Is Robotic Surgery the Future for Resectable Esophageal Cancer?: A Systematic Literature Review of Oncological and Clinical Outcomes

This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) protocols in observational studies and randomised trials,¹⁹ and was registered on the international prospective register of systematic reviews (PROSPERO), registration number CRD42022370983. A review protocol was not prepared.

A search of the MEDLINE (PubMed), Embase and Cochrane databases was performed by two authors (NMP and PHP), identifying all studies published up to 1 August 2023. The Medical Subject Heading (MeSH) terms ‘robotic surgery’, ‘minimally invasive surgery’, ‘esophageal cancer’, and ‘outcomes’ were included. Conference proceedings and articles not published in English were excluded.

Two reviewers (NMP and PHP) screened articles independently by title and abstract before reading the full text of eligible studies. Relevant data including demographics and parameters on perioperative outcomes, including lymph node yield (LNY), anastomotic leak (AL) rate and length of stay (LoS), were collated.

Single-arm meta-analyses of oncological and clinical outcomes were performed using RStudio version 4.3.2 (Boston, MA, USA) [Table 1].²⁰ Weighted mean (95% confidence interval) and heterogeneity (I²) were calculated for all studies. Statistical significance was confirmed at p < 0.05. Forest plots were constructed for all outcomes, examples of which are demonstrated in Figs. 1, 2 and 3.

Bild vergrößern Bild vergrößern

Bild vergrößern

Bild vergrößern

Risk of bias was assessed using the modified Newcastle–Ottawa scale for non-randomised studies, and the modified Jadad scale for randomised trials.^21,22

Retrospective cohort (79%) was the most common study type, with four clinical trials eligible for inclusion.^1,24‐26 Several trials comparing robotic esophagectomy with open and conventional MIS are awaiting publication of the results;^27‐29 therefore, limited data on perioperative outcomes are currently available. Most studies were from North American (30%), and Chinese centres (20%). The wider adoption of robotic surgery throughout the United States reflects greater availability of robotic platforms and supporting infrastructure.³⁰ Furthermore, China accounts for nearly half the global disease burden of esophageal squamous cell carcinoma (ESCC), enabling centres to undertake more resections compared with the West.^31,32

Eighty-five (75%) studies reported a robot-assisted thoracic phase, with 64 studies (57%) performing both robotic abdominal and thoracic phases. A hybrid minimally invasive approach involving laparoscopic abdominal and robotic thoracic phases was reported in 19 studies (17%).^1,27,33 Open thoracic or abdominal phases were used in combination with robotic surgery in five studies (4%).^34‐38 Initially, the literature reported equivalent oncological outcomes and shorter procedure length in open esophagectomy when compared with thoraco-laparoscopic approaches.¹³ Therefore, many surgeons may lack experience in thoraco-laparoscopic esophageal resection, moving immediately to the robotic console without developing skills in what may be perceived as an ‘intermediate step’ in MIS.^14,34

Three studies reinforce the notion of learning curves associated with developing proficiency with novel surgical technologies, manifested by analysing learning curves in robotic esophagectomy.^39‐41 These identified the mean number of cases required before surgeons experienced significant improvements in outcomes. Park et al. suggested a change point of 28 robot-assisted esophagectomies for an observed increase in LNY from 25 to 45 (p < 0.001);³⁹ however, other factors, including marked reduction in the incidence of complications, for example reduction in AL rate, were reported after 80 and 85 cases, respectively. These findings are supported by the cumulative sum (CUSUM) learning curves derived by Kingma et al., where 22 robotic esophagectomy cases were performed before a plateau in estimated blood loss (EBL) and operative time was noticed, suggesting certain components of the procedure take a greater number of cases for expert credentialing.⁴⁰

Robotic esophagectomy has been associated with longer operating times than open surgery.¹ This is partly due to time spent ‘docking’ instruments, requiring familiarisation of theatre teams with the robotic platform. This literature review reported a weighted mean operative time of 372.16 min (range 168–808 min) for robotic esophageal cancer resections, taken as the total operating time and not solely time spent on the robotic console.

Kingma et al. identified that after 23 cases, surgeons noticed a reduction in operating time for both the thoracic and abdominal phases, plateauing at case number 70.⁴⁰ Park et al. confirmed that temporal improvement is seen with accumulated experience, but this occurred after 80 cases.³⁹ It can be hypothesised that with greater experience comes reduced operating times, reiterating the presence of a learning curve. With sufficient experience, centres may then be able to match higher-volume American and Chinese units.^42‐49

An advantage of the robotic platform is the ability to perform finer dissection within challenging anatomical areas, with reduced EBL and rate of visceral injury.⁵⁰ Minimising these complications may allow for shorter recovery times and reduced length of hospitalisation.

