Open versus minimally invasive total gastrectomy after neoadjuvant chemotherapy: results of a European randomized trial

A non-inferiority, multicenter, international, randomized trial between January 2015 and June 2018 was performed in 13 hospitals in six European countries. Hospitals in the Netherlands, Sweden, Spain, Italy, Germany and the United Kingdom participated in this trial. The medical ethics board of all participating hospitals approved the trial protocol prior to inclusion of patients.

Eligible patients had resectable adenocarcinoma of the stomach and were treated with neoadjuvant chemotherapy, and based on location of the tumor, total gastrectomy was indicated. Patients had to be above 18 years of age and an American Society of Anesthesiologist classification of 3 or lower. Patients with early gastric cancer suitable for local treatment, distal gastric cancer suitable for distal gastrectomy, gastro-esophageal junction tumors, previous surgery of the stomach and previous or co-existing other malignancies were excluded [13].

For quality assurance, the principal investigator visited all centers interested in trial participation for a site initiation. With a low incidence of gastric cancer in the West, all gastrectomies are performed in specialized hospitals to maintain quality of care for these patients. To prevent surgeon bias, both procedures were done by surgeons experienced in open and minimally invasive gastrectomy. Surgeons eligible for participation had to have experience with at least 20 laparoscopic gastrectomies, consisting of at least ten total gastrectomies. Because this study reflected current practice, no additional quality measures were deemed necessary.

Diagnosis and staging were established before neoadjuvant chemotherapy by esophagogastroscopy, with biopsies and CT scan. Additional diagnostics such as PET-CT, MRI, ultrasound or diagnostic laparoscopy could be performed if deemed necessary. A diagnostic laparoscopy before the start of neoadjuvant chemotherapy was advised if there was doubt regarding resectability. Surgeons at the outpatient clinic informed eligible patients of the treatment regimen. Written informed consent was obtained from included patients.

A computer-generated randomization tool was used to randomize the patients in a 1:1 ratio to either an open or a minimally invasive procedure. Randomization was stratified per study center and performed in blocks of six. Randomization took place after assessment of the inclusion and exclusion criteria and after signing the informed consent form. Both patient and the surgeon were not blinded for the allocated procedure. In one hospital a coin flip on the operating room was used for randomization of several patients due to an error with the online randomization tool at that time. These patients were added as soon as the tool was working again, to ensure block randomization was still correct. To assure the quality of the primary outcome regarding lymphadenectomy and radicality, the pathologist was blinded for the type of surgery.

All patients received neoadjuvant chemotherapy in accordance with local protocols. An overview of types of neoadjuvant chemotherapy are depicted in Table 1.

Anaesthesia was similar for patients in open or minimally invasive surgery. All patients received a single, prophylactic gift of antibiotics according to the local protocol in the participating centers 30 min preoperatively.

Gastrectomy was performed 6–8 weeks after completion of neoadjuvant chemotherapy. Access to the abdomen in the OTG group was obtained via a median laparotomy. In the MITG group, the placement and number of trocars were based on surgeon preferences. An overview of number of trocars and site of extraction are depicted in Table 2.

Recommended lymph node dissection consisted of a D2 lymphadenectomy, being lymph node stations 1–7, 8a, 9, 10, 11p and 12a [14]. After extraction of the specimen the surgeon attached numbered tags in correspondence with the number of the lymph node station. The specimen was analysed by the pathologist for radicality, the number of resected lymph nodes with the corresponding lymph node stations and response to neoadjuvant chemotherapy according to Mandard [15]. The pathologists had separate access to the online case report forms to enter their results.

Postoperative treatment was similar in both groups. A nasojejunal tube could be inserted according to local preference. Enteral feeding could be resumed depending on the clinical status of the patient. Patients were encouraged to get out of bed and start walking with the help of a physical therapist or nurse. Follow-up took place 3, 6 and 12 months postoperatively, or sooner if deemed necessary.

The number of resected lymph nodes with a D2 lymphadenectomy according to the definition of the Japanese Gastric Cancer Association, alongside radicality, is a marker for the quality of care and, therefore, our primary outcome [14]. A standardized report regarding tumor size, radicality, number of resected lymph nodes, resected lymph node stations and Mandard scores was conducted [15].

Secondary outcomes regarded operative results such as operation duration, blood loss and perioperative complications. Postoperative outcomes regarded morbidity and mortality with postoperative mortality being defined as 30-day mortality or in hospital mortality. Complications were graded according to the Clavien–Dindo classification and all complications per patient were assessed using the comprehensive complication index [16, 17]. Complications requiring invasive intervention of admission to an intensive care unit, as measured by a Clavien–Dindo score of 3A or higher, were deemed major complications.

In addition, postoperative recovery such as time to first intake, time to first flatus and hospital stay was measured. In the overall analysis, textbook outcome was determined for each patient. A textbook outcome was achieved if: a curative resection was performed, with no perioperative complications, a pathological radical resection (R0) with more than 15 lymph nodes resected, no major complications (Clavien–Dindo 3A or higher), no reinterventions, a length of hospital stay less than 21 days, no mortality in the first 30 days and no readmission in the first 30 days [18].

Data regarding 1-year disease free and overall survival was collected.

The sample size was calculated based on two previous studies. The first sample size was based on the Dutch cancer registry of 2013, with a median yield of 19 lymph nodes, for non-inferiority with a power of 90% and one-sided t test with a significance level (α) of 0.05 and a non-inferiority margin of − 4, a total of 66 patients per group were deemed necessary [19, 20].

