Yang LB, Gao X, Xu M, Li Y, Dong L, Huang XD, She X, Zhang DY, Zhang QW, Liu CY, Fan ST, Wang Y. Noninvasive model based on liver and spleen stiffness for predicting clinical decompensation in patients with cirrhosis. World J Gastroenterol 2025; 31(33): 107408 [DOI: 10.3748/wjg.v31.i33.107408]
Corresponding Author of This Article
Yan Wang, MD, Assistant Professor, Department of Gastroenterology, The Second Affiliated Hospital of Xi’an Jiaotong University, No. 157 of Xiwu Road, Xi’an 710004, Shaanxi Province, China. sarrye@163.com
Research Domain of This Article
Gastroenterology & Hepatology
Article-Type of This Article
Retrospective Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Long-Bao Yang, Xin Gao, Yong Li, Lei Dong, Xin-Di Huang, Xiao She, Dan-Yang Zhang, Qian-Wen Zhang, Chen-Yu Liu, Yan Wang, Department of Gastroenterology, The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, Shaanxi Province, China
Meng Xu, Department of General Surgery, The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, Shaanxi Province, China
Shu-Ting Fan, Department of Supply, The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, Shaanxi Province, China
Author contributions: Yang LB and Gao X contributed equally to this work as co-first authors; Xu M, Li Y, and Dong L designed the research study; Liu CY and Gao X performed the research; Huang XD, She X, and Zhang DY contributed new reagents and analytic tools; Zhang QW, Fan ST, and Wang Y analyzed the data and wrote the manuscript; all the authors have read and approved the final manuscript.
Supported by Xi’an Science and Technology Plan, No. 23YXYJ0172.
Institutional review board statement: The study was approved by the Ethics Committee of The Second Affiliated Hospital of Xi’an Jiaotong University, No. 2017-445.
Informed consent statement: The informed consent was waived because of its retrospective nature.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
Data sharing statement: The dataset generated and analyzed during the current study is available from the corresponding author on reasonable request.
Open Access: This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: https://creativecommons.org/Licenses/by-nc/4.0/
Corresponding author: Yan Wang, MD, Assistant Professor, Department of Gastroenterology, The Second Affiliated Hospital of Xi’an Jiaotong University, No. 157 of Xiwu Road, Xi’an 710004, Shaanxi Province, China. sarrye@163.com
Received: March 23, 2025 Revised: May 13, 2025 Accepted: August 12, 2025 Published online: September 7, 2025 Processing time: 162 Days and 23 Hours
Abstract
BACKGROUND
The hepatic venous pressure gradient serves as a crucial parameter for assessing portal hypertension and predicting clinical decompensation in individuals with cirrhosis. However, owing to its invasive nature, there has been growing interest in identifying noninvasive alternatives. Transient elastography offers a promising approach for measuring liver stiffness and spleen stiffness, which can help estimate the likelihood of decompensation in patients with chronic liver disease.
AIM
To investigate the predictive ability of the liver stiffness measurement (LSM) and spleen stiffness measurement (SSM) in conjunction with other noninvasive indicators for clinical decompensation in patients suffering from compensatory cirrhosis and portal hypertension.
METHODS
This study was a retrospective analysis of the clinical data of 200 patients who were diagnosed with viral cirrhosis and who received computed tomography, transient elastography, ultrasound, and endoscopic examinations at The Second Affiliated Hospital of Xi’an Jiaotong University between March 2020 and November 2022. Patient classification was performed in accordance with the Baveno VI consensus. The area under the curve was used to evaluate and compare the predictive accuracy across different patient groups. The diagnostic effectiveness of several models, including the liver stiffness-spleen diameter-platelet ratio, variceal risk index, aspartate aminotransferase-alanine aminotransferase ratio, Baveno VI criteria, and newly developed models, was assessed. Additionally, decision curve analysis was carried out across a range of threshold probabilities to evaluate the clinical utility of these predictive factors.
