Prognostic value of acute liver injury in cirrhotic patients after the transjugular intrahepatic portosystemic shunt procedure

Abstract

BACKGROUND

Acute liver injury (ALI) is a potential complication after transjugular intrahepatic portosystemic shunt (TIPS) in patients with cirrhosis, which may result from hemodynamic shear stress and direct mechanical needle trauma. However, the incidence, risk factors, and prognostic significance of post-TIPS ALI remain poorly defined. Such injury may contribute to postoperative morbidity and adverse outcomes, yet its clinical impact remains underrecognized.

AIM

To investigate the incidence, risk factors, and prognostic impact of post-TIPS ALI in liver cirrhosis patients.

METHODS

Data of 346 patients who had liver cirrhosis and had undergone TIPS procedure were included for a 6-year period for analysis of ALI. Using Kaplan-Meier curves, we analyzed the rates of overt hepatic encephalopathy (HE) and overall survival (OS) between patients with ALI and those without it. Then, we elucidated mortality- and ALI-related risk factors using logistic regression and Cox regression model, and according to the findings of the latter, we generated a nomogram.

RESULTS

Seventy-four patients (21.4%, 74/346) developed ALI at 72 hours after TIPS. The patients with ALI had higher overt HE and lower OS compared to patients without ALI after TIPS. Serum creatinine was independently associated with ALI. Sarcopenia, the Model for End-Stage Liver Disease score, age, and post-TIPS ALI were independent mortality predictors. Finally, we established a nomogram with modest survival prediction performance (C-index = 0.743).

CONCLUSION

In cirrhotic patients, post-TIPS ALI can be a serious complication and may predict adverse postoperative outcomes.

Key Words: Transjugular intrahepatic portosystemic shunt; Acute liver injury; Liver cirrhosis; Hepatic encephalopathy; Prognosis; Risk factors

Core Tip: Acute liver injury (ALI) is a common and serious complication following the transjugular intrahepatic portosystemic shunt (TIPS) procedure in cirrhotic patients. This study highlights the incidence, risk factors, and prognostic implications of ALI after TIPS. Our findings demonstrate that post-TIPS ALI significantly increases the risk of overt hepatic encephalopathy and reduces overall survival. Independent predictors of mortality include sarcopenia, Model for End-Stage Liver Disease score, age, and post-TIPS ALI. A nomogram was developed based on these key factors, offering a reliable tool for survival prediction and clinical decision-making in managing cirrhotic patients undergoing TIPS.

INTRODUCTION

The transjugular intrahepatic portosystemic shunt (TIPS) procedure represents a minimally invasive treatment approach for portal hypertension-related complications like variceal bleeding and refractory ascites[1]. However, TIPS creation reduces portal perfusion by 80%-100%, a reduction that is partially compensated by increased flow from the hepatic artery[2,3]. This acute hemodynamic alteration exposes sinusoidal endothelial cells to abnormal shear stress, which may lead to endothelial dysfunction and hepatocyte injury[4,5]. In addition to hemodynamic changes, the mechanical injury from needle puncture and intrahepatic tract creation during TIPS can directly damage hepatocytes and sinusoidal endothelial cells, further contributing to postoperative acute liver injury (ALI)[6]. Although liver function parameters stabilize in most cases, an exaggerated increase of liver enzymes in the acute phase indicates irreversible liver injury and progression to acute liver failure[7]. A previous study has reported that acute transient elevation of liver enzymes following TIPS increases the risk of early mortality[8].

Serum alanine aminotransferase (ALT) is a standard biomarker for ALI[9]. According to established diagnostic criteria, ALI can be defined as an ALT level exceeding 3 times the upper limit of normal (ULN), a threshold that is consistent with the ALT criterion of Hy’s Law[10,11]. In our institution, the ULN for ALT is 40 U/L; thus, ALT > 120 U/L (3 × ULN) was considered indicative of ALI. Early detection of abnormal liver function within few days since TIPS can facilitate effective early management and better clinical observation. A recent study showed that ALT levels peaked earlier than bilirubin and aspartate aminotransferase (AST) after TIPS (1 week, 3 months, and 1 month, respectively)[12]. Therefore, ALT may be a superior screening tool for detecting post-TIPS ALI.

