Hiatal hernia after open versus minimally invasive transthoracic oesophagectomy for cancer

Abstract

Background

Hiatal hernia (HH) after oesophagectomy is a potentially life-threatening complication, more commonly observed after minimally invasive procedures. The aim of the study was to compare the incidence of HH after open versus minimally invasive oesophagectomy (MIO) for cancer, to identify risk factors for its onset, and analyse the technical differences between the approaches.

Methods

This was a retrospective study of patients who underwent transthoracic oesophagectomy for cancer over a 15-year period. Open and minimally invasive procedures were compared according to demographics, and operative and perioperative parameters. MIO included both laparoscopic and robotic operations. Risk factors for HH after oesophagectomy were analysed by calculating odds ratios of uni- and multivariable generalized linear models.

Results

A total of 898 patients operated on between 2008 and 2023 were included in the study. HH was observed in 1 of 490 patients (0.2%) in the open group and in 21 of 408 patients (5.2%) in the minimally invasive group (P < 0.001). At multivariable analysis, patients with an ASA score of II and III within the MIO group had a significantly lower risk of HH compared with ASA I subjects (P = 0.002 and P < 0.001, respectively). Omentectomy was performed in all open procedures but in none of the MIO.

Conclusion

The rate of HH was significantly lower in patients who underwent open oesophagectomy. Omentectomy may prevent postoesophagectomy HH as it was the only additional technical difference between the groups. Multicentric randomized clinical trials are needed to assess whether omentectomy during MIO may reduce the occurrence of paraconduit HH.

Hiatal hernia (HH) after oesophagectomy is more commonly observed after minimally invasive (MI) procedures; however, its physiopathological mechanism is still unclear and controversially discussed. After Institutional Review Board approval, open and MI procedures were compared according to several parameters. Risk factors for HH after oesophagectomy were analysed. A total of 898 patients were included. HH was observed in 1 of 490 patients (0.2%) in the open group and in 21 of 408 patients (5.16%) in the MI group (P < 0.001). Omentectomy was performed in all open procedures but in none of the MI operations. Despite the MI approach, omentectomy may play a role in the prevention of postoesophagectomy HH as in the authors’ practice it was the only additional technical difference between the groups.

Introduction

Hiatal hernia (HH) after oesophagectomy for cancer is a rare but potentially life-threatening complication with an overall weighted incidence of 5%¹. This delayed complication is significantly more common after minimally invasive (MI) procedures^2–4. In a recent systemic review with a meta-analysis of 19 included studies by Murad H et al., the incidence of postoesophagectomy hiatal hernia (PHH) ranged from 0 to 10% for open oesophagectomy (OO) and from 1 to 26% for MI oesophagectomy (MIO)¹. In recent years, several potential mechanisms and risk factors have been identified. These include fewer adhesions after MIO compared with OO, younger age, upfront surgery, pre-existing HH on a pre-operative computed tomography (CT) scan, previous hiatal surgery, tumours of the esophagogastric junction, and neoadjuvant radiochemotherapy (nRCT)^3–5. Despite these findings, the pathogenesis of PHH remains unknown and is likely multifactorial. Fetzner and colleagues^6,7 have proposed several ‘good clinical practice’ measures to prevent PHH. These include avoiding extensive radical incision of the diaphragmatic crura unless necessary, avoiding opening the left pleural cavity, leaving sufficient omentum on the greater curvature to fill the space in the mediastinum, and performing an anterior hiatoplasty from the abdomen to be completed from the chest during the thoracic phase. However, there are currently no retrospective or prospective results available to prove their feasibility, safety, and efficacy, and hence they can only be considered as experts’ opinions.

At the authors’ centre, MIO has been implemented as the standard of care for patients with resectable oesophageal carcinoma, with laparoscopic techniques introduced in 2014 and robotic-assisted techniques in 2016⁸. The aim of this retrospective cohort study was to investigate the incidence of PHH after both oncological OO and MIO in the authors’ centre and to identify risk factors for its onset. Furthermore, the authors aimed to investigate potential mechanisms, particularly technical variations that might account for the differences in PHH incidence between the open and MI approaches.