There is a significant, multifactorial physiological stress response to major surgery, and open esophagectomy has a significant impact on patients.⁵ The degree of postoperative haemodynamic and respiratory support required typically results in admitting patients to Level 1 care postoperatively.⁵¹

Weighted mean critical care and total inpatient LoS for the included studies were 1.92 days (range 0.85–23) and 15.2 days (range 7–24), respectively. This indicates significant variation among units, which may be associated with perioperative complications. In their single-centre cohort study of 321 patients, Angehern et al. reported shorter duration of hospitalisation (18.5 days) among their open esophagectomy cohort compared with those who had robotic procedures (20 days, p = 0.368).³⁴ This contradicts the notion that MIS is associated with shorter LoS. However, given that this is a novel surgical technology, the surgeons may have felt inclined to keep patients in under observation for longer, in anticipation of delayed postoperative complications. This is despite reduced rates of re-intervention among the robotic cohort (5.3%) compared with patients who had open surgery (7.9%).³⁴ Pneumonia is a common cause of morbidity after esophagectomy and poor pain control is a major causative factor.^42,52 Smaller incisions required in robotic esophagectomy and the reduced nerve injury result in less pain after surgery, better respiratory effort and reduced risk of pneumonia. Tsunoda et al. demonstrated a lower rate of pulmonary complications (18%, p = 0.006) among patients who underwent robot-assisted esophagectomy compared with conventional minimally invasive esophagectomy (44%).⁵³ The incidence of postoperative pneumonia ranged from 0% to 45.4%; however, the literature varied in its definition and criteria influencing treatment decisions.⁴² Notably, Meredith et al. reported no significant difference in the incidence of pneumonia between patients undergoing open, robotic or conventional minimally invasive esophagectomy.⁵² In comparison, three studies reported an incidence of postoperative pneumonia of >30%, despite all patients undergoing laparoscopic and robotic phases, suggesting a multifactorial aetiology for postoperative pneumonia.^33,54,55

Twenty studies reported total inpatient LoS of < 10 days; none of these studies used an open approach for the thoracic or abdominal phases.^{25,41,43,44,52,56‐69} In comparison, 27 studies reported total LoS > 14 days, of which six used an open approach in either the abdominal or thoracic phases.^{5,24,25,33,42,45,46,55,70‐85} This indicates the robotic and conventional minimally invasive approaches are associated with shorter LoS, however clinical trials are required to validate this.^1,26,27

Factors contributing to reduced LoS include less postoperative pain and nausea, earlier introduction of oral intake, and mobilisation.^5,81,86 Enhanced Recovery After Surgery (ERAS) programmes or ‘fast-track protocols’ in elective upper GI resection have led to improvements in patient outcomes, including LoS and postoperative morbidity, by implementing a standardised pathway for patients and care providers.^87,88 No studies commented on the use of ERAS. Since robotic esophagectomy is a recent adoption for many units, surgeons may implement a tailored postoperative recovery programme instead of a goal-orientated ERAS pathway.⁸⁹

Twenty-eight studies reported on rate of reoperation, ranging from 0 to 35%. This may reflect varying levels of experience with robotic esophagectomy, and may also be explained by the availability of endoscopy and interventional radiology, which could be used as alternatives to manage selected complications.²⁸ Of note, when comparing the open approach with all minimally invasive approaches, the rates of return to theatre did not differ significantly.^34,52,61

Additional comparative perioperative measures, including time to mobilisation, quantitative data on postoperative pain, and hospital readmission, would be beneficial to describe tangible representative outcomes across studies.

Weighted mean EBL across the included studies was 197.7 mLs (range 35–598 mLs), however blood loss per operative phase was not specified. The two studies with the lowest mean blood loss, 35 mLs in total, are also the two where a totally minimally invasive esophagectomy was performed.^63,78 These studies highlight key advantages offered by MIS through smaller incisions and reduced surgical trauma, giving robotic surgery the advantage over open approaches in resectable esophageal cancer.