The exact incidence and mortality rates of ALI (i.e., ALT > 3 × ULN) following a TIPS procedure are not well known, but clinical observations have revealed that post-TIPS ALI may negatively impact clinical outcome and mortality[13]. In the current study, we assessed the incidence and possible contributing factors for ALI after TIPS. Furthermore, a nomogram was established on the basis of independent prognosticators to predict how ALI will impact overall survival (OS) following a TIPS procedure.

MATERIALS AND METHODS

Study population

Patients having undergone TIPS at the First Affiliated Hospital of the University of Science and Technology of China from January 2016 to March 2022 were considered for inclusion.

Inclusion criteria: (1) Portal hypertension due to cirrhosis; and (2) Variceal rebleeding and/or refractory ascites.

Exclusion criteria: (1) Baseline ALT > ULN (40 U/L); (2) TIPS technical failure; (3) Loss to follow-up within three months; (4) Lack of preoperative abdominal computed tomography (CT) scan or poor quality of CT; (5) Severe comorbidities like pulmonary/renal insufficiency, cerebrovascular/cardiovascular disease; and (6) Lack of baseline data.

Clinical data from 346 patients (74 patients, ALI group; 272 patients, no-ALI group) were included and analyzed (Supplementary Figure 1).

In this study, we ensured that the protocol complied with the Declaration of Helsinki and obtained approval from the Institutional Review Board of the First Affiliated Hospital of the University of Science and Technology of China (Approval No. 2023-RE-157). Given the study’s retrospective design, the need for informed consent was waived by the Institutional Review Board of the First Affiliated Hospital of the University of Science and Technology of China.

The TIPS procedure

The TIPS procedure was carried out under the guidance of ultrasonography and digital subtraction angiography. Through the right internal jugular vein, the hepatic vein was catheterized, and for the puncture, we used a transjugular liver access set (RUPS-100; Cook Incorporated, Bloomington, IN, United States). We used the TIPS needle to puncture the bifurcation of the left and right portal vein branches from the right hepatic vein. Notably, for portosystemic shunting, we used an 8-mm stent [Fluency stent (Angiomed subsidiary of C.R. Bard, NJ, United States) or polytetrafluoroethylene-covered Viatorr stent (W. L. Gore & Associates, Flagstaff, AZ, United States)]. For dilatation, we used an 8-mm balloon catheter. A postoperative portosystemic pressure gradient ≤ 12 mmHg, or a reduction of > 50%, was considered a hemodynamically successful TIPS therapy.

Clinical variables and follow-up

All data were retrieved from the electronic records of the hospital and included information on sex, age, cause of liver cirrhosis, blood urea nitrogen (BUN), hemoglobin, creatinine, international normalized ratio (INR), platelets, white blood cell (WBC) counts, prealbumin, albumin, globulin, AST, ALT, prothrombin time (PT), total bilirubin (TBIL), presence of ascites, hypertension, diabetes, and imaging evidence indicating a diagnosis of hepatocellular carcinoma (HCC).

The Child-Pugh score, Child-Pugh class, and Model for End-Stage Liver Disease (MELD) score were derived from the data collected on the day of admission. Sarcopenia was defined as a transverse psoas muscle thickness (TPMT) of < 7.8 mm/m (females) or < 10.7 mm/m (males)[14]. The TPMT was measured using the abdominal CT data. To calculate this, we used the largest axial diameter of the right psoas muscle at the level of L3, and the measurements (mm/m) were normalized to height. ALI was considered when ALT was ≥ 3-fold ULN at 72 hours after TIPS, where the baseline ALT level was normal.

Clinical evaluations and laboratory tests were completed prior to discharge (usually 72 hours after TIPS) and again during follow-up at either the inpatient or outpatient clinic at the following time points: 1 month, 3 months, and 6 months postoperatively, followed by regular, scheduled visits at 6-month intervals. Overt hepatic encephalopathy (HE) was considered if the condition of grade 2 or higher West-Haven criteria was met or when signs and symptoms indicative of disorientation and asterixis were noted[15]. The end of the study period was either: (1) March 2023; (2) Liver transplantation; (3) Two consecutive follow-ups without response; or (4) Death.

Statistical analysis

Statistical analyses were performed using R software (version 4.2.2, The R Development Core Team) and SPSS 26 (IBM Corp., Armonk, NY, United States). Normality of the variables was verified. For comparisons of quantitative variables, we used the Mann-Whitney U test or Student’s t-test, and similarly, for comparing qualitative variables, we used the χ² test. Diagnostic accuracy assessments were performed with receiver operating characteristic (ROC) analyses. Kaplan-Meier curves and log-rank tests were used to compare differences in survival and overt HE between patients with ALI and those without ALI. The associations of risk factors and ALI were assessed using multivariate and univariate logistic regression models.

