Development and validation of a nomogram predicting rebleeding in cirrhotic patients with patent transjugular intrahepatic portosystemic shunt

Abstract

BACKGROUND

Even with stent patency following transjugular intrahepatic portosystemic shunt (TIPS) placement, some patients still experience recurrent variceal bleeding.

AIM

To identify risk factors and develop a predictive model for rebleeding in patients with patent TIPS stents.

METHODS

Cirrhotic patients (n = 647) undergoing TIPS (2010-2024) were retrospectively analyzed and grouped by rebleeding status. Propensity score matching (1:3) was performed for age, sex, and model for end-stage liver disease score. Independent predictors identified via univariate and multivariate logistic regression were used to construct a nomogram. The model was internally validated using bootstrapping and evaluated via area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis.

RESULTS

Of the 73 patients with rebleeding following patent TIPS placement, variceal rebleeding was the primary etiology (n = 45). Univariate analysis revealed that platelet count (PLT), pre-procedural pressure gradient (pre-PPG), spleen dimensions (length, width, volume), and splenic vein parameters (diameter, flow) were significantly associated with rebleeding (all P < 0.001). Multivariate regression identified PLT [odds ratio (OR) = 0.980, 95% confidence interval (CI): 0.963-0.996], spleen length (OR = 1.177, 95%CI: 1.002-1.381), spleen width (OR = 1.425, 95%CI: 1.088-1.866), pre-PPG (OR = 1.182, 95%CI: 1.067-1.309), and splenic vein diameter (OR = 1.501, 95%CI: 1.074-2.097) as independent predictors (all P < 0.05). The nomogram exhibited strong discrimination (AUC = 0.883, 95%CI: 0.831-0.936) and favorable calibration (mean absolute error = 0.046). Decision curve analysis confirmed a positive clinical net benefit.

CONCLUSION

The nomogram developed in this study enables the precise prediction of post-TIPS placement rebleeding risk, assisting clinicians in individualized risk stratification and postoperative therapeutic strategies.

Key Words: Transjugular intrahepatic portosystemic shunt; Oesophagogastric variceal rebleeding; Nomogram; Predictive model; Portal hypertension; Stent patency; Propensity score matching

Core Tip: Rebleeding following technically patent transjugular intrahepatic portosystemic shunt (TIPS) placement remains a significant clinical challenge, especially with the use of reduced-diameter stents aimed at mitigating overt hepatic encephalopathy. Although smaller stents decrease procedure-related complications, they may increase the risk of insufficient shunt flow and subsequent variceal rebleeding. This study identified patients experiencing rebleeding despite shunt patency and evaluated pre-procedural factors associated with this adverse outcome. By establishing a predictive nomogram, this research provides a practical tool for screening high-risk individuals, ultimately supporting individualized clinical decision-making and the selection of optimal TIPS diameters for cirrhotic patients.

INTRODUCTION

Transjugular intrahepatic portosystemic shunt (TIPS) placement is currently regarded as an important method to decrease portal pressure and avoid portal hypertension-related complications in selected patients[1-4]. Although the use of covered stents has significantly reduced the rate of rebleeding caused by stent dysfunction, clinical observations indicate that recurrent bleeding may still occur post-TIPS placement, which is primarily attributable to shunt dysfunction or insufficient shunt flow in a patent TIPS[5,6]. Moreover, because of concerns about severe hepatic encephalopathy and liver failure following TIPS placement, a trend towards the use of stents with a reduced diameter, either underdilated or smaller (6 mm in diameter), has occurred; however, this strategy may precipitate rebleeding because of insufficient flow, even in the absence of stent dysfunction[7-9].