Negative resection margins (R0) and LNY were collated to assess perioperative oncological outcomes of the included studies. Aside from reduced postoperative pain and shorter LoS, local disease control must be achieved to potentially improve OS and reduce the chances of recurrence. Although resection margin involvement in the surgical specimen was reported by 87 studies (77.0%), it was not specified whether this related to longitudinal or circumferential margins (CRMs). Median positive margin status from the included studies was 3.48%, demonstrating high rates of ‘curative’ resection were achieved with robotic surgery. In comparison, six studies reported positive resection margin rates of 10% or higher.^{40,56,61,90‐92}

Comparisons with national registries should be performed for contextualisation. The UK National Oesophago-Gastric Cancer Audit reported a 4.2% positive longitudinal and 20.3% positive CRM status for all esophageal resections performed from April 2018 to May 2021.⁹³ Eight years of data from the Dutch Upper Gastrointestinal Cancer Audit (DUCA) reported a positive CRM rate ranging from 3.7 to 6.8%, noting the higher utilisation of neoadjuvant chemoradiotherapy, and 8.7% in Swedish registries.^51,94 The higher R0 rate reported in this review compared with contemporary registry data, suggests the technical benefits offered by the robotic platform may contribute to greater R0 rates by improving dissection in difficult anatomical locations.⁵⁰ However, other factors, including access to neoadjuvant therapies and disease stage at presentation, may also impact on achieving clear resection margins.

Furthermore, case selection may influence reported outcomes, especially for centres new to performing robot-assisted esophagectomy. Less complex cases may be chosen when testing a novel technique, which may influence outcomes, including R0 resection rate.

None of the included studies reported on CRM status in the resected esophageal specimens. The literature has highlighted the importance of CRM as an independent prognostic factor for local disease recurrence and survival in esophageal cancer.^95‐98 Although the literature suggests that robotic platforms can improve perioperative outcomes, including pulmonary complications and LoS, oncological outcomes are crucial to improving survival for potentially curative disease and should be recorded as standard practice.³⁶

Standardised lymphadenectomy is a key component of esophagectomy for accurate disease staging, local disease control and prognostication for OS.⁹⁹ Current guidelines indicate at least 15 LNs must be submitted for pathological examination according to the National Comprehensive Cancer Network (NCCN) and National Oesophago-Gastric Cancer Audit (NOGCA);^93,100,101 however, standards of and experience in histopathological analysis may vary between centres. This is reflected by the studies included in this review, with a range in LNY of 8–69 nodes, despite a median of 25 nodes. This suggests a significant variability in the extent of lymphadenectomy performed in robotic esophagectomy. Four studies have highlighted that the quality of lymphadenectomy in thoracoscopic esophageal resection is inferior to robotic surgery or thoracotomy, which may explain why uptake of thoracoscopy in the esophagectomy is limited, especially now that robotic surgery is increasingly available.^73‐75,82

Three studies reported significantly lower LNY than the median and the recommended minimum.^14,102,103 Of these, Washington et al. also reported a positive resection margin rate of 5.56%, above the median of 3.48%.¹⁴ Furthermore, they reported equivalent LNY when comparing laparoscopic (13.9) and robotic (14.3) esophagectomy.¹⁴ This was corroborated by Zhang et al., i.e. 19.1 nodes during thoraco-laparoscopic McKeown esophagectomy compared with 19.3 in robotic.⁵⁸ This highlights the importance of following key principles of oncological surgery. In particular, that quality of lymphadenectomy should not be compromised when using a novel surgical technology, even though said new technology may offer other benefits to patients.

Factors influencing the use of oncological therapies include prevalence of different tumour types and recognised standard of care among units. This systematic review highlights international variation in practice, for example, giving definitive chemoradiotherapy in ESCC followed by salvage esophagectomy, versus neoadjuvant chemotherapy followed by surgery.^{26,36,46,54,67,70,82,91,92,101}

The published literature supports the use of adjuvant therapy after neoadjuvant treatment and esophagectomy with clear resection margins, citing an improved OS up to 5 years.^104,105 However, just 14 studies (12%, n = 380 patients) reported giving adjuvant therapy. Perioperative therapy, followed by a radical robotic esophagectomy with clear resection margins, without postoperative complications, may enable patients to proceed on to complete adjuvant therapy, improving OS and RFS.^105,106

Reported morbidity in esophagectomy can be as high as 50%, with AL and chyle leak (CL) responsible for the greatest risk of prolonged hospitalisation and mortality.^49,107 Weighted mean reported AL and CL rates were 10% and 4%, respectively.