RESULTS
Univariate and multivariate analyses demonstrated that SSM, LSM, and the spleen length diameter (SLD) were linked to clinical decompensation in individuals with viral cirrhosis. On the basis of these findings, a predictive model was developed via logistic regression: Ln [P/(1-P)] = -4.969 - 0.279 × SSM + 0.348 × LSM + 0.272 × SLD. The model exhibited strong performance, with an area under the curve of 0.944. At a cutoff value of 0.56, the sensitivity, specificity, positive predictive value, and negative predictive value for predicting clinical decompensation were 85.29%, 88.89%, 87.89%, and 86.47%, respectively. The newly developed model demonstrated enhanced accuracy in forecasting clinical decompensation among patients suffering from viral cirrhosis when compared to four previously established models.
CONCLUSION
Noninvasive models utilizing SSM, LSM, and SLD are effective in predicting clinical decompensation among patients with viral cirrhosis, thereby reducing the need for unnecessary hepatic venous pressure gradient testing.
Core Tip: In this study, we developed a novel noninvasive predictive model using the liver stiffness measurement, spleen stiffness measurement, and spleen length diameter to evaluate the risk of clinical decompensation in individuals with viral cirrhosis. This is a new model that has not been reported before. Our findings indicate that it outperforms existing prediction models, demonstrating greater accuracy in identifying patients at risk. As a result, it holds significant potential for supporting clinical decision-making processes.
Citation: Yang LB, Gao X, Xu M, Li Y, Dong L, Huang XD, She X, Zhang DY, Zhang QW, Liu CY, Fan ST, Wang Y. Noninvasive model based on liver and spleen stiffness for predicting clinical decompensation in patients with cirrhosis. World J Gastroenterol 2025; 31(33): 107408
Cirrhosis represents the terminal phase of chronic liver disease, with its natural progression moving from asymptomatic, compensated cirrhosis to the symptomatic stage known as decompensated cirrhosis. In contrast to the compensated phase, decompensated cirrhosis is characterized by portal hypertension (PH) and impaired liver function, both of which are associated with a significantly increased risk of patient mortality[1]. Patients diagnosed with compensated advanced chronic liver disease can be categorized into different risk groups depending on whether they have clinically significant PH (CSPH)[2]. CSPH is typically defined as a hepatic venous pressure gradient (HVPG) exceeding 10 mmHg[3]. In patients with decompensated cirrhosis, various complications can arise, including bleeding from esophageal and gastric varices (EGVB), hepatic encephalopathy (HE), and the accumulation of ascitic fluid. The prognosis varies significantly between disease stages, with patients suffering from decompensated cirrhosis having a median survival time of under 2 years, compared with approximately 12 years for those without decompensation[1]. Patients with compensatory cirrhosis generally have a more favorable prognosis and longer survival than do those with decompensated cirrhosis. Liver transplantation remains the sole effective therapeutic option for individuals with end-stage liver disease[4]. Therefore, the early detection of morphological or hemodynamic alterations in cirrhotic patients, along with the assessment of decompensation risk, plays a crucial role in minimizing mortality linked to complications of PH. Additionally, such early intervention facilitates the application of optimal preventive measures, which can potentially affect the progression of the disease.
HVPG measurement is widely regarded as the most reliable method for evaluating the severity of PH and predicting disease decompensation in individuals with cirrhosis. Existing clinical guidelines advocate the use of the HVPG as a key marker for assessing prognosis and monitoring treatment response in such patients[5,6]. Nevertheless, the widespread use of the HVPG is limited by its invasive nature, considerable cost, potential complications, and low patient acceptance[7]. As a result, there is a pressing need for a noninvasive and effective alternative that can accurately diagnose PH and predict decompensation in cirrhotic patients. In recent years, various noninvasive approaches, including blood biomarkers, biochemical parameters, imaging techniques, and scoring systems, have been developed and utilized to assess liver fibrosis and, more recently, to predict outcomes in patients with chronic liver disease[8-10]. Transient elastography (TE) is an innovative method that allows for fast, noninvasive, and cost-effective assessment of tissue elasticity, demonstrating high accuracy in identifying cirrhosis[11]. Multiple studies have confirmed its effectiveness in predicting CSPH, suggesting its potential as a valuable tool for evaluating the risk of decompensation in individuals with chronic liver disease[4]. Research has also indicated that liver stiffness and spleen stiffness, both derived from TE, are closely related to portal pressure and have emerged as dependable indicators for detecting PH[3,12]. Nevertheless, the application of liver stiffness measurement (LSM) in patients with HVPG values exceeding 12 mmHg remains insufficiently validated, largely because of the increasing impact of extrinsic hepatic factors on PH progression. Emerging evidence suggests that spleen stiffness measurement (SSM) may offer a more accurate approach for diagnosing PH, as it better reflects the hemodynamic changes associated with advanced cirrhosis[6,13].
