The Impact of Preoperative Variables on Intraoperative Blood Loss and Transfusion Requirements During Orthotopic Liver Transplant

Application of machine learning model in predicting the likelihood of blood transfusion after hip fracture surgery

OBJECTIVE

Anemia is one of the common adverse reactions after hip fracture surgery. The traditional method to solve anemia is allogeneic transfusion. However, the transfusion may lead to some complications such as septicemia and fever. So far, few studies have reported roles of machine learning in predicting whether blood transfusion is needed or not after hip fracture surgery. Therefore, the purpose of this study is to develop machine learning models to predict the likelihood of postoperative blood transfusion in patients undergoing hip fracture surgery.

METHODS

This study enrolled 1355 patients who underwent hip fracture surgery at the Affiliated Hospital of Qingdao University from January 2016 to December 2021. Among all patients, 210 cases received postoperative blood transfusion. All patients were randomly divided into a training group and a testing group at a ratio of 7:3. In the training group, univariate and multivariate logistic regression analyses were used to determine independent risk factors for the postoperative transfusion. Then, based on these independent risk factors, tenfold cross-validation method was utilized to develop five machine learning models, including logistic, multilayer perceptron (MLP), extreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM). The receiver operating characteristic (ROC) curve, area under ROC curve (AUC), and Matthews correlation coefficient (MCC) were generated to evaluate the performance of the models. Calibration plot and decision curve analysis (DCA) were used to test the performance, stability, and clinical applicability of the models. The models were validated using the testing group; and the ROC curve, MCC, calibration plot, and DCA curves were also generated to validate the performance, stability, and clinical applicability of the models. To further verify the robustness of the model, we randomly grabbed 70% of the samples in the testing set, performed 1000 iterations, and calculated the AUC and confidence interval of the five models. Finally, we used SHapley Additive exPlanations (SHAP) to explain these models.

RESULTS

Multivariate logistic regression analysis showed that there were 8 independent risk factors, including age, blood transfusion history, albumin (ALB), globulin (GLO), total bilirubin (TBIL), indirect bilirubin (IBIL), hemoglobin (HB), and blood loss > 200 ml. We finally selected five independent risk factors including HB, GLO, age, IBIL, and blood loss > 200 ml. Based on these five independent risk factors, we generated six characteristic variables, namely HB, HB × HB, HB × blood loss, GLO × HB, age, age × IBIL, and established five machine learning models using a tenfold cross-validation method. In the training group, the AUC values of logistic, RF, MLP, SVM, and XGB were 0.9320, 0.8911, 0.9327, 0.9225, and 0.8825, respectively, and the average AUC was 0.9122 ± 0.0212. The MCC values were 0.65, 0.77, 0.65, 0.66, and 0.68, respectively, and the calibration plot and DCA performed well. In the testing group the AUC values of logistic, RF, MLP, SVM, and XGB were 0.8483, 0.7978, 0.8576, 0.8598, and 0.8216, respectively. The average AUC was 0.8370 ± 0.0238, and the MCC values were 0.41, 0.35, 0.40, 0.41, and 0.41, respectively. The calibration plot and DCA in the testing group also showed good performance. The AUC values and confidence intervals of the 1000-iteration model were: logistic (AUC, min confidence interval [CI]-max confidence interval [CI] 0.848, 0.804-0.903), RF (AUC, minCI-maxCI 0.797, 0.734-0.857), MLP (AUC, minCI-maxCI 0.858, 0.812-0.902), SVM (AUC, minCI-maxCI 0.859, 0.819-0.910), and XGB (AUC, minCI-maxCI 0.821, 0.764-0.894). The model performed well. Finally, according to SHAP, among all five models, HB played the most important role in model prediction and interpretation.

CONCLUSION

The five models we developed all performed well in predicting the likelihood of blood transfusion after hip fracture surgery. Therefore, we believed that the prediction model based on machine learning had great application prospects in clinical practice, which could help clinicians better predict the risk of blood transfusion after hip fracture surgery.

Liver Transplantation and the Elderly Candidate: Perioperative Considerations

Pioneered by Thomas Starzl in the early 1970s, liver transplant (LT) is nowadays often considered a final intervention and standard of care to cure many forms of acute and chronic end-stage liver diseases. Started in recipients younger than 60 years old, LT indications are now much broader, and at least, one-fifth of the candidates are older than 65 years. Problems associated with ageing and frailty in LT recipients and their impact on the entire perioperative course are discussed according to a modern anesthesiological perspective and the anesthesiologist covering the role of the perioperative (transplant) physician.