Methods

Study criteria and population

This retrospective single-centre cohort study was conducted at the tertiary surgical centre of Nuremberg University Hospital (Klinikum Nürnberg, Paracelsus Medical University), Nuremberg, Germany, in accordance with the Declaration of Helsinki. After approval by the local Institutional Review Board (IRB number: FMS_W_015.24-I-1, IRB approval date: 25 March 2024), the authors queried their prospectively maintained oesophageal database for patients who underwent transthoracic oesophagectomy for cancer (both Ivor Lewis and McKeown oesophagectomy) between January 2008 and July 2023. Informed consent was waived. Patients aged 18 and older were included in the study. Patients who underwent transhiatal extended gastrectomy or oesophagectomy for benign diseases were excluded.

Patients were divided into open and MI groups accordingly to the chosen abdominal approach. The MI approach included both laparoscopic- and robotic-assisted operations. If conversion to laparotomy was needed, the patient was included in the open group.

The groups were compared according to the following 28 demographic, operative, histologic, and postoperative parameters: age, sex, American Society of Anesthiologists (ASA) physical status classification, pre-existing cardiac condition, pre-existing pulmonary condition, history of smoking, history of alcohol abuse, neoadjuvant treatment (both nRCT and chemotherapy), neoadjuvant chemotherapy alone, type of anastomosis (for example, cervical versus intrathoracic), harvested lymph nodes, histology, TNM-pT ≤ 2 versus TNM-pT > 2, TNM-pN ≤ 2 versus TNM-pN > 2, TNM-M0 versus TNM-M1, R-status, simplified pathological Union for International Cancer Control (pUICC) stage, in-hospital mortality, anastomotic leak, early anastomotic leak (within postoperative day 5), postoperative bleeding, chylothorax, pneumonia, sepsis, pulmonary embolism, myocardial infarction, arrhythmia, and need for blood transfusion.

A pre-existing cardiac condition was defined as any cardiac illness diagnosed before surgery, including arrhythmia, conditions following a myocardial infarction, conditions following coronary bypass surgery, and chronic cardiac failure.

A pre-existing pulmonary condition was defined as any pulmonary illness diagnosed before surgery, such as chronic obstructive pulmonary disease, asthma, lung fibrosis, conditions following major lung resection, and conditions following lung cancer treatment.

Neoadjuvant treatments comprised chemotherapy alone and nRCT, tailored according to the tumour type (histology) and as recommended by the local tumour boards.

For study purposes, the pUICC stage was simplified (sUICC) to 0 (full responder), I, II, III, and IV.

Within the MIO group, 23 potential risk factors for PHH were analysed: age, sex, pre-existing cardiac condition, pre-existing pulmonary condition, history of smoking, history of alcohol abuse, ASA classification, neoadjuvant therapy (RCT or chemotherapy), McKeown operation, harvested lymph nodes, R-status, histology, pT, pN, intensive care unit stay, hospital stay, anastomotic leak, postoperative bleeding, chylothorax, pneumonia, pulmonary embolism, arrhythmia, and need for blood transfusion.

PHH was identified with a CT scan during the follow-up. Every patient with a PHH in this study was sent for HH repair to the authors’ centre.

As far as available, the CT scans of the MIO group performed before HH repair were independently reviewed by two authors (L.B. and L.G.) in order to define in how many patients the omentum was herniated.