In addition, the Baveno VI, which is based on the LSM (LSM ≥ 20 kPa) and platelet count (PLT ≤ 150 × 109/L), can be used in place of HPVG measurement[14-16]. Currently, models based on the LSM and SSM combined with other noninvasive indicators are effective tools for predicting liver cirrhosis and EGVB. However, few studies have reported the usefulness of the LSM, SSM, and other indicators for predicting clinical decompensation in patients with cirrhosis. Thus, our objective was to assess the value of integrating the LSM and SSM with additional markers in independently forecasting clinical decompensation among patients suffering from viral cirrhosis.
MATERIALS AND METHODS
Participants
This study was a retrospective analysis conducted at a single center. It adhered to the principles outlined in the 1965 Declaration of Helsinki, and all research procedures followed applicable guidelines. The Ethics Committee of The Second Affiliated Hospital of Xi’an Jiaotong University granted approval for the study, and the requirement for written informed consent was waived because of the retrospective design. We retrospectively collected clinical data from 200 patients who were admitted to the gastroenterology department of The Second Affiliated Hospital of Xi’an Jiaotong University between March 2020 and November 2022. All included patients tested positive for serum hepatitis B virus/hepatitis C virus DNA or RNA through polymerase chain reaction testing. The diagnosis of liver cirrhosis was made on the basis of a combination of liver histology, imaging findings, endoscopic results, biochemical parameters, and physical examination. According to the Baveno VI consensus, decompensated cirrhosis is characterized by a current or prior occurrence of EGVB, ascites, and HE.
The patient selection criteria were as follows: (1) Diagnosis of cirrhosis associated with hepatitis B or C virus infection; (2) Availability of laboratory test results and esophagogastroduodenoscopy reports within a 3-month window around the TE assessment; and (3) Age above 18 years. Patients were excluded from the study if they met any of the following conditions: (1) Had liver tumors during the study period; (2) Had prior treatment with endoscopic variceal ligation, sclerotherapy, or transjugular intrahepatic portal shunting; (3) Had undergone previous splenectomy or partial splenic embolization; (4) Had cirrhosis resulting from etiologies other than viral hepatitis, such as alcoholic or nonalcoholic fatty liver disease; (5) Had prior use of nonselective beta-blockers; (6) Were pregnant; (7) Had severe comorbidities involving other organ systems; (8) Had alcohol consumption within the past six months; (9) Had congestive heart failure, abdominal trauma at the location of TE measurement, alanine aminotransferase (ALT) levels five times or higher than the upper limit of normal, or total bilirubin ≥ 2 mg/dL, which could interfere with TE accuracy; and (10) Were obese, defined as a body mass index of 30 kg/m2 or higher. Ultimately, 200 patients were enrolled in the study, with 140 assigned to the model-building cohort and 60 to the external validation cohort.
Determination of liver and spleen volume
A 128-slice GE spiral computed tomography (CT) scanner (GE Healthcare, MA, United States) was used to assess the true liver volume and actual spleen volume, with reconstructed slices of 5 mm thickness and a 5-second scanning interval. The spleen length diameter (SLD) was determined as the maximum distance between the upper and lower poles of the spleen on its largest cross-sectional image. The portal vein diameter was measured at the central point of the section located between the bifurcation of the portal vein and the site of venous convergence. During the measurement process, care was taken to exclude major blood vessels, the gallbladder, and anatomical fissures[17-19].
