A Risk-Prediction Model for Hepatorenal Syndrome in Patients with Liver Failure: A Retrospective Analysis

doi:10.21203/rs.3.rs-4836869/v1

Download PDF

Article

A Risk-Prediction Model for Hepatorenal Syndrome in Patients with Liver Failure: A Retrospective Analysis

https://doi.org/10.21203/rs.3.rs-4836869/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Objectives

To identify the risk factors and develop a risk-prediction model for hepatorenal syndrome (HRS) in patients with liver failure (LF).

Methods

A retrospective case-control study involving 372 inpatients with LF admitted to The First Affiliated Hospital of Guangxi University of Chinese Medicine between July 2012 and July 2022 was performed. Univariate and multifactorial logistic stepwise regression analyses were employed to identify risk factors for HRS. A risk-prediction model was constructed, and its predictive value was evaluated using the receiver operating characteristic (ROC) curve, calibration curve, and decision curve analyses.

Results

Combined ascites, combined spontaneous bacterial peritonitis, and high serum levels of gamma-glutamyl transpeptidase, uric acid, and cystatin C were independent risk factors for HRS. The areas under the ROC curve for the training and validation sets were 0.877 and 0.828, respectively. The logistic model demonstrated a good fit. In the decision curve analysis, the curves for both the training and validation sets were well-positioned away from the two extreme treatment strategies (all patients treated or untreated).

Conclusions

The risk-prediction model developed in this study for HRS in LF patients exhibits robust predictive capability, offering a valuable tool for timely clinical intervention and effective treatment of HRS.

Health sciences/Gastroenterology/Hepatology

Health sciences/Risk factors

Health sciences/Medical research/Biomarkers

Health sciences/Medical research/Experimental models of disease

Health sciences/Diseases/Gastrointestinal diseases/Liver diseases

liver failure

hepatorenal syndrome

risk prediction

machine-learning

Liver failure (LF) is an advanced stage in the progression of many liver diseases. It is characterized by a short window for effective clinical intervention, rapid consumption of medical resources, and an extremely high mortality rate, embodying the typical features of “short, rapid, and high.” Common symptoms of LF include generalized weakness, abnormal coagulation function, jaundice, ascites, and multi-organ failure. Despite advancements in comprehensive medical therapy, artificial liver support systems, and liver transplantation, efficacy remains limited, with a 90-day mortality rate of > 50%^1,2.

Hepatorenal syndrome (HRS) is one of the most serious complications in LF patients. It is difficult to treat and has a poor prognosis. This condition encompasses a group of progressive, functionally renal-injury–based complications occurring in patients with severe liver disease without primary renal lesions³ and manifests high morbidity and mortality rates. Timely diagnosis and treatment of HRS crucially impact patient prognosis⁴. However, due to its complex pathogenesis, the clinical diagnosis of HRS is challenging. By the time it is diagnosed, patients often reach an irreversible stage, severely threatening their health and even their lives.

The onset of HRS in LF patients signifies the simultaneous failure of two critical organs, amplifying the challenges in patient management and serving as a significant contributor to mortality. Nevertheless, prophylactic intervention in LF patients at HRS risk may forestall their progression to HRS. Although the model for end-stage liver disease (MELD) is a widely used tool for the assessment of short-term prognosis in end-stage liver disease, its clinical applicability is limited because it does not consider certain complications and relevant biochemical indices^5,6. Importantly, MELD is not efficient in predicting HRS risk in LF patients. Moreover, the etiology, pathogenesis, and epidemiological characteristics of LF in China significantly differ from those in Europe and the United States. Therefore, this study aimed to develop a risk-prediction model for HRS in LF patients by considering a wide range of factors, with the ultimate aim to provide a reference for early identification of LF patients at risk of developing HRS and facilitate the adoption of appropriate preventive measures.