Fifty-one studies (45%) specified the type of anastomosis created when reporting AL rates. However, AL rate did not vary considerably between circular (8.55%), linear stapled (8.75%) or hand-sewn (8.6%) anastomoses. Six studies reported performing either a robot-assisted hand-sewn or stapled intrathoracic esophagogastric anastomosis, with a leak rate ranging from 0 to 16%.^{58,66,90,108‐110} Five studies were carried out in American and Chinese institutions; both were associated with more experience in robotic esophagectomy, which may explain their lower AL rates. In comparison with established national registries, the DUCA reported incidence ranging from 18.2 to 19.3% for all intrathoracic and cervical anastomoses, regardless of technique, and the UK Oesophago-Gastric Anastomosis Audit (OGAA) reported rates of 12.2% and 20.1% for intrathoracic and cervical anastomoses, respectively.^51,111 As with other outcomes of robotic esophagectomy, volume and experience in performance of the procedural steps directly influence outcomes.

Although most studies reported performing a hand-sewn or stapled extracorporeal anastomosis, there was no appreciable difference in the AL rate between the two subgroups.^38,103,112 Exteriorising the proximal esophagus and gastric conduit to form a hand-sewn or stapled anastomosis remains the preferred means of restoring continuity as it is technically less challenging than an anastomosis formed entirely within the body cavity through minimally invasive approaches.⁸³ Follow-up data from the included studies did not report the incidence of anastomotic strictures and therefore it was not possible to make further comparisons between techniques.

Incidence of CL was reported by 72 studies (64%). Although the average reported rates ranged from 1.1 to 3.8%, the incidence of CL among the included studies was as high as 29%.^33,107 As suggested by Dezube et al., experience may be the determining factor influencing the risk of CL in esophagectomy. As such, this may be an important parameter to assess for competence in performing robotic esophagectomy in learning curve analyses alongside parameters including operative time.^39,107 The ramifications of a persistent CL are associated with infection, electrolyte imbalance, hypovolaemia, and nutritional derangement, causing prolonged hospitalisation, delayed oral intake and impact on QOL.

A total of 51 and 10 studies reported on two- and three-stage esophagectomy using a robotic platform for both abdominal and thoracic phases, respectively. More cases of ESCC were managed with three-stage esophagectomy (25 cases) than two-stage (21 cases), consistent with the preponderance of SCC in the proximal esophagus.¹¹³ Table 6 highlights that robotic three-stage esophagectomy was associated with longer average operating time, and greater blood loss and AL rate compared with two-stage procedures. The addition of a third phase may explain the prolonged time taken to perform this resection and the associated higher blood loss.

A robotic cervical phase was reported in four studies, compared with six studies performing an open lymphadenectomy and anastomosis. GI surgeons may begin developing robotic skills by operating within the abdominal cavity, an area more familiar to them given likely previous experience with laparoscopy, before progressing to the thorax. However, uptake of robotic surgery for neck pathology and the cervical esophagus is currently limited according to the published literature.^18,114 This may explain the greater use of an open approach among the included studies.

In four studies, a stapled esophagogastric anastomosis was formed within the cervical wound to restore continuity of the digestive tract.^42,44,46,115 In comparison, two studies reported a hand-sewn anastomosis—one formed using the robotic platform and one extracorporeal.^42,115 The reported AL rate was higher among those who underwent three-stage esophagectomy (12.5%) compared with two-stage (8.10%). This fits with the reported literature, that the incidence of AL is lower in two-stage than three-stage esophagectomy, regardless of whether open, thoraco-laparoscopic or robotic procedures are performed.¹¹⁶

This is a comprehensive review of outcomes of robotic esophagectomy for resectable esophageal cancer, as evidence by the evaluation of over 100 studies that fulfilled the inclusion criteria. The range of key clinical parameters analysed cover the entirety of the patient’s hospital admission. Assessment of oncological outcomes scrutinises the potential benefits of robot-assisted esophageal resection further by taking into consideration the impact of radical surgery and lymphadenectomy on disease- and recurrence-free survival. Meta-analysis of clinical and oncological outcomes objectively validates the findings of the included studies; however, as demonstrated in Table 2, there was a significant degree of heterogeneity between the included studies in a number of outcomes. Uptake of robotic surgery is not consistent internationally and this is reflected in the reported outcomes. Furthermore, there were only four clinical trials on robotic esophageal cancer resection. These may limit the conclusions that can be drawn from current evidence. Data from prospective trials comparing open, thoraco-laparoscopic and hybrid procedures with robot-assisted esophagectomy are therefore required in order to make more definitive conclusions on the advantages of robotic surgery for resectable esophageal cancer.