Patients were randomly assigned to the training set (n = 242) and validation set (n = 104) in a 7:3 ratio using simple randomization implemented in R software. The training set served the purposes of variable screening and model formulation, while the validation set was designated for corroborating the results derived from the training set. Baseline characteristics of the two sets were compared and are shown in Supplementary Table 1. No significant differences were observed (all P > 0.05).

Univariable Logistic/Cox regression was first performed for each candidate variable. Variables with a P value < 0.10 were then entered into the LASSO regression for further selection. The final selected variables were fitted in a multivariable Logistic/Cox regression model. Using multivariate regression, we generated a nomogram (rms package in R software). The discriminative ability was evaluated using the area under the ROC curve (AUC) and the concordance index (C-index), and the calibrating ability was assessed with the use of calibration plots. The model’s clinical utility was assessed using decision curve analysis (DCA). A nomogram was constructed to analyze the risk scores. According to the cut-off value acquired by the survival ROC, patients were divided into low- and high-risk groups. The log-rank and Kaplan-Meier tests were used for examining the difference between survival curves. Statistically significant results were indicated by P value of < 0.05 indicated.

RESULTS

Patient demographics

In the present study, we enrolled 346 cirrhotic patients [mean age: 54.5 ± 11.3 years; males, 246 (71.1%)] who underwent TIPS. The cause of cirrhosis in 209 (60.4%) patients was a viral hepatitis infection. These were the mean values for the following parameters: Body mass index: 22.6 ± 3.0 kg/m²; the Child-Pugh score: 7.3 ± 1.5; and the MELD score: 9.7 ± 4.1 (Table 1). Of the 346 patients, 236 (68.2%), 38 (11.0%), 58 (16.8%), and 46 (13.3%) were complicated with ascites, hypertension, diabetes, and HCC, respectively. Overall, 74 (21.4%) patients were subsequently diagnosed with ALI after TIPS (at 72 hours). The occurrence of HCC was higher in patients with ALI (P = 0.048) than in those without ALI. Patients with ALI had lower TBIL (P = 0.008) and PT (P = 0.042) and higher creatinine (P = 0.002) and BUN (P < 0.001) levels, when compared with no-ALI individuals.

Table 1 Patient characteristics, mean ± SD/n (%).

Variables	All patients (n = 346)	Acute liver injury (n = 74)	No-acute liver injury (n = 272)	P value
Age (year)	54.5 ± 11.3	56.7 ± 12.1	53.9 ± 11.1	0.051
Male gender	246 (71.1)	57 (77.0)	189 (69.5)	0.204
BMI (kg/m2)	22.6 ± 3.0	22.1 ± 2.9	22.7 ± 3.0	0.092
Virus-related cirrhosis	209 (60.4)	42 (56.8)	167 (61.4)	0.469
Ascites	236 (68.2)	25 (33.8)	84 (30.9)	0.648
Hypertension	38 (11.0)	10 (13.5)	28 (10.3)	0.438
Diabetes	58 (16.8)	10 (13.5)	48 (17.6)	0.392
HCC	46 (13.3)	15 (20.3)	31 (11.4)	0.048
Hb (g/L)	81.8 ± 23.6	84.1 ± 21.6	81.2 ± 24.1	0.277
WBC (× 109/L)	4.3 ± 3.1	4.8 ± 4.2	4.1 ± 2.7	0.574
Platelets (× 109/L)	86.7 ± 64.1	89.4 ± 74.4	85.9 ± 61.4	0.552
Albumin (g/L)	33.2 ± 5.4	33.3 ± 5.8	32.7 ± 5.3	0.361
Prealbumin (g/L)	98.4 ± 38.9	106.7 ± 39.5	96.4 ± 38.6	0.082
Globulin (g/L)	28.0 ± 7.0	27.4 ± 5.9	28.2 ± 7.2	0.394
TBIL (µmol/L)	23.6 ± 15.7	21.1 ± 17.5	24.3 ± 14.4	0.008
ALT (U/L)	20.7 ± 8.2	22.2 ± 8.5	20.2 ± 8.0	0.067
AST (U/L)	30.7 ± 12.5	30.6 ± 12.6	30.0 ± 12.5	0.661
Creatinine (µmol/L)	70.9 ± 54.6	87.2 ± 101.6	66.4 ± 30.2	0.002
BUN (mmol/L)	6.7 ± 4.4	7.8 ± 3.8	6.5 ± 4.6	< 0.001
INR	1.3 ± 0.3	1.2 ± 0.2	1.3 ± 0.3	0.084
PT (second)	15.3 ± 2.8	14.7 ± 2.6	15.4 ± 2.8	0.042
Child-Pugh score	7.3 ± 1.5	7.1 ± 1.6	7.4 ± 1.5	0.076
Child-Pugh class
A	87 (25.1)	31 (41.9)	84 (30.9)	0.075
B	200 (57.8)	36 (48.6)	165 (60.7)	0.063
C	59 (17.1)	7 (9.5)	23 (8.5)	0.786
MELD score	9.7 ± 4.1	9.0 ± 4.1	9.9 ± 4.1	0.099
TPMT (mm/m)	11.4 ± 3.4	11.6 ± 3.5	11.4 ± 3.4	0.762
Sarcopenia (yes/no)	103 (46.5)	21 (28.4)	82 (30.1)	0.731