[Using Liver Function Indicators to Predict Massive Blood Transfusion in Orthotopic Liver Transplantation]

Objective

To explore the predictive effect of preoperative liver function indicators for intraoperative massive blood transfusion in orthotopic liver transplantation and to establish a prediction model.

Methods

We retrospectively analyzed the relevant data of 607 patients who underwent orthotopic liver transplantation in the Department of Liver Surgery, West China Hospital, Sichuan University between January 1, 2015 and June 30, 2021. According to the intraoperative transfusion volume of leukocyte-reduced red blood cells in additive solution, the patients were divided into a massive blood transfusion (MBT) group and a non-massive blood transfusion (NMBT) group. Univariate and multivariate logistic regressions were performed to analyze the risk factors of intraoperative MBT in orthotopic liver transplantation, the calibration of the predictive model was assessed by Hosmer-Lemeshow test, and the discrimination power of the predictive model was measured by area under the curve ( AUC) of the receiver operating characteristic (ROC) curve.

Results

According to the results of logistic regression, alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), direct bilirubin (DBIL), albumin (ALB), and Child-Pugh score showed no correlation with the risk of MBT in orthotopic liver transplantation operation. Platelet count (PLT) (odds ratio [ OR]=0.90, 95% confidence interval [ CI]: 0.09-0.19, P=0.02), international normalized ratio (INR) ( OR=19.43, 95% CI: 7.64-19.44, P<0.01), prothrombin time (PT) ( OR=1.43, 95% CI: 1.25-1.63, P<0.01), and activated partial thromboplastin time (APTT) ( OR=0.92, 95% CI: 0.90-0.95, P<0.01) were identified as the risk factors of intraoperative MBT in orthotopic liver transplantation. The Hosmer-Lemeshow test showed that the predictive model had good calibration ( χ 2=9.06, P=0.48) and discrimination power ( AUC=0.80, 95% CI 0.766-0.834, P<0.01).

Conclusion

A predictive model based on the preoperative PLT, INR, PT, and APTT of patients undergoing orthotopic liver transplantation was established and can be used to predict the risk of intraoperative MBT in liver transplantation patients.

Risk factors analysis of surgical complications of hepatic hemangioma: a modified Clavien-Dindo classification-based study

PURPOSE

There are few studies on the risk factors of postoperative complications after surgical treatment of hepatic hemangioma (HH). This study aims to provide a more scientific reference for clinical treatment.

METHODS

The clinical characteristics and operation data of HH patients undergoing surgical treatment in the First Affiliated Hospital of Air Force Medical University from January 2011 to December 2020 were retrospectively collected. All enrolled patients were divided into two groups based on the modified Clavien-Dindo classification: Major group (Grade II/III/IV/V) and Minor group (Grade I and no complications). Univariate and multivariate regression analysis was used to explore the risk factors for massive intraoperative blood loss (IBL) and postoperative Grade II and above complications.

RESULTS

A total of 596 patients were enrolled, with a median age of 46.0 years (range, 22-75 years). Patients with Grade II/III/IV/V complications were included in the Major group (n = 119, 20%), and patients with Grade I and no complications were included in the Minor group (n = 477, 80%). The results of multivariate analysis of Grade II/III/IV/V complications showed that operative duration, IBL, and tumor size increased the risk of Grade II/III/IV/V complications. Conversely, serum creatinine (sCRE) decreased the risk. The results of multivariate analysis of IBL showed that tumor size, surgical method, and operative duration increased the risk of IBL.

CONCLUSIONS

Operative duration, IBL, tumor size, and surgical method are independent risk factors that should be paid attention to in HH surgery. In addition, as an independent protective factor for HH surgery, sCRE should attract more attention from scholars.