2.1 Research Data

This retrospective cohort study involved data from 372 hospitalized LF patients at The First Affiliated Hospital of Guangxi University of Chinese Medicine between July 2012 and July 2022. The main observational endpoint was the occurrence of HRS within 30 days of admission. Inclusion criteria were 1) adherence to the diagnostic criteria for acute, subacute, chronic, and slow plus acute LF as outlined in the Guidelines for Diagnosis and Treatment of Liver Failure (2018 edition)⁷; 2) meeting the diagnostic criteria for HRS as outlined in the Guidelines for Diagnosis and Treatment of Cirrhosis Ascites and Related Complications⁸; and 3) age ≥ 18 years. Exclusion criteria were 1) previous organic renal impairment or obstructive urinary-tract disease; 2) malignancies outside the liver; 3) recent usage of nephrotoxic drugs; 4) cases of severe immunosuppression, including those resulting from immunosuppressive therapies or diseases, such as severe tuberculosis and AIDS; 5) cases of concurrent severe systemic diseases or severe psychiatric disorders; 6) pregnancy or lactation; and 7) clinical studies with > 20% missing data. Using the R software, random numbers were generated to allocate the 297 LF patients included in the study into a training set (70%, n = 208) and a validation set (30%, n = 89). The training set was employed to develop a risk-prediction model, and the validation set was utilized to assess the predictive accuracy of the model. Patients were stratified into an HRS group (n = 30) and a non-HRS group (n = 267) based on HRS occurrence within 30 days of admission.

2.2 Ethics Statement

This study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi University of Chinese Medicine (approval number: 2020-046-02) and conducted in accordance with the principles of the Declaration of Helsinki and its subsequent amendments. All the methods used in this study were performed in accordance with relevant guidelines and regulations. Due to the retrospective nature of the study, Institutional Review Board waived the need of obtaining informed consent.

2.3 Data Collection

Basic patient information, results of initial laboratory tests, and treatment details were obtained from the electronic medical record system. Basic patient information encompassed gender, age, history of smoking, and history of alcohol consumption. Laboratory tests encompassed routine blood tests (white blood cell count [WBC], neutrophil count [NEUT], lymphocyte count [LYM], red blood cell count [RBC], hemoglobin level [HGB], mean corpuscular volume [MCV], mean corpuscular hemoglobin level [MCH], mean corpuscular hemoglobin concentration [MCHC], and platelet count [PLT]), coagulation tests (prothrombin time [PT], activated partial thromboplastin time [APTT], prothrombin time activity [PTA], and international normalized ratio [INR]), and blood biochemical indices (blood levels of albumin [ALB], globulin [GLB], glutamic pyruvic transaminase [ALT], glutamic oxaloacetic transaminase [AST], total bilirubin [TBil], gamma-glutamyl transpeptidase [GGT], alkaline phosphatase [AKP], uric acid [UA], blood urea nitrogen [BUN], creatinine [SCr], cystatin C [CysC], potassium [K], and sodium [Na]). Additionally, complications, such as ascites, infections, spontaneous bacterial peritonitis (SBP), and digestive bleeding were documented.

2.4 Statistical Analysis

2.4.1 Statistical description and missing-data treatment

The SPSS 25.0 software was utilized to assess the variability of each clinical study variable in the training and validation sets. Patient data within the HRS and non-HRS groups were compared. Normally distributed continuous variables were presented as mean ± standard deviation (Mean ± SD) and assessed using the t-test. Non-normally distributed variables were expressed as median and interquartile range (Median [P25, P75]) by using the Mann-Whitney U test. Categorical variables were presented as frequencies (percentages) and compared between groups by using Pearson's chi-square test or Fisher's test. Statistical significance was set at P < 0.05. Missing data were imputed based on the relationship among available variables by using the random-forest algorithm (default setting) implemented in the "missForest" package of the R software. This package is adaptable to data with non-linear relationships, robust to outliers, and provides accurate substitutions for missing values.

2.4.2 Establishment and visualization of a risk-prediction model

Single-factor logistic regression was performed to analyze the relationship between each variable and the occurrence of HRS by using the R 4.4.0 software. Variables exhibiting statistically significant differences in the single-factor analysis underwent multifactorial logistic stepwise regression, with a significance threshold set at P = 0.05. A prediction model was then established based on the β coefficients derived from the multifactor logistic regression results. Additionally, the "rms" package in the R software was employed to construct a line graph model for predicting the HRS risk in LF patients. This transformation of complex regression equations into simple and visual graphs enhances the readability and clinical utility of the risk-prediction model.

