Development and Validation of a Nomogram Based on Perioperative Factors to Predict Post-hepatectomy Liver Failure

Realizing Textbook Outcomes Following Liver Resection for Hepatic Neoplasms with Development and Validation of a Predictive Nomogram

BACKGROUND

'Textbook Outcome' (TO) represents an effort to define a standardized, composite quality benchmark based on intraoperative and postoperative endpoints. This study aimed to assess the applicability of TO as an outcome measure following liver resection for hepatic neoplasms from a low- to middle-income economy and determine its impact on long-term survival. Based on identified perioperative predictors, we developed and validated a nomogram-based scoring and risk stratification system.

METHODS

We retrospectively analyzed patients undergoing curative resections for hepatic neoplasms between 2012 and 2023. Rates of TO were assessed over time and factors associated with achieving a TO were evaluated. Using stepwise regression, a prediction nomogram for achieving TO was established based on perioperative risk factors.

RESULTS

Of the 1018 consecutive patients who underwent liver resections, a TO was achieved in 64.9% (661/1018). The factor most responsible for not achieving TO was significant post-hepatectomy liver failure (22%). Realization of TO was independently associated with improved overall and disease-free survival. On logistic regression, American Society of Anesthesiologists score of 2 (p = 0.0002), perihilar cholangiocarcinoma (p = 0.011), major hepatectomy (p = 0.0006), blood loss >1500 mL (p = 0.007), and presence of lymphovascular emboli on pathology (p = 0.026) were associated with the non-realization of TO. These independent risk factors were integrated into a nomogram prediction model with the predictive efficiency for TO (area under the curve 75.21%, 95% confidence interval 70.69-79.72%).

CONCLUSION

TO is a realizable outcome measure and should be adopted. We recommend the use of the nomogram proposed as a convenient tool for patient selection and prognosticating outcomes following hepatectomy.

Online interpretable dynamic prediction models for clinically significant posthepatectomy liver failure based on machine learning algorithms: a retrospective cohort study

BACKGROUND

Posthepatectomy liver failure (PHLF) is the leading cause of mortality in patients undergoing hepatectomy. However, practical models for accurately predicting the risk of PHLF are lacking. This study aimed to develop precise prediction models for clinically significant PHLF.

METHODS

A total of 226 patients undergoing hepatectomy at a single center were recruited. The study outcome was clinically significant PHLF. Five preoperative and postoperative machine learning (ML) models were developed and compared with four clinical scores, namely, the MELD, FIB-4, ALBI, and APRI scores. The robustness of the developed ML models was internally validated using fivefold cross-validation (CV) by calculating the average of the evaluation metrics and was externally validated on an independent temporal dataset, including the area under the curve (AUC) and the area under the precision-recall curve (AUPRC). SHapley Additive exPlanations analysis was performed to interpret the best performance model.

RESULTS

Clinically significant PHLF was observed in 23 of 226 patients (10.2%). The variables in the preoperative model included creatinine, total bilirubin, and Child-Pugh grade. In addition to the above factors, the extent of resection was also a key variable for the postoperative model. The preoperative and postoperative artificial neural network (ANN) models exhibited excellent performance, with mean AUCs of 0.766 and 0.851, respectively, and mean AUPRC values of 0.441 and 0.645, whereas the MELD, FIB-4, ALBI, and APRI scores reached AUCs of 0.714, 0.498, 0.536, and 0.551, respectively, and AUPRC values of 0.204, 0.111, 0.128, and 0.163, respectively. In addition, the AUCs of the preoperative and postoperative ANN models were 0.720 and 0.731, respectively, and the AUPRC values were 0.380 and 0.408, respectively, on the temporal dataset.

CONCLUSION

Our online interpretable dynamic ML models outperformed common clinical scores and could function as a clinical decision support tool to identify patients at high risk of PHLF preoperatively and postoperatively.

