Side-to-side esophagogastric anastomosis for minimally invasive Ivor-Lewis esophagectomy: operative technique and short-term outcomes

Esophageal resection offers the best chance for cure in patients with locally advanced esophageal cancer. Ivor-Lewis esophagectomy (IL) is the universally accepted technique for disease located in the middle–distal esophagus and gastro-esophageal junction [1]. However, this operation is a complex surgical procedure and associated to higher rates of post-operative complications and morbidities. With the advent of minimally access surgery, IL has rapidly evolved towards totally minimally invasive (laparoscopy and thoracoscopy) approach [2] to minimize surgical trauma and reduce perioperative complications (particularly pulmonary infections), decrease length of hospital-stay and improve quality of life compared to open esophagectomy [3‐5]. However, outcomes of minimally invasive esophagectomy (MIE) have been often discordant when considering the incidence of anastomotic leakage (AL) [6‐10] probably as a consequence of the technical difficulty of intra-thoracic anastomosis and a long proficiency gain curve for MIE.

Intra-thoracic anastomosis in totally minimally invasive IL (TMIIL) is technically challenging and lacks detailed, generally accepted standardized technique. It is associated with a learning curve, being the refinement of surgical technique an important part of this curve [11, 12]. Individual surgeons starting implementing TMIIL in regular practice refined their technique during implementation, leading to heterogeneous surgical procedures. Thus, a range of options have been described over the years for intra-thoracic anastomosis [2, 13]. Common goal of all techniques is the creation of a safe anastomosis to reduce the risk of leakage and related complications. Nevertheless, there is no accepted ideal approach to perform gastroesophageal anastomosis and search for the optimal procedure is still under study.

In this work, we present our experience with TMIIL utilizing a stapled side-to-side anastomosis in 36 patients with esophageal or esophagogastric junction malignancies. This technique, first reported by Ben-David et al. [14], was modified in our center by introducing the use of a 30 mm linear stapler to perform the anastomosis. The adopted technique was described and discussed in light of the post-operative outcomes.

TMIIL was introduced in clinical practice in our department in December 2016 using intra-thoracic side-to-side stapled gastro-esophageal anastomosis. A total of 88 IL were performed from January 2017 to September 2020 (15 open, 34 hybrid with only the abdominal phase performed by a minimally invasive approach and 39 planned totally minimally invasive). All patients had a resectable middle–lower esophageal cancer or Siewert type 1 or 2 esophagogastric junction carcinoma. All patients were initially scheduled for TMIIL except in case of: (i) previous major abdominal or thoracic surgery; (ii) abdominal phase lasting more than 210 min; (iii) bulky tumors (relative contraindication). Of the 39 patients undergoing planned TMIIL, one converted to open approach (side-to-side anastomosis performed during thoracoscopy reinforced in thoracotomy) was included. Instead, in three patients, converted to a hybrid approach (due to, respectively: the transection of tPICC catheter inside the azygous vein at the moment of its division; an unsafe anastomosis; lung injury during trocar insertion due to massive pleural adhesions) a circular end-to-side anastomosis was performed during thoracotomy phase; for this reason, these patients were excluded. Therefore, a total of 36 patients were included in this study and a prospectively collected database was retrospectively reviewed. Data recorded included: demographic characteristics, comorbidities, pre-operative staging, neoadjuvant treatment, intra-operative data, postoperative outcomes and complications, length of hospital stay and mortality, re-admission rate and short-term oncological outcome.

All cases were subjected to a standardized, pre-operative evaluation, discussed in a multidisciplinary setting and indications generally followed international guidelines. Perioperative chemotherapy or neoadjuvant chemoradiotherapy were offered to patients with cT ≥ 3 or node-positive disease. The intervention was performed after 3 weeks and 8 weeks in patients treated with chemotherapy or chemoradiotherapy, respectively. Surgical procedures were performed by a single surgeon skilled in minimally invasive surgery. Pre-operative nutritional jejunostomy was placed during a staging laparoscopy in selected patients candidate to neoadjuvant therapy (for pre-operative nutritional supplementation), or at the time of esophagectomy in the other cases. Hospital stay was calculated from the date of surgery to discharge. Post-operative complications were graded according both to the ECCG and Clavien–Dindo Classification. Comprehensive Complications Index (CCI) was also calculated. Readmission and mortality were recorded for the first 90 days after surgery. The study was approved by Institutional Review Board as does not include any patient identifying information.

