A risk-based framework to support conversion decision-making during minimally invasive left pancreatectomy: results from a multicentre registry

Minimally invasive left pancreatectomy (MILP) is increasingly adopted worldwide, but conversion to open surgery still occurs and is associated with poorer postoperative outcomes. Tools to identify patients at increased risk of conversion are limited. This multicentre registry-based study included patients undergoing laparoscopic (LLP) or robot-assisted (RLP) MILP within the IGOMIPS registry. Predictors of conversion were analysed using multivariable logistic regression. A simple preoperative risk score, the modified Conversion Risk Score (mCRS), was derived from independent predictors of conversion. We explored the interaction between preoperative risk, intraoperative complexity, and center-level variability. Of the 2127 MILPs included, 1235 (58.1%) were performed laparoscopically, and 892 (42.9%) were robot-assisted. Conversion occurred in 180 cases (8.5%) and was more frequent during LLP than RLP (11.8% vs 3.8%). Age >= 61 years, tumour diameter >= 35 mm, and preoperative suspicion of pancreatic malignancy were independently associated with conversion and formed the mCRS (range 0-5). Conversion rates increased progressively with higher mCRS values (ranging from 3.5% to 20.6%). When a high preoperative risk was combined with intraoperative complexity, the probability of conversion increased further (up to 19%). Differences in conversion rates between centres were primarily explained by case volume. In this large multicentre cohort, conversion during MILP was associated with patient age, tumour size, and suspected malignancy. The proposed mCRS, combined with intraoperative cues, may support risk stratification and timely intraoperative decision-making.