Beyond patient-specific factors, shunt diameter is significantly correlated with the incidence of postprocedural overt hepatic encephalopathy (OHE) and hepatic insufficiency[10]. Consequently, small-diameter stents are under active investigation to minimize procedure-related adverse events. Evidence suggests that 8 mm stents effectively reduce OHE rates while maintaining rebleeding rates compared with those of larger stents[11-14]. This raises the question of what the optimal TIPS diameter is to minimize OHE and improve survival. To date, the smallest stent diameter that has been systematically evaluated is 6 mm. A comparative study demonstrated that the 2-year cumulative incidence of OHE was significantly greater for 8 mm stents than for 6 mm stents (42.0% vs 20.3%), and the rebleeding rates were 35.2% for 6 mm stents vs 24.1% for 8 mm stents[7].

The risk of variceal rebleeding stemming from inadequate shunt flow represents a major constraint in the adoption of smaller-diameter stents. Therefore, pre-TIPS screening for patients who will remain at high risk for variceal rebleeding, even with patent shunts, may reveal specific subgroups of patients able to experience survival benefits from stents of a reduced diameter. This possibility warrants further investigation.

On the basis of the above considerations, this study identified patients who experienced rebleeding despite receiving patent TIPS stents in a post-TIPS placement cohort. These patients were rigorously compared with their non-rebleeding counterparts to identify independent risk factors for variceal rebleeding and to develop a clinically applicable nomogram prediction model utilizing pre-TIPS procedure indicators.

MATERIALS AND METHODS

Patients

This study was registered and approved by our Institutional Review Committee, and informed consent was obtained from all study participants prior to performing the TIPS procedures. The data of all patients with cirrhosis who were admitted and treated with TIPS placement for the prevention of variceal rebleeding at our hospital (a tertiary university hospital) between January 2010 and October 2024 were retrospectively reviewed. The inclusion criteria were as follows: (1) Liver cirrhosis (diagnosed by clinical presentation, laboratory tests, imaging, or liver biopsy); (2) Age > 18 and < 75 years; and (3) The indication for TIPS was to prevent variceal rebleeding. The exclusion criteria included lost to follow-up, hepatocellular carcinoma or other extrahepatic malignancies diagnosed before the TIPS procedure, shunt dysfunction confirmed during follow-up, unknown stent status, previous TIPS placement or shunt surgery, and incomplete baseline demographic data. A total of 1111 patients who met the inclusion criteria underwent TIPS placement. After applying the exclusion criteria, 464 patients were excluded, and 647 patients were ultimately included.

Clinical laboratory data, imaging parameters and the occurrence of variceal rebleeding were retrospectively reviewed. All patients underwent computed tomography (CT) before TIPS implantation. Image analyses were performed by two experienced radiologists who were blinded to the clinical information and patient outcome, and the size of the spleen was evaluated as previously described[15]. Splenic height/Length (L) was measured from the number of consecutive CT sections through the spleen. The maximal width (W) of the spleen was defined as the maximum diameter on the transverse sections. The thickness at the hilum (Th) was defined as the distance between the inner and outer borders of the spleen on a plane perpendicular to the width of the spleen and through the hilum[15]. The volume of the spleen (S Vol) was calculated with the following formula: S Vol = 30 + 0.58 (W × L × Th) (cm³)[15] (Figure 1). The diameter of the splenic vein was defined as the maximum diameter measured transversely to the vessel’s longitudinal axis in the portal venous phase[16]. The flow velocity and flow volume of the splenic vein were measured simultaneously using Doppler ultrasound. Abdominal vascular ultrasound was performed to measure the flow velocity in the stent on the hepatic vein side at 1 month, 3 months, 6 months and 12 months after TIPS placement[17].

Figure 1 Computed tomography images. A: Measurements of the length of the spleen (a-b); B: The maximal width (c-d); C: The thickness at the hilum (e-f) determined on a plane perpendicular to the maximal splenic width and through the hilum.