2.4.3 Evaluation of the predictive value of the model

The "pROC" package in the R software was utilized to generate the receiver operating characteristic (ROC) curve and calculate the area under the curve (AUC), sensitivity, and specificity to assess the discriminative performance of the model. An AUC value > 0.75 is considered to indicate high predictive capability, whereas an AUC value < 0.5 suggests a lack of diagnostic value⁹. Model accuracy was assessed by plotting a calibration curve based on 1000 bootstrap samples by using the "caret" package. This curve compares the predicted HRS risk with the actual risk and was plotted using the "vcdExtension" package. The Hosmer-Lemeshow goodness-of-fit test, conducted using the "vcdExtra" package, assessed model fit, with a P-value > 0.05 indicating good fit¹⁰. Decision curve analysis (DCA) was performed using the "rmda" package to evaluate the clinical utility of the model. A high distance from the "All" and "None" curves (the two extreme curves) indicates high clinical significance and net benefit conferred by the model¹¹.

3.1 General Patient Information

We collected data on 372 hospitalized patients with LF who were observed within 30 days of hospitalization period. This cohort included 6 patients with malignant tumors outside the liver, 27 patients who were admitted to hospital in a state of shock, and 7 patients with renal parenchymal injury. In addition to these patients, 35 patients with ≥ 20% missing clinical data, totaling 75 patients were excluded from the study (resulting in a 20.2% missing rate). The final analyzed patient population consisted of 297 LF patients, and only 30 of them developed HRS within 30 days of admission. The inclusion criteria encompassed demographic characteristics, relevant medical history, and admission-related examination results (Table 1).

Table 1

General information of the included patients.
Categorical variables	N (%) / Mean ± SD / Median (P₂₅,P₇₅)	Continuous variables	N (%) / Mean ± SD / Median (P₂₅,P₇₅)
Hepatorenal syndrome		WBC	6.10[4.05;9.45]
No	267 (89.9%)	NEUT	4.10[2.43;7.5]
Yes	30(10.1%)	LYM	0.97[0.61;1.36]
Sex		RBC	3.16[2.48;3.87]
Male	241(81.1%)	HGB	96[77;118.5]
Female	56(18.9%)	MCV	95[85.55;103.9]
Age (years)		MCH	32.10[27.75;34.9]
<55years old	152(51.2%)	MCHC	336[324;344]
≥ 55years old	145(48.8%)	PLT	56[23.05;103.5]
History of alcohol consumption		K	3.87[3.44;4.3]
No	248(83.5%)	Na	135[132;138]
Yes	49(16.5%)	PT	20.8[17.7;24.4]
History of smoking		PTA	44[35;57]
No	253(85.2%)	INR	1.78[1.46;2.21]
Yes	44(14.8%)	APTT	49.6[43.75;56.4]
Ascites		ALB	29.1[24.55;33.5]
No	248(83.5%)	GLB	32.1[26.225;37.95]
Yes	49(16.5%)	TBil	115.3[65.1;252.05]
Gastrointestinal bleeding		AKP	141.5[100;202]
No	274(92.3%)	GGT	67[32;156]
Yes	23(7.7%)	ALT	42[27.25;95.5]
Infection		AST	85[49;167]
No	288(97.0%)	BUN	4.9[3.4;8.78]
Yes	9(3.0%)	SCr	71[57;102]
Spontaneous bacterial peritonitis		UA	227[140;349]
No	230(77.4%)	CysC	1.23[0.93;1.8]
Yes	67(22.6%)

3.2 Comparison between the Training and Validation Sets

The 297 LF patients included in the study were randomly allocated into a training set (n = 208, 70%) and a validation set (n = 89, 30%). The training set was used for developing the risk-prediction model, whereas the validation set was employed to assess the predictive efficacy of the model. In the training and validation sets, 24 and 6 cases had HRS, respectively, accounting for 11.5% and 6.7% of the corresponding sets, respectively. Statistical analyses regarding general information, clinical characteristics, and laboratory indices revealed no significant differences between the two sets, except for MCH and MCHC (P > 0.05). These results indicate that the clinical study variables were evenly distributed between the training and validation sets following group randomization .

3.3 Comparison between the HRS and Non-HRS Groups

Among the 297 LF patients, 30 (10.1%) developed HRS within 30 days of admission. Significant differences were observed between the HRS and non-HRS groups in terms of HRS, WBC, NEUT, PLT, Na, AKP, GGT, ALT, AST, SCr, BUN, UA, CysC, ascites, and SBP (P < 0.05).

