J Robot Surg. 2026 Jul 23;20(1):746. doi: 10.1007/s11701-026-03712-y
Live-streaming robotic liver resection for surgical education: is it safe? a multicenter matched cohort analysis
Alessia Fassari1, Benedetto Ielpo2, Vito de Blasi3, Francesca Marcucci4, Alexandru Amariutei4, Antonio Iannelli5,6, Edoardo Rosso7
Affiliations
1Centre Hospitalier Universitaire de Rennes, Rennes, France.
2Department of Surgery, Hepato‑Biliary and Pancreatic Surgery Unit, Hospital del Mar, Pompeu Fabra University, Barcelona, Spain.
3Department of General Surgery, Centre Hospitalier de Luxembourg, Luxembourg City, Luxembourg.
4Institut de Chirurgie Foie et Pancréas, Pôle Santé Sud, rue de Gutteloup, Le Mans, 72100, France.
5Institut de Biologie Valrose, Nice, 06000, France.
6ALMAVIVA – SEDNA Clinique du Parc Impérial, Nice, 06000, France.
7Institut de Chirurgie Foie et Pancréas, Pôle Santé Sud, rue de Gutteloup, Le Mans, 72100, France. edoardo_rosso@hotmail.com.
Abstract
Live-streaming surgery is increasingly used as an educational tool in the surgical community. The safety of live-streamed robotic liver resection has not been formally evaluated. We assessed perioperative outcomes of robotic live-streaming liver resections compared with not-live streaming liver resections. This multicenter retrospective matched cohort study reviewed 311 liver resections performed between October 2024 and December 2025 at three European HPB centers. Forty-two consecutive patients who underwent robotic live-streaming liver resection (L-SS group) were matched 1:1 to patients who underwent not-live liver resection (not-L-SS group) during the same period, generating a matched cohort of 84 patients. The primary endpoint was the rate of severe complications (Clavien-Dindo ≥ III). Secondary endpoints included conversion to open surgery, intraoperative blood transfusion, postoperative bile leak, posthepatectomy liver failure, length of hospital stay, R1 resection rate, and 90-day mortality. Intraoperative variables, including use and duration of the Pringle maneuver, were also recorded. After matching on ASA score, presence of cirrhosis, tumour laterality, lesion contact with the cavo-caval confluence, and surgical approach, baseline characteristics showed reduced imbalance between groups, although the L-SS group retained significantly larger tumours and a higher proportion of technically advanced resections. The primary endpoint, severe complications (Clavien-Dindo ≥ III), was similar between groups (4.8% vs. 7.1%, p = 1.000), as were conversion rate, transfusion rate, bile leak, posthepatectomy liver failure, readmission, reintervention and 90-day mortality. The L-SS group had a significantly higher frequency (83.3% vs. 48.8%, p = 0.001) and longer duration (median 45 vs. 25 min, p = 0.004) of the Pringle maneuver, a shorter hospital stay (7 vs. 10.5 days, p = 0.001), and a lower rate of any-grade 30-day complications (7.3% vs. 33.3%, p = 0.005). These findings were unchanged after excluding patients with a Pringle maneuver duration exceeding 60 min (sensitivity analysis). In this multicenter matched cohort study, robotic live-streaming liver resection did not detect an increase in severe postoperative complications in highly selected patients, with severe complication rates comparable to non-streamed resections. The higher Pringle maneuver use likely reflects a didactic intraoperative strategy rather than increased risk. Prospective studies should evaluate educational efficacy and trainee skill acquisition.