Nomogram based on liver stiffness and spleen area with ultrasound for posthepatectomy liver failure: A multicenter study

BACKGROUND

Liver stiffness (LS) measurement with two-dimensional shear wave elastography (2D-SWE) correlates with the degree of liver fibrosis and thus indirectly reflects liver function reserve. The size of the spleen increases due to tissue proliferation, fibrosis, and portal vein congestion, which can indirectly reflect the situation of liver fibrosis/cirrhosis. It was reported that the size of the spleen was related to posthepatectomy liver failure (PHLF). So far, there has been no study combining 2D-SWE measurements of LS with spleen size to predict PHLF. This prospective study aimed to investigate the utility of 2D-SWE assessing LS and spleen area (SPA) for the prediction of PHLF in hepatocellular carcinoma (HCC) patients and to develop a risk prediction model.

AIM

To investigate the utility of 2D-SWE assessing LS and SPA for the prediction of PHLF in HCC patients and to develop a risk prediction model.

METHODS

This was a multicenter observational study prospectively analyzing patients who underwent hepatectomy from October 2020 to March 2022. Within 1 wk before partial hepatectomy, ultrasound examination was performed to measure LS and SPA, and blood was drawn to evaluate the patient's liver function and other conditions. Least absolute shrinkage and selection operator logistic regression and multivariate logistic regression analysis was applied to identify independent predictors of PHLF and develop a nomogram. Nomogram performance was validated further. The diagnostic performance of the nomogram was evaluated with receiver operating characteristic curve compared with the conventional models, including the model for end-stage liver disease (MELD) score and the albumin-bilirubin (ALBI) score.

RESULTS

A total of 562 HCC patients undergoing hepatectomy (500 in the training cohort and 62 in the validation cohort) were enrolled in this study. The independent predictors of PHLF were LS, SPA, range of resection, blood loss, international normalized ratio, and total bilirubin. Better diagnostic performance of the nomogram was obtained in the training [area under receiver operating characteristic curve (AUC): 0.833; 95% confidence interval (95%CI): 0.792-0.873; sensitivity: 83.1%; specificity: 73.5%] and validation (AUC: 0.802; 95%CI: 0.684-0.920; sensitivity: 95.5%; specificity: 52.5%) cohorts compared with the MELD score and the ALBI score.

CONCLUSION

This PHLF nomogram, mainly based on LS by 2D-SWE and SPA, was useful in predicting PHLF in HCC patients and presented better than MELD score and ALBI score.

Application and impact of Lasso regression in gastroenterology: A systematic review

Least absolute shrinkage and selection operator (Lasso) regression is a statistical technique that can be used to study the effects of clinical variables in outcome prediction. In this study, we aimed at systematically reviewing the application of Lasso regression in gastroenterology for developing predictive models and providing a method of performing Lasso regression. A comprehensive search strategy was conducted in PubMed, Embase and Cochrane CENTRAL databases (Keywords: lasso regression; gastrointestinal tract/diseases) following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Studies were screened for eligibility based on pre-defined selection criteria and the data was extracted using a standardized form. Total 16 studies were included, comprising a diverse range of gastroenterological disease-related outcomes. Sample sizes ranged from 134 to 8861 subjects. Eleven studies reported liver disease-related prediction models, while five focused on non-hepatic etiology models. Lasso regression was applied for variable selection, risk prediction and model development, with various validation methods and performance metrics used. Model performance metrics included Area Under the Receiver Operating Characteristics (AUROC), C-index and calibration plots. In gastroenterology, Lasso regression has been used in various diseases such as inflammatory bowel disease, liver disease and esophageal cancer. It is valuable for complex scenarios with many predictors. However, its effectiveness depends on high-quality and complete data. While it identifies important variables, it doesn't provide causal interpretations. Therefore, cautious interpretation is necessary considering the study design and data quality.

A comprehensive preoperative predictive score for post-hepatectomy liver failure after hepatocellular carcinoma resection based on patient comorbidities, tumor burden, and liver function: the CTF score

BACKGROUND

Post-hepatectomy liver failure (PHLF) is a dreaded complication following liver resection for hepatocellular carcinoma (HCC) with a high mortality rate. We sought to develop a score based on preoperative factors to predict PHLF.