3.4 Single-Factor and Multifactor Logistic Regression Analyses of the Training Set

To identify factors influencing the occurrence of HRS in LF patients, one-way logistic regression analysis was performed with the occurrence of HRS as the dependent variable, controlling for confounding factors. The results showed that the following 11 variables were influencing factors (P < 0.05): ascites, SBP, PLT, K, Na, APTT, GGT, BUN, SCr, UA, and CysC (Table 2). Multifactorial logistic stepwise regression analysis revealed ascites (P = 0.024), SBP (P = 0.048), GGT (P = 0.014), UA (P = 0.044), and CysC (P < 0.01) as independent risk factors affecting the development of HRS in LF patients.

Table 2

Results of Multifactor Logistic Regression Analysis.
Variables	β	S.E	Z	P	OR(95%CI)
Intercept	-6.159	0.919	-6.704	< .001	0.002(0.000 ~ 0.013)
Ascites
No
Yes	1.212	0.538	2.252	0.024	3.361(1.170 ~ 9.653)
Spontaneous bacterial peritonitis
No
Yes	1.052	0.532	1.978	0.048	2.865(1.010 ~ 8.127)
GGT	0.003	0.001	2.463	0.014	1.003(1.001 ~ 1.005)
UA	0.003	0.001	2.010	0.044	1.003(1.001 ~ 1.006)
CysC	1.042	0.316	3.302	< .001	2.836(1.527 ~ 5.266)

3.5 Establishment of a Risk-Prediction Model for HRS in LF Patients

A prediction model was established based on the results of the logistic regression analysis, yielding the following formula (with GGT in u/L, UA in µmol/L, and CysC in mg/L):

Logit (P) = − 6.159 + 1.212 × ascites + 1.052 × SBP + 0.003 × GGT + 0.003 × UA + 1.042 × CysC

3.6 Visualization of the Model

A nomogram model was established using these five variables to generate a visual chart suitable for clinical application. Each indicator, such as GGT, was aligned with corresponding endpoints, and a vertical line was drawn upward to the scoring axis to determine the individual scoring value. These values were then summed to obtain the total score, which corresponded to the risk for HRS occurrence in LF patients. In this model, a high total score indicates a high HRS risk. For example (Fig. 1), consider an LF patient with ascites (26.5 points), SBP (22.5 points), 150 u/L GGT (10 points), 670 µmol/L UA (40 points), and 2 mg/L CysC (45 points). The total score of this patient would be 144 points, indicating a high risk for HRS occurrence, with a probability of > 0.6. Accordingly, medical staff should implement preventive measures against HRS in this patient.

3.7 Evaluation and Validation of the Model

3.7.1 Distinguishing ability of the model

The distinguishing ability of the model was evaluated using the ROC curve analysis. The model had an AUC of 0.877 (95% CI: 0.822–0.933) in the training set (Fig. 2A). At the optimal cutoff point of 0.077, it demonstrated high specificity (61.7%) and sensitivity (87.5%). In the validation set, its AUC was 0.828 (95% CI: 0.706–0.950) (Fig. 2B), and at the optimal cutoff point of 0.235, it showed high specificity (70.1%) and sensitivity (95.8%). These results indicate the good distinguishing ability of the model.

3.7.2 Calibration of the model

The bootstrap resampling method was employed to repeatedly sample the training and validation sets 1000 times each. Subsequently, calibration curves were plotted after validation (Fig. 3A and Fig. 3B). The fitted logistic curves of both sets closely overlapped with the standard curves, suggesting a strong agreement between the predicted and actual risks and indicating minimal prediction error. Additionally, the goodness-of-fit Hosmer-Lemeshow test yielded the following results: χ² = 4.4172, P = 0.8177 (for the training set) and χ² = 4.8204, P = 0.6819 (for the validation set). These results indicate that there is no significant difference between the predicted and actual observed values in both groups, demonstrating good goodness of fit.

3.7.3 DCA

DCA was employed to analyze the net benefit of the model for clinical applications. The curves for both the training and validation sets were observed to be far from the two extreme curves (Fig. 4A and Fig. 4B), namely, all the patients treated and none treated. This observation indicates that the model holds significant clinical value.