TIPS procedure

All TIPS procedures were performed under local anaesthesia by the same team of hepatologists with extensive intervention experience at our medical centre, as previously described[18]. A 6 mm diameter stent was preferred if patients met both of the following criteria, while a 7 mm stent was used for those who met only one, and an 8 mm stent was used for patients who met neither of them: (1) Age > 65 years; and (2) A Child-Pugh score ≥ 10. A polytetrafluoroethylene covered stent (Fluency; Bard, Murray Hill, NJ, United States) combined with a bare stent or a Gore Viatorr stent (Gore, Newark, DE, United States) was used, depending on the operation date and the patient’s personal preference. The latter has been used since 2015 at our centre. The coronary vein was embolized by coils (Cook, Bloomington, IN, United States) combined with tissue-adhesive glue (octyl-α-cyanoacrylate; Bai Yun Shan, Guangzhou, Guangdong Province, China) during the procedure. The portacaval pressure gradient (PPG) was measured before and immediately after shunt establishment during surgery.

Follow-up and endoscopic diagnosis

Follow-up visits were scheduled at 1 month, 3 months, 6 months and 12 months; every 6 months for the second year; and every year thereafter or whenever the complications of portal hypertension recurred. The follow-up period was defined as the time interval between the initial TIPS stent insertion and liver transplantation, death, or the end of the study. All patients with clinically significant all-cause rebleeding after TIPS underwent endoscopy within 48 hours of admission. Variceal rebleeding observed by endoscopy was defined as blood emanating from a varix or as blood clots seen on the varices if no active bleeding was found, while other causes needed to be ruled out[19]. Portal hypertensive gastropathy was diagnosed on the basis of its typical endoscopic appearance, such as a mucosal mosaic pattern, cherry red spots, and black-brown spots[20]. Other causes of gastrointestinal bleeding were defined according to the corresponding recognized diagnostic criteria[19,21-23].

Definitions

Clinically significant all-cause rebleeding was identified according to Baveno V[24]. Shunt dysfunction was considered in the absence of flow in the shunt on Doppler ultrasound imaging or shunt stenosis > 50% confirmed by TIPS revision[25]. If patients had peak intra-shunt velocity > 250 cm/second or maximum portal vein velocity < two-thirds of the baseline value[26] or hepatofugal-to-hepatopetal flow in the intrahepatic portal branches, TIPS revision was conducted to further confirm the stent dysfunction.

Statistical analysis

SPSS and R software packages were used for all the statistical analyses. Continuous variables are expressed as the mean ± SD for normally distributed data or as the median (interquartile range) for nonnormally distributed data and were compared using the independent-sample t test or Mann-Whitney U test as appropriate, whereas qualitative variables were compared using the χ² test or Fisher’s exact test. The patients who did not experience gastrointestinal rebleeding during follow-up were matched to the variceal rebleeding patients (3:1) and served as a control group according to age, sex, and model for end-stage liver disease (MELD) score, with a calliper of 0.02. In the univariate analysis, variables with P < 0.05 were first subjected to collinearity diagnostics to ensure that the variance inflation factor of all the variables was less than 5. Logistic regression analysis was performed by incorporating variables with P < 0.05 from the univariate analysis, and a nomogram model was constructed on the basis of the independent predictors. This was followed by internal validation using the bootstrap resampling method with 1000 repetitions to evaluate the stability of the model. Calibration curves were used to assess the agreement between the predicted probabilities and actual outcomes. A receiver operating characteristic (ROC) curve was used to evaluate the discriminative ability of the model, and the area under the curve (AUC) was calculated. Decision curve analysis was applied to estimate the clinical net benefit of the model.

RESULTS

Patient characteristics

A total of 647 patients were enrolled between 2010 and 2024, with a mean age of 56.1 ± 10.9 years and 379 were men. The mean MELD score was 10.3 ± 2.8, and Child-Pugh classes A, B, and C were seen in 35.9%, 59.8%, and 4.3% of the patients, respectively, prior to TIPS placement. Clinical all-cause gastrointestinal rebleeding occurred in 73/647 (11.3%) patients during follow-up (Figure 2). Postoperative bleeding occurred from oesophagogastric varices in 45/73 (61.6%) patients; duodenal ulcers in 10/73 (13.7%) patients; portal hypertensive gastropathy in 3/73 (4.1%) patients; idiopathic haemorrhage in 4/73 (5.5%) patients; glue injection-induced ulceration in 4/73 (5.5%) patients; Dieulafoy disease in 2/73 (2.7%) patients; and haemorrhoids, oesophageal erosion, compound ulcer, capillary telangiectasia, and ischaemic bowel disease in 1/73 (1.4%) patients each.

