Since 2007, robotic gastrectomy (RG) has been considered an option for minimally invasive surgery (MIS)1. In recent years, several randomised trials and retrospective studies have compared short-term efficacy and oncological outcomes between RG and laparoscopic gastrectomy (LG)2-5. Compared with traditional laparoscopic gastric surgery, RG may yield better short-term and comparable long-term outcomes4-6.

According to previous studies7, 8, American Society of Anesthesiologists (ASA) grade ≥3, body mass index (BMI) ≥ 30 kg/m2, age ≥80 years, and clinical T4 stage are considered to be sensitive indicators for assessing surgical risk among patients with cancer. Patients with any of these factors were defined as high risk patients who constitute a special group. Previous research has shown that obesity, advanced age, comorbid conditions, and tumour cT stage are factors associated with mortality and complication rates in MIS for gastrointestinal malignancies6,7. Therefore, it is crucial to explore the impact of high-risk factors on prognosis. Many elderly patients have comorbidities with a high incidence of cardiopulmonary disease(s)8-10. Moreover, patients with poor baseline conditions often struggle to tolerate major surgery, and their comorbid state and ASA score are high-risk factors that affect postoperative complications and prognosis11-14. High BMI, especially among obese patients in whom excessive abdominal fat obscures anatomical regions, limits the flexibility of laparoscopic instruments, hinders complex perivascular lymph node dissection, and increases the difficulty of D2 lymphadenectomy, may increase postoperative complications15-17. Advanced-stage tumours, particularly those at the cT4 stage, present with more enlarged metastatic lymph nodes, making the exposure of vessels and dissection more challenging6,7. All of these factors are considered to be high-risk factors for surgical procedures. With the increasing prevalence of aging populations and obesity18-20, more high-risk patients with resectable gastric cancer (GC) are undergoing MIS.

To further explore this increasing concern about clinical practice, our study is the first to compare the short- and long-term outcomes of RG and LG in a high-risk gastric cancer population, with the aim of providing high-level, evidence-based medicine for the widespread application of RG in this population. To the best of our knowledge, this is the first multicentre, large-sample study to compare the advantages of RG and LG in a high-risk GC population.

A retrospective collection of clinical, pathological, and follow-up data was conducted on 1 103 patients who underwent radical RG and 2 515 who underwent radical LG at eight high-volume tertiary hospitals in China between August 2016 and June 2019.

The inclusion criteria were as follows: postoperative pathological staging of T1-4aN0-3bM0 gastric adenocarcinoma; and radical gastrectomy with gastric lymph node dissection. Exclusion criteria included: residual gastric cancer (n=52); concurrent/past malignancies (n = 58); loss to follow-up (n = 123); previous indocyanine green-guided lymph node dissection (n = 125); undergoing neoadjuvant chemotherapy (n = 144 [RG group, n = 7]; LG group, n = 137]); and patients without high-risk factors (n = 1617). These inclusion criteria identified 2 001 patients for baseline analysis. The study was approved by the institutional review board of each participating centre.

Propensity score matching and sample size Propensity score matching (PSM) was used to adjust for the different baseline characteristics between the RG and LG groups. A 1:1 ratio PSM was performed using the nearest neighbour matching method with a caliper of 0.2. Fourteen factors, including age, ASA class, previous abdominal surgery, tumour size, tumour location, resection extent, extent of lymphadenectomy, histology, cT, cN, pT, pN, and adjuvant chemotherapy were included.Ultimately, data from 1 068 patients were included in the analysis (eFig 1). The balance of matching between the two groups was assessed using the standardised mean difference (SMD), with a value < 0.1 considered to be acceptable.

Surgical quality control and perioperative management All surgeons participating in this study were from high-capacity tertiary referral centres and surpassed the learning curve associated with LG. The da Vinci robotic system (Intuitive Surgical, Inc., Sunnyvale, CA, USA) was used for all RGs, with surgeons receiving standardised RG training and operational certification.