4.1 Association between LF and HRS

HRS is one of the serious complications of end-stage liver disease and is clinically divided into type I (HRS-I) and type II (HRS-II). HRS-I predominantly manifests in severe LF cases, often triggered by various factors such asTablerointestinal hemorrhage, major surgeries, and severe bacterial infections, including SBP. Its onset is typically rapid, with a median survival period of merely 2 weeks¹². LF, as the culmination of progressive or sudden deterioration across all liver diseases, frequently leads to HRS complications. The pathological mechanism of HRS is intricate and involves multifactorial interactions, which remain to be clarified. Early-stage HRS typically lacks evident renal histological alterations, mainly presenting as reversible functional renal impairment. However, clinical diagnosis proves challenging, with patients often reaching an irreversible stage post-diagnosis. Once HRS complicates LF, the patient prognosis significantly deteriorates, posing limitations and challenges to clinical pre-intervention and early diagnosis. Thus, identification of relevant clinical indicators is crucial for the early detection of HRS in LF patients.

Currently, the precise mechanism of HRS in ACLF remains unclear, although hemodynamic disorder is posited as the primary driver. Schrier et al. have observed that HRS develops in 27.7% (28/101) of ALF patients within 4 weeks¹³. Additionally, Huang et al. have shown that HRS is an independent risk factor affecting the prognosis of patients with HBV-associated ACLF, with a relative-risk value of 2.084¹⁴.

In this study, we developed a diagnostic model to predict HRS by retrospectively analyzing the clinical data of LF patients complicated with HRS. Our findings revealed that five factors—SBP, ascites, GGT, UA, and CysC—significantly influenced the development of HRS. Comparison between the HRS and non-HRS groups highlighted significantly higher serum levels of the hepatic and renal function indices PLT, ALT, AST, TBil, AKP, BUN, and SCr in the HRS group than in the non-HRS group. This finding suggests a close association of HRS risk with hepatic and renal function levels in LF patients. Hence, in the management of LF patients, vigilant monitoring of these organ functions is essential against the risk for HRS development.

4.2 Analysis of the Influencing Factors of HRS in LF Patients

4.2.1 Complications

Approximately 10–40% of patients with ascites or advanced cirrhosis develop HRS¹⁵. Additionally, ascites has been suggested to be a significant inducer of HRS, as the accumulation of substantial ascites exerts pressure on renal blood vessels, elevating renal vascular resistance and diminishing effective circulating blood volume. Clinical interventions, such as diuretic use or peritoneal drainage further reduce blood volume, resulting in inadequate renal perfusion and decreased glomerular filtration rate, thereby instigating renal damage. The likelihood of HRS induction due to ascites correlates with the volume of ascites¹⁶. Our study identifies ascites as an independent risk factor, with a risk factor of 3.361 for HRS induction.

Due to the continuous decline in liver function and the loss of compensatory mechanisms, LF patients frequently develop SBP¹⁷. Liver pathology in HRS patients often exhibits extensive hepatocyte necrosis, severely compromising the monocyte-macrophage system and leading to impaired hepatic reticuloendothelial immunity. This immune dysfunction, coupled with alterations in the intestinal microbiota and increased intestinal mucosal permeability, facilitates bacterial migration into the peritoneal cavity. Concurrent peritoneal effusions provide an optimal environment for SBP development¹⁸. Once SBP ensues, it exacerbates the condition of the patient and, in the absence of specific interventions, can progress to life-threatening multiorgan failure¹⁹. A study by Ruiz-Del-Arbol et al. has indicated that SBP increases the risk of death in HRS patients²⁰. In line with this finding, the present study found a 2.865-fold increase in HRS risk among LF patients with SBP.

Clinical presentation in LF patients complicated with SBP often lacks typical symptoms, leading to delays in diagnosis and treatment. Even in confirmed cases with positive ascites bacterial culture and significantly elevated bacterial cell counts, some patients may exhibit atypical symptoms. This variability may be due to the immunocompromised state of patients with liver disease, their diminished responsiveness to inflammatory stimuli, the dilution effect of abundant ascites on inflammatory mediators, and the overshadowing of symptoms by a pre-existing liver disease^21,22. Ascites itself serves as a favorable medium for bacterial proliferation, which further increases the HRS risk in LF patients.