BMI: Body mass index; HCC: Hepatocellular carcinoma; Hb: Hemoglobin; WBC: White blood cell; TBIL: Total bilirubin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; BUN: Blood urea nitrogen; INR: International normalized ratio; PT: Prothrombin time; MELD: Model for End-Stage Liver Disease; TPMT: Transverse psoas muscle thickness.

Potential risk factors for ALI after TIPS

Univariable Logistic regression was initially performed for each of the 24 candidate clinical variables. Using a relaxed threshold of P < 0.10, 9 variables were identified as potentially associated with ALI after TIPS (Table 2). These 9 variables were then subjected to LASSO regression with 10-fold crossvalidation. The lambda.min value was selected, which yielded a final set of 7 variables with nonzero coefficients. The LASSO-selected variables were age, HCC, WBC, prealbumin, ALT, creatinine, and PT. Subsequently, a standard multivariable Logistic regression model was fitted including only these 7 variables. The multivariable logistic regression analysis identified only creatinine (OR = 1.011; 95%CI: 1.000-1.023; = 0.046) as an independent predictive variable for ALI after TIPS. The AUC of creatinine for predicting post-TIPS ALI was 0.615 (95%CI: 0.541-0.688, P = 0.003). The optimal cutoff value determined by the Youden index was 62.85, with a sensitivity of 66.2% and a specificity of 52.9%.

Table 2 Univariable and multivariable logistic regression analysis of risk factors for acute liver injury after transjugular intrahepatic portosystemic shunt.

Variables	Univariate			Multivariate
	OR	95%CI	P value	OR	95%CI	P value
Age (year)	1.022	0.999-1.046	0.064	1.031	0.999-1.064	0.061
Gender (male/female)	1.472	0.808-2.683	0.206
BMI	0.933	0.850-1.019	0.132
Etiology (viral hepatitis/other)	1.212	0.717-2.036	0.470
Ascites (yes/no)	0.891	0.520-1.554	0.678
Hypertension (yes/no)	1.362	0.602-2.867	0.434
Diabetes (yes/no)	0.711	0.324-1.430	0.363
HCC	1.976	0.981-3.848	0.049	1.69	0.685-4.17	0.255
Hb	1.005	0.994-1.016	0.340
WBC	1.068	0.987-1.15	0.092	0.894	0.742-1.077	0.237
Platelets	1.001	0.997-1.005	0.685
Albumin	1.022	0.975-1.072	0.360
Prealbumin	1.007	1.001-1.014	0.033	1.004	0.994-1.014	0.452
Globulin	0.982	0.946-1.019	0.355
TBIL	0.984	0.963-1.002	0.105
ALT	1.029	0.997-1.062	0.071	1.021	0.980-1.064	0.321
AST	1.004	0.983-1.024	0.703
Creatinine	1.009	1.002-1.017	0.026	1.011	1.000-1.023	0.046
BUN	1.070	1.015-1.130	0.012	0.865	0.731-1.023	0.090
INR	0.376	0.123-1.149	0.086
PT	0.910	0.817-1.005	0.074
Child-Pugh score	0.871	0.733-1.036	0.119
MELD score	0.948	0.888-1.012	0.107
Sarcopenia (yes/no)	1.106	0.622-1.965	0.731

BMI: Body mass index; HCC: Hepatocellular carcinoma; Hb: Hemoglobin; WBC: White blood cell; TBIL: Total bilirubin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; BUN: Blood urea nitrogen; INR: International normalized ratio; PT: Prothrombin time; MELD: Model for End-Stage Liver Disease.

