophageal surgery compared to gastrectomy, for several reasons. First, the features of delicate tissue-handling, improved degrees of movement and ergonomic force inherent to the robotic system lend themselves favourably to thoracic dissection, especially given the confined space.^33,34 Second, oesophagectomy involves two or three-compartment surgery, and the thoracotomy in particular predisposes to a higher rate of morbidity. Thus, there is a pressing need for a minimally invasive technique to be applied, promoting improved postoperative pain control and early mobilisation with smaller incisions. Finally, many randomised studies assessing laparoscopy demonstrated improved short-term outcomes compared with open surgery, but with several areas where a robotic approach would potentially further improve outcome. The MIRO trial, for example, demonstrated much-improved overall and respiratory complication rates with the addition of a laparoscopic abdominal phase, rather than a completely open procedure.³⁵ Similarly, the TIME trial compared totally minimally invasive oesophagectomy (laparoscopy and thorascopy) and conventional open surgery, and noted a 28% improvement in respiratory complications with the former. Furthermore, a shorter hospital stay (by three days) and better short-term patient satisfaction was achieved in the minimally invasive group. Oncological outcomes, anastomotic leak rates and margin positivity were equivalent.³⁶ However, the operations were noted to be exceptionally long and technically difficult.

In the minimally invasive era, there are two main types of operations that have been used to perform a robotic oesophagectomy. The first is a trans-thoracic approach (also known as Ivor Lewis oesophagectomy) for distal and middle-third oesophageal tumours. This is a two-stage procedure involving an abdominal phase and right thoracotomy. The anastomosis is constructed and lies within the chest. The second operation is a three-stage McKeown-type procedure involving an abdominal and thoracic phase, and a cervical incision to perform an anastomosis. Advocates of the latter procedure refer to the oncological advantage of removing the whole oesophagus with extended lymphadenectomy, with the placement of the anastomosis in the neck rather than the chest. Although leak rates in the neck are historically higher, they are perceived to be less morbid compared with a leak in the thorax. Of note, there are reports of the tranhiatal robotic approach, but these are usually reserved for early tumours or patients unable to withstand single-lung ventilation.^33,34,37

The abdominal phase is performed to mobilise the stomach and form a conduit, based on the right gastro-epiploic and right gastric vessels. Port placement is similar to the gastrectomy (Fig 1). Dissection begins in the gastro-colic ligament, separating the stomach from the transverse colon. Next, the short gastric vessels are divided and the left crus mobilised. On the right, the lesser omentum is divided, disconnecting the stomach and liver. The right crus is dissected and the lower oesophagus mobilised. Often a cuff of diaphragm and the pre-cardiac fat-pad are taken en bloc with the oesophagus to achieve a negative circumferential margin. Finally, the stomach is lifted supero-anteriorly and the left gastric vessels securely ligated. Given that no resection and specimen retrieval is required, many authors perform the abdominal phase laparoscopically and reserve the robotic system for the chest phase. This mainly relates to a faster operating time. For example, a Dutch team reported a reduction in operating time by almost four hours by adopting such a hybrid approach.³⁸

The thoracic phase is performed in either the left lateral decubitus or prone position. The prone position is considered to require less lung deflation, which in part contributes to the purported lower respiratory complication rates.^34,37 The lateral position is preferred by some because, in the unfortunate event of a major vascular injury mandating immediate conversion to open surgery, access is faster and more familiar. As a compromise, many authors use a semi-prone position.^33,34 Generally, most units use between 4 and 5 ports between 5–12mm (Fig 2); aside from the camera, 2 to 3 are used for the robot arms, with a single port used by the laparoscopic assistant. Depending on the type of operation, either a mini-thoracotomy or neck incision is used to extract the specimen and aid in forming the anastomosis.

Figure 2.

Port placements for robot-assisted oesophagectomy in the left lateral decubitus position. The camera () is usually placed in the eighth intercostal space. Three robot arms () are then placed in the third intercostal space, just above the diaphragm and postero-inferiorly – although these can be adjusted based on the anatomy of the thoracic cage in order to avoid clashing of the arms. Finally, a port for the assistant () is placed between the camera and first robot arm (although this can be swapped with the second robot arm).