LSM and SSM
TE was used to obtain the LSM and SSM in patients with cirrhosis. TE was performed on the basis of the FibroScan system (Echosens, France) of one-dimensional echocardiography technology, and the results are presented in kPa. The SSM was measured with the same probe as the LSM. Each patient was evaluated by an experienced operator after fasting for at least 4 hours. The interquartile range indicates the variability among measurements. Tests are deemed reliable if they include a minimum of 10 valid measurements and achieve a success rate exceeding 60% for each patient. TE and SSM tests were performed by the same physician on the same day, and the SSM was measured with the same criteria as the LSM. A valid measurement is defined as an interquartile range/median ratio ≤ 30%[20-22].
Laboratory parameters
A history of esophageal varices rupture and bleeding was obtained via medical records retrieval and telephone inquiry. We collected demographic information, medication history, body mass index, virological parameters, and laboratory data. The following biochemical parameters were measured: White blood cell count, red blood cell count, PLT, ALT, aspartate aminotransferase (AST), total bilirubin, alkaline phosphatase, gamma-glutamyl transferase, albumin, total cholesterol, coagulation prothrombin time, international prothrombin ratio, and prothrombin activity. The Child-Pugh scoring method was used to evaluate the scores of patients with cirrhosis. Blood samples were analyzed via an XN-9000 hematology analyzer (Shanghai, China), coagulation profiles were assessed via the Sysmex Cocs-1500 system, and liver function was evaluated via a Cobas 8000 analyzer (Roche Diagnostics, Mannheim, Germany).
Prediction rules and models
The noninvasive prediction models selected for comparison in this study included the liver stiffness-spleen diameter-platelet ratio (LSPS), calculated as [LSM (kPa) × SLD (cm)]/PLT (× 109/L)[14], and the variceal risk index (VRI), determined by the formula: -4.364 + 0.538 × SLD - 0.049 × PLT - 0.044 × LSM + 0.001 × (LSM × PLT)[23]. Additionally, the AST-ALT ratio (AAR), expressed as AST/ALT[8], was also evaluated. Furthermore, we considered the Baveno VI criteria, which are defined as LSM values below 20 kPa combined with PLT exceeding 150 × 109/L, as well as the extended Baveno VI criteria, characterized by an LSM less than 25 kPa and a PLT greater than 110 × 109/L[15]. Finally, the Baveno VII criteria were applied, which include individuals meeting the Baveno VI criteria or those who do not meet them but have an SSM value below -40 kPa[16,24].
Statistical analysis
Statistical analysis was conducted via SPSS version 26.0 (SPSS Inc., Chicago, IL, United States). For comparing measurement data, t tests were applied. Qualitative variables were analyzed via the χ2 test. In cases where the data distribution was not normal, nonparametric tests were employed. A P value less than 0.05 was considered statistically significant. Multivariate logistic regression was carried out for variables that were significant (P < 0.05) in the univariate analysis to identify independent risk factors. The area under the curve (AUC) was calculated for all noninvasive predictive models, along with other diagnostic metrics such as sensitivity, specificity, positive predictive value, negative predictive value, and the Youden index. The AUC was also utilized to assess the overall diagnostic performance of each model. The optimal cutoff values for predicting clinical decompensation in individuals with viral cirrhosis were determined on the basis of the highest combined sensitivity and specificity. Additionally, an external validation cohort comprising 60 subjects who fulfilled the inclusion criteria was used to evaluate the model’s calibration, discriminatory power, and clinical applicability. Decision curve analysis (DCA) was further employed to evaluate the potential clinical impact of applying these noninvasive predictive models in medical decision-making.
RESULTS
Patient features
Based on the predefined inclusion and exclusion criteria, 140 eligible participants were ultimately selected for the modeling group, including 72 patients with compensated cirrhosis and 68 patients with decompensated cirrhosis. The external validation group consisted of 60 eligible subjects, including 28 patients with compensated cirrhosis and 32 with decompensated cirrhosis. No statistically significant differences were observed in age or sex between the modeling groups (P > 0.05), indicating that the two groups were comparable. Similarly, no significant differences were found in age, sex, or underlying causes within the external validation group (P > 0.05). The population characteristics and baseline clinical attributes of the participants included in the study are presented in Tables 1 and 2, respectively.