Figure 2 Flowchart of patient enrolment and study design. TIPS: Transjugular intrahepatic portosystemic shunt; PSM: Propensity score matching.

Oesophagogastric variceal rebleeding

Given that, in our study, oesophagogastric variceal rebleeding (EGVB) was the main cause of post-TIPS rebleeding in patients without shunt dysfunction, we explored the risk factors for EGVB by comparing the baseline characteristics of the variceal rebleeding and non-rebleeding groups (Table 1). The patients who did not experience gastrointestinal rebleeding (non-rebleeding group) during follow-up were matched to the variceal rebleeding patients (3:1) according to age, sex, and MELD score. The characteristics of the patients in the two groups are detailed in Table 1. According to the univariate analysis, a low preoperative platelet (PLT) was associated with a risk of variceal rebleeding with TIPS patency (P < 0.001), and this association persisted at 1 and 12 months after surgery. A variceal rebleeding event was more common in patients with a higher PPG before TIPS placement (P < 0.001), and the difference in PPG between the two groups remained after stent implantation (P < 0.001). The occurrence of variceal rebleeding was not associated with the severity of liver disease (as assessed by liver function) and was independent of the underlying liver disease and the velocity of blood flow in the stent at 1 and 12 months.

Table 1 Clinical characteristics of patients with variceal rebleeding (n = 45) and the matched non-rebleeding group (n = 135), mean ± SD/n (%).

