Prediction of Post-Operative Liver Dysfunction by Serum Markers of Liver Fibrosis in Hepatocellular Carcinoma

Application of Interpretable Machine Learning Models to Predict the Risk Factors of HBV-Related Liver Cirrhosis in CHB Patients Based on Routine Clinical Data: A Retrospective Cohort Study

Chronic hepatitis B (CHB) infection represents a significant global public health issue, often leading to hepatitis B virus (HBV)-related liver cirrhosis (HBV-LC) with poor prognoses. Early identification of HBV-LC risk is essential for timely intervention. This study develops and compares nine machine learning (ML) models to predict HBV-LC risk in CHB patients using routine clinical and laboratory data. A retrospective analysis was conducted involving 777 CHB patients, with 50.45% (392/777) progressing to HBV-LC. Admission data consisted of 52 clinical and laboratory variables, with missing values addressed using multiple imputation. Feature selection utilized Least Absolute Shrinkage and Selection Operator (LASSO) regression and the Boruta algorithm, identifying 24 key variables. The evaluated ML models included XGBoost, logistic regression (LR), LightGBM, random forest (RF), AdaBoost, Gaussian naive Bayes (GNB), multilayer perceptron (MLP), support vector machine (SVM), and k-nearest neighbors (KNN). The data set was partitioned into an 80% training set (n = 621) and a 20% independent testing set (n = 156). Cross-validation (CV) facilitated hyperparameter tuning and internal validation of the optimal model. Performance metrics included the area under the receiver operating characteristic curve (AUC), Brier score, accuracy, sensitivity, specificity, and F1 score. The RF model demonstrated superior performance, with AUCs of 0.992 (training) and 0.907 (validation), while the reconstructed model achieved AUCs of 0.944 (training) and 0.945 (validation), maintaining an AUC of 0.863 in the testing set. Calibration curves confirmed a strong alignment between observed and predicted probabilities. Decision curve analysis indicated that the RF model provided the highest net benefit across threshold probabilities. The SHAP algorithm identified RPR, PLT, HBV DNA, ALT, and TBA as critical predictors. This interpretable ML model enhances early HBV-LC prediction and supports clinical decision-making in resource-limited settings.

A Multi-Institutional Study on Ultrasound Image Analysis for Staging HBV-Derived Liver Fibrosis: A Potential Noninvasive Alternative to Liver Stiffness Measurement

INTRODUCTION

Liver stiffness measurement is principal for staging liver fibrosis but not included in routine examinations. We investigated whether comparable diagnostic performance can be achieved by mining ultrasound images and developing a novel serum index (NSI).

METHODS

Texture features were extracted from ultrasound images. Spearman correlation and logistics regression selected independent variables for significant (F ≥ 2) and advanced (F ≥ 3) fibrosis. We compared the diagnostic performance of transient elastography (TE), ultrasound image biomarker, conventional serum indices (aspartate aminotransferase-to-platelet ratio index, fibrosis-4 index, gamma-glutamyl transpeptidase-to-platelet ratio), and NSI in 365 patients with chronic hepatitis B.

RESULTS

Among patients, 52.1% had significant fibrosis and 24.2% had advanced fibrosis. PLT, gamma-glutamyl transferase, prealbumin, and globulin were incorporated into NSI. In the validation group, TE achieved the best performance (area under the curve [AUC]: 0.765 [0.690-0.849] for significant fibrosis; 0.812 [0.745-0.878] for advanced fibrosis), followed by ultrasound image biomarker (AUC: 0.712 [0.629-0.795]; 0.678 [0.595-0.763]) and NSI (AUC: 0.630 [0.534-0.725]; 0.659 [0.572-0.745]), outperforming conventional indices.

DISCUSSION

Texture analysis enhances ultrasound's diagnostic utility, but TE remains superior. When TE is unavailable, ultrasound image analysis and NSI, incorporating prealbumin, can serve as alternative tools for fibrosis staging.

The preoperative M2BPGi score predicts operative difficulty and the incidence of postoperative complications in laparoscopic liver resection

BACKGROUND

Liver fibrosis or cirrhosis frequently makes parenchymal transection more difficult, but the difficulty score of laparoscopic liver resection (LLR), including the IWATE criteria, does not include a factor related to liver fibrosis. Therefore, this study aimed to evaluate M2BPGi as a predictor of the difficulty of parenchymal transection and the incidence of postoperative complications in LLR.