The restricted, narrow space of the hemi-thorax, the mitigation of an exaggerated fulcrum effect, and the need for delicate dissection around major vascular and airway structures have lent themselves favourably to the robotic technique compared with thorascopy. However, akin to the robotic gastrectomy experience, most studies are confined to specialist centres with an interest in minimally invasive surgery. The first robotic oesophagectomy was described in 2003, and since then there have been several large series from various centres.³⁹ In 2007, Boone described 47 consecutive cases of a McKeown robotic approach with a 44.6% morbidity rate.³⁸ Other initial series also highlighted similar morbidity. The Utrecht (Netherlands) group, for example, in their early experience of 18 cases had a respiratory complication rate of 48%, an anastomotic leak rate of 14%, median blood loss of 400mls and an operating time of 450 minutes.⁴⁰ However, in a follow-up study of 108 cases 9 years later, the respiratory complication rate had fallen to 33%, blood loss reduced to 340mls, with an operating time of 381 minutes and an earlier hospital discharge.⁴¹ This example highlights how results do improve over time and with increasing experience.

There is little doubt that the robotic approach is technically and oncologically safe to perform, and is likely more beneficial in terms of morbidity and recovery than other approaches

More contemporary studies are highlighted in Table 3. These generally demonstrate better morbidity-related outcomes when compared with open surgery.^41–48 Longer-term data suggest that survival is comparable to established practice. As the studies highlighted in Table 3 reveal, lymph node yields are generally around 20. Although fewer than expected with open surgery, this has not manifested as a difference in disease-free survival. The resection margin negativity rate in all these studies is well above 90%; most have rates of above 98%. The operating times are variable, with 400 minutes representing a mean value. This is slightly longer than open surgery, but appreciably quickly than a combined laparoscopic and thorascopic approach. However, it should be mentioned that in some studies it is not clear whether the initial set-up, docking and patient position changing (between phases) time is included, and may account for some data where the procedure appears to be completed exceptionally fast. Intraoperative blood loss is likely lower than open surgery, with few reports of conversion to open surgery. This likely relates to the more careful dissection that the 3D view and magnification that is inherent in the robotic system facilitates. Furthermore, a scrubbed laparoscopic assistant is used routinely, where any bleeding can be controlled swiftly and repaired with the robotic instruments. In our experience, controlling major bleeding with a laparoscopic set-up without a robot is much more challenging and likely to indicate the requirement for conversion to open surgery. The overall complication rates are variable, but generally appear much improved compared with open surgery.

Table 3.

Short-term outcomes in contemporary reports of robot-assisted oesophagectomy. *Transhiatal approach; †Includes two cervical anastomosis; ‡ Includes chylothorax where data supplied.

Study	n	Anastomosis site	Operation time (mins)	Nodal yield	Blood loss (ml)	‡Respiratory complications (%)	Complications (%)	Anastomotic leak (%)	Hospital stay (days)
de la Fuente42 2013	50	Thorax	445	18.5	146	10	28	2	9
Puntambekar43 2015	83	Neck	205	18.4	86.8	6	19.3	3.6	10.4
van der Sluis41 2015	108	Neck	381	26	340	33	34	19	16
Cerfolio44 2016	85	Thorax	360	22	35	13	34	4.3	8
Park45 2016	114	Cervical	420	43.5	208	11.4	34	14.9	16
Chiu46 2016	20	Cervical	499	18	355	0	-	15	13
Dunn47 2017	100	*Cervical	264	17	75	11	-	16	8
Okusanya48 2017	25	†Thorax	661	26	250	32	-	4	8

Without the benefit of classification of morbidity, it is difficult to ascertain the true impact of this apparent lower complication rate. However, in most studies it is clear that earlier hospital discharge is also achieved, thus it is likely that the lower complication rate is a true effect of the minimally invasive approach. Discharge on day 8 following an oesophagectomy seems remarkable, but it should be noted that many North American units do institute a policy of early discharge with little or no oral intake and home jejunal feeding, followed by regular clinical reviews in the outpatient setting. In the UK, generally speaking, most units will ensure a complete resolution of complications, drain(s) removal and oral alimentation prior to discharge. However, there is no doubt that a minimally invasive approach with smaller incisions results in better postoperative pain management compared with a standard open thoracotomy, with downstream effects on mobilisation, drain and tube removal, and discharge. Similarly, the most common complication following a thoracotomy is respiratory infection. Atelectasis is common and intercostal drains are often cited as causing significant pain, affecting respiratory toilet. The respiratory complication rates in Table 3 are much lower than reported with open surgery.

