An algorithm based on the postoperative decrease of albumin (ΔAlb) to anticipate complications after liver surgery

Abstract

Abstract Background Perioperative decrease of albumin (ΔAlb) appeared as a promising predictor of complications after digestive surgery, but its role after liver surgery remains unclear. This study aimed to analyze whether and how ΔAlb can be used to predict complications after liver surgery. Methods A bicentric retrospective analysis of patients undergoing liver surgery (2010–2016) was performed, following TRIPOD guidelines. The preoperative and postoperative difference of albumin was calculated on POD 0 and defined as ΔAlb. Patients with any missing variable were excluded. The primary endpoint was overall complications according to the Clavien classification. A multiparametric algorithm based on ΔAlb was generated to optimize prediction performance. Results A total of 110 patients were analyzed. At least one complication occurred in 66 (60%) patients. Patients with and without complication showed a ΔAlb of 15.8 vs. 9.5 g/L (p<0.001). Area under ROC curve (AUC) of ΔAlb was 0.75 (p<0.01.). The ΔAlb-based algorithm showed an AUC of 0.84 (p<0.01), significantly improving performance (p=0.03). Multivariable analysis identified ΔAlb as independent predictor of complications (HR, 1.12; 95% CI, 1.01–1.07; p = 0.002). Conclusions ΔAlb appeared as a promising predictor independently associated with the risk of complication after liver surgery. The study presents a novel decision-tree based on ΔAlb to anticipate complications.

Keywords

• Partial hepatectomy
• Liver resection
• Decision-making
• Morbidity
• Biomarkers
• Predictors