METHODS

Data from 54 patients who underwent laparoscopic partial liver resection (LLR-P) and 24 patients who underwent laparoscopic anatomical liver resection between 2017 and 2019 in our institution were retrospectively analyzed. All cases were classified according to M2BPGi scores, and reserve liver function, intraoperative blood loss, and postoperative complications were compared among these groups.

RESULTS

Sixteen cases (29.6%) were M2BPGi negative (cut-off index < 1.0), 25 cases (46.3%) were 1+ (1.0 ≤ cut-off index < 3.0), and 13 cases (24.1%) were 2+ (cut-off index ≥ 3.0). M2BPGi-positive cases had significantly worse hepatic reserve function (K-ICG: 0.16 vs 0.14 vs 0.08, p < 0.0001). Intraoperative bleeding was significantly greater in M2BPGi-positive cases [50 ml vs 150 ml vs 200 ml, M2BPGi (-) or (1+) vs M2BPGi (2+), p = 0.045]. Postoperative complications (Clavien-Dindo ≥ II) were significantly more frequent in M2BPGi-positive cases [0% vs 4% vs 33%, M2BPGi (-) or (1+) vs M2BPGi (2+), p = 0.001].

CONCLUSION

M2BPGi could predict surgical difficulty and complications in LLR-P. In particular, it might be better not to select M2BPGi (2+) cases as teaching cases because of the massive bleeding during parenchymal transection.

Development and validation of a non-invasive model for diagnosing HBV-related liver cirrhosis

BACKGROUND

Liver cirrhosis is closely related to the abnormal liver function and occurrence of liver cancer. Accurate non-invasive assessment of liver cirrhosis is of great significance for preventing disease progression and treatment decision-making. We aim to develop and validate a non-invasive diagnostic model for liver cirrhosis in patients with chronic hepatitis B (CHB).

METHODS

From July 2015 to April 2017, seven-hundred fifty-four patients with primary HBV-related liver cancer who underwent hepatectomy were reprospectively recruited. All patients were examined with 2D-SWE and serologic testing preoperatively, which were utilized for measurement of liver stiffness and serum fibrosis models. The stage of liver fibrosis was evaluated using a resected liver specimen. Least absolute shrinkage and selection operator (Lasso) regression was used for feature selection and binary logistic regression analysis was chosen to build a diagnostic model, which was presented as a nomogram and evaluated for calibration, discrimination and clinical usefulness. The performance of noninvasive model was then prospectively validated in an independent cohort (361 patients) by the ROC curve analysis.

RESULTS

The diagnostic model, which consists of 5 selected clinical characteristics (PIII-NP, IV-C, Hyaluronan, Platelet and Liver stiffness), showed the strongest correlation with liver fibrosis stage (ρ = 0.702, P < 0.05). Compared with APRI, FIB-4, King's Score, and Forns Index, the model presented the optimal discrimination and the best predictive performance with the highest AUC in the training cohort (0.866, 95%CI 0.840-0.892, P < 0.05) and validation cohorts (0.852, 95%CI 0.813-0.890, P < 0.05). Decision curve analysis demonstrated that nomogram based on the model was extremely useful for diagnosing cirrhosis in patients with chronic hepatitis B.

CONCLUSION

This study proposes a non-invasive diagnostic model that incorporates the clinical predictors which can be conveniently used in the individualized diagnosis of HBV-related liver cirrhosis.

Preoperative Serum Prealbumin Level and Adverse Prognosis in Patients With Hepatocellular Carcinoma After Hepatectomy: A Meta-Analysis

Background

Prealbumin is a sensitive indicator of liver function and nutritional status.

Objectives

This meta-analysis aimed to examine the association of the serum prealbumin level with the prognosis of patients with hepatocellular carcinoma (HCC) undergoing hepatectomy.

Methods

We comprehensively searched the PubMed, Embase, Wanfang, China Academic Journals (CNKI), and SinoMed databases up to September 1, 2021. Eligible studies should report the association of the serum prealbumin level with prognosis and provide the multivariable-adjusted risk estimates of the outcomes of interest in HCC patients undergoing hepatectomy.

Results

A total of 11 studies with 7,442 HCC patients were identified and analyzed. Meta-analysis of a fixed effects model showed that a low serum prealbumin level was associated with poor overall survival [hazard ratio (HR) = 1.54, 95% confidence interval (CI) = 1.42-1.68], recurrence-free survival (HR = 1.34, 95% CI = 1.17-1.52), and a higher risk of postoperative hepatic insufficiency (HR = 2.21; 95% CI = 1.36-3.60) in HCC patients. Sensitivity and subgroup analyses confirmed the robustness of low serum prealbumin in predicting poor overall survival.