Variable	EGVB (n = 45)	Non-rebleeding (n = 135)	P value
Male sex	26 (57.8)	79 (58.5)	0.933
Age (years), median (IQR)	54.6 ± 10.1	55 (48-66)	0.562
Aetiology of cirrhosis			0.783
Hepatitis B virus	20 (44.4)	68 (50.4)
Alcoholic liver disease	7 (15.6)	24 (17.8)
Autoimmune	12 (26.7)	27 (20.0)
Others	6 (13.3)	16 (11.9)
CTP score, median (IQR)	7.0 (6.0-8.0)	7.0 (6.0-8.0)	0.202
CTP class			0.244
A	12 (26.7)	53 (39.3)
B	32 (71.1)	79 (58.5)
C	1 (2.2)	3 (2.2)
MELD score, median (IQR)	10.0 (9.0-12.0)	10.0 (8.9-13.0)	0.962
WBC (× 109/L), median (IQR)	2.5 (1.6-3.8)	2.9 (1.9-4.8)	0.0922
PLT (× 109/L), median (IQR)	41.0 (22.0-61.5)	76.0 (58.0-116.0)	< 0.0012,c
ALT (U/L), median (IQR)	19.2 (14.8-26.2)	21.2 (15.0-28.3)	0.682
AST (U/L), median (IQR)	25.6 (19.5-36.2)	28.3 (21.9-37.3)	0.322
ALB (g/L)	32.8 ± 4.4	33.2 ± 5.0	0.661
TB (μmol/L), median (IQR)	20.4 (12.9-26.4)	21.0 (14.8-29.6)	0.462
Cr (mmol/L), median (IQR)	59.0 (48.5-75.5)	59.0 (50.0-74.0)	0.972
PT (second)	14.5 ± 1.4	14.5 ± 2.4	0.991
INR, median (IQR)	1.3 ± 0.2	1.3 (1.2-1.4)	0.652
Pre-PPG (mmHg), median (IQR)	24.7 ± 3.5	21.0 (18.0-23.7)	< 0.0012,c
Post-PPG (mmHg), median (IQR)	11.8 ± 4.0	8.8 (7.9-12.0)	< 0.0012,c
PPG improvement rate (%), median (IQR)	51.9 ± 15.7	56.6 (44.4-63.2)	0.362
Diameter of stent			0.294
6 mm	7 (15.6)	10 (7.4)
7 mm	2 (4.4)	4 (3.0)
8 mm	36 (80.0)	120 (88.9)
10 mm	0 (0)	1 (0.7)
Length of spleen (cm), median (IQR)	20.3 ± 3.7	15.8 (13.6-18.3)	< 0.0012,c
Width of spleen (cm)	15.4 ± 2.5	12.7 ± 2.5	< 0.0011,c
Thickness of spleen (cm), median (IQR)	5.6 ± 1.5	5.6 (4.8-6.6)	0.781
Volume of spleen (cm3), median (IQR)	1019 (743-1354)	749 (454-924)	< 0.0012,c
Diameter of splenic vein (mm), median (IQR)	14.5 ± 3.1	12.0 (10.0-13.3)	< 0.0012,c
Spleen vein velocity (cm/second), median (IQR)	30.8 (26.8-37.1)	33.3 (28.2-37.5)	0.282
Flow volume of splenic vein (L/minute), median (IQR)	3.0 (2.1-4.5)	2.3 (1.4-3.2)	< 0.0012,c
Post 1 month WBC (× 109/L), median (IQR)	2.6 (1.6-3.6)	3.3 (2.2-4.4)	0.0142,a
Post 1 month PLT (× 109/L), median (IQR)	49 (32-67)	65 (39-100)	0.0092,b
Post 1 month velocity of blood flow in stent (cm/second), median (IQR)	199 (179-214)	189 (175-204)	0.0642
Post 12 months velocity of blood flow in stent (cm/second), median (IQR)	192 ± 33	182 (171-199)	0.0762
Post 12 months WBC (× 109/L), median (IQR)	2.7 (1.9-3.6)	3.7 (2.6-4.3)	< 0.0012,c
Post 12 months PLT (× 109/L), median (IQR)	38 (32-72)	77 (48-113)	< 0.0012,c
Portal vein thrombosis			0.124
No	31 (68.9)	85 (63.0)
Partial	12 (26.7)	49 (36.3)
Completely occluded or cavernoma	2 (4.4)	1 (0.7)
Gastrorenal shunt	6 (13.3)	14 (10.4)	0.583

^aP < 0.05.

^bP < 0.01.

^cP < 0.001.

¹Independent-sample t-test.

²Mann-Whitney U test.

³χ² test.

⁴Fisher’s exact test.

EGVB: Oesophagogastric variceal rebleeding; IQR: Interquartile range; CTP: Child-Turcotte-Pugh; MELD: Model for end-stage liver disease; WBC: White blood cell; PLT: Platelet; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALB: Albumin; TB: Total bilirubin; Cr: Creatinine; PT: Prothrombin time; INR: International normalized ratio; PPG: Portal pressure gradient.

The results concerning the spleen parameters and other portal hypertension-related manifestations on CT are detailed in Table 1. The presence of portal vein thrombosis and a gastrorenal shunt were not associated with post-TIPS placement variceal rebleeding. However, the spleen size and flow volume of the splenic vein were strongly associated with the risk of variceal rebleeding. The length, width and volume of the spleen measured pre-TIPS placement were significantly greater in patients with variceal rebleeding, at 20.3 ± 3.7 cm vs 15.8 (13.6-18.3) cm for the splenic length (P < 0.001), 15.4 ± 2.5 cm vs 12.7 ± 2.5 cm for the splenic width (P < 0.001), and 1019 (743-1354) cm³ vs 749 (454-924) cm³ for the splenic volume (P < 0.001) in patients with and without variceal rebleeding, respectively.

