Introduction

Gastric cancer (GC) has been shown to be the fifth most common cancer and the third leading cause of cancer-related death worldwide [1]. Surgical resection is the only curative treatment approach, and regional lymphadenectomy is recommended as a part of radical gastrectomy [2, 3]. Laparoscopic gastrectomy (LG) is being increasingly used because of its beneficial short-term effects when compared with open gastrectomy [4].

The da Vinci surgical system (DVSS) was developed to overcome several disadvantages that were identified in standard minimally invasive surgery using a laparoscope [2]. In the field of gastric surgery, most laparoscopic surgeons expect that the use of the DVSS will help to overcome the technical difficulties of LG, improving its safety, reproducibility, and long-term outcomes [2, 5, 6]. However, the only large nonrandomized prospective study (NCT01309256) comparing the DVSS with LG demonstrated that the use of the DVSS increased operative time and cost and that no difference in morbidity was observed, suggesting that the use of the DVSS might reduce cost-effectiveness [7]. In the meantime, at Fujita Health University, robotic gastrectomy (RG) was introduced in 2009 and actively used this approach in operable patients with resectable GC at the patient’s own expense [2, 8]. The double-bipolar method, the da Vinci’s plane theory, and the monitor quadrisection theory, which are universally available for gastrointestinal and hepato-pancreato-biliary surgeries, were established by I.U. and K.S., and the technique was standardized [9]. Consequently, it was demonstrated that the morbidity (overall complications) rate with RG was about one-fifth the rate with LG and that reduced morbidity, including postoperative pancreatic fistula (POPF), with RG led to an improvement in the short-term postoperative course [2]. The safety, feasibility, and economical efficiency of RG for GC remain unclear.

Therefore, the aim of the present study (UMIN000015388) was to determine if the advantageous short-term outcomes achieved at Fujita Health University could be well reproduced in a prospective multicenter setting, so that insured use of RG for GC may be approved by the Ministry of Health, Labour, and Welfare (MHLW). We hereby present the data on the short-term outcomes. Survival analysis is scheduled to take place in January 2020.

Materials and methods

Study design and endpoints

This was a multi-institutional, single-arm prospective phase II study. RG was performed prospectively in consecutive patients with cStage I or II GC. The primary endpoint was morbidity determined by Clavien–Dindo classification (C-D) Grade IIIa or higher [10]. We selected this endpoint, because postoperative complications requiring surgical, endoscopic, or radiological intervention, which correspond to C-D Grade IIIa, remarkably extend the admission period, threaten the patient’s life, and increase medical cost [2, 10,11,12]. The secondary endpoints were morbidity rate (C-D Grade ≥ II); health outcomes measured with EuroQol 5 Dimension (EQ-5D) [13] and total medical cost; RG completion rate; open surgery conversion rate; intraoperative adverse events; surgical outcomes, including operative time, console time, estimated blood loss, and number of dissected lymph nodes; short-term outcomes, including time to oral intake, postoperative hospital stay, and mortality; and long-term outcomes, including 3-year overall and relapse-free survival rates.

For sample size calculation, we used data from historical controls in three leading institutions (Fujita Health University, Saga University, and Kyoto University). In these institutions, 998 LG procedures were performed between 2009 and 2012. Of these 998 LG procedures, 801 were performed on patients preoperatively diagnosed with cStage I or II. Morbidity (C-D Grade ≥ IIIa) occurred in 51 of these 801 patients (6.4%). We hypothesized that RG could reduce the morbidity rate to less than half of that with LG, as RG has many advantages to reduce postoperative complications. Therefore, the threshold level was set as 6.4% and the expected level was set as 3.2%. The required sample size was calculated to be 330 (one-sided alpha of 0.05 and power of 80%). In addition, in these three hospitals, morbidity (C-D Grade ≥ IIIa) in patients with cStage IA diseases who underwent laparoscopic distal gastrectomy was only 4.7%. The planned patient enrollment period was 2 years and the follow-up period was 3 years. Interim analysis was scheduled at registration of the initial 220 cases to evaluate the efficacy of RG using Bayesian inference.

