Using Machine Learning to Predict Postoperative Liver Dysfunction After Aortic Arch Surgery.

OBJECTIVES: The study compared machine-learning models with traditional logistic regression to predicting liver outcomes after aortic arch surgery. DESIGN: Retrospective review from January 2013 to May 2017. SETTING: Fuwai Hospital. PARTICIPANTS: The study comprised 672 consecutive patients who had undergone aortic arch surgery. MEASUREMENTS AND MAIN RESULTS: Three machine-learning methods were compared with logistic regression with regard to the prediction of postoperative liver dysfunction (PLD) after aortic arch surgery. The perioperative characteristics, including the patients' baseline medical condition and intraoperative data, were analyzed. The performance of the models was assessed using the area under the receiver operating characteristic curve. Naïve Bayes had the best discriminative ability for the prediction of PLD (area under the receiver operating characteristic curve = 0.77) compared with random forest (0.76), support vector machine (0.73), and logistic regression (0.72). The primary endpoint of PLD was observed in 185 patients (27.5%). The cardiopulmonary bypass time, long surgery time, long aortic clamp time, high preoperative bilirubin value, and low rectal temperature were strongly associated with the development of PLD after aortic arch surgery. CONCLUSION: The machine-learning method of naïve Bayes predicts PLD after aortic arch surgery significantly better than traditional logistic regression.

Postoperative liver dysfunction after total arch replacement combined with frozen elephant trunk implantation: incidence, risk factors and outcomes.

OBJECTIVES: The authors aimed to clarify the incidence and risk factors of postoperative liver dysfunction (PLD) in patients undergoing total arch replacement combined with frozen elephant trunk implantation and to determine the association of PLD with short-term outcomes. METHODS: Data from 672 adult patients undergoing total arch replacement with frozen elephant trunk from January 2013 until December 2016 at Fuwai Hospital were analysed retrospectively. A multivariable logistic regression model was used to identify the risk factors for PLD. RESULTS: The overall incidence of PLD was 27.5%, which was associated with higher in-hospital mortality (PLD 4.9% vs No PLD 0.8%, P = 0.002) and 30-day mortality (PLD 9.2% vs No PLD 2.5%, P < 0.001) and a higher incidence of major adverse events (PLD 54.6% vs No PLD 23.4%, P < 0.001). In the multivariable analysis, preoperative hypotension [odds ratio (OR) 1.97, 95% confidence interval (CI) 1.14-3.41; P = 0.02), coronary artery disease (OR 2.64, 95% CI 1.17-5.96; P = 0.02), prolonged cardiopulmonary bypass duration (OR 1.01, 95% CI 1.00-1.01; P < 0.001), increased preoperative alanine transferase (OR 1.01, 95% CI 1.00-1.01; P < 0.001), preoperative platelet count <100 × 109/l (OR 3.99, 95% CI 1.74-9.14; P = 0.001) and increased intraoperative erythrocyte transfusion (OR 1.07, 95% CI 1.01-1.12; P = 0.02) were identified as independent risk factors for PLD. CONCLUSIONS: PLD was associated with increased mortality and morbidity. Among the independent risk factors for PLD, cardiopulmonary bypass duration and erythrocyte transfusion could be modifiable. A skilled surgical team and an ideal blood protection strategy may be helpful to protect liver function.

Integration of Child-Pugh score with future liver remnant yields improved prediction of liver dysfunction risk for HBV-related hepatocellular carcinoma following hepatic resection.

Assessment of hepatic functional reserve is important to enable the selection of appropriate treatment methods and safe hepatic resection in hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC). In the present study, an evaluation was made of the clinical value of combining the Child-Pugh score (CPS) with the standardized future liver remnant (sFLR) measurement to predict postoperative liver dysfunction (PLD). A total of 61 HBV-related HCC patients undergoing liver volumetry prior to hepatectomy were enrolled in the study. The sFLR was calculated as the ratio of FLR volume to standardized liver volume. PLD was defined as a prothrombin time of >18 sec or a peak serum bilirubin level of >51.3 µmol/l for 7 days after surgery. Univariate analysis and multivariate logistic regression analysis were performed to identify risk factors associated with PLD. The correlation between PLD and the combination of sFLR and CPS was analyzed. In total, 18 out of 61 patients developed PLD (29.5%), with a significantly higher PLD incidence for a CPS of 6 than a CPS of 5 (P<0.05). Multivariate logistic regression analysis revealed that a prothrombin time of <13.3 sec and an sFLR of <0.55 were independent risk factors for PLD. Receiver operating characteristic (ROC) curve analysis revealed that the cut-off values of sFLR and sFLR/CPS for predicting PLD were 54.5% and 0.0916, respectively, with areas under the ROC curve of 0.820 and 0.860, respectively. The combination of CPS and sFLR appears to yield improved prediction of the occurrence of PLD compared with either CPS or sFLR alone.