General features of the 36 patients and oncological pre-operative characteristics are reported in Table 1. Most patients were male (69%), with a median age of 65 years (29–83 years). The median body mass index (BMI) was in the healthy range (23 kg/m²). The median Charlson Comorbidity Index (CCI) was 4, being arterial hypertension, diabetes mellitus and cardiopathies the most frequent comorbidities. Most esophageal neoplasms were adenocarcinomas (ADC) (80.5%) or squamous cell carcinomas (SCC) (16.7%). Only one patient was affected by primary malignant melanoma of the esophagus. Most of the cases were classified in the clinical stage II and III. Regarding pre-operative therapy, patients with ADC were treated with chemotherapy (10) or chemoradiotherapy (7), while all patients with SCC received chemoradiotherapy (except one psychiatric case). Four patients (11.1%) had undergone a previous endoscopic submucosal dissection (ESD), without criteria for radicality at the final histopathological examination.

Operative data and histopathological results are shown in Tables 2 and 3, respectively.

The median operative time was 365 min (ranging from 240 to 480 min), with a median blood loss of 100 ml. There was no intra-operative mortality. Median number of lymph nodes harvested was 24. Complete R0 resection was achieved in all patients except in 3 (8.3%) (one patient with focal extension of the neoplasm to the proximal resection margin despite the esophageal resection was done under intra-operative esophagoscopic control and the other two patients with the circumferential resection margins (CRM) involved despite preoperative therapy). Among 22 patients who received pre-operative therapy, a complete pathological response was reported in seven patients (31.8%), (five ADC and two SCC), six of them treated by neoadjuvant chemo-radiotherapy.

Short-term outcomes and post-operative complications are shown in Table 4. According to ECCG Classification, two patients (5.6%) had an anastomotic leakage (without subsequent anastomotic stricture), both successfully treated endoscopically with a self-expandable esophageal stent, after 1 week course of Esosponge® treatment.

One patient developed an anastomotic stricture (about 1 month after the operation), resolved with a single endoscopic dilation. No conduit necrosis was recorded.

Abdominal chylous leakage developed in two patients, both treated conservatively; chylothorax occurred in three patients, two of these (of A and B types, respectively) required a surgical intervention. In these two patients, prophylactic duct ligation was performed during esophagectomy: one patient was operated on before the introduction of ICG thoracic duct visualization; whereas, in the other patient only minimal duct fluorescence was visible at surgery. A lesion of duct and a lesion of the cisterna chyli were diagnosed at re-thoracoscopy, respectively. The first was sutured and the second healed after percutaneous embolization of the cisterna.

Pulmonary complications occurred in six patients (16.7%): two of them had a pleural effusion requiring percutaneous drainage while four had a pneumonia requiring antibiotic therapy. Of the two patients with pleural effusion, one also developed acute respiratory distress syndrome and died on POD 64 due to this respiratory complication (after two left thoracotomies for hemothorax).

Cardiac complications, as atrial fibrillation occurred in four patients without heart and renal failure, and resolved after amiodarone administration. Other complications included two cases of delayed conduit emptying and one case of urinary retention, requiring reinsertion of urinary catheter. An antibiotic therapy was required in 27.8% of patients for infection (wound, or central IV line or BAL without pneumonia).

The median overall length of stay was 13 (7–64) days. In-hospital mortality rate was 2.8% (n = 1) and re-admission rate was 11.4% (n = 4). Causes of readmission were: recurrent episodes of vomiting and dysphagia for delayed gastric-tube emptying, resolved with erythromycin administration and diet modifications (two patients); a recurrent chyloperitoneum requiring percutaneous drainage; general asthenia and iron-deficiency anemia (with stable hemodynamics) resolved with blood transfusions.

Based on Comprehensive Complications Index (CCI), median values of complications were 27.9 (ranging from 20.9 to 100) (Table 5).

A retrospective experience on totally minimally-invasive Ivor-Lewis esophagectomy (TMIIL) using side-to-side linear-stapled anastomosis in patients with esophageal malignancies is described in this study.

Open esophagectomy has been long the standard surgical approach, although minimally invasive techniques (MIE) have been gradually gaining favor among surgeons. The appropriateness of oncologic resection and short-term benefits of MIE versus open approach have been supported by robust randomized multi-center trials showing improved QoL, lower rates of pulmonary infection, with no significant differences in margin status, nodal yield, mortality, or survival [3, 16]. At the beginning, McKeown technique was largely used to avoid intra-thoracic anastomosis, although the Ivor-Lewis procedure represents the current indication in case of tumors located in the middle, lower esophagus and esophagogastric junction (S1 and 2), leaving McKeown for cervical and upper thoracic esophageal malignancy [1, 17]. Nevertheless, the adaptation of MIE procedure for intra-thoracic anastomosis proceeds at slow pace among surgeons, mainly due to the difficulty and heterogeneity of surgical techniques for the intra-thoracic anastomosis.