4.2.2 Clinical indices

Serum CysC is widely present in nucleated cells of various tissues and bodily fluids, and it is cleared from the circulation via glomerular filtration. Its levels promptly change in response to early-stage renal impairment, rendering it a sensitive marker for the assessment of renal function. Studies have indicated significantly higher serum CysC levels in patients with renal disease than in healthy individuals, underscoring the diagnostic utility of this index^23,24. Gharaibeh et al. have demonstrated that although both SCr and serum CysC levels increase in the early stages of HRS, the latter occurs earlier²⁵. Likewise, serum CysC level starts to normalize earlier than SCr level during recovery from HRS, suggesting the potential of CysC as an early monitoring indicator for HRS onset and treatment efficacy. In our study, CysC emerged as an independent risk factor, increasing the HRS risk in LF patients by 2.836 fold.

GGT is a mitochondrial enzyme widely distributed in the body and is one of the most prominent extracellular enzymes. It hydrolyzes glutathione while utilizing glutamyl transfer to form the γ-glutamyl cycle²⁶. GGT is mainly expressed in the kidney, liver, and pancreas, and upregulated in hepatobiliary damage, especially in biliary obstructive diseases^27,28. It has long been recognized as a sensitive indicator of hepatobiliary diseases²⁹. Additionally, its upregulation is recognized as a sign of liver injury, cancer, or low-grade chronic inflammation³⁰. In LF patients, increased GGT levels indicate increased liver injury, necessitating early intervention against HRS. Our study revealed a 1.003-fold increase in HRS risk among LF patients per unit increase in serum GGT level.

UA is the end product of purine metabolism, and plasma urate level > 7 mg/dL in humans are typically referred to as hyperuricemia³¹. This condition can lead to renal damage through various mechanisms, including the direct deposition of urate crystals in the kidney, alterations in renal hemodynamics, and the activation of immune and pro-inflammatory responses. The synthesis of UA involves xanthine oxidase, which is primarily found in the liver and catalyzes the oxidation of xanthine and hypoxanthine³². In LF patients, the increased UA levels due to hepatic metabolic dysfunction exacerbate renal damage, thereby increasing the risk for HRS. UA serves as an indicator of renal-injury severity, and our study revealed a 1.003-fold increase in HRS risk among LF patients per unit increase in serum UA level.

4.3 Clinical Significance of the Risk-Prediction Model

In this study, a nomogram prediction-model was developed to assess the risk for HRS development in LF patients. To this end, multifactorial analysis was used, and the identified risk factors were analyzed and visualized. The AUCs for both the training and validation sets were approximately 0.85, indicating good model differentiation. Notably, the actual curve of the column-line diagram model overlapped well with the ideal curve, further validating the accuracy of the model. Additionally, the results of the DCA demonstrated the clinical applicability of the model. Therefore, this model provides an effective tool to predict HRS development in LF patients and may aid clinical practice.

This study has some limitations. Firstly, it is a single-center retrospective study with a modest sample size and a limited number of clinical factors. Although an internal validation set was utilized, broader sample sizes and external validations across multiple centers are imperative to enhance the robustness and generalizability of this risk prediction model.

In summary, ascites, SBP, UA, GGT, and CysC irregularities emerge as independent risk factors for HRS complications in LF patients. These parameters hold substantial diagnostic value for assessing HRS risk. Moreover, the column-line diagram model for HRS risk prediction, built upon these variables, exhibits commendable differentiation, calibration, and clinical utility. It serves as a valuable tool for healthcare practitioners in formulating early, tailored interventions to mitigate HRS incidence, thereby ameliorating patient prognosis, elevating quality of life, and alleviating healthcare burdens.

Ethics approval and consent to participate:

This study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi University of Chinese Medicine (approval number: 2020-046-02) and conducted in accordance with the principles of the Declaration of Helsinki and its subsequent amendments. All the methods used in this study were performed in accordance with relevant guidelines and regulations.Due to the retrospective nature of the study, Institutional Review Board waived the need of obtaining informed consent.

Competing interests:
The authors declare that they have no competing interests, and all authors should confirm its accuracy.

Funding:

This study was supported by the National Natural Science Foundation of China (grant no.82160881; 82360912; 82274434), Natural Science Foundation of Guangxi Zhuang Autonomous Region (grant no. 2023GXNSFAA026176; 2021GXNSFAA220112) and National Traditional Chinese Medicine Inheritance and Innovation Center Construction Project.

