Integrating surgical complexity and nutritional parameters to enhance prediction of postoperative complications in liver resection

In patients undergoing liver resection, postoperative complications remain high. We hypothesized that the incidence of postoperative complications after liver resection would be predicted well by liver resection complexity and nutritional status. We retrospectively assessed patients undergoing liver resection at The University of Tokyo Hospital from 2011 to 2021. Liver resection procedures were categorized by surgical complexity using a 3-level complexity classification. Nutritional parameters (including cholinesterase and albumin levels) were evaluated together with well-known nutritional indexes, including the modified Glasgow Prognostic Score, prognostic nutritional index, platelet-to-lymphocyte ratio, neutrophil-to-lymphocyte ratio, and controlling nutritional status. Of 1,258 patients, 570 (44.5%) experienced postoperative complications, with 506 (39.9%) requiring treatment (Clavien–Dindo grade II or greater). Multivariate logistic regression model analyses showed that cholinesterase and albumin levels, complexity classification, and open approach were associated with postoperative complications. The cholinesterase–liver resection complexity/approach model (area under the curve, 0.634) performed significantly better in predicting complications than the prognostic nutritional index (area under the curve, 0.560; P