Conclusions

This meta-analysis indicated that a low preoperative serum prealbumin level was significantly associated with adverse prognosis in HCC patients undergoing hepatectomy.

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

BACKGROUND AND AIMS

Post-hepatectomy liver failure (PHLF) is a severe complication and main cause of death in patients undergoing hepatectomy. The aim of this study was to build a predictive model of PHLF in patients undergoing hepatectomy.

METHODS

We retrospectively analyzed patients undergoing hepatectomy at Zhongshan Hospital, Fudan University from July 2015 to June 2018, and randomly divided them into development and internal validation cohorts. External validation was performed in an independent cohort. Least absolute shrinkage and selection operator (commonly referred to as LASSO) logistic regression was applied to identify predictors of PHLF, and multivariate binary logistic regression analysis was performed to establish the predictive model, which was visualized with a nomogram.

RESULTS

A total of 492 eligible patients were analyzed. LASSO and multivariate analysis identified three preoperative variables, total bilirubin (p=0.001), international normalized ratio (p<0.001) and platelet count (p=0.004), and two intraoperative variables, extent of resection (p=0.002) and blood loss (p=0.004), as independent predictors of PHLF. The area under receiver operating characteristic curve (referred to as AUROC) of the predictive model was 0.838 and outperformed the model for end-stage liver disease score, albumin-bilirubin score and platelet-albumin-bilirubin score (AUROCs: 0.723, 0.695 and 0.663, respectively; p<0.001 for all). The optimal cut-off value of the predictive model was 14.7. External validation showed the model could predict PHLF accurately and distinguish high-risk patients.

CONCLUSIONS

PHLF can be accurately predicted by this model in patients undergoing hepatectomy, which may significantly contribute to the postoperative care of these patients.

Quantitative Ultrasound Elastography Methods in Focal Liver Lesions Including Hepatocellular Carcinoma: From Diagnosis to Prognosis

ABSTRACT

The ability of ultrasound elastography to diagnose focal liver lesions and determine their prognoses including hepatocellular carcinoma (HCC) is unclear. At present, radiofrequency ablation and liver resection are the most common treatments for HCC. However, the survival rate remains disappointing because of recurrences and postoperative liver failure, necessitating the development of noninvasive approaches. There is currently no systematic definition of an elastic technique for measuring liver stiffness to predict the recurrence of HCC after radiofrequency ablation and postoperative liver failure. In this review, recent advances in ultrasound elastography for the diagnosis and prognosis of focal liver lesions are discussed including HCC.

Comparison of the Extent Classification and the New Complexity Classification of Hepatectomy for Prediction of Surgical Outcomes: a Retrospective Cohort Study

BACKGROUND

In predicting the risk for posthepatectomy complications, hepatectomy is traditionally classified into minor or major resection based on the number of resected segments. Recently, a new hepatectomy complexity classification was proposed. This study aimed to compare the value of the traditional and that of the new classification in perioperative outcomes prediction.

METHODS

Demographics, perioperative laboratory tests, intraoperative and postoperative outcomes, and follow-up data of patients with hepatocellular carcinoma who underwent liver resection were retrospectively analyzed.

RESULTS

A total of 302 patients were included in our study. Multivariable analysis of intraoperative variables showed that the complexity classification could independently predict the occurrence of blood loss > 800 mL, operation time > 4 h, intraoperative transfusion, and the use of Pringle's maneuver (all p < 0.05). For postoperative outcomes, the high-complexity group was independently associated with severe complications, and hepatic-related complications (all p < 0.05); the traditional classification was independently associated only with posthepatectomy liver failure (PHLF) (p = 0.004).

CONCLUSIONS

Complexity classification could be used to assess the difficulty of surgery and was independently associated with postoperative complications. The traditional classification did not reflect operation complexity and was associated only with PHLF.