Operative technique (abdominal phase)

In the open approach, a transverse laparotomy is performed above the umbilicus. A large retractor ring with four valves and a rib retractor is positioned, followed by abdominal exploration. An oesophagogastric tube (42 Fr) is placed in the oesophagus for guidance. Dissection starts at the greater curvature, dividing the gastrocolic ligament in the anatomic avascular plane beneath the greater omentum. The short gastric vessels are then divided and a subtotal omentectomy is performed (Fig. 1a), preserving the gastroepiploic arcade, in order to permit adequate pull-up of gastric conduit later on. Both diaphragmatic crura are identified, and the oesophagus is circularly mobilized. If oncologically indicated, a partial en bloc resection of the diaphragmatic muscle is conducted for an R0 resection margin. The lesser omentum is mobilized, the right gastric vessels are divided, and an en bloc D2/D2+ lymphadenectomy is performed. The left gastric vessels are ligated and a gastric tube is fashioned using an Echelon™ 60 mm, gold load (Ethicon, Johnson & Johnson, New Brunswick, NJ, USA) three-row linear stapler. The gastric conduit is completed in the chest after pull-up. Feeding jejunostomy is only placed for patients undergoing the McKeown procedure. For instance, in the case of a McKeown procedure the gastric conduit is pulled up anatomically through the hiatus (for example, following the former oesophageal route).

Fig. 1.

Oesophagectomy performed with the open and minimally invasive oesophagectomy approach

Whereas during open oesophagectomy the omentum is resected (a), this is left in place attached to the transverse colon during a minimally invasive oesophagectomy (b). The remaining part of the greater omentum attached to the transverse colon may migrate upwards to the hiatus (c) and trigger or facilitate a colonic pull-up, hence a postoesophagectomy hiatal hernia.

The laparoscopic (or robotic) approach uses a five-port technique, as previously described by the authors⁸. Dissection begins at the lesser omentum, identifying the right crus. The oesophagus is partially mobilized from the diaphragm. If diaphragmatic infiltration is suspected, a partial en bloc resection is performed, similar to the open technique. The lesser omentum is mobilized, an en bloc D2/D2+ lymphadenectomy is performed, and the left gastric vessels are divided. Next, the greater curvature is mobilized. Unlike the open approach, the gastrocolic ligament is divided through the greater omentum below the gastroepiploic arcade, preserving it while leaving the greater omentum attached to the transverse colon (Fig. 1b). Short gastric vessels are divided, and mobilization of the hiatus is completed. The gastric conduit is created. In robotic-assisted MI oesophagectomy, the gastric conduit is completed in the abdomen and reattached to the specimen with a suture. In patients converted to laparotomy, a subtotal omentectomy is performed.

Statistical analysis

Data were checked for consistency and normal distribution. Continuous variables were described using mean(standard deviation (s.d.)) and compared using a generalized log-normal model. Fisher's exact test or Pearson's χ² test was used to compare percentages of categorial variables. Univariable generalized linear models based on the binomial distribution were used to analyse risk factors, and corresponding odds ratios (ORs) with 95% confidence intervals (c.i.) were used to estimate effect sizes. A backward stepwise variable selection algorithm was applied for multivariable modelling of risk factors together with multivariable ORs and 95% c.i. If necessary, bootstrap t-tests based on Monte Carlo simulations were used to analyse data. All reported tests were two-sided, and P < 0.05 was considered statistically significant. All statistical analyses in this report were performed using NCSS™ (NCSS™ 2022, NCSS™, LLC. Kaysville, UT, USA) and STATISTICA™ 13 (Hill, T. & Lewicki, P. Statistics: Methods and Applications. StatSoft, Tulsa, OK, USA).

Results

General demographics

In total, 898 subjects were included in the study. Mean(s.d.) age was 64.4(10.2) years. There were 738 (82.3%) male and 159 (17.7%) female patients. Adenocarcinoma was the most represented tumour type with 652 (72.6%) cases, followed by squamous cell carcinoma with 228 (25.4%) cases. There were 2 (0.2%) gastrointestinal stromal tumours and 16 (1.7%) other malignant entities.