Outcomes following TIPS

In this study, 149 (43.1%) patients had overt HE after TIPS. For overt HE, the cumulative incidence rate was 45.8% and 28.0% at 3 months, 52.7% and 35.0% at 6 months, 54.3% and 39.0% at 12 months, and 55.9% and 41.1% at 18 months in the ALI and no-ALI groups, respectively (Supplementary Table 2). Evidently, its incidence was found to be significantly elevated in the former than in the latter (log-rank P = 0.005; Figure 1).

Figure 1 Kaplan-Meier curves for cumulative incidence of overt hepatic encephalopathy and overall survival after the transjugular intrahepatic portosystemic shunt procedure. A and B: Kaplan-Meier curves for cumulative incidence of overt hepatic encephalopathy after the transjugular intrahepatic portosystemic shunt (TIPS) procedure in all patients (A), and in acute liver injury (ALI) and no-ALI groups (B); C and D: Kaplan-Meier curves for overall survival after TIPS in all patients (C), and in ALI and no-ALI groups (D). HE: Hepatic encephalopathy.

The median follow-up time was 28.1 (15.0-48.0) months. The probability of OS in the ALI and no-ALI groups was 77.0% and 94.5% at 6 months, 70.3% and 90.8% at 1 year, 68.4% and 80.4% at 2 years, and 68.4% and 77.1% at 3 years, respectively (Supplementary Table 2). Findings of the Kaplan-Meier survival analysis showed that patients with ALI had a significantly lower cumulative OS rate than patients without ALI (log-rank P = 0.025; Figure 1).

Construction and verification of the nomogram

To establish a nomogram, we randomly assigned patients into training (n = 242) and validation (n = 104) sets in a 7:3 ratio. Patient characteristics for sets are shown in Supplementary Table 1. The univariate Cox regression analysis using the training set suggested that age (HR = 1.039; 95%CI: 1.015-1.063; P = 0.002), HCC (HR = 2.621; 95%CI: 1.512-4.546; P = 0.001), prealbumin (HR = 0.989; 95%CI: 0.982-0.996; P = 0.002), globulin (HR = 1.037; 95%CI: 1.001-1.075; P = 0.043), BUN (HR = 1.043; 95%CI: 1.006-1.081; P = 0.022), ALI (HR = 1.746; 95%CI: 1.008-3.024; P = 0.047), AST (HR = 1.033; 95%CI: 1.015-1.052; P < 0.001), sarcopenia (HR = 2.685; 95%CI: 1.631-4.420; P < 0.001), Child-Pugh score (HR = 1.370; 95%CI: 1.171-1.603; P < 0.001), and the MELD score (HR = 1.117; 95%CI: 1.058-1.180; P < 0.001) were potentially associated with mortality. These 10 variables were then subjected to LASSO regression with 10fold crossvalidation (using a relaxed threshold of P < 0.10). Using the lambda.min value, the model retained the following 8 variables: Age, HCC, ALI, AST, BUN, Child-Pugh score, sarcopenia, and MELD score. The multivariable Cox regression analysis indicated age (HR = 1.041; 95%CI: 1.016-1.066; P = 0.001), sarcopenia (HR = 2.257; 95%CI: 1.356-3.756; P = 0.002), ALI (HR = 2.379; 95%CI: 1.405-4.028; P = 0.001), HCC (HR = 2.157; 95%CI: 1.144-4.066; P = 0.018), and MELD score (HR = 1.096; 95%CI: 1.012-1.178; P = 0.013) as independent risk factors for post-TIPS survival (Supplementary Table 3).

