BackgroundPost-hepatectomy liver failure (PHLF) can significantly compromise outcomes, especially in cirrhotic patients. The identification of accurate and non-invasive pre-operative predictors is of paramount importance to appropriately stratify patients according to their estimated risk and select the best treatment strategy. Materials and methodsConsecutive patients undergoing liver resection for HCC on cirrhosis between 1-2015 and 12–2020 at 10 international Institutions were enrolled and their pre-operative CT scans were evaluated for the presence of spontaneous portosystemic shunts (SPSS) to identify predictors of PHLF and develop a nomogram. ResultsThe analysis of the CT scans identified SPSS in 74 patients (17.4 %). PHLF was developed in 27 out of 425 cases (6.4 %), with grades B/C observed in 17 patients (4 %). At the multivariable analysis, the presence of SPSS resulted an independent risk factor for all-grades PHLF (OR 6.83, 95%CI 2.39–19.51, p < 0.001) and clinically significant PHLF development (OR 7.92, 95%CI 2.03–30.85, p = 0.003) alongside a patient's age ≥74 years, a pre-operative platelets count <106x103/μL, a multiple-segments liver resection, and an intraoperative blood loss ≥1200 mL. The 30- and 90-days mortality in patients with and without SPSS resulted 2.7 % vs 0.3 % (p = 0.024) and 5.4 % vs 1.1 % (p = 0.014). The accuracy of SPSS in predicting PHLF development was 0.847 (95%n CI 0.809–0.880). The internally validated nomogram showed excellent performance in predicting grades B/C PHLF (c-statistic = 0.933 (95%CI 0.888–0.979)). ConclusionThe pre-operative presence of SPSS assessed on the pre-operative imaging proved to be a valuable radiological biomarker able to predict PHLF development in patients undergoing liver resection for HCC.
Read full abstract