Open versus MI surgery

Demographic and perioperative parameters

There were 490 (54.7%) patients in the open and 408 (45.3%) patients in the MI group. Of note, there were 25 (2.7%) conversions from MI to open technique during the abdominal phase and 24 (2.6%) conversions in the thoracic phase. Mean(s.d.) age was similar, with 64.0(10.4) and 64.9(10.0) in the open and MI groups, respectively. There was no significant difference between the open and the MI groups with regards to ASA classification, pre-existing cardiac condition, pre-existing pulmonary condition, and history of alcohol abuse. History of smoking was more common in the MIO group.

The proportion of patients undergoing neoadjuvant treatment (regardless of nRCT or chemotherapy) was higher in the MI (58.9%) compared with the open (50.1%) group. Neoadjuvant chemotherapy without radiation was administered in 95 (19.4%) and 90 (22.1%) patients in the open and MI groups, respectively (P = 0.30). A cervical anastomosis was performed more often in the open group, whereas the number of harvested lymph nodes was higher in the MI group. Please refer to Table 1 for details.

Table 1.

Demographic and perioperative parameters of open versus MIO group

	Open (n = 490)	MIO (n = 408)	P*
Age (years), mean(s.d.)	64.0(10.4)	64.9(10.0)	0.21
Sex
Male	400 (81.3%)	339 (83.4%)	0.57
Female	90 (18.7%)	69 (16.6%)
ASA category
I	30 (6.1%)	28 (7.0%)	0.22
II	275 (56.1%)	195 (48.9%)
III	177 (36.1%)	170 (42.6%)
IV	7 (1.4%)	6 (1.5%)
V	1 (0.1%)	0 (0.0%)
Missing	0	9
Pre-existing cardiac condition	276 (56.1%)	226 (55.7%)	0.89
Pre-existing pulmonary condition	137 (27.8%)	99 (24.4%)	0.24
History of alcohol abuse	44 (9.0%)	51 (12.6%)	0.08
History of smoking	171 (34.9%)	207 (51.5%)	< 0.001
Neoadjuvant treatment	246 (50.2%)	239 (59.1%)	0.008
Upfront surgery	244 (49.7%)	166 (40.9%)
Missing	0	3
nCT	95 (19.4%)	90 (22.1%)	0.30
Cervical anastomosis	118 (24.0%)	35 (8.5%)	< 0.001
No. of LN, mean(s.d.)	27.4(11.1)	24.9(11.2)	< 0.001

MIO, minimally invasive oesophagectomy; s.d., standard deviation; ASA, American Society of Anesthesiologists; nCT, neoadjuvant chemotherapy; LN, lymph nodes. *Bootstrap t-tests and Pearson´s χ2 test were used to obtain P-values.

Histology and stage

Distribution of the histological subtypes was comparable between the open and MI groups. A pT ≤ 2 stage was observed in 53.4% and 58.8% of patients in the open and MI groups, respectively. A pN ≤ 2 stage was observed significantly more often in the MI group (P = 0.005). Patients with oligometastatic disease (for example, pM1) were treated more commonly with open surgery. A R0 resection margin was achieved in 94.3% and 96.7% of patients in the open and MI groups, respectively (P = 0.19). Regarding the pathological sUICC, MIO was used more often in earlier tumour stages. Please refer to Table 2 for details.

Table 2.

Histologic parameters and stages of open versus MIO group

	Open (n = 490)	MIO (n = 408)	P*
Adenocarcinoma	337 (68.8%)	319 (78.2%)	0.17
SCC	145 (29.6%)	84 (20.6%)
GIST	2 (0.4%)	0 (0.0%)
Other malignant	6 (1.2%)	5 (1.2%)
pT status
≤ 2	261 (53.4%)	237 (58.8%)	0.1
> 2	228 (26.6%)	166 (41.2%)
Missing	1	5
pN status
≤ 2	426 (87.8%)	379 (93.6%)	0.005
> 2	59 (12.2%)	26 (6.4%)
Missing	5	3
pM1	41 (8.4%)	12 (3.0%)	< 0.001
R-status
R0	446 (94.3%)	380 (96.7%)	0.19
R1	23 (4.9%)	13 (3.3%)
R2	3 (0.6%)	0 (0.0%)
Missing	18	15
sUICC status
0	51 (10.4%)	64 (16.1%)	0.01
I	135 (27.7%)	86 (21.7%)	0.04
II	98 (20.1%)	107 (27.0%)	0.02
III	164 (33.6%)	118 (29.7%)	0.24
IV	40 (8.2%)	22 (5.5%)	0.15
Missing	2	11