METHODS

Patients who underwent resection for HCC between 2000 and 2020 were identified from an international multi-institutional database. Factors associated with PHLF were identified and used to develop a preoperative comorbidity-tumor burden-liver function (CTF) predictive score.

RESULTS

Among 1785 patients, 106 (5.9%) experienced PHLF. On multivariate analysis, several factors were associated with PHLF including high Charlson comorbidity index (CCI ≥ 5) (OR 2.80, 95%CI, 1.08-7.26), albumin-bilirubin (ALBI) (OR 1.99, 95%CI, 1.10-3.56), and tumor burden score (TBS) (OR 1.06, 95%CI, 1.02-1.11) (all p < 0.05). Using the beta-coefficients of these variables, a weighted predictive score was developed and made available online ( https://alaimolaura.shinyapps.io/PHLFriskCalculator/ ). The CTF score (c-index = 0.67) performed better than Child-Pugh score (CPS) (c-index = 0.53) or Barcelona clinic liver cancer system (BCLC) (c-index = 0.57) to predict PHLF. A high CTF score was also an independent adverse prognostic factor for survival (HR 1.61, 95%CI, 1.12-2.30) and recurrence (HR 1.36, 95%CI, 1.08-1.71) (both p = 0.01).

CONCLUSION

Roughly 1 in 20 patients experienced PHLF following resection of HCC. Patient (i.e., CCI), tumor (i.e., TBS), and liver function (i.e., ALBI) factors were associated with risk of PHLF. These preoperative factors were incorporated into a novel CTF tool that was made available online, which outperformed other previously proposed tools.

Machine learning prediction model for post- hepatectomy liver failure in hepatocellular carcinoma: A multicenter study

Introduction

Post-hepatectomy liver failure (PHLF) is one of the most serious complications and causes of death in patients with hepatocellular carcinoma (HCC) after hepatectomy. This study aimed to develop a novel machine learning (ML) model based on the light gradient boosting machines (LightGBM) algorithm for predicting PHLF.

Methods

A total of 875 patients with HCC who underwent hepatectomy were randomized into a training cohort (n=612), a validation cohort (n=88), and a testing cohort (n=175). Shapley additive explanation (SHAP) was performed to determine the importance of individual variables. By combining these independent risk factors, an ML model for predicting PHLF was established. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, and decision curve analyses (DCA) were used to evaluate the accuracy of the ML model and compare it to that of other noninvasive models.

Results

The AUCs of the ML model for predicting PHLF in the training cohort, validation cohort, and testing cohort were 0.944, 0.870, and 0.822, respectively. The ML model had a higher AUC for predicting PHLF than did other non-invasive models. The ML model for predicting PHLF was found to be more valuable than other noninvasive models.

Conclusion

A novel ML model for the prediction of PHLF using common clinical parameters was constructed and validated. The novel ML model performed better than did existing noninvasive models for the prediction of PHLF.

Development and validation of a nomogram for patients with stage II/III gastric adenocarcinoma after radical surgery

BACKGROUND

We aimed to construct nomograms based on clinicopathological features and routine preoperative hematological indices to predict cancer-specific survival (CSS) and disease-free survival (DFS) in patients with stage II/III gastric adenocarcinoma (GA) after radical resection.

METHODS

We retrospectively analyzed 468 patients with stage II/III GA after curative gastrectomy between 2012 and 2018; 70% of the patients were randomly assigned to the training set (n = 327) and the rest were assigned to the validation set (n = 141). The nomogram was constructed from independent predictors derived from the Cox regression in the training set. Using the consistency index, the calibration and the time-dependent receiver operating characteristic curves were used to evaluate the accuracy of the nomogram. Decision curve analysis was used to assess the value of the model in clinical applications. Patients were further divided into low- and high-risk groups based on the nomogram risk score.

