Machine learning improves the accuracy of graft weight prediction in living donor liver transplantation.

Precise graft weight (GW) estimation is essential for planning living donor liver transplantation to select grafts of adequate size for the recipient. This study aimed to investigate whether a machine-learning model can improve the accuracy of GW estimation. Data from 872 consecutive living donors of a left lateral sector, left lobe, or right lobe to adults or children for living-related liver transplantation were collected from January 2011 to December 2019. Supervised machine-learning models were trained (80% of observations) to predict GW using the following information: donor's age, sex, height, weight, and body mass index; graft type (left, right, or left lateral lobe); computed tomography estimated graft volume and total liver volume. Model performance was measured in a random independent set (20% of observations) and in an external validation cohort using the mean absolute error (MAE) and the mean absolute percentage error and compared with methods currently available for GW estimation. The best-performing machine-learning model showed an MAE value of 50 ± 62 g in predicting GW, with a mean error of 10.3%. These errors were significantly lower than those observed with alternative methods. In addition, 62% of predictions had errors <10%, whereas errors >15% were observed in only 18.4% of the cases compared with the 34.6% of the predictions obtained with the best alternative method ( p < 0.001). The machine-learning model is made available as a web application ( http://graftweight.shinyapps.io/prediction ). Machine learning can improve the precision of GW estimation compared with currently available methods by reducing the frequency of significant errors. The coupling of anthropometric variables to the preoperatively estimated graft volume seems necessary to improve the accuracy of GW estimation.

Successful use of venous graft from native liver with hepatocellular carcinoma during living donor liver transplantation with no impact on recurrence rate: A retrospective cohort study.

Introduction: There are still debates regarding using portal vein (PV) from liver with hepatocellular carcinoma (HCC) for vascular reconstruction. This study aimed to assess the feasibility and patency of PV venous graft from an explanted liver with HCC for the reconstruction of the hepatic veins tributaries or PV in living donor liver transplantation (LDLT) and to see if it has any risk on recurrence of HCC. Patient and methods: We conducted a retrospective study on 81 patients with HCC who underwent LDLT from April 2004 to July 2022. Results: Venous graft from native liver PV was used for vascular reconstruction in 31 patients as follows; reconstruction of V5 in 7 patients, V8 in 4 patients, V6 in 3 patients, combined V5 and V8 in 4 patients, V6 with V5/V8 in 5 patients, and as Y shape venous graft for 2 PV reconstruction in 8 patients. The implantation of the new conduit PV graft after reconstruction of the anterior sector tributaries was direct to the IVC in 8 patients, and to the common orifice of the left and middle hepatic veins in 12 patients. The 1 month, 3 months, and 1-year overall patency of the venous graft was 93.5%, 90.3%, and 84%, respectively. Nine patients had recurrent HCC. In multivariate analysis, the independent risk factors for HCC recurrence were AFP >400 ng/mL (HR = 1.47, 95% CI: 1.69-2.31, P = 0.01), moderate/poor differentiated tumor (HR = 3.06, 95% CI: 2.58-6.29, P = 0.02), and microvascular invasion (HR = 2.51, 95% CI: 1.05-1.93, P = 0.01). Using a PV venous graft had no risk factor for HCC recurrence (P = 0.9). Conclusion: The use of PV venous graft of native liver with HCC for venous reconstruction is a feasible and valuable option in LDLT with good patency rates and no risk of HCC recurrence.