MIO, minimally invasive oesophagectomy; SCC, squamous cell carcinoma; GIST, gastrointestinal stromal tumour; sUICC, simplified UICC. *Pearson´s χ2 test.

Postoperative parameters and complications

Regarding postoperative complications, there were significant differences for sepsis and blood transfusion requirements, favouring the MI group over the open group. No significant differences were observed for anastomotic leak, early anastomotic leak, postoperative bleeding, chylothorax, pneumonia, pulmonary embolism, myocardial infarction, and cardiac arrhythmia. Of note, in-hospital mortality was lower in the MI group. Findings are summarized in Table 3.

Table 3.

Postoperative parameters and major complications of open versus MIO group

	Open (n = 490)	MIO (n = 408)	P*
Hiatal hernia	1 (0.2%)	21 (5.2%)	< 0.001
In-hospital mortality	12 (3.0%)	36 (7.3%)	0.003
Sepsis	82 (16.7%)	37 (9.1%)	< 0.001
Need for blood transfusion	187 (38.1%)	85 (20.1%)	< 0.001
Anastomotic leak	55 (11.2%)	46 (11.3%)	0.65
Postoperative bleeding	16 (3.3%)	13 (3.2%)	0.96
Chylothorax	15 (3.0%)	19 (4.7%)	0.20
Pneumonia	144 (29.3%)	100 (24.6%)	0.11
Pulmonary embolism	7 (1.4%)	3 (0.7%)	0.33
Myocardial infarction	2 (0.4%)	1 (0.2%)	0.68
Cardiac arrythmia	99 (20.1%)	84 (20.7%)	0.83

MIO, minimally invasive oesophagectomy.

PHH in open versus MI approaches

A PHH occurred in 1 of 490 patients (0.2%) in the open group and in 21 of 408 patients (5.2%) in the MI group (OR 26.6, 95% c.i. 3.6 to 198.7; P < 0.001). Notably, the patient in the open group required HH repair 21.9 months after the primary surgery, whereas in the MI group, HH repair was needed at a median of 2.6 (interquartile range 0.6–6.3) months after oesophagectomy.

Risk factors for hiatal hernia after MIO

Both univariable and multivariable analyses showed that, within the MIO-group, patients with ASA II and ASA III had a significant lower risk of developing a PHH compared with ASA I patients, whereas there was no significant difference between ASA I and ASA IV. All other 22 investigated risk factors were not significantly correlated with occurrence of PHH after MIO (P > 0.05). Please refer to Table 4 for details.

Table 4.