Viscoelastic Management of Coagulopathy during the Perioperative Period of Liver Transplantation

Viscoelastic testing (VET) in liver transplantation (LT) has been used since its origin, in combination with standard laboratory testing (SLT). There are only a few, small, randomized controlled trials that demonstrated a reduction in transfusion rates using VET to guide coagulation management. Retrospective analyses contrasting VET to SLT have demonstrated mixed results, with a recent concern for overtreatment and the increase in postoperative thrombotic events. An oversight of many studies evaluating VET in LT is a single protocol that does not address the different phases of surgery, in addition to pre- and postoperative management. Furthermore, the coagulation spectrum of patients entering and exiting the operating room is diverse, as these patients can have varying anatomic and physiologic risk factors for thrombosis. A single transfusion strategy for all is short sighted. VET in combination with SLT creates the opportunity for personalized resuscitation in surgery which can address the many challenges in LT where patients are at a paradoxical risk for both life-threatening bleeding and clotting. With emerging data on the role of rebalanced coagulation in cirrhosis and hypercoagulability following LT, there are numerous potential roles in VET management of LT that have been unaddressed.

Blood products and liver transplantation: A strategy to balance optimal preparation with effective blood stewardship

BACKGROUND

Unanticipated transfusion requirements during liver transplantation can delay lifesaving intraoperative resuscitation and strain blood bank resources. Risk-stratified preoperative blood preparation can mitigate these deleterious outcomes.

STUDY DESIGN AND METHODS

A two-tiered blood preparation protocol for liver transplantation was retrospectively evaluated. Eleven binary variables served as criteria for high-risk (HR) allocation. Primary outcomes included red blood cell (RBC), plasma (FFP), and platelet (Plt) utilization. Secondary outcomes included product under- and overpreparation. Contingency tables for transfusion requirements above the population means were generated using 15 clinical variables. Modified protocols were developed and retrospectively optimized using the study population.

RESULTS

Of 225 recipients, 102 received HR preoperative orders, which correlated to higher intraoperative transfusion requirements. However, univariate analysis identified only two statistical risk factors per product: Hgb ≤7.8 g/dl (p < .001) and MELD ≥38 (p = .035) for RBCs, Hgb ≤7.8 g/dl (p = .002) and acute alcoholic hepatitis (p = 0.015) for FFP, and Hgb ≤7.8 g/dl (p = .001) and normothermic liver preservation (p = .037) for Plts. Based on these findings, we developed modified protocols for individual products, which were evaluated retrospectively for their effectiveness at reducing under-preparatory events while limiting product overpreparation. Cohort statistics were used to define the preparation strategy for each protocol. Retrospective comparative analysis demonstrated the superiority of the modified protocols by improving the under-preparation rate from 24% to <10% for each product, which required a 1.56-fold and 1.44-fold increase in RBC and FFP overpreparation, respectively. Importantly, there was no difference in Plt overpreparation.

DISCUSSION

We report translatable data-driven blood bank preparation protocols for liver transplantation.

Advancing Prediction of Risk of Intraoperative Massive Blood Transfusion in Liver Transplantation With Machine Learning Models. A Multicenter Retrospective Study

Background

Liver transplantation surgery is often accompanied by massive blood loss and massive transfusion (MT), while MT can cause many serious complications related to high mortality. Therefore, there is an urgent need for a model that can predict the demand for MT to reduce the waste of blood resources and improve the prognosis of patients.

Objective

To develop a model for predicting intraoperative massive blood transfusion in liver transplantation surgery based on machine learning algorithms.

Methods

A total of 1,239 patients who underwent liver transplantation surgery in three large grade lll-A general hospitals of China from March 2014 to November 2021 were included and analyzed. A total of 1193 cases were randomly divided into the training set (70%) and test set (30%), and 46 cases were prospectively collected as a validation set. The outcome of this study was an intraoperative massive blood transfusion. A total of 27 candidate risk factors were collected, and recursive feature elimination (RFE) was used to select key features based on the Categorical Boosting (CatBoost) model. A total of ten machine learning models were built, among which the three best performing models and the traditional logistic regression (LR) method were prospectively verified in the validation set. The Area Under the Receiver Operating Characteristic Curve (AUROC) was used for model performance evaluation. The Shapley additive explanation value was applied to explain the complex ensemble learning models.

Results

Fifteen key variables were screened out, including age, weight, hemoglobin, platelets, white blood cells count, activated partial thromboplastin time, prothrombin time, thrombin time, direct bilirubin, aspartate aminotransferase, total protein, albumin, globulin, creatinine, urea. Among all algorithms, the predictive performance of the CatBoost model (AUROC: 0.810) was the best. In the prospective validation cohort, LR performed far less well than other algorithms.

Conclusion

A prediction model for massive blood transfusion in liver transplantation surgery was successfully established based on the CatBoost algorithm, and a certain degree of generalization verification is carried out in the validation set. The model may be superior to the traditional LR model and other algorithms, and it can more accurately predict the risk of massive blood transfusions and guide clinical decision-making.