Increased splenic vein diameter and flow volume were also associated with the occurrence of variceal rebleeding. The splenic vein diameter and flow volume in the group with variceal rebleeding was 14.5 ± 3.1 mm and 3.0 (2.1-4.5) L/minute, respectively, vs 12 (10.0-13.3) mm and 2.3 (1.4-3.2) L/minute, respectively, in the nonbleeding group (all P < 0.001).

Logistic regression and variable selection

Prior to multivariate analysis, collinearity diagnostics confirmed the absence of multicollinearity among the candidate variables, with all variance inflation factor values being strictly less than 5 (Table 2). Multivariate logistic regression analysis of the variables significant in univariate analysis (P < 0.05) revealed the following independent predictors: Spleen length [odds ratio (OR) = 1.177, 95% confidence interval (CI): 1.002-1.381, P = 0.047], spleen width (OR = 1.425, 95%CI: 1.088-1.866, P = 0.010), splenic vein diameter (OR = 1.501, 95%CI: 1.074-2.097, P = 0.017), pre-PPG (OR = 1.182, 95%CI: 1.067-1.309, P = 0.001), and PLT (OR = 0.980, 95%CI: 0.963-0.996, P = 0.017) (Table 2). A predictive model and a nomogram were subsequently developed using these variables.

Table 2 Logistic regression analysis of risk factors for rebleeding.

Variable	β	SE	P value	Exp (B)	VIF
Length of spleen (cm)	0.163	0.08	0.047	1.177 (1.002, 1.381)	2.551
Width of spleen (cm)	0.354	0.14	0.010	1.425 (1.088, 1.866)	2.066
Volume of spleen (cm3)	0.001	0.00	0.108	0.998 (0.996, 1.000)	1.541
Diameter of splenic vein (mm)	0.406	1.71	0.017	1.501 (1.074, 2.097)	4.432
Flow volume of splenic vein (L/minute)	0.366	0.24	0.123	0.694 (0.435, 1.104)	3.587
Pre-PPG (mmHg)	0.167	0.05	0.001	1.182 (1.067, 1.309)	1.105
PLT	0.021	0.01	0.017	0.980 (0.963, 0.996)	1.170

PLT: Platelet; VIF: Variance inflation factor; PPG: Portal pressure gradient.

Nomogram prediction model

The nomogram developed for predicting the probability of rebleeding, which was constructed on the basis of the independent predictors identified from multivariate logistic regression analysis, is shown in Figure 3. The model incorporates five key variables: PLT (ranging from 600 to 0), spleen length (from 5 units to 40 units), spleen width (from 7 units to 22 units), splenic vein diameter (from 6 units to 21 units), and PPG (from 10 units to 38 units). Each variable is assigned points on a scale from 0 to 100 on the basis of its measured value. The sum of these points yields a total score (ranging from 0 to 200), which is then mapped to the corresponding predicted probability of rebleeding (from 0.05 to 0.95).

Figure 3 Nomogram for predicting the probability of rebleeding. This nomogram integrates five predictors platelet count, spleen length, spleen width, splenic vein diameter, and portal pressure gradient to estimate the individual risk of rebleeding. For each predictor, the corresponding patient value is located on its respective axis and converted to points using the “points” scale at the top. The sum of all points yields the “total points”, which maps to the predicted probability of re-bleeding shown on the bottom scale. This model provides a quantitative tool to sup-port clinical risk assessment. PLT: Platelet; PPG: Portacaval pressure gradient.

Model performance evaluation

Internal validation using 1000 bootstrap resamples demonstrated that the model is robust and has strong discriminative performance. The concordance index was 0.86, indicating an excellent ability to distinguish between patients with and without the outcome of interest. The calibration curve showed close agreement between the predicted probabilities and the observed event rates, with a mean absolute error of 0.046 (Figure 4), suggesting good calibration and minimal bias.

Figure 4 Calibration plot of the nomogram model. This figure illustrates the agreement between the nomogram-predicted probability of rebleeding and the actual observed rebleeding rate. The dotted line (apparent) represents the model’s performance on the original dataset, while the solid line (bias-corrected) shows the boot-strap-corrected calibration curve based on 1000 resamples, indicating the model’s internal stability. The diagonal line denotes perfect calibration. The mean absolute error was 0.046 (n = 180), demonstrating good concordance between predicted and observed outcomes.