Author Contribution

H.L: Writing-original draft. Y.L&K.Z: Data curation and investigation. M.W&R.Z: Methodology and resources. H.L&K.Z: Software and validation. M.W&D.M: Writing-review and editing. All authors have read and agreed to the published version of the manuscript.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Morrison, M. & Artru, F. Predicting the development of acute-on‐chronic liver failure. United. European. Gastroenterol. J. 11, 813–4 (2023).
Kok, B. & Karvellas, C. J. Management of cerebral edema in acute liver failure. Semin. Respir. Crit. Care. Med. 38, 821–9 (2017).
Moreau, R., et al. Acute on-chronic liver failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenterology. 144, 1426–37 (2013).
Ginès, A., et al. Incidence, predictive factors, and prognosis of the hepatorenal syndrome in cirrhosis with ascites. Gastroenterology, 105, 229–36 (1993).
Habas, E., et al. Update on hepatorenal Syndrome: definition, Pathogenesis, and management. Arab. J. Gastroenterol. 23, 125–33 (2022).
Ai Sibae, M. R. & Cappell, M. S. Accuracy of MELD scores in predicting mortality in decompensated cirrhosis from variceal bleeding, hepatorenal syndrome, alcoholic hepatitis, or acute liver failure as well as mortality after non-transplant surgery or TIPS. Dig. Dis. Sci. 56, 977–87 (2011).
Chinese Medical Association. The guidelines of prevention and treatment for chronic hepatitis B (2019 version). Chin. J. Hepatol. 27, 938–61 (2019).
Chinese Society of Hepatology, Chinese Medical Association. Guidelines on the management of ascites and complications in cirrhosis. J. Clin. Hepatol. 25, 664–77 (2017).
Silva, T. B., et al. Development and validation of a nomogram to estimate the risk of prostate cancer in Brazil. Anticancer. Res. 35, 2881–6 (2015).
Niu, X. K., et al. Developing anew PI-RADS v2-based nomogram for forecasting high-gradeprostate cancer. Clin. Radiol. 72, 458–64 (2017).
Nattino, G., Finazzi, S. & Bertolini, G. A new test and graphical tool to assess the goodness of fit of logistic regression models. Stat. Med. 35, 709–20 (2016).
Dundar, H. Z. & Yilmazlar, T. Management of hepatorenal syndrome. World. J. Nephrol. 4, 277 – 86 (2015).
Schrier, R. W., et al. Peripheral arterial vasodilation hypothesis, a proposal for the initiation of renal sodium and water retention in cirrhosis. Hepatology. 8, 1151–7 (1988).
Huang, K., et al. Survival and prognostic factors in hepatitis B virus-related acute-on-chronic liver failure. World. J. Gastroenterol. 17, 3448–52 (2011).
Bertino, G., et al. Emerging hepatic syndromes: pathophysiology, diagnosis and treatment. Intern. Emerg. Med. 11, 905–16 (2016).
Rossle, M. & Gerbes, A. L. TIPS for the treatment of refractory ascites, hepatorenal syndrome and hepatic hydrothorax: a critical update. Gut. 59, 988–1000 (2010).
Mittal, H., Wyawahare, M. & Sistla, S. Microbiological profile of pathogens in spontaneous bacterial peritonitis secondary to liver cirrhosis: a retrospective study. Trop. Doct. 50, 138–41 (2020).
Bajaj, J. S., Kamath, P. S. & Reddy, K. R. The evolving challenge of infections in cirrhosis. N. Engl. J. Med. 384, 2317–30 (2021).
Devani, K., et al. Trends in hospitalization, acute kidney injury, and mortality in patients with spontaneous bacterial peritonitis. J. Clin. Gastroenterol. 53, 68–74 (2019).
Lenz, K., Kapral, C., Gegenhuber, A., Buder, R. & Wewalka, F. Systemic, renal, and hepatic hemodynamic derangement in cirrhotic patients with spontaneous bacterial peritonitis. Hepatology. 38, 1210–8 (2003).
Angeloni, S., et al. Efficacy of current guidelines for the treatment of spontaneous bacterial peritonitis in the clinical practice. World. J. Gastroenterol. 14, 2757–62 (2008).
Riggio, O. & Angeloni, S. Ascitic fluid analysis for diagnosis and monitoring of spontaneous bacterial peritonitis. World. J. Gastroenterol. 15, 3845–50 (2009).
Gustot, T., et al. Clinical Course of acute-on-chronic liver failure syndrome and effects on prognosis. Hepatology. 62, 243–52 (2015).
Bernsmeier, C., et al. Patients with acute-on-chronic liver failure have increased numbers of regulatory immune cells expressing the receptor tyrosine kinase MERTK. Gastroenterology. 48, 603–15 (2015).
Gharaibeh, K. A., et al. Cystatin C predicts renal recovery earlier than creatinine among patients with acute kidney injury. Kidney. Int. Rep. 3, 337–42 (2018).
Reddy, K. R., et al. Acute Liver Failure Study Group. Determinants of outcome among patients with acute liver failure listed for liver transplantation in the United States. Liver. Transpl. 22, 505–15 (2016).
Reichel, C., et al. Elevated soluble tumour necrosis factor receptor serum concentrations and short-term mortality in liver cirrhosis without acute infections. Digestion. 62, 44–51 (2000).
Anadón, A., Castellano, V. & Martínez-Larrañaga, M. R. Biomarkers Of Drug Toxicity. Pittsburgh: Academic Press (2014).
Dufour, D. R., et al. Diagnosis and monitoring of hepatic injury. I. Performance characteristics of laboratory tests. Clin. Chem. 46, 2027–49 (2000).
Mitrić, A. & Castellano, I. Targeting gamma-glutamyl transpeptidase: a pleiotropic enzyme involved in glutathione metabolism and in the control of redox homeostasis. Free. Radic. Biol. Med. 208, 672–83 (2023).
Meng, X., Li, X., Yang, B. & Zhou, H. Pathogenesis and treatment of chronic hyperuricemic nephropathy. J. Physiol. Stud. 4, 1–9 (2016).
Chen-Xu, M., Yokose, C., Rai, S. K., Pillinger, M. H. & Choi, H. K. Contemporary prevalence of gout and hyperuricemia in the United States and decadal trends: the National Health and Nutrition Examination Survey, 2007–2016. Arthritis. Rheumatol. 71, 991–9 (2019).