Clinicopathological findings and tumor stages were recorded according to the 14th edition of the Japanese Classification of Gastric Carcinoma (JCGC) [14]. Cancer staging was performed according to the findings of contrast-enhanced computed tomography (CECT), gastrography, endoscopic study, and endosonography. Lymph nodes > 8 mm in the short-axis diameter in < 5-mm-thickness CECT images were considered to indicate positive lymph-node metastasis. The extent of lymph-node dissection, as well as gastric resection, was determined according to the Japanese Gastric Cancer Treatment Guidelines ver. 3 [15]. The reconstruction approach was selected according to surgeon preference. Intra- or extra-corporeal anastomosis was allowed in this study.

Patient eligibility

Patients who met the following criteria were included: operable under general anesthesia; histologically proven gastric adenocarcinoma; cStage I or II disease not indicated for endoscopic resection according to the Japanese Gastric Cancer Treatment Guidelines ver. 3 [15]; curably treated with total, distal, or proximal gastrectomy involving D1+ or D2 lymph-node dissection; age ≥ 18 years. Patients who met the following criteria were excluded: preoperative use of chemotherapy and presence of mental disorders. Informed consent was obtained from all patients. This study was conducted in accordance with the amended Declaration of Helsinki under approval by the institutional review board of each participating institution. In addition, this clinical trial was approved for Advanced Medical Technology (“Senshiniryo”) B, which is managed by the MHLW. “Senshiniryo” is a special interim system for newly developed uninsured medical technologies approved by the MHLW, in which patients are requested to pay for the designated technology [16]. Using this system, the MHLW determines whether it is feasible to cover the designated technology under the universal medical insurance program from the standpoint of technical and social validity including a cost-effectiveness point of view [17].

Criteria for participating institutions and quality control of surgery

To control for the safety and quality of robotic operations, the institutions and operating surgeons were required to meet certain criteria. The criteria for the institutions were as follows: (1) introduced RG ≥ 1 year previously; (2) managed ≥  20 RG procedures (including ≥ 5 total gastrectomies); (3) managed ≥  50 LG procedures over the past 4 years; (4) achieved a morbidity rate (C-D Grade ≥ IIIa) for LG procedures of ≤ 12% over the past 4 years. The criteria for operating surgeons were as follows: (1) considered a qualified surgeon (Japan Society for Endoscopic Surgery) according to the endoscopic surgical skill qualification system; (2) achieved board certification in gastroenterology (The Japanese Society of Gastroenterological Surgery); (3) received a certificate for da Vinci Surgical System Off-site Training as a Console Surgeon; (4) performed ≥  10 RG procedures (including ≥ 1 total gastrectomies).

Initially, three institutions participated in this trial. However, during the enrollment period, the study group gradually expanded, and finally, 15 institutions registered patients. Before the inclusion of an operating surgeon in this study, the principle investigator (PI; I.Uyama) shared the generalized technique of RG, including the double-bipolar method, the da Vinci’s plane theory, and the monitor quadrisection theory [2, 9], with each surgeon via a proctorship program provided by Intuitive Surgical, Inc. (Sunnyvale, CA, USA), and reviewed the surgeon’s adherence to technical details using a non-edited video of an entire robotic total gastrectomy procedure. Any critical adverse effect (C-D Grade IV or V) was required to be reported to the PI within a week after its occurrence. If the total number of patients who experienced any critical adverse effect related to the robotic operation reached ≥  10, it was decided that this study would be suspended according to the decision of the data and safety monitoring board. All intracorporeal manipulations during the robotic operation were required to be recorded on a Digital Versatile Disc (DVD) and provided to the data center.

Protocol treatment

Patients who met the aforementioned inclusion criteria were registered after obtaining informed consent. RG was performed according to the Japanese Gastric Cancer Treatment Guidelines ver. 3 [15]. If distant metastasis, including peritoneal dissemination, was found during the operation, the treatment deviated from this protocol. Patients who underwent curative operation completed this protocol regardless of conversion to the laparoscopic or open method.