The anastomotic leak rates, at initial glance, appear high in some cases. However, they are comparable with open surgery. The transthoracic approach with a chest anastomosis suffers a leak rate of 5–7% with open surgery and a cervivcal anastomosis carrying a rate of 15–20%.⁴⁹ Although most cervical anastomosis are hand-sewn, many authors favour a robotic-sewn anastomosis with the transthoracic approach. The freedom of movement and articulation of the robot wrists make this far easier than with conventional thorascopy. However, most authors still perform a stapled anastomosis in the chest, often using a mini-thoracotomy made to extract the specimen. It must be noted that the advent of high-quality circular staplers has greatly facilitated the minimally invasive approach.³⁷

The MIRO and TIME trials conclusively demonstrate improved in-hospital outcomes with a thorascopic approach, but many institutions failed to adopt this technique. In large part, this was because thorascopic oesophagectomy is a difficult and long operation. The latter point is especially important as some degree of single-lung ventilation is required, and this insult to pulmonary physiology often affects patient recovery. Furthermore, good open skills do not automatically translate to endoscopic skills, and learning a new procedure requires a great deal of motivation and patient acquiescence. However, the robotic thorascopic approach appears to overcome some of the difficulties encountered with the former approach. It reportedly has an easier learning curve, with as few as 17 cases required to attain sufficient proficiency.³⁴

The Utrecht group established the aptly named ROBOT trial in 2012, comparing open and robotic oesophagectomy.⁵⁰ One hundred and twelve patients were randomised and included in this study. Initial results have recently been presented, and indicate significantly reduced pulmonary, cardiac and overall complication rates, with reduced blood loss in favour of the robotic group. Importantly, they have also demonstrated improvement of objective postoperative pain scores in the robotic cohort, with better functional and quality of life scores at six weeks post-discharge. At 38 months median follow-up, oncological outcomes are equivalent. This well-designed, controlled trial represents a benchmark for robotic oesophago-gastric surgery.⁵¹ By demonstrating better short-term outcomes, including quality of life and a return to reasonable activity, a strong case for encouraging this technique is made. The avoidance of a thoracotomy and the morbidity it carries provides sufficient incentive to adopt a minimally invasive approach. The robotic-assisted procedure appears to be an effective and more readily learnable method to address this.

Future direction

There is sufficient evidence to suggest that robotic gastrectomy and oesophagectomy are, at least, equivocal to established practice – both in terms of postoperative and oncological outcomes. In emerging reports, including Level 1 and 2 evidence, the outcomes with robotic surgery are in fact superior. This is particularly evident for oesophagectomy, where the potential morbidity of a thoracotomy appears to some extent to be mitigated by this new approach. Indeed, many patients who were precluded from curative surgery on the basis of respiratory comorbidity may benefit from a robotic approach.

For a new procedure to replace established practice requires substantial motivation on the part of the surgeon and hospital management. Robotic surgery is not effortless to learn, access to training is limited, and robotic systems remain expensive. A cost-benefit analysis may not always be favourable, especially when the challenges of the initial learning curve are included. Within the construct of a resource-limited health service, the liberty of introducing such innovation is not always straightforward. Although the ROBOT trial has demonstrated a cost-saving of almost €4,000 per case when factors such as complications, length of stay and quality of life are incorporated,⁵¹ a broad, long-term approach is required, which is a challenging strategy to implement in the present financial environment.

Although previous laparoscopic experience is useful for the surgeon and wider theatre team, it does not automatically translate into good robotic skills. One of the concerns has been training of the next generation of surgeons. Oesophago-gastric cancer surgery is a small specialty, with resections limited to a few centralised units. Would such units attract high-quality trainees and senior fellows if the consultant cohort were themselves still learning how to perform a procedure? It is, perhaps, for these reasons, along with the considerable financial commitment, that robotic – and laparoscopic – oesophago-gastric practice remains limited to specific units. Within this context, it is difficult to generalise results; some excellent outcomes are not necessarily replicable. It must also be considered that much innovation with robotics – particularly with gastrectomy – emerged from the Far East, where factors such as healthcare system, body habitus and patient endorsement are different to UK practice.

As robotic surgery has demonstrated clinical benefit – particularly for oesophagectomy – it is evident that there will be wider acceptance of this technique in the coming years. This approach will be further supported by a reduction in costs through the introduction of newer and less expensive robotic systems. In a specialty with improving oncological outcomes, there is more scope now than previously to explore, innovate and adopt new operative approaches, supported by robust randomised evidence.