No competing interests reported.

Download PDF

Reviews received at journal
24 Oct, 2024
Reviewers agreed at journal
16 Oct, 2024
Reviewers agreed at journal
16 Oct, 2024
Reviewers agreed at journal
07 Aug, 2024
Reviewers invited by journal
06 Aug, 2024
Editor assigned by journal
06 Aug, 2024
Editor invited by journal
06 Aug, 2024
Submission checks completed at journal
05 Aug, 2024
First submitted to journal
31 Jul, 2024

You are reading this latest preprint version

A Risk-Prediction Model for Hepatorenal Syndrome in Patients with Liver Failure: A Retrospective Analysis

Status:

Version 1

Abstract

Objectives

Methods

Results

Conclusions

Figures

1 INTRODUCTION

2 MATERIALS and METHODS

2.1 Research Data

2.2 Ethics Statement

2.3 Data Collection

2.4 Statistical Analysis

2.4.1 Statistical description and missing-data treatment

2.4.2 Establishment and visualization of a risk-prediction model

2.4.3 Evaluation of the predictive value of the model

3 RESULT

3.1 General Patient Information

3.2 Comparison between the Training and Validation Sets

3.3 Comparison between the HRS and Non-HRS Groups

3.4 Single-Factor and Multifactor Logistic Regression Analyses of the Training Set

3.5 Establishment of a Risk-Prediction Model for HRS in LF Patients

3.6 Visualization of the Model

3.7 Evaluation and Validation of the Model

3.7.1 Distinguishing ability of the model

3.7.2 Calibration of the model

3.7.3 DCA

4 DISCUSSION

4.1 Association between LF and HRS

4.2 Analysis of the Influencing Factors of HRS in LF Patients

4.2.1 Complications

4.2.2 Clinical indices

4.3 Clinical Significance of the Risk-Prediction Model

Declarations

Ethics approval and consent to participate:

Competing interests: The authors declare that they have no competing interests, and all authors should confirm its accuracy.

Funding:

Author Contribution

Data Availability

References

Additional Declarations

Status:

Version 1

Competing interests:
The authors declare that they have no competing interests, and all authors should confirm its accuracy.