Risk factors for HH after MIO

	Hazard ratio	P*
Univariable analysis
Age	0.9 (0.9, 1.0)	0.06
Sex	0.8 (0.2, 2.8)	0.73
Pre-existing cardiac condition	0.9 (0.4, 2.3)	0.88
Pre-existing pulmonary condition	0.9 (0.3, 2.7)	0.94
History of smoking	0.7 (0.3, 2.8)	0.55
History of alcohol abuse	2.9 (0.4, 22.7)	0.26
ASA status
II	0.2 (0.06, 0.5)	0.002
III	0.1 (0.03, 0.5)	< 0.001
IV (ASA I is reference)	0.7 (0.07, 7.5)	0.79
Neoadjuvant therapy   (RCT or CT)	0.7 (0.3, 1.8)	0.45
McKeown operation (Ivor Lewis is reference)	†	0.14
Harvested lymph nodes	1.0 (0.9, 1.0)	0.65
R-status	†	0.39
SCC histology (adenocarcinoma is reference)	†	0.08
pT status
1	†	0.92
2	†	0.46
3	†	0.55
4 (pT0 is reference)	†	> 0.99
pN status
1	0.4 (0.1, 1.7)	0.22
2	0.6 (0.1, 2.7)	0.51
3 (pN0 is reference)	0.1 (0.01, 1.3)	0.08
ICU stay	0.9 (0.9, 1.0)	0.96
Hospital stay	1.0 (0.9, 1.0)	0.32
Anastomotic leak	0.4 (0.05, 2.9)	0.32
Postoperative bleeding	†	0.39
Chylothorax	†	0.29
Pneumonia	1.6 (0.6, 4.0)	0.34
Pulmonary embolism	†	0.68
Arrhythmia	0.6 (0.2, 2.2)	0.46
Need for blood transfusion	1.5 (0.6, 4.1)	0.38
Multivariable analysis (only significant factors)
ASA status
II	0.2 (0.05, 0.2)	0.002
III	0.1 (0.03, 0.4)	< 0.001

Values in parentheses are 95% confidence intervals. HH, hiatal hernia; MIO, minimally invasive oesophagectomy; ASA, American Society of Anesthesiologists; RCT, radiochemotherapy; CT, chemotherapy; SCC, squamous cell carcinoma; ICU, intensive care unit. *Univariable generalized linear models based on the binomial distribution were used to analyse risk factors and a backward stepwise variable selection algorithm was applied for multivariable modelling of risk factors; †OR and 95% c.i. could not be computed.

Review of technical differences between the procedures

The retrospective review of both operative techniques revealed only one substantial difference: whereas during the open approach the omentum majus is lifted upwards and the posterior layer of the gastrocolic ligament is divided in the anatomical avascular plane directly attached to the colon transversum and, consequently, a subtotal omentectomy is performed to allow adequate pull-up of the gastric conduit, this step is omitted during MI oesophagectomy, where the bursa omentalis is entered via a direct incision of the anterior layer of the gastrocolic ligament between the stomach and transverse colon.

Retrospective review of the CT scans before HH repair in the MIO group

In total, 14 of 21 CT scans were available for retrospective review. In all the reviewed scans, the omentum was identified in the chest, either alone or together with the colon and the small intestine.

Discussion

The results, based on a large dataset of patients treated in one of the largest high-volume units in Germany, showed a significant higher risk for PHH after MIO than after OO. Notably, PHH in the single patient from the OO group was diagnosed late after index oesophagectomy compared with PHH in the MIO group, which became apparent early after oesophagectomy. This discrepancy suggests that differing pathogenic mechanisms may be involved.

With both univariable and at multivariable analysis (Table 4), only one significant risk factor for PHH in the MIO group was identified: ASA I status. However, the authors’ main finding was, aside from the surgical approach (open versus MIO), that the key procedural difference was the omission of omentectomy in the MIO group. This leads to the proposal of a new theory regarding the pathogenesis of PHH. A multifactorial mechanism is suggested, likely unfolding in several stages: a cytokine-mediated response may prompt migration of the omental remnant toward the hiatus and lymphadenectomy area, both of which are inflamed after dissection. Previous studies^9–11 have noted a similar response of the omentum to abdominal injury, inflammation, or sepsis. Once in the hiatal region, the omentum may herniate into the mediastinum, potentially pulling up the colon through the hiatus (Fig. 1c). This process could be facilitated in MIO due to fewer adhesions (as suggested by Willer et al., 2012⁵), reduced pain, and faster mobilization, where patients are encouraged to breathe deeply and cough to prevent pneumonia under Enhanced Recovery After Surgery (ERAS) or ERAS-like protocols¹². Supporting this theory, the findings indicate that ASA I patients—who generally recover more quickly—have a higher risk of PHH compared with ASA II and III patients, who may experience a slower recovery. In the retrospective analysis of pre-hiatoplasty CT scans of the MIO group, the authors found that, in all reviewed available scans (14 of 21), the omentum was always herniated in the chest (Figs S1, S2). Notably, a 33-year-old female patient (ASA I) who developed an early PHH (1.2 weeks after oesophagectomy) presented only omentum in the left chest (Fig. S1), thus supporting the theory that the omental remnant could be the main trigger and first herniated part of PHH. However, in the authors’ extensive experience with open oesophagectomy, PHH is an exceptionally rare complication. The incidence of PHH in the open group (0.2%) was significantly lower than the average incidence reported by Murad et al.¹ in a meta-analysis, where 112 of 4250 patients (2.6%) developed PHH. This difference may reflect the authors’ specific technical approach, which includes omentectomy during open oesophagectomy. To the authors’ knowledge, most centres do not mobilize the gastrocolic ligament in the avascular plane (below the greater omentum) during either MIO or OO^13,14.