Further ROC analysis revealed an AUC of 0.883 (95%CI: 0.831-0.936) (Figure 5), indicating strong discriminative ability. The optimal cut-off value determined by the Youden index corresponded to a sensitivity of 0.733 and a specificity of 0.904, demonstrating that the model achieves high specificity while maintaining satisfactory sensitivity.

Figure 5 Receiver operating characteristic curve of the rebleeding risk pre-diction model. This receiver operating characteristic curve evaluates the discriminatory ability of the model to distinguish patients with rebleeding from those without. The area under the curve was 0.883, with a 95% confidence interval of 0.831-0.936, indicating excellent discriminative performance. The dashed line represents the reference line of a random classifier. ROC: Receiver operating characteristic; AUC: Area under the curve; CI: Confidence interval; TIPS: Transjugular intrahepatic portosystemic shunt.

Decision curve analysis (Figure 6) revealed that the model provided greater net benefit than either the “treat-all” or “treat-none” strategies across the entire threshold probability range of 0-1.0. Notably, the model yielded the most substantial clinical net benefit when the threshold probability was between approximately 0.2 and 0.8. Within this range, the corresponding cost-benefit ratio (risk threshold) ranged from approximately 1:4 to 4:1, indicating strong potential clinical utility in guiding decision-making.

Figure 6 The blue curve represents the net benefit of the prediction model incorporating platelet count, spleen length, spleen width, splenic vein diameter, and portal pressure gradient. The grey and black lines correspond to the “treat all” and “treat none” strategies, respectively. Across a wide range of high-risk thresholds, the model demonstrates a consistently greater standardized net benefit compared with the two default strategies, indicating its potential clinical utility in guiding individualized risk-based decision-making.

DISCUSSION

Individualized selection of the shunt diameter implanted during TIPS is important for balancing the prevention of postoperative rebleeding and reducing the risk of hepatic encephalopathy. In this study, we investigated risk factors predictive of variceal rebleeding in patients with a patent TIPS, aiming to facilitate optimal shunt diameter selection for TIPS and more rigorous postoperative patient management in high-risk patients, such as more frequent follow-up, combined use of nonselective beta-blockers and endoscopic treatment, or even splenectomy or splenic artery embolization.

The primary findings are threefold: First, variceal rebleeding constituted the predominant cause of gastrointestinal rebleeding in patients with a patent TIPS; Second, the preoperative PLT, PPG, spleen length, spleen width, and splenic vein diameter were identified as independent risk factors; And third, a validated nomogram incorporating these five predictive factors demonstrated robust predictive performance. Propensity score matching was employed to mitigate selection bias and control for confounding variables, including age, sex, and liver disease severity, thereby enhancing intergroup comparability.

The findings of this study indicate that splenic factors significantly influence post-TIPS implantation portal haemodynamics and clinical outcomes. Although a TIPS effectively reduces portal hypertension by decreasing resistance within the portal venous system, its decompressive efficacy is not solely dependent on stent patency; rather, it is modulated by the haemodynamic load from the spleen[27-29]. In patients with significant splenomegaly or splenic vein dilatation, substantial splenic inflow to the portal vein may persist, potentially attenuating the effective diversion achieved by a TIPS. Consequently, residual portal pressure may remain elevated, and complications related to portal hypertension, such as rebleeding, may occur even when the stent remains patent postoperatively[27,29].

Furthermore, thrombocytopenia a key clinical manifestation of hypersplenism exhibits a strong correlation with splenic enlargement and the severity of portal hypertension[28,30]. Several studies have demonstrated that PLT may improve in some patients following TIPS placement as portal pressure decreases; however, such improvement is not universally observed. Notably, patients who present with large preoperative splenic volumes or high splenic inflow demonstrate only limited postoperative PLT recovery[27,28]. This observation suggests that persistent hypersplenism may reflect either an insufficient reduction in portal pressure or the irreversible nature of hypersplenism even after TIPS placement, thus potentially increasing the risk of rebleeding.