Five variables (age, HCC, ALI, sarcopenia, and MELD score) were selected based on LASSO Cox regression to construct the nomogram (Figure 2). The nomogram findings showed acceptable discrimination and good calibration ability for the prediction of OS in this series. The C-index in the validation set was 0.714 (95%CI: 0.609-0.819), and in the training set, it was 0.743 (95%CI: 0.675-0.811). The AUC was > 0.7 for the prediction of 1-, 2-, and 3-year OS rates. The observed and predicted survival probabilities showed good consistency in both sets. In addition, the DCA revealed that the nomogram could predict OS and has a clinical benefit (Figure 3). Stratification of patients on the basis of the optimal cutoff of risk scores (117 points on OS nomogram) showed good discrimination between the high-risk and low-risk groups in both sets (Figure 4). To further elucidate the contribution of each independent predictor to postTIPS mortality and to enhance clinical interpretability beyond conventional hazard ratios, we applied SHapley Additive exPlanations to the final Cox model containing the five selected variables (age, HCC, ALI, sarcopenia, and MELD score; Supplementary Figure 2).

Figure 2 Nomogram for the prediction and receiver operating characteristic curve of prediction of 1-, 2- and 3-year overall survival. A-C: Nomogram for the prediction of 1-, 2- and 3-year overall survival (OS) after transjugular intrahepatic portosystemic shunt procedure (A), training set (B) and validation set (C); D and E: Receiver operating characteristic curve predicting one-, two- and three-year OS in the training set (D) and validation set (E). HCC: Hepatocellular carcinoma; MELD: Model for End-Stage Liver Disease; OS: Overall survival; AUC: Area under the receiver operating characteristic curve.

Figure 3 Decision curve analysis to predict survival for cirrhotic patients after transjugular intrahepatic portosystemic shunt procedure. A-C: 1-, 2- and 3-year survival rates for training set; D-F: 1-, 2- and 3-year survival rates for validation set. OS: Overall survival.

Figure 4 Kaplan-Meier curve for overall survival of patients after the transjugular intrahepatic portosystemic shunt procedure in low-risk and high-risk groups stratified according to the nomogram prediction model. A-C: The survival curves for the total set (A), the training cohort (B), and validation set (C), based on an optimal cutoff point for low- and high-risk groups. OS: Overall survival.

DISCUSSION

In the present study, patients who underwent TIPS and subsequently developed an ALI had poor prognosis, with a higher rate of overt HE and lower survival rates. Apart from the expected prognostic markers such as age, sarcopenia, HCC, and MELD score, ALI also emerged as an independent risk factor for post-TIPS mortality. The current study developed and validated a nomogram to show that ALI can be used as a prognostic marker, and it was useful to evaluate the post-TIPS survival of patients with cirrhosis to reduce portal pressure.

Currently, there is no uniform definition for postoperative liver dysfunction. The hepatic injuries that may occur after a TIPS procedure should not be overlooked[16]. Serum bilirubin, ALT, AST, INR, PT and MELD are often used to identify abnormalities in liver function[2,17-19]. Casadaban et al[8] reported that the mean peak ALT and AST levels were increased three-fold higher than baseline just two days after a TIPS procedure. The incidence of peak ALT for grades 1 or higher (for ALT ≥ 3 × ULN when baseline value was normal and ALT ≥ 1.5 × baseline value when baseline value was abnormal) was 15.6% (17/109) after TIPS. In the current study, when ALI was defined as ALT ≥ 3 × ULN at 72 hours after TIPS, the overall incidence of post-TIPS ALI was 21.4% (74/346). The potential risk factors for developing ALI after TIPS included HCC, prealbumin, creatinine, and BUN levels. Multivariate analyses indicated that creatinine levels were a predictor for ALI. This can help clinicians identify high-risk candidates for the TIPS procedure.

Serum creatinine levels are an important renal function indicator in cirrhotic patients[20]. However, elevated creatinine signals more than renal impairment; it may reflect a decompensated circulatory and inflammatory state that predisposes patients to postTIPS ALI. First, elevated creatinine indicates advanced hemodynamic derangement (systemic vasodilation and reduced effective arterial volume), which predisposes the liver to ischemiareperfusion injury after TIPS[21]. Second, elevated creatinine often coexists with systemic inflammation and endothelial dysfunction, increasing the vulnerability of sinusoidal endothelial cells to shear stressinduced damage[22]. Third, elevated creatinine reflects reduced clearance of inflammatory cytokines and ammonia, thereby exacerbating liver injury and leading to poor outcomes[23]. In patients with cirrhosis, a serum creatinine level ≥ 1.5 mg/dL reportedly leads to worse outcomes than a serum creatinine level < 1.5 mg/dL[24,25]. A study of 90 patients after TIPS placement showed that creatinine had the highest predictive power for predicting 30-day mortality[26]. A recent prospective observation of 411 patients reported serum creatinine and sodium as independent risk factors for post-TIPS liver-related death in older adults, which shows the importance of assessing kidney function[27]. Therefore, high preoperative creatinine levels should be viewed as a risk factor for ALI development and poor prognosis after TIPS, and such patients warrant close monitoring and aggressive management.