Table 1 Comparison of the general characteristics of the modeling group.
Parameter
Patients with compensatory stage of cirrhosis, n = 72
Patients with decompensation of cirrhosis, n = 68
T value/χ2value
P value
Age, years
52.83 ± 12.2
53.16 ± 10.9
-0.168
0.867
Male (%)
40 (55.6)
40 (58.8)
0.153
0.696
Etiology, HBV/HCV
52/20
62/6
8.308
0.004
Table 2 Comparison of general characteristics in the external validation group.
Parameter
Patients with compensatory stage of cirrhosis, n = 28
Patients with decompensation of cirrhosis, n = 32
T value/χ2value
P value
Age, years
52.54 ± 13.7
54.97 ± 10.4
-0.780
0.438
Male (%)
15 (53.6)
14 (43.8)
0.577
0.448
Etiology, HBV/HCV
25/3
30/2
0.024
0.876
Clinical deterioration predictors among individuals suffering from viral cirrhosis
Univariate analysis was conducted via t tests and nonparametric rank sum tests to evaluate differences in indicators between the compensated and decompensated groups. Significant differences were observed between the two groups in terms of SSM, PLT, LSM, ALT, gamma-glutamyl transferase, SLD, albumin, total cholesterol, prothrombin time, international prothrombin ratio, prothrombin activity, portal vein diameter, actual liver volume measured by CT, and actual spleen volume measured by CT (P < 0.05). The parameters that showed statistically significant differences were further analyzed via reverse Wald regression. The results indicated that SSM, LSM, and SLD were independently associated with clinical decompensation in patients with viral cirrhosis (P < 0.05). These three variables were subsequently incorporated into the predictive model. The outcomes of both the univariate and multivariate analyses for the two groups are summarized in Tables 3 and 4, respectively.
Table 3 Univariate analysis of parameters in patients with compensated and decompensated cirrhosis.
Parameter
Patients with compensatory stage of cirrhosis, n = 72
Patients with decompensation of cirrhosis, n = 68
t/Z
P value
SSM, kPa
23.50 ± 6.1
18.32 ± 4.2
5.866
< 0.001
PLT, × 109/L
110.51 ± 69.1
61.81 ± 25.7
5.463
< 0.001
LSM, kPa
15.53 ± 5.7
23.86 ± 5.2
-9.023
< 0.001
ALT, IU/L
36.00 (24.00, 47.75)
27.50 (17.00, 42.25)
-2.490
0.013
AST, IU/L
40.50 (26.75, 58.75)
38.00 (29.00, 56.75)
-0.452
0.651
ALP, IU/L
105.00 (77.50, 143.00)
98.50 (79.25, 135.50)
-0.709
0.478
GGT, IU/L
61.00 (28.00, 120.00)
34.50 (19.25, 59.50)
-3.718
< 0.001
SLD, mm
12.99 ± 3.0
15.68 ± 3.1
-5.282
< 0.001
TBIL, μmol/L
29.64 ± 27.0
31.27 ± 16.2
-0.430
0.668
ALB, g/dL
38.69 ± 6.4
34.43 ± 7.4
3.691
< 0.001
TCHO, mmol/L
3.67 ± 1.3
3.09 ± 0.9
2.939
0.004
PT, seconds
12.08 ± 1.9
13.70 ± 2.5
-4.322
< 0.001
INR
1.10 ± 0.2
1.24 ± 0.2
-4.029
< 0.001
PTA,%
84.91 ± 18.9
72.37 ± 17.0
4.123
< 0.001
PVD, mm
12.03 ± 1.6
13.78 ± 2.4
-5.072
< 0.001
CTLV, cm3
1055.93 ± 344.8
898.64 ± 257.3
3.045
0.003
CTSV, cm3
519.10 ± 296.0
842.19 ± 416.3
-5.315
< 0.001
Table 4 Multivariate analysis of parameters in patients with compensated and decompensated cirrhosis.