Procedure-related bleeding risk in patients with cirrhosis and severe thrombocytopenia

BACKGROUND

Gaps of knowledge still exist about the potential association between severe thrombocytopenia and increased risk of procedure-associated bleeding in patients with liver disease.

METHODS

In this narrative review, we aimed at examining the association between procedure-related bleeding risk and platelet count in patients with cirrhosis and severe thrombocytopenia in various settings. We updated to 2020 a previously conducted literature search using MEDLINE/PubMed and EMBASE. The search string included clinical studies, adult patients with chronic liver disease and thrombocytopenia undergoing invasive procedures, any interventions and comparators, and haemorrhagic events of any severity as outcome.

RESULTS

The literature search identified 1276 unique publications, and 15 studies met the inclusion criteria and were analysed together with those identified by the previous search. Most of the new studies included in our analysis did not assess the association between post-procedural bleeding risk and platelet count alone in patients with chronic liver disease. Furthermore, some results could have been biased by prophylactic platelet transfusions. A few studies found that severe thrombocytopenia may be predictive of bleeding following percutaneous liver biopsy, dental extractions, percutaneous ablation of liver tumours and endoscopic polypectomy.

CONCLUSIONS

Currently available literature cannot support definitive conclusions about the appropriate target platelet counts to improve the risk of bleeding in cirrhotic patients who underwent invasive procedures; moreover, it showed enormous variability in the use of prophylactic platelet transfusions.

The Association between Prognostic Nutritional Index (PNI) and Intraoperative Transfusion in Patients Undergoing Hepatectomy for Hepatocellular Carcinoma: A Retrospective Cohort Study

BACKGROUND

PNI is significantly associated with surgical outcomes; however, the association between PNI and intraoperative transfusions is unknown.

METHODS

This study retrospectively analyzed 1065 patients who underwent hepatectomy. We divided patients into two groups according to the PNI (<44 and >44) and compared their transfusion rates and surgical outcomes. We performed multivariate logistic and Cox regression analysis to determine risk factors for transfusion and the 5-year survival. Additionally, we found the net reclassification index (NRI) to validate the discriminatory power of PNI.

RESULTS

The PNI <44 group had higher transfusion rates (adjusted odds ratio [OR]: 2.20, 95%CI: 1.06-4.60, p = 0.035) and poor surgical outcomes, such as post hepatectomy liver failure (adjusted [OR]: 3.02, 95%CI: 1.87-4.87, p < 0.001), and low 5-year survival (adjusted OR: 1.68, 95%CI: 1.17-2.24, p < 0.001). On multivariate analysis, PNI <44, age, hemoglobin, operation time, synthetic colloid use, and laparoscopic surgery were risk factors for intraoperative transfusion. On Cox regression analysis, PNI <44, MELD score, TNM staging, synthetic colloid use, and transfusion were associated with poorer 5-year survival. NRI analysis showed significant improvement in the predictive power of PNI for transfusion (p = 0.002) and 5-year survival (p = 0.004).

CONCLUSIONS

Preoperative PNI <44 was significantly associated with higher transfusion rates and surgical outcomes.

Blood Cell Salvage and Autotransfusion Does Not Worsen Oncologic Outcomes Following Liver Transplantation with Incidental Hepatocellular Carcinoma: A Propensity Score-Matched Analysis

BACKGROUND

Intraoperative blood cell salvage and autotransfusion (IBSA) during liver transplantation (LT) for hepatocellular carcinoma (HCC) is controversial for concern regarding adversely impacting oncologic outcomes.

OBJECTIVE

We aimed to evaluate the long-term oncologic outcomes of patients who underwent LT with incidentally discovered HCC who received IBSA compared with those who did not receive IBSA.

METHODS

Patients undergoing LT (January 2001-October 2018) with incidental HCC on explant pathology were retrospectively identified. A 1:1 propensity score matching (PSM) was performed. HCC recurrence and patient survival were compared. Kaplan-Meier survival analyses were performed, and univariable Cox proportional hazard analyses were performed for risks of recurrence and death.