The preoperative PPG is a key determinant of TIPS efficacy. It is widely accepted that if the post-TIPS implantation PPG does not decrease to < 12 mmHg or by ≥ 50% from baseline, the risk of rebleeding remains high[31-33]. Consistent with previous studies, patients with high baseline PPG in this study often required a greater degree of shunting to reach the physiologic “safety threshold”. Consequently, they were more likely to experience residual portal hypertension or recurrent bleeding after TIPS placement[31,32]. Therefore, patients with elevated preoperative PPG should undergo more rigorous risk assessment, with careful intraoperative and early postoperative pressure monitoring to ensure the adequacy of shunting.

Although the post-TIPS placement PPG reduction in our cohort reached the predefined therapeutic target and no significant differences were observed among different stent diameters, the phenomenon of “relative under-shunting” may still have occurred in patients with marked splenomegaly. Anatomical studies have demonstrated that splenomegaly is frequently accompanied by extensive portosystemic collateralization around the splenic hilum, including the short gastric veins, posterior gastric veins, and splenorenal shunts; these collateral vessels play a role in the formation of gastric varices[34]. Given these established anatomical features, it is reasonable to hypothesize that, even after a technically successful TIPS placement with adequate PPG reduction, a substantial proportion of splenic inflow may continue to be diverted through these preexisting collaterals rather than through the TIPS tract. Theoretically, this could lead to insufficient relative shunt volume.

From a therapeutic perspective, accumulating evidence substantiates the concept that “splenic-portal inflow burden” warrants consideration as a factor in TIPS decision-making. When the splenic volume is significantly enlarged or the splenic inflow is disproportionately elevated, TIPS monotherapy may prove insufficient to achieve optimal portal decompression. Combined or sequential splenic interventions such as partial splenic artery embolization or splenectomy have demonstrated potential benefits in reducing rebleeding risk and ameliorating thrombocytopenia in selected patients[35,36]. Such approaches may be particularly valuable for patients who are unable to tolerate surgical procedures or those who exhibit a suboptimal haemodynamic response to TIPS placement.

The primary innovation of this study lies in the integration of the splenic size and splenic vein diameter into the predictive model. These parameters directly reflect the volumetric load within the splenoportal system, thereby enhancing its predictive performance. To optimize personalized management, high-risk patients should undergo more frequent endoscopic surveillance with shorter follow-up intervals (e.g., every 3 months). Additionally, for patients with higher nomogram scores, clinicians may consider TIPS combined with splenic embolization or splenectomy, the utilization of larger-diameter stents, or more stringent postoperative β-blocker therapy.

This study also has several limitations. First, its single-centre, retrospective design and limited sample size render selection bias unavoidable. Second, the inclusion of only patients with cirrhosis undergoing TIPS placement necessitates multicentre studies to validate the applicability of the model to other hepatic pathologies. Furthermore, the model underwent only internal validation and lacked external cohort validation. Future work should encompass multicentre prospective studies incorporating external validation and attempts to develop a dynamically updating model. Finally, the splenic size quantification relied on imaging-based measurements, introducing potential measurement variability.

CONCLUSION

In this study, five independent preoperative predictors PLT, PPG, splenic length, splenic width, and splenic vein diameter were identified and used to develop a nomogram for predicting the risk of rebleeding after TIPS placement. The model exhibited robust discrimination, calibration, and clinical utility, offering clinicians a simple and intuitive tool for individualized risk assessment, thereby facilitating the optimization of post-TIPS placement management strategies. Future investigations with larger sample sizes and multicentre external validation are warranted to refine and generalize this model.

ACKNOWLEDGEMENTS

The authors would like to thank all the study participants for their voluntary participation.