Parameter
Patients with compensatory stage of cirrhosis, n = 72
Patients with decompensation of cirrhosis, n = 68
t/Z
P value
SSM, kPa
23.50 ± 6.1
18.32 ± 4.2
5.866
< 0.001
PLT, × 109/L
110.51 ± 69.1
61.81 ± 25.7
5.463
0.575
LSM, kPa
15.53 ± 5.7
23.86 ± 5.2
-9.023
< 0.001
ALT, IU/L
36.00 (24.00, 47.75)
27.50 (17.00, 42.25)
-2.490
0.597
GGT, IU/L
61.00 (28.00, 120.00)
34.50 (19.25, 59.50)
-3.718
0.566
SLD, mm
12.99 ± 3.0
15.68 ± 3.1
-5.282
0.037
ALB, g/dL
38.69 ± 6.4
34.43 ± 7.4
3.631
0.207
TCHO, mmol/L
3.67 ± 1.3
3.09 ± 0.9
2.939
0.224
PT, seconds
12.08 ± 1.9
13.70 ± 2.5
-4.322
0.480
INR
1.10 ± 0.2
1.24 ± 0.2
-4.029
0.584
PTA,%
84.91 ± 18.9
72.37 ± 17.0
4.123
0.623
PVD, mm
12.03 ± 1.6
13.78 ± 2.4
-5.072
0.773
CTLV, cm3
1055.93 ± 344.8
898.64 ± 257.3
3.045
0.967
CTSV, cm3
519.10 ± 296.0
842.19 ± 416.3
-5.315
0.895
Establishment of the predictive model
After excluding parameters that did not significantly differ between the two groups, logistic regression analysis revealed that SSM, LSM, and SLD were significant independent predictors of clinical decompensation (P < 0.05). On the basis of these three factors, a noninvasive predictive model was developed: Ln [P/(1-P)] = -4.969 - 0.279 × SSM + 0.348 × LSM + 0.272 × SLD. Clinical decompensation was negatively associated with SSM and positively associated with both LSM and SLD. The outcomes of the logistic regression analysis are summarized in Table 5.
Table 5 Parameters used to establish the noninvasive prediction model.
Parameter
B
SE
Wald
Sig
Exp (B)
95%CI of exp (B)
SSM
-0.279
0.066
17.768
< 0.001
0.756
0.664-0.861
LSM
0.348
0.067
26.728
< 0.001
1.416
1.241-1.616
SLD
0.272
0.110
6.053
0.014
1.312
1.057-1.630
Constant
-4.969
2.245
4.899
0.027
0.007
Performance of the predictive model
The diagnostic efficacy of the noninvasive predictive model is presented in Table 6. The newly developed model was compared with previously established consensus models, including the LSPS, VRI, AAR, and Baveno VI. The AUC, sensitivity, specificity, and Youden index of all the models were calculated. A model is considered to have favorable discriminative ability when its AUC exceeds 0.7. The AUC of the model developed in this research was 0.944, whereas the AUCs for the LSPS, VRI, AAR, and Baveno VI models were 0.834, 0.824, 0.670, and 0.680, respectively, as illustrated in Figure 1. These findings demonstrate that the newly constructed model outperforms the existing models in terms of predictive accuracy. Within this model, the optimal threshold value was determined to be 0.56. If the P value derived using the proposed formula surpasses this threshold, it suggests that clinical decompensation has occurred in individuals with viral cirrhosis. The accuracy and positive predictive value indicate that the model can effectively identify patients at risk of clinical decompensation, with higher values reflecting greater diagnostic reliability. As detailed in Table 7, the precision of the newly developed model is 85%, and its positive predictive value reaches 87.89%, both of which surpass the corresponding values of the LSPS, VRI, AAR, and Baveno VI models. The findings demonstrate that the newly developed model exhibits excellent diagnostic accuracy.
Figure 1 Area under the curve of various models for predicting clinical decompensation in patients with viral cirrhosis.
A: Modeling group; B: External validation group. The area under the curve of the new model in predicting clinical decompensation in patients with viral cirrhosis was 0.944 in the modeling group, which was greater than that of the liver stiffness-spleen diameter-to-platelet ratio score, variceal risk index, aspartate transaminase-to-platelet ratio index, and aspartate transaminase/alanine aminotransferase ratio, which was 1 in the external validation group.