RESULTS

Multivariate Cox model identified depth of invasion, lymph node invasion, tumor differentiation, adjuvant chemotherapy, CA724, and platelet-albumin ratio as covariates associated with CSS and DFS. CA199 is a risk factor unique to CSS. The nomogram constructed using the results of the multivariate analysis showed high accuracy with a consistency index of 0.771 (CSS) and 0.771 (DFS). Moreover, the area under the curve values for the 3-and 5-year CSS were 0.868 and 0.918, and the corresponding values for DFS were 0.872 and 0.919, respectively. The nomogram had a greater clinical benefit than the TNM staging system. High-risk patients based on the nomogram had a worse prognosis than low-risk patients.

CONCLUSION

The prognostic nomogram for patients with stage II/III GA after radical gastrectomy established in this study has a good predictive ability, which is helpful for doctors to accurately evaluate the prognosis of patients to make more reasonable treatment plans.

A Novel Nomogram for Predicting Postoperative Liver Failure After Major Hepatectomy for Hepatocellular Carcinoma

Background

Post-hepatectomy liver failure (PHLF) is the most common cause of mortality after major hepatectomy in hepatocellular carcinoma (HCC) patients. We aim to develop a nomogram to preoperatively predict grade B/C PHLF defined by the International Study Group on Liver Surgery Grading (ISGLS) in HCC patients undergoing major hepatectomy.

Study Design

The consecutive HCC patients who underwent major hepatectomy at the Eastern Hepatobiliary Surgery Hospital between 2008 and 2013 served as a training cohort to develop a preoperative nomogram, and patients from 2 other hospitals comprised an external validation cohort. Least absolute shrinkage and selection operator (LASSO) logistic regression was applied to identify preoperative predictors of grade B/C PHLF. Multivariable logistic regression was utilized to establish a nomogram model. Internal and external validations were used to verify the performance of the nomogram. The accuracy of the nomogram was also compared with the conventional scoring models, including MELD and ALBI score.

Results

A total of 880 patients who underwent major hepatectomy (668 in the training cohort and 192 in the validation cohort) were enrolled in this study. The independent risk factors of grade B/C PHLF were age, gender, prothrombin time, total bilirubin, and CSPH, which were incorporated into the nomogram. Good prediction discrimination was achieved in the training (AUROC: 0.73) and validation (AUROC: 0.72) cohorts. The calibration curve also showed good agreement in both training and validation cohorts. The nomogram has a better performance than MELD and ALBI score models.

Conclusion

The proposed nomogram showed more accurate ability to individually predict grade B/C PHLF after major hepatectomy in HCC patients than MELD and ALBI scores.

Nomogram to Predict the Survival of Chinese Patients with Alcohol-Related Liver Disease

OBJECTIVES

Alcohol-related liver disease is an increasing public health burden in China, but there is a lack of models to predict its prognosis. This study established a nomogram for predicting the survival of Chinese patients with alcohol-related liver disease (ALD).

METHODS

Hospitalized alcohol-related liver disease patients were retrospectively enrolled from 2015 to 2018 and followed up for 24 months to evaluate survival profiles. A total of 379 patients were divided into a training cohort (n = 265) and validation cohort (n = 114). Cox proportional hazard survival analysis identified survival factors of the patients in the training cohort. A nomogram was built and internally validated.

RESULTS

The 3-month, 6-month, 12-month, and 24-month survival rates for the training cohort were 82.6%, 81.1%, 74.3%, and 64.5%, respectively. The Cox analysis showed relapse (P=0.001), cirrhosis (P=0.044), liver cancer (P < 0.001), and a model for end-stage liver diseases score of ≥21 (P=0.041) as independent prognostic factors. A nomogram was built, which predicted the survival of patients in the training cohort with a concordance index of 0.749 and in the internal validation cohort with a concordance index of 0.756.

CONCLUSION

The long-term survival of Chinese alcohol-related liver disease patients was poor with a 24-month survival rate of 64.5%. Relapse, cirrhosis, liver cancer, and a model for end-stage liver disease score of ≥21 were independent risk factors for those patients. A nomogram was developed and internally validated for predicting the probability of their survival at different time points.