Robot-Assisted Versus Conventional Minimally Invasive Oesophagectomy

Key Takeaways

• Five-year overall survival was higher with robot-assisted surgery, and the prespecified non-inferiority criterion was met; an exploratory superiority analysis gave p=0.032.
• The trial randomly assigned 362 patients in a 1:1 ratio across six hospitals in China, with allocation stratified by neoadjuvant therapy.
• Intraoperative conversion, grade 3 or higher postoperative complications, and treatment-related deaths were comparable between groups.

In the RAMIE phase 3 trial, robot-assisted oesophagectomy was associated with higher 5-year overall survival than thoracoscopic oesophagectomy, with a hazard ratio of 0.71 (95% CI, 0.51-0.97). Across six hospitals in China, adults with resectable oesophageal squamous cell carcinoma were enrolled in this multicenter, randomized phase 3 non-inferiority trial, which used overall survival in the intention-to-treat population as the primary endpoint. Robot-assisted surgery met the prespecified non-inferiority criterion for overall survival versus the thoracoscopic approach. Perioperative findings for conversion to open surgery, severe postoperative complications, and treatment-related deaths were similar between groups.

RAMIE was a multicenter, open-label, randomized, controlled, phase 3, non-inferiority trial conducted at six hospitals in China and registered as NCT03094351. Eligible adults were aged 18-75 years, had biopsy-proven squamous cell carcinoma, ECOG scores of 0-2, and resectable cT1-4a, N0-2, M0, or M1 disease limited to supraclavicular nodal metastasis. Between Aug 2, 2017, and Dec 23, 2019, computer-generated 1:1 allocation stratified by neoadjuvant therapy assigned 183 patients to robot-assisted oesophagectomy and 179 to thoracoscopic oesophagectomy. Both operations included at least two-field lymphadenectomy, and overall survival in the intention-to-treat population was the primary endpoint. The randomized cohort included 309 men and 53 women.

Median follow-up at the planned final analysis was 71.5 months, with an interquartile range of 63.9 to 81.8 months. Five-year overall survival was 69.4% (95% CI, 62.1-75.6) after robot-assisted oesophagectomy and 56.2% (48.5-63.2) after thoracoscopic oesophagectomy. These estimates came from the intention-to-treat analysis prespecified for the primary overall survival endpoint. The non-inferiority framework used a 9% margin for 5-year overall survival and an upper 95% confidence boundary of 1.33 for the hazard ratio. One-sided pnon-inferiority was 0.0001, and an exploratory superiority analysis yielded p=0.032.

Harms were assessed in the per-protocol population, defined as eligible participants who underwent resection after trial assignment. Two patients in each group did not undergo resection and were excluded from these perioperative analyses. Conversions to open surgery occurred in 7 of 181 patients (4%) in the robot-assisted group and 6 of 177 (3%) in the thoracoscopic group. Grade 3 or higher postoperative complications were reported in 22 of 181 patients (12%) and 18 of 177 patients (10%), respectively. One treatment-related death occurred in each group, with broadly comparable results across these measured perioperative endpoints.