The path toward optimal results faces a steep learning curve, even in case of surgeons already skilled in MIE techniques. Since the initial reports of TMIIL [18, 19], technical execution of intra-thoracic anastomosis has been a major challenge and associated to surgical learning curve. As evidenced by Van Workum et al. [12] in a multicenter retrospective analysis, operative time and incidence of anastomotic leakage represent the key elements of IL MIE-associated learning curve. The length of this curve ranged from 35 to 40 based on operative time and duration of hospital stay [11], although a longer curve (50–119 case) seems to be required to reduce AL rate, as mean incidence decrease from 18.8% to 4.5% after the plateau had been reached, as recently reviewed by Claassen et al. [20]. Thus, a substantial extra number of patients seem to be exposed et al. risk, with possibly devastating sequelae. Also, the choice of anastomotic techniques and subsequent refinements play a pivotal role in establishing a MIE program. Not uncommonly, some surgeons have changed or modified their initial technique until reaching a certain degree of experience and technical confidence with a specific one [21]. Becoming familiar with a surgical technique rather than putative performance differences among techniques seems to be a major determinant of learning curve.

In our case history using a side-to-side linear-stapled technique, a low leakage rate (5.6%) was found. This outcome is comparable with those reported in a literature survey of studies using the same anastomotic technique, ranging from 2.9 to 15.6% (Table 6). Also the median operative time (365 min) is in line with the range reported in other reports, although longer than the 270 min plateau reported by Van Workum et al. [12]. The small cohort size, pre-selection of patients undergoing TMIIL and conversion to hybrid procedure in case of a prolonged abdominal phase (i.e. > 210 min) may bias the reliability of our observed outcomes; however, total case number (close to the lower limit of the learning curve range), leakage rates and operative time collectively indicate the achievement of a certain surgical proficiency in TMIIL in our center. Which of the most common performed esophageal anastomotic techniques (i.e. circular-stapled, linear-stapled, intra-thoracic or Orvil, hand-sewn) has the lowest leakage rates remains controversial, as little consensus exists and varying outcomes are reported from their comparison. A recent meta-analysis found lower anastomotic leak rates with a linear-stapled esophagogastric anastomosis compared to a completely hand-sewn technique [22], although data derived from mixed studies (i.e. IL or McKeown). Analysis of EsoBenchmark database reported lower AL rate in side-to-side linear-stapled (15.6%) and end-to-side purse-string (13.9%) intra-thoracic esophagogastrostomies compared to end-to-side double-stapling anastomosis (23.3%) [36]. Other literature studies comparing the different mechanical approaches suggest no significant differences among stapling techniques for TMIIL (i.e. side-to-side, end-to-side, or end-to-end) [13, 23]. Irrespective of the small sample size, the low incidence of leakage in our study would seem to suggest an apparently shorter learning curve in the side-to-side compared to other techniques. This hypothesis is also supported in other studies, showing similar relationship between leakage incidence and cohort size. However, available data derives from retrospective or prospective single center studies while the question of the best MIE anastomotic technique should be addressed by accruing a randomized control trial.

The occurrence of anastomotic strictures can be relevant, until 18% in a meta-analysis of TMIIL including both end-to-end and side-to-side techniques [1]. In our case, the incidence of anastomotic stenosis was low (2.8%) and comparable to other studies using LS technique (Table 6). However, stricture required endoscopic dilation within the short-term follow-up period. While the contribution of linear-stapled anastomosis to reduce stricture rate compared to hand-sewn has been supported by various studies [22], differences with other mechanical techniques are less clear. A meta-analysis suggested an increased risk of anastomotic strictures using circular compared to linear stapler [30], probably as a consequence of the diameter of the annular stapler [31].

Regarding other post-operative complications, our experience is comparable with those reported in the literature [24, 25]. The incidence of chyle leak (12.8%) is higher compared to other studies using either side-to-side or others techniques (Table 6, [26, 27]) but comparable with recent reports [28, 29] adopting the more stringent definition proposed by ECCG classification. In our case, this complication is probably due to the extended mediastinal lymphadenectomy performed and, thus, not associated to anastomotic technique.

The results of our study suggest that TMIIL with a 3-cm linear-stapled anastomosis seems to be safe and effective, with favorable outcomes, low rates of post-operative anastomotic leak and stricture.

This technique seems to be easy to learn and perform and it could be a promising technique for beginners training in minimally invasive esophagectomies.