The second sample size was based on a large western study regarding patients that underwent gastrectomy for cancer with curative intent and described lymph node yield following neoadjuvant therapy in a Western population [21]. For a non-inferiority test with 90% power, and one-sided t test with a significance level (α) of 0.05, a standard deviation of 7.8 and non-inferiority margin of − 5 a total of 48 patients per group were to be included. Although the second sample size was more in fit with the study, we were confident that we could reach our first sample size. Accrual started according to the expected scheme, but accrual was slow. Based on the advice of the statistician and medical ethical review board, the research group decided to reassess the sample size.

The main reason for this slow accrual was a large group of patients that did not meet the inclusion criteria, mainly patients were not deemed fit enough for neoadjuvant chemotherapy.

Statistical analysis was performed using SPSS statistical package, version 22 (IBM software). Open and minimally invasive total gastrectomy were compared with continuous variables described as means and standard deviation for normal distributions and medians and interquartile ranges for non-normal distributions. Analysis was performed as intention to treat. Comparison tests were performed with Student’s t test and Mann–Whitney U tests as appropriate.

A total of 110 patients of 134 eligible patients were included in this trial and randomly assigned to undergo an open or minimally invasive gastrectomy. In ten patients, gastrectomy was not performed due to peritoneal carcinomatosis at the time of surgery. These patients were excluded from further analyses. Four patients retracted from the trial and therefor the data could not be used for analyses, thus resulting in 96 patients. A flow chart of the included patients is depicted in Fig. 1.

In the OTG group 14 patients (28.6%) had a proximal tumor, 25 patients (51.0%) a middle tumor and ten patients (20.4%) a distal tumor. In the MITG group 13 patients (27.7%) had a proximal tumor, 25 (53.2%) a middle tumor and nine (19.1%) a distal tumor. There was no significant difference in distribution between both groups (p = 0.979). Baseline characteristics and preoperative workup are depicted in Table 1.

Oncological safety of the procedure was measured as the number of retrieved lymph nodes and radicality of the procedure.

Pathological examination of the resected specimen showed that the numbers of retrieved lymph nodes and completeness of resection (i.e., resection margin (R0)) were similar between both groups. The mean number of resected lymph nodes was 43.4 ± 17.3 in OTG and 41.7 ± 16.1 in MITG (p = 0.612), mean difference 1.728 (95% CI − 5.02–8.47). For all lymph node stations data was collected as to whether the surgeon reported the station as resected. Next the pathologist confirmed resection for each tagged station if lymphoid tissue was found in that station. The number of harvested lymph nodes per station were also reported. No differences were observed in resected lymph node stations between both groups. A full D2 lymphadenectomy was performed in 49% of OTG versus 36.2% of MITG (p = 0.223). A splenectomy was performed in 12.2% OTG cases, versus 4.3% in MITG cases (p = 0.269).

There was no significant difference in radicality. Four patients had positive resection margins, one in the OTG and three in the MITG group. In 89.6% of all patient’s histological report showed adenocarcinoma with 51.0% intestinal type in the OTG group versus 34.0% in the MITG group and 38.8% diffuse type in the OTG group versus 55.3% in the MITG group, this was not significantly different.

An overview of all oncological results is depicted in Tables 3 and 4.

Regression techniques determined no confounding or effect modification by sex, age, BMI, T-stage and N-stage for these outcomes.

Although duration of surgery was significantly longer in the MITG group, with a median length of 244 min (IQR 198–293 min) in comparison to 200 min (IQR 164–245 min) in the OTG group (p = 0.005), blood loss was similar in both groups, with a median of 200 ml in the OTG and 171 ml in the MITG group (p = 0.454). In three patients the minimally invasive procedure was converted to an open procedure. Reasons for conversions were technical difficulty due to tumor infiltration, adhesions and perioperative hemodynamic instability.

In nine patients perioperative bleeding was reported, five in the MITG and four in the OTG group and one hospital reported an iatrogenic perforation of the colon which was directly repaired.

Operative details are depicted in Table 2.

Regarding postoperative recovery, outcomes were similar in both groups. Mean hospital stay was 8 days (IQR 7–11 days) in both groups if no complications occurred (p = 0.338). There was no significant difference in postoperative complications between OTG and MITG, with a total of 42.9% in OTG and 34.0% in MITG (p = 0.408). No differences were observed between OTG and MITG for major complications, being Clavien–Dindo grade 3 or higher (p = 0.377). Closer assessment of complications like postoperative bleeding, abscesses, anastomotic leakages (five in the OTG group, 10.2% and four in the MITG, 8.5%), and wound dehiscence, one in every group, were not statistically different between OTG and MITG. Both circular stapled (54,2%) and linear stapled (45,8%) techniques were applied equally in OTG and MITG for the formation of the esophagojejunostomy (p = 0.546). No differences were observed in anastomotic leak rates between the two techniques (p = 0.501). In the OTG group two patients died after anastomotic leakage due to multi organ failure and sepsis, two other patients had a reoperation, and one was treated with antibiotics only. In the MITG group one patient had a reoperation, one had percutaneous drainage and two were treated with antibiotics. Median Comprehensive Complication Index was not significantly different between both groups with a mean of 20.9 (IQR 20.9–31.56) in the OTG and 26.22 (IQR 20.9–33.7) in the MITG group (p = 0.242).

The time to first intake was 5 days in both groups, as was the time to first passage of stool. Postoperative results are depicted in Table 5.

No differences were seen in the 1-year survival between the OTG and MITG group. In the OTG group 90.4% and in the MITG group 85.5% was alive 1 year after surgery (p = 0.701). One-year survival curves are depicted in Fig. 2.