Surgical intervention was initiated after obtaining informed consent from all patients, which included the additional costs of robotic surgery, a comprehensive explanation of surgical risks, and potential alternative treatments, thus facilitating informed surgical approach choices (i.e., LG or RG) within the participating institutions. Adherence to the "Japanese Gastric Cancer Treatment Guidelines" was maintained for determining the extent of gastrectomy and lymphadenectomy.21,22 RG operating procedures are similar to LG in terms of trocar placement, surgical phases, and anastomosis techniques, with detailed procedural delineations available elsewhere.2,23 Postoperative morbidity and mortality were assessed within a 30-day framework using the Clavien-Dindo classification system24. Complications were categorised as surgical and systemic, and were defined based on previous descriptions25, with intraoperative blood loss ≥400 ml defined as massive haemorrhage in accordance with the existing literature26.

Each surgical specimen was subjected to macroscopic and histological examination by two experienced pathologists, with TNM staging according to the 8th Edition of the American Joint Committee on Cancer staging system27. Any disparities were resolved by a third pathologist. Patients with stage ≥ II disease without contraindications to chemotherapy were recommended postoperative adjuvant chemotherapy based on 5-fluorouracil (i.e., "5-FU")21,28, with all adjuvant chemotherapy regimens and dosages formulated by experienced oncologists adaptable to patient responsiveness and chemotherapy toxicity.

Follow-up and surveillance Follow-up evaluations included physical examinations, comprehensive laboratory investigations, chest computed tomography (CT), abdominal ultrasound/CT scans with contrast enhancement, and annual gastroscopy. Patients were followed up once every 3 to 6 months in the first two years and every 6 to 12 months within the next 3 to 5 years. Follow-up telephone calls were used as supplements to outpatient reviews and medical insurance systems to monitor patient survival, recurrence and mortality. The multicentre inclusion resulted in variable final follow-up durations across centres. Surviving patients achieved follow-up durations ≥ 36 months.

Primary outcomes and definitions The main endpoints of this study were the 3-year cumulative incidence function (3y-CIF) of three-year mortality under a competing risk model, and 3-year disease-free survival (3y-DFS) with secondary outcomes, including three-year recurrence patterns.

The CIF refers to the adjusted mortality rate of non-gastric cancer-related deaths from the date of surgery to the date of the last follow-up29, 30. DFS was defined as the time from the date of surgery to the date of recurrence, death, or last follow-up without recurrence. The diagnosis of recurrence was based on radiologically or biopsy-confirmed suspicious lesions. Recurrence types were categorised as local (including local sites such as the gastric bed or anastomotic site), peritoneal, hepatic, multisite, and other uncertain locations. Multisite recurrence involves simultaneous recurrence at ≥ 2 sites, including the peritoneum, liver, lungs, bones, brain, distant lymph nodes, and other sites of haematogenous spread. Other uncertain locations encompassed vascular recurrences outside the liver (e.g., lungs, bones, brain, adrenal glands), distant lymph node recurrences, and recurrences of uncertain origin31,32.

Inclusion Criteria:

• postoperative pathological staging of T1-4aN0-3bM0 gastric adenocarcinoma; and radical gastrectomy with gastric lymph node dissection. According to previous studies7, 8, American Society of Anesthesiologists (ASA) grade ≥3, body mass index (BMI) ≥ 30 kg/m2, age ≥80 years, and clinical T4 stage are considered to be sensitive indicators for assessing surgical risk among patients with cancer

Exclusion Criteria:

• Exclusion criteria included: residual gastric cancer ; concurrent/past malignancies ; loss to follow-up ; previous indocyanine green-guided lymph node dissection ; undergoing neoadjuvant chemotherapy; and patients without high-risk factors . These inclusion criteria identified 2 001 patients for baseline analysis.

Since 2007, robotic gastrectomy (RG) has been considered an option for minimally invasive surgery (MIS)1. In recent years, several randomised trials and retrospective studies have compared short-term efficacy and oncological outcomes between RG and laparoscopic gastrectomy (LG)2-5. Compared with traditional laparoscopic gastric surgery, RG may yield better short-term and comparable long-term outcomes4-6.

According to previous studies7, 8, American Society of Anesthesiologists (ASA) grade ≥3, body mass index (BMI) ≥ 30 kg/m2, age ≥80 years, and clinical T4 stage are considered to be sensitive indicators for assessing surgical risk among patients with cancer. Patients with any of these factors were defined as high risk patients who constitute a special group. Previous research has shown that obesity, advanced age, comorbid conditions, and tumour cT stage are factors associated with mortality and complication rates in MIS for gastrointestinal malignancies6,7. Therefore, it is crucial t