A nomogram for prediction of posthepatectomy liver failure in patients with hepatocellular carcinoma: A retrospective study

To generate a nomogram to predict posthepatectomy liver failure (PHLF), we attempted to elucidate salient risk factors in patients with hepatocellular carcinoma (HCC).We performed a retrospective review of 665 patients with HCC who received hepatectomy in 2 academic institutions in China. Independent risk factors for PHLF were identified from putative demographic, intrinsic, biochemical, surgery-related, and volumetric data. A predictive nomogram was formulated based on relevant risk factors, and we compared this with existing models.We identified clinical signs of portal hypertension (P = .023), serum total bilirubin (P = .001), serum creatinine (P = .039), and intraoperative hemorrhage (P = .015) as being important risk factors in predicting PHLF. The nomogram had a C-index of 0.906 for the externally validated data. The nomogram displayed better predictive value than 2 of the other most cited models (C-indices of 0.641 and 0.616, respectively) in the current cohort. Additionally, we were able to patients into low- (<10%), intermediate- (10-30%), and high-risk (≥30%) groups based on the nomogram. This allows us to facilitate person-specific management.Here, we constructed a simple nomogram for prediction of PHLF in patients with HCC weighted by independent risk factors. Further prospective studies are required to confirm the predictive ability of our nomogram.

Serum Prealbumin is Negatively Associated with Survival in Hepatocellular Carcinoma Patients after Hepatic Resection

Objective: Low serum prealbumin levels are associated with poor prognoses in some type of cancers. However, the role of prealbumin in patients with hepatocellular carcinoma (HCC) is unknown. The present study aimed to investigate the role of serum prealbumin levels in long-term survival for HCC patients after hepatic resection.

Methods: HCC patients who underwent hepatic resection from June 2007 to December 2015 were retrospectively analyzed in a tertiary liver center. Patients were classified as having normal or reduced serum prealbumin based on a cut-off value of 200 mg/L. Overall survival and recurrence rate were analyzed between groups. Propensity score analysis was used to reduce bias due to other patient differences at baseline.

Results: A total of 1349 HCC patients who underwent hepatic resection were enrolled on this study, including 1168 (86.6%) male and 181 (13.4%) female. Patients with normal serum prealbumin had significantly higher overall survival than those with reduced serum prealbumin (P < 0.001). Similar findings were observed after propensity analysis and subgroup analysis based on liver cirrhosis. Moreover, patients with normal serum prealbumin had a significantly lower recurrence rate than those with reduced serum prealbumin (P < 0.001).

Conclusions: Low preoperative level of serum prealbumin is associated with poor long-term survival in patients with HCC after hepatic resection. Low serum prealbumin may be a marker to identify patients at high risk of poor prognosis after hepatic resection.

Developing and validating a pre-operative risk score to predict post-hepatectomy liver failure

BACKGROUND

Post hepatectomy liver failure (PHLF) is a serious complication in patients undergoing liver resection. This study hypothesized that a new pre-operative risk score developed through statistical modeling to predict PHLF could be used to stratify patients at higher risk of PHLF.

METHODS

Patients who underwent hepatectomy between 2008 and 2016 were included in the derivation and validation cohorts. A multivariable binary logistic regression model was performed to identify predictors of PHLF, and a prognostic score was derived.

RESULTS

A total of 1269 patients were included in the derivation cohort. PHLF was encountered in 13.1% and was associated with significantly increased 90-day mortality and prolonged post-operative hospital stay (both p < 0.001). Multivariable analysis identified the extent of surgery (p < 0.001) and pre-operative bilirubin (p = 0.015), INR (p < 0.001), and creatinine (p = 0.048) to be independent predictors of PHLF. A risk score derived from these factors returned an area under the ROC curve (AUROC) of 0.816 (p < 0.001) for an internal validation cohort (N = 453), significantly outperforming the MELD score (AUROC: 0.643).

CONCLUSION

The PHLF risk score could be used to stratify the risk of PHLF among patients planned for hepatectomy.

Predictors of post-hepatectomy liver failure in patients undergoing extensive liver resections for hepatocellular carcinoma

Backgrounds/Aims

To determine the prevalence of post-hepatectomy liver failure/insufficiency (PHLF/I) in patients undergoing extensive hepatic resections for hepatocellular carcinoma (HCC) and to assess the predictive value of preoperative factors for post-hepatectomy liver failure or insufficiency (PHLF/I).

Methods

A retrospective review of patients who underwent liver resections for HCC between 2001 and 2013 was conducted. Preoperative parameters were assessed and analyzed for their predictive value of PHLF/I. Definitions used included the 50–50, International Study Group of Liver Surgery (ISGLS) and Memorial Sloan Kettering Cancer Centre (MSKCC) criteria.

Results

Among the 848 patients who underwent liver resections for HCC between 2001 and 2013, 157 underwent right hepatectomy (RH) and extended right hepatectomy (ERH). The prevalence of PHLF/I was 7%, 41% and 28% based on the 50–50, ISGLS and MSKCC criteria, respectively. There were no significant differences in PHLF/I between RH and ERH. Model for End-Stage Liver Disease (MELD) score and bilirubin were the strongest independent predictors of PHLF/I based on the 50–50 and ISGLS/MSKCC criteria, respectively. Predictive models were developed for each of the criteria with multiple logistic regression.