Of note, HH following laparoscopic total gastrectomy is an exceptionally rare complication. Urabe et al. conducted a retrospective study¹⁵, published in 2019, reporting a long-term incidence of 0.4%. The literature^16–18 on this topic is sparse, and largely limited to case reports from Eastern countries. One possible reason for the rarity of HH after gastrectomy may be the routinely performed omentectomy (at least as a partial resection), which, despite ongoing debate^19,20, remains a standard procedure in cancer-related gastrectomies^21,22. Additionally, Gong et al.²³ analysed 490 consecutive patients treated by a single surgeon (2011–2017) using two anastomotic techniques. When performing the ‘functional method’, the crus muscle was divided to create a wider space for oesophagojejunostomy, a step omitted in the ‘overlap method’. HH developed only in patients treated with the functional technique, an observation also noted by Urabe et al¹⁵. Therefore, contrary to some beliefs^6,7, extensive hiatal dissection with preservation of the diaphragmatic pillars (for example, as mostly performed during oesophagectomy) does not necessarily increase HH risk.

The present study has limitations. In addition to its retrospective design, as a single-centre cohort study, it has a major confounding factor: there are two simultaneous differences for the two group (that is, open/MI approach and omentectomy/non-omentectomy). However, considering all the facts and observations, such as the lower overall PHH rate in the present study compared with the meta-analysis by Murad et al¹, the highly significant difference between the two groups in term of PHH, and the data from the gastrectomy literature, the authors hypothesize that the omitted omentectomy may play an important role in the onset of PHH. Furthermore, the sample size was large, and all surgeries were performed by experienced oesophageal surgeons following a standardized operative protocol (as described in the Methods section). Furthermore, OO was performed as standard of care in the early years and MIO in the later years of the present study, with a consequent minimization of the patients’ selection bias.

In conclusion, the authors propose that omentectomy, even as a subtotal resection, may play a critical role in preventing PHH after oesophagectomy for cancer. Based on their findings, they are currently planning a randomized clinical trial to investigate this aspect further with the aim to conclusively validate or refute this hypothesis.

Funding

The authors have no funding to declare.

Author contributions

Luca Giulini (Conceptualization, Methodology, Writing—original draft), Irina Avramovska (Data curation, Investigation, Writing—original draft), Melissa Kemeter (Data curation, Investigation), Lisa Bernhardt (Data curation, Writing—review & editing), Lucas Thumfart (Supervision, Visualization), Felix J. Hüttner (Supervision, Visualization, Writing—review & editing), Patrick Heger (Methodology, Supervision), Wolfgang Hitzl (Formal analysis), Markus K. Diener (Project administration, Validation, Writing—review & editing), and Attila Dubecz (Supervision, Validation, Writing—review & editing)

Disclosure

The authors declare no conflict of interest.

Supplementary material

Supplementary material is available at BJS Open  online.

Data availability

The raw data are not available for the public, but can be obtained upon reasonable request from the corresponding author. For instance, the statistical analysis was performed by W.H. and additionally by K.T. (see Acknowledgements), who are both professional statisticians.