Data management

The data center (Center for Research Promotion and Support, Fujita Health University) performed data management and site-visit monitoring. The center provided monitoring reports twice during the study and the reports were submitted to and reviewed by the PI. The center collected all data from the participating patients in case report forms (CRFs) in a linkable anonymized fashion and stored the data for 5 years after the completion of the study. In detail, a case registry form was faxed a day before the operation. A perioperative CRF, the DVDs onto which the operation was recorded, a discharge reporting form, and a medical expense reporting form were sent within 28 days after patient discharge. Quality-of-life (QOL) data (EQ-5D) were sent within 120 days after the operation. Prognosis reporting forms were sent within 1 month after 1, 2, and 3 postoperative years.

Follow-up and postoperative management

Patients with pathological stage II or III cancer received adjuvant chemotherapy using S-1 for 1 year, while those with pathological stage IV cancer received S-1-based palliative chemotherapy according to the Japanese Gastric Cancer Treatment Guidelines ver. 3 [15]. Data on QOL and prognosis were investigated after 1 and 3 months and after 12, 24, and 36 months, respectively.

Statistical analysis

The primary endpoint (morbidity C-D Grade ≥ IIIa) was analyzed using the one-tailed binomial test. Independent continuous variables were compared using the Mann–Whitney U test or Kruskal–Wallis test. Categorical variables were compared using the Chi-square test. Data are expressed as median (range), unless otherwise stated. All statistical analyses related to the study design and endpoints were conducted by Sogo Rinsho Medefi Co., Ltd. Within-group analyses and subanalyses were performed by K.S. using IBM SPSS Statistics 22 (IBM Corporation, Armonk, NY, USA). A p value < 0.05 (two-tailed) was considered statistically significant.

Results

Patient background

Between October 2014 and December 2016, 330 patients from 15 institutions were registered into this study (Supplementary Table 1). The flow diagram of the study process is shown in Fig. 1. All registered patients met the criteria of this study. Among the 330 patients, peritoneal dissemination was noted during the operation in 2 patients, and they underwent non-curative operation. No patient had CY1P0 status. In 2 patients, after these patients were registered, the credential surgeon for RG (K.E.) sustained such severe finger injuries before the scheduled date of RG that he could not manipulate the robot. Thus, protocol treatment was discontinued and LG was performed instead. Therefore, 326 patients completed the protocol treatment (full analysis set; FAS) and their data were analyzed. Patient background data are summarized in Table 1. Age and BMI were 66 (26–85) years and 22.4 (15.0–31.5) kg/m2, respectively. Among the 326 patients, 216 (66.3%) had preoperative comorbidity with ≥ 2 in the ASA performance status classification. Most of the patients had cStage I disease located at the middle or lower third of the stomach.

Table 1 Patient background
Full size table
Fig. 1
figure 1

Flow diagram of the study process

Full size image

Operative findings

The operative findings are summarized in Table 2. All patients underwent R0 resection. With regard to the extent of gastric resection, distal gastrectomy, total gastrectomy, and proximal gastrectomy were performed in 253 (77.6%), 47 (14.4%), and 26 (8%) patients, respectively. D2 lymph-node dissection was performed in 118 (36.2%) patients. The operative time and estimated blood loss were 313 (167–587) min and 20 (0–612) mL, respectively. Intraoperative adverse events were noted in 5 patients (redo of reconstruction: 2, injury to surrounding organs: 2, and others: 1). No severe intraoperative adverse event extended the duration of hospital stay or resulted in irreversible deterioration of QOL. Conversion to the open method was noted in 1 patient (0.3%) because of extended resection combined with splenectomy for tumor invasion. Pathological JGCG stage IA, IB, IIA, IIB, IIIA, IIIB, IIIC, and IV cancers were noted in 201 (61.7%), 46 (14.1%), 25 (7.7%), 35 (10.7%), 10 (3.1%), 4 (1.2%), 5 (1.5%), and 0 (0%) patients, respectively.