Value of postoperative indocyanine green retention rate at 15 minutes combined with standard remnant liver volume in predicting liver dysfunction after hepatectomy / 中南大学学报(医学版)

Objective:To investigate the value of indocyanine green retention rate at 15 minutes (ICG R15) on postoperative day 3 combined with standard remnant liver volume (SRLV) in predicting the occurrence of liver dysfunction after hepatectomyin hepatocellular carcinoma (HCC).Methods:The clinical data of 61 HCC patients undergone hepatectomy in Xiangya Hospital of Central South University from January 2015 to February 2016 were collected and analyzed.The patients were divided into 2 groups:a normal liver function group (n=40) and a liver dysfunction group (n=21).Univariate analysis was used to evaluate the risk factors for postoperative liver dysfunction.Logistic regression was used to assess the independent risk factors for postoperative liver dysfunction,and the regression equation between independent risk factors and postoperative liver dysfunction was established,The receiver operating characteristic (ROC) curve was used to examine the regression equation and compare the value difference in predicting postoperative liver dysfunction between single and combined independent risk factors.Results:Postoperative liver dysfunction occurred in 21 of the 61 patients,with an incidence rate at 34.4％.There was no significant difference in the time of operation,time of hepatic portal occlusion,volume of tumor and volume of resected liver between the 2 groups (all P＞0.05),but there were significant differences in the ICG R15 on postoperative day 3,intraoperative blood loss and SRLV between the 2 groups (all P＜0.05).The ICG R15 on postoperative day 3,intraoperative blood loss,SRLV were the risk factors for postoperative liver dysfunction.Logistic regression analysis showed ICG R15 on postoperative day 3 and SRLV were the independent risk factors for postoperative liver dysfunction,and the regression equation between independent risk factors and postoperative liver dysfunction was as follows:logit(P)=1.277+0.140×ICG R15 on postoperative day 3-5.125×SRLV.The area under the ROC curve ofICG R15 on postoperative day 3 combined with SRLV was more than that of single ICG R15 and single SRLV.Conclusion:ICG R15 on postoperative day 3 and SRLV are the independent risk factors for postoperative liver dysfunction.The regression equation,which is established by combination of ICG R15 with SRLV,can predict the occurrence of postoperative liver dysfunction.The accuracy of ICG R15 on postoperative day 3 combined with SRLV is better than that of single ICG R15 or single SRLV.

Evaluation of different types of inferior vena cava cannulae placement by transesophageal echocardiography and its impact on hepatic dysfunction.

BACKGROUND: Postoperative hepatic dysfunction may occur in an otherwise uncomplicated open heart surgery. One of the reasons is malpositioning of the inferior vena cava (IVC) cannula in the hepatic vein (HV) or beyond. A straight cannula is considered more likely to be malpositioned compared to the angled cannula and a malpositioned cannula can lead to hepatic dysfunction. METHODS: In this prospective study, forty adult patients undergoing atrial septal defect repair were randomized into two groups as: straight cannula group (n=20) and angled cannula group (n=20). The cannula position was assessed by transesophageal echocardiography (TEE) (hepatic vein view). Alanine aminotransferase levels (ALT) and bilirubin levels were measured immediately, at 6 hours and on day 1, day 2 and day 7 after surgery as a marker of hepatic injury. RESULTS: TEE localization of the IVC cannula was achieved in all patients except one. Visualization was good in 85% of patients. A cannula in the HV or beyond the HV in the IVC was considered malpositioned. The number of cases of cannula malposition was 10 (50%) and 4 (20%) in the straight and angled cannula groups, respectively. The pattern of change in serum bilirubin and liver enzymes levels in the postoperative period was similar in both the groups (p>0.05). The mean distance between the right atrium (RA) - inferior vena cava (IVC) junction to the hepatic vein was 1.94±0.56 cm and the mean diameters of the IVC and HV were 1.95±0.5 and 1.31±0.33 cm, respectively. CONCLUSION: TEE can be used to monitor IVC cannula position. A higher frequency of cannula malposition was observed with the straight cannula compared to the angled cannula, but was not found to be associated with hepatic dysfunction.