RESULTS

Overall, 110 patients were identified (IBSA, n = 76 [69.1%]; non-IBSA, n = 34 [30.9%]). Before matching, the groups were similar in terms of demographics, transplant, and tumor characteristics. Overall survival was similar for IBSA and non-IBSA at 1, 3, and 5 years (96.0%, 88.4%, 83.0% vs. 97.1%, 91.1%, 87.8%, respectively; p = 0.79). Similarly, the recurrence rate at 1, 3, and 5 years was not statistically different (IBSA 0%, 1.8%, 1.8% vs. non-IBSA 0%, 3.2%, 3.2%, respectively; p = 0.55). After 1:1 matching (26 IBSA, 26 non-IBSA), Cox proportional hazard analysis demonstrated similar risk of death and recurrence between the groups (IBSA hazard ratio [HR] of death 1.26, 95% confidence interval [CI] 0.52-3.05, p = 0.61; and HR of recurrence 2.64, 95% CI 0.28-25.30, p = 0.40).

CONCLUSIONS

IBSA does not appear to adversely impact oncologic outcomes in patients undergoing LT with incidental HCC. This evidence further supports the need for randomized trials evaluating the impact of IBSA use in LT for HCC.

Intravenous transfusion of iron sucrose reduces blood transfusions and improves postoperative anaemia after a second thoracotomy: a propensity-score matching study

Objective

To evaluate the efficacy of intravenous iron sucrose (IS) in the perioperative period of a second thoracotomy after open-heart surgery.

Methods

This study enrolled patients that underwent a second thoracotomy after previous open-heart surgery. Propensity-score matching was used to match patients that received oral iron therapy with those that received IS after admission. The baseline demographic and clinical characteristics, perioperative data and adverse outcomes were compared between these two matched groups.

Results

The study enrolled 447 patients: of these, 148 received oral iron therapy and effective dietary guidance; and 299 patients received 300 mg IS per week. After matching the propensity scores, there were no significant differences between the two groups in terms of age, sex, smoking history, incidence of major cardiac adverse events, Euro-score, time between the two thoracotomies, application of recombinant erythropoietin, haemoglobin (Hb) on admission and red blood cell (RBC) count on admission. The differences between the two groups after propensity-score matching in terms of the amount of RBC input and Hb after the operation were significant.

Conclusion

The administration of IS might be useful for improving postoperative Hb levels and reducing intraoperative RBC input in patients undergoing a second thoracotomy.

Risk Factors for and Outcomes of Intraoperative Blood Loss in Liver Resection for Hepatocellular Tumors

Intraoperative blood loss (IBL) during liver resection is a predictor of morbidity, mortality, and tumor recurrence after hepatectomy; however, there have been few reports on patient factors associated with increased IBL. We enrolled consecutive patients who underwent liver resection for primary liver malignancies, and evaluated the predictors of IBL using a data set in which factors that might influence IBL, such as surgical devices, methods and anesthetic technique, were all standardized. We studied 244 patients. A multivariate analysis revealed that higher IBL was an independent risk factor for post-hepatectomy liver failure grade ≥B and overall survival. Multiple linear regression analyses showed serum creatinine, clinically significant portal hypertension (CSPH), tumor size, and major hepatectomy were all significant predictors of IBL. In conclusion, higher IBL was significantly associated with increased morbidity and mortality in patients with primary HCC who underwent liver resection. The risk of IBL was related to several factors including tumor size, serum creatinine, CSPH, and major hepatectomy.

Prediction of perioperative transfusions using an artificial neural network

BACKGROUND

Accurate prediction of operative transfusions is essential for resource allocation and identifying patients at risk of postoperative adverse events. This research examines the efficacy of using artificial neural networks (ANNs) to predict transfusions for all inpatient operations.

METHODS

Over 1.6 million surgical cases over a two year period from the NSQIP-PUF database are used. Data from 2014 (750937 records) are used for model development and data from 2015 (885502 records) are used for model validation. ANN and regression models are developed to predict perioperative transfusions for surgical patients.

RESULTS

Various ANN models and logistic regression, using four variable sets, are compared. The best performing ANN models with respect to both sensitivity and area under the receiver operator characteristic curve outperformed all of the regression models (p < .001) and achieved a performance of 70-80% specificity with a corresponding 75-62% sensitivity.

CONCLUSION

ANNs can predict >75% of the patients who will require transfusion and 70% of those who will not. Increasing specificity to 80% still enables a sensitivity of almost 67%. The unique contribution of this research is the utilization of a single ANN model to predict transfusions across a broad range of surgical procedures.