Table 2 Operative and pathological findings
Full size table

Postoperative findings

Postoperative findings are shown in Table 3. During the 30 days following the operation, the morbidity rate (C-D grade ≥ IIIa) with RG was 2.45% (8/326). As the morbidity rate was significantly lower than that of the historical control (6.4%), the null hypothesis was rejected [95% confidence interval 0.9755 (0.9522–0.9893) and p = 0.0018]. Intra-abdominal bleeding was noted in 3 (0.9%) patients. A pancreatic fistula with intra-abdominal bleeding and an internal hernia with small bowel obstruction were each noted in 1 (0.3%) patient. The reoperation rate was 1.2% (n = 4). Complication of C-D grade IVa was seen in 1 (0.3%) patient (intra-abdominal bleeding). No C-D grade IVb complication or 30-day mortality was noted. One of the ten patients, who were preoperatively scheduled to undergo distal gastrectomy but actually underwent total gastrectomy, developed C-D grade I complication (constipation, classified as Local-Others in Table 3). The total medical cost, including the operative fee and admission expense, was 1,799,628 (1,530,170–5,173,706) JPY. The surgical cost per procedure was 1,063,800 (950,000–1,158,970) JPY. The changes in the QOL score measured using EQ-5D were 1.0 (0.5920–1.0) (preoperative), 0.8040 (0.3940–1.0) (postoperative day 7), and 1.0 (0.3940–1.0) (postoperative day 30).

Table 3 Postoperative complications
Full size table

Impact of postoperative complications on outcomes

Postoperative hospital stay was longer, total medical cost was higher, and the QOL score at postoperative day 30 was lower in patients with morbidity (C-D grade ≥ IIIa, n = 8) than in those without morbidity (n = 318) [morbidity + vs. − : postoperative hospital stay, 25 (12–62) vs. 9 (6–56) days, p = 0.014; total medical cost, 2,936,159 (2,522,180–5,173,706) vs. 1,795,506 (1,530,170–3,268,218) JPY, p = 0.004; QOL score at postoperative day 30, 0.786 (0.444–1.000) vs. 1.000 (0.394–1.000), p = 0.015]. In addition, operative time was longer and the reoperation rate was higher in patients with morbidity than in those without morbidity [morbidity + vs. − : operative time, 374.5 (299–587) vs. 312 (167–580) min, p = 0.010; reoperation rate, 50% vs. 0%, p = 0.033]. However, there were no differences in age, sex, BMI, blood loss, comorbidities, history of laparotomy, institution, type of reconstruction, extent of resection, extent of lymph-node dissection, number of dissected lymph nodes, number of metastatic lymph nodes, tumor size, cStage, pStage, and conversion rate.

Subanalyses of patient characteristics and surgical outcomes in the historical control

The results of the subanalyses of patient characteristics and surgical outcomes in the historical control are presented in the supplementary text and Tables 2 and 3.

Discussion

The present “Senshiniryo” B clinical trial clearly demonstrated not only the feasibility and safety but also the clinical advantage (reduction in morbidity) and disadvantage (higher cost) of RG. To our knowledge, this is the first study to demonstrate the benefit of RG for postoperative complications, with a multi-institutional prospective approach.

In this study, our hypothesis that RG reduces the morbidity rate to less than half of that with LG was confirmed and the primary endpoint was met. The reduction in morbidity reduced the total medical cost, reduced the duration of postoperative hospital stay, and improved the QOL of the patient. These data suggested that RG would likely show improved cost-effectiveness if the surgical cost per procedure for RG was reduced to about the same amount as that for LG. Thus, the MHLW finally decided to recognize RG as a part of LG from the standpoint of medical insurance coverage as of April 2018. Thus, an operable patient with resectable GC could undergo RG, similar to LG, for 80,000–100,000 JPY/month using the High-cost Medical Expense Benefit system. However, no additional fee is reimbursed to the hospital, although RG is performed instead of the conventional LG. Currently, in the Japanese universal medical insurance system, amounts of 796,200, 787,300, and 960,900 JPY are reimbursed for laparoscopic distal gastrectomy, proximal gastrectomy, and total gastrectomy, respectively. On the other hand, the median surgical cost for RG was 1,063,800 JPY in this study, suggesting that not only competition among rival robots developed by Medicaroid, Inc. [18], Medtronic, Inc. [19], Verb Surgical, Inc. [20], TransEnterix, Inc. [21], and Meere Company, Inc. [22], but also efforts by Intuitive Surgical Inc. to lower prices are desirable to reduce cost from an administrative point of view.