In addition to patient-related factors, procedural variables during TIPS creation may also contribute to post-TIPS ALI. Mechanical injury from needle puncture and intrahepatic tract creation directly damages hepatocytes. Specifically, a higher number of puncture attempts increases the risk of capsular perforation and parenchymal trauma[28]; peripheral (vs central) puncture sites result in a longer intrahepatic tract, potentially causing more extensive liver injury[6]; and a larger stent diameter leads to excessive portosystemic shunting, thereby aggravating hepatic ischemia[29]. To minimize such injury, we recommend limiting puncture attempts (≤ 3), central puncture approach, and smallest effective stent diameter. Future studies should systematically record these procedural details to better clarify their impact on ALI.

Post-TIPS liver dysfunction is associated with adverse clinical outcomes, although the deterioration of hepatic function is usually a transient event. A retrospective study of 211 patients indicated that TIPS could precipitate acute hepatic decompensation (assessed by eight liver disease scoring systems), which may be more pronounced in patients with 30-day and 90-day mortality[30]. One study on 216 TIPS patients reported that early liver failure (death, liver transplantation, or MELD > 18 points within 3 months) after TIPS also had lower survival rates at 6 months (37% versus 95%) and 12 months (24% vs 86%)[31]. Similarly, the current study revealed that ALI was associated with higher overt HE rates (3-month: 45.8% vs 28.0%; 6-month: 52.7% vs 35.0%; 12-month: 54.3% vs 39.0%) and lower OS rates (1-year: 70.3% vs 90.8%; 2-year: 68.4% vs 80.4%; 3-year: 68.4% vs 77.1%) compared to patients without ALI after TIPS. Therefore, early postoperative monitoring of liver function and treatment aimed at liver protection are important to improve clinical outcomes.

Previous studies have proposed that age, HCC, sarcopenia, and MELD score can affect survival rates after TIPS[32-36]. In the present study, aside from the risk factors mentioned above, ALI was also an independent predictor for post-TIPS mortality, which has rarely been previously reported. Measured by standard deviation along nomogram scales, the MELD score was the most important prognostic factor, followed by age, HCC, sarcopenia, and ALI. In the current study, internal validation of the nomogram showed an AUC of > 0.7 and a C-index value of 0.714 (95%CI: 0.609-0.819) for predicting 1-, 2-, and 3-year OS. This upholds the acceptable discrimination ability of the nomogram. In addition, the predictions based on the nomogram and actual observations were consistent, indicating satisfactory and reliable performance of the nomogram. As such, the nomogram has major clinical utility and will allow clinicians to easily and accurately identify patients at higher risk of mortality after TIPS (for example, older age, malnutrition, or presence of HCC). The improvement of preoperative MELD scores, postoperative early monitoring, and sufficiently preserved liver function may help decrease the risk of mortality and improve outcomes after TIPS.

This study had several limitations. First, owing to its retrospective nature, data concerning the portal pressure gradients were not routinely available in clinical practice before 2020. Second, the nomogram model was established using single-center data and may require studies using larger multicenter cohorts for clinical validation of the results. Third, the definition of ALI (ALT ≥ 3 × ULN at 72 hours post-TIPS) may not capture all forms of post-procedural liver injury, and the optimal threshold requires further investigation. To mitigate bias, we used consecutive enrollment, multivariate regression with standardized definitions, bootstrap validation (1000 repetitions), and dual independent data extraction. Despite these efforts, unmeasured confounders and the single-center design remain inherent limitations, and external validation in prospective multicenter cohorts is warranted.

CONCLUSION

The current study showed a relatively high incidence (21.4%) of ALI (defined as ALT ≥ 3 × ULN) after TIPS. Serum creatinine was associated with ALI development after TIPS. The ALI group had poorer clinical outcomes, such as higher overt HE rates and lower OS, compared to the group without ALI. Finally, the nomogram model based on sarcopenia, age, HCC, the MELD score, and ALI had acceptable discrimination and good calibration. Clinically, ALI should be routinely assessed after TIPS, and the nomogram can be used as a point-of-care tool to stratify patients by mortality risk, thereby guiding early intensive followup and liverprotective interventions in highrisk individuals.