Table 6 Comparison of various parameters of each model.
The calibration performance of the newly developed model was assessed by applying the Hosmer-Lemeshow test to the χ2 values from both the computational modeling cohort and the external validation cohort. The results show that the modeling group χ2 is -14.17, and the external verification group χ2 is 0.03. The respective P values were 0.999 and 0.98, suggesting that there was no significant statistical difference between the two groups, since both values were above the 0.05 significance level. As illustrated in Figure 2, the calibration scatter plots for both groups varied around the reference line, showing no notable divergence. This conclusion is further substantiated by the P values of both groups, which are greater than 0.05, indicating that the observed differences between the groups were did not reach statistical significance. These findings indicate that the new model is capable of precisely predicting clinical decompensation in individuals suffering from viral cirrhosis. Moreover, there is a strong alignment between the model’s predictions and the actual instances of decompensation observed in these patients.
Figure 2 Calibration scatter plot of patient data.
A: Modeling group; B: External validation group. In predicting patients in the modeling group and external validation group, the scattered points fluctuated around the reference line without significant deviations.
Clinical practicability of the noninvasive model
In this research, DCA was employed to assess the clinical effectiveness of the newly developed model. Within the DCA curve, the black line represents an extreme scenario where the new model predicts the absence of clinical decompensation in all individuals diagnosed with decompensated cirrhosis, resulting in no significant clinical advantage. Moreover, the gray line illustrates the opposite extreme, where the model predicts clinical decompensation in every patient within the same cohort. In the modeling group (Figure 3A), the red line depicts the DCA of the new model across a risk threshold range of 5% to 100%. In the external validation group (Figure 3B), the DCA curve of the new model outperformed both the black and gray reference lines, suggesting that the application of this model may yield clinical advantages for the population under study. The DCA curve spans a higher risk threshold range of 30% to 100%, and the new model’s curve similarly outperforms the two extreme lines. This suggests that patients in the external validation group could also derive clinical benefits, highlighting the model’s strong practical value in real-world clinical settings. Overall, the findings demonstrate that the application of the new model to both cohorts results in clinical benefits for patients in each group.
Figure 3 Adjusted decision curve analysis of patient data.
A: Modeling group; B: External validation group. The black line indicates that in extreme cases, the new model predicted that there was no clinical decompensation in patients with viral cirrhosis, and the clinical net benefit was 0. The gray curve indicates that in extreme cases, the new model predicts that there is clinical decompensation in all patients with viral cirrhosis, and the clinical net benefit is the negative slope. The red line indicates that the new model has a net clinical benefit. The red line is greater than the black and gray lines are, indicating that patients in the modeling group can benefit from the new model.
DISCUSSION
PH is a prevalent and severe complication observed in individuals with cirrhosis[25,26]. CSPH, i.e., an HVPG ≥ 10 mmHg, is considered a key marker of cirrhosis progression to decompensation. CSPH status is closely related to clinical decompensation status in patients with liver cirrhosis, which may lead to EGVB, ascites, HE and other complications and has a serious impact on the prognosis and survival rate of patients with liver cirrhosis[19,27]. Consequently, CSPH has emerged as a critical element in predicting the progression of decompensation among individuals with cirrhosis. At present, global guidelines advocate HVPG measurements as the primary method for diagnosing CSPH[28]. Research conducted by Ripoll et al[29] demonstrated that an HVPG exceeding 10 mmHg serves as the most significant indicator for predicting clinical decompensation in individuals with compensated cirrhosis. Moreover, the likelihood of clinical decompensation increases as the HVPG value increases. However, owing to the invasive nature of HVPG measurement, its high cost, and the need to perform it in highly specialized centers, its practical clinical application is somewhat limited. Therefore, it is important to use noninvasive tools to assess PH status and severity in daily clinical practice. To date, TE has demonstrated its effectiveness as a noninvasive technique for accurately detecting the existence of CSPH and assessing the likelihood of decompensation[9,30]. Therefore, this study sought to establish a noninvasive model with predictive strength no weaker than or even better than the Baveno VI consensus for identifying clinical decompensation status among cirrhosis patients, including factors such as ascites, esophageal variceal bleeding, and HE.