For GC, robotic procedures have been developed according to procedures in the conventional LG, which has been more commonly performed. Thus, the impact of the use of the DVSS for GC treatment has been assessed mostly in comparison with LG [2, 8, 23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39]. Most previous reports have failed to demonstrate clear clinical advantages of RG over LG [7, 40]. However, the benefits of RG for reducing morbidity, including POPF, were identified in the present multi-institution prospective study, as well as in a previous single-institution retrospective study performed in Japan [2, 41]. This may be at least partly because common technical principles for robotic set-up and dissection [2, 9] were used across the institutions. In fact, following the death of a patient soon after RG, the Japan Society for Endoscopic Surgery took the initiative to safely introduce robotic surgery nationwide according to the “proposal on the introduction of robotic surgery” issued in July 2011 [9]. In accordance with this proposal, I.U., as a proctor, made the best efforts to mentor the potential collaborating surgeons in this study for their initial few cases. Another possible reason may be the strict requirements for participating institutions and operating surgeons. In particular, operating surgeons were included if they had performed ≥ 10 RG procedures, including ≥ 1 total gastrectomies, and institutions were included if they had managed ≥ 20 RG procedures, including ≥ 5 total gastrectomies, considering the learning curve of RG for experienced laparoscopic gastric surgeons [9]. Thus, since April 2018, RG with health insurance coverage has been performed by the qualified surgeons (Japan Society for Endoscopic Surgery) according to the endoscopic surgical skill qualification system who had previously performed ≥ 10 RG procedures in institutions that managed ≥ 50 gastrectomies, including ≥ 20 LG or RG procedures annually.

The present study had several limitations. First, this study was conducted as a single-arm study. A historical control was used to assess the statistical hypothesis. In this study, blood loss and duration of stay improved when compared to the findings in the historical control; however, some parts of the patient background significantly differed. Therefore, there might have been data bias, and the overall results should be interpreted with caution. We decided to use such a study design instead of a randomized-controlled trial mainly because of our limited budget. Randomization was technically difficult, as each patient who underwent RG needed to pay approximately 700,000 JPY, even when using the “Senshiniryo” B system in combination with the reimbursement of 500,000 JPY from Intuitive Surgical, Inc., whereas the use of LG involved a cost of only approximately 100,000 JPY per patient. In fact, according to the results of the subanalyses, considerable differences in the patient background were observed. We plan to perform propensity score-matched analyses once the long-term outcomes are obtained in the prospective cohort of this study. Second, the quality of the historical control may be a concern. While the design of this study was considered, MHLW requested a historical control with higher level of evidence. However, no multi-institutional large randomized-controlled trial on LG for advanced GC has been completed worldwide [42]. The Nationwide Survey of Endoscopic Surgery in Japan biennially conducted by the Japan Society of Endoscopic Surgery [43] could not be used, because all types of postoperative complications were not registered and severity of those was not graded. National Clinical Database [44] was also lined up as a candidate, but it had been just a little, while since registration had started. Thus, after a long discussion, MHLW finally allowed us to use short- and long-term (3-year) outcomes from historical controls in the three leading institutions (Fujita Health University, Saga University, and Kyoto University) between 2009 and 2012, although the data were unpublished. As a matter of fact, the morbidity rate (C-D Grade ≥ IIIa) in patients with cStage I disease who underwent laparoscopic distal gastrectomy with D1+ lymph-node dissection was only 4.7% in the historical control, which was less than the rate of 5.1% noted in the Japan Clinical Oncology Group 0703 trial [4], suggesting excellent quality with regard to LG. Third, as GC is commonly found in Korea and Japan, where population-based screening is being undertaken [45], more reports, predominantly on the early GC, are being published in these countries than in Western countries. Most of the patients in this study had cStage I disease. It may be challenging to extend the present findings to the Western population. As our previous study suggested that larger extents of gastric resection and lymph-node dissection are associated with more effective use of the robot [2], a multi-institutional prospective study with a considerable number of patients with advanced GC is warranted.

In conclusion, RG reduced the morbidity rate when compared to the findings with LG. RG might be safe, feasible, and effective for GC. Medical insurance coverage in combination with a reduction in the cost of the robotic system would accelerate the spread of robotic surgery in Japan.