Conclusions

MELD score, bilirubin, alpha-fetoprotein and platelet count showed significant predictive value for PHLF/I (all p<0.05). A composite score based on these factors serves as guideline for physicians to better select patients undergoing extensive resections to minimize PHLF.

Liver Stiffness Assessed by Shear Wave Elastography Predicts Postoperative Liver Failure in Patients with Hepatocellular Carcinoma

BACKGROUND

Cirrhosis increases a patient's risk of developing postoperative liver failure (PLF). Liver stiffness (LS), assessed by two-dimensional shear wave elastography (SWE), indicates liver fibrosis with high accuracy. Whether LS is superior to portal hypertension (PHT) in predicting PLF remains to be studied.

METHODS

The study enrolled 280 patients who underwent hepatectomy for hepatocellular carcinoma from July 2015 to July 2016. All patients received preoperative assessments for LS, PHT, and serum markers of liver fibrosis in addition to other clinicopathological tests. Risk factors for grade A and grade B (or greater) PLF were subjected to univariate and multivariate analysis and receiver operating characteristic curve analysis.

RESULTS

Fifty-five patients (19.6%) experienced PLF. The cutoff value of LS for predicting cirrhosis was 10.1 kPa. Multivariate analysis identified LS, hyaluronic acid, IV collagen, and the presence of splenomegaly as independent predictors of PLF. The cutoff value of LS for predicting PLF and grade B (or greater) PLF was 11.75 and 11.9 kPa, respectively. LS was superior to PHT in predicting PLF or greater than grade B PLF (0.72 vs. 0.60, 0.76 vs. 0.59, P < 0.05).

CONCLUSION

LS measured by SWE can predict risk of PLF with greater accuracy than PHT.

Postoperative Liver Failure

Technical innovations in surgical techniques, anaesthesia, critical care and a spatial understanding of the intra-hepatic anatomy of the liver, have led to an increasing number of liver resections being performed all over the world. However, the number of complications directly attributed to the procedure and leading to inadequate or poor hepatic functional status in the postoperative period remains a matter of concern. There has always been a problem of arriving at a consensus in the definition of the term: postoperative liver failure (PLF). The burgeoning rate of living donor liver transplants, with lives of perfectly healthy donors involved, has mandated a consensual definition, uniform diagnosis and protocol for management of PLF. The absence of a uniform definition has led to poor comparison among various trials. PLF remains a dreaded complication in resection of the liver, with a reported incidence of up to 8 % [1], and mortality rates of up to 30–70 % have been quoted [2]. Several studies have quoted a lower incidence of PLF in eastern countries, but when it occurs the mortality is as high as in the West [3].

Liver failure after hepatectomy: A risk assessment using the pre-hepatectomy shear wave elastography technique

OBJECTIVE

To determine the efficacy of liver stiffness (LS) measurements utilizing the Shear Wave Elastography (SWE) technique for predicting post-hepatectomy liver failure (PHLF) among patients with hepatocellular carcinoma (HCC).

METHODS

Data from eighty consecutive patients who were undergoing hepatectomy for HCC were prospectively identified and evaluated with preoperative SWE. The SWE was measured with advanced ultrasound equipment (Philips EPIQ7; Philips Healthcare, Seattle, WA, USA). PHLF classification was defined according to the International Study Group of Liver Surgery Recommendations (ISGLS).

RESULTS

SWE was successfully performed in 77 patients. According to the ISGLS criteria, PHLF occurred in 35.1% of patients (27 patients), including 2/25 patients with Grade A/B, respectively. Elevated SWE values (P=0.002) and histological cirrhosis (P=0.003) were independent predictors of PHLF according to the multivariate analysis. Patients with SWE values higher than or equal to 6.9kPa were identified at higher risk of PHLF (area under the curve: 0.843, sensitivity: 77.8% and specificity: 78.0%). Postoperative dynamic course of the median the Model For End-stage Liver Disease (MELD) score showed irregular changes among patients with an SWE >6.9kPa. Patients with an SWE <6.9kPa, postoperative dynamic course of the median MELD score gradually decreased.

CONCLUSION

LS measured with SWE is a valid and reliable method for the prediction of PHLF grade A/B among patients with HCC. SWE could become a routine examination for the preoperative evaluation of PHLF.