Validation of the Southampton Difficulty Scoring System in Robotic Liver Surgery and Development of the International RoboLiver Difficulty Scoring System.

Abstract

Robotic liver surgery (RLS) provides technical advantages over laparoscopic liver surgery, but no validated robotic-specific difficulty scoring system (DSS) exists. We evaluated the applicability of the Southampton DSS to RLS and developed a dedicated RLS difficulty model. This multicenter retrospective cohort study included adults undergoing planned RLS across 24 international hepatobiliary centers. The Southampton DSS was assessed for calibration and discrimination in predicting intraoperative complications. Given limited performance, a robotic-specific model (International RoboLiver DSS) was developed using multivariable logistic regression with prolonged operative time (more than 280 minutes; 75th percentile) as a surrogate of technical difficulty. Model discrimination, calibration, and bootstrap internal validation were performed. Among 1,497 RLS patients, higher Southampton DSS categories were associated with increased intraoperative complications (p = 0.003); however, discrimination was poor (area under the curve [AUC] 0.571, 95% CI 0.530 to 0.612) with miscalibration (slope 0.43; intercept 0.06). Independent predictors of prolonged operative time included neoadjuvant chemotherapy, earlier extrahepatic surgery, lesion greater than 50 mm, multiple lesions, bilobar disease, and technically or anatomically major resection. The International RoboLiver DSS demonstrated moderate discrimination (AUC 0.719, 95% CI 0.686 to 0.751) with excellent calibration (intercept 0.00; slope 1.14). Bootstrap validation confirmed model stability (corrected AUC 0.719). Difficulty factors in RLS partially overlap with laparoscopic liver surgery but are not directly transferable. The International RoboLiver DSS provides a calibrated, robot-specific tool for preoperative complexity stratification and operative planning in RLS. External validation is required.