This study was carried out on a cohort of 140 patients diagnosed with viral cirrhosis, with no significant differences observed across racial groups. We found that the SSM, LSM, and SLD independently predict the risk of decompensation in individuals with virus-related compensated cirrhosis. The findings of this study are consistent with the conclusions drawn from numerous previous studies, suggesting a significant association between LSM and PH. Singh et al[4] conducted a comprehensive analysis of 17 prospective cohort studies involving 7058 individuals diagnosed with chronic liver disease and demonstrated a significant association between LSM and the likelihood of decompensation in patients with cirrhosis. Corpechot et al[31] demonstrated that increased LSM is associated with liver decompensation in patients with primary biliary cirrhosis[31,32]. In contrast, the LSM primarily reflects compensatory mechanisms associated with intrahepatic resistance-related PH. Consequently, relying solely on the LSM index is insufficient to accurately predict the development of PH in individuals suffering from viral cirrhosis. When the HVPG is ≥ 12 mmHg, the correlation between the LSM and HVPG tends to weaken. This may be explained by the increasing influence of extraliver factors on portal venous pressure regulation at HVPG levels ≥ 12 mmHg. In contrast, the LSM reflects PH-related compensatory factors associated with intrahepatic resistance[4,33,34]. Therefore, relying solely on the LSM index is insufficient for accurately predicting the progression of PH in patients with viral cirrhosis. Currently, researchers such as Giannini et al[35,36] have explored the integration of the SLD with other noninvasive markers to assess the severity of PH. Hepatosplenomegaly is frequently observed in individuals suffering from hepatic cirrhosis, and its underlying mechanisms may include elevated portal pressure, increased resistance to splenic venous outflow, angiogenesis, and exacerbated fibrosis. However, the correlation between spleen size and the severity of PH remains controversial[14,28,32]. Compared with other parameters used to reflect PH, the SSM has shown superior performance in the diagnosis of PH, and it has a wide range of applications; in particular, it can accurately reflect the hemodynamic changes in the visceral circulation in patients with advanced cirrhosis[37,38]. In this study, we observed a significant association between SSM or HVPG and the presence of PH. Furthermore, our findings indicate that the SSM serves as an independent indicator for predicting clinical decompensation in individuals with cirrhosis. As a result, we suggest incorporating SSMs in the evaluation of patients with viral cirrhosis. However, Stefanescu et al[39] noted that the predictive power of the SSM may be weakened when HVPG values exceed 19 mmHg. Therefore, to ensure the accuracy and reliability of clinical decompensation risk assessment of viral hepatitis, the above three indicators should be comprehensively considered and evaluated.
Both the new model established in this study and the Baveno VI consensus have negative predictive values greater than 85%, which means that patients who have not undergone decompensation events may be excluded. However, the new model demonstrated a significantly greater positive predictive value than did the Baveno VI consensus, indicating its superior ability to predict clinical decompensation in individuals with compensated cirrhosis. The AUC value of the new model is 0.944, and that of the Baveno VI consensus is 0.680, indicating that the new model has better discrimination ability. In particular, when the cutoff value for predicting the occurrence of decompensated cirrhosis was 0.56, the new model demonstrated higher specificity, positive predictive value, and overall accuracy than the other four models did. Therefore, this new model shows potential for effectively identifying patients at risk of decompensated cirrhosis. In addition, we used 60 patients as an external validation group, and the results demonstrate that the new model exhibits strong discriminatory capability. Furthermore, when DCA is employed to demonstrate the clinical advantages of the proposed model, both the development cohort and the external validation cohort show benefits from the new approach. This study revealed that the occurrence of clinical decompensation among individuals suffering from viral cirrhosis was negatively associated with the SSM and positively associated with the LSM, a finding similar to that of a study in large chronic hepatitis C patients, in which higher LSMs and lower SSMs were associated with a greater risk of cirrhosis decompensation[40]. This association is not surprising, as the associations of the LSM and SSM with adverse clinical outcomes are mediated primarily by their associations with PH development and progression. More interestingly, however, we found that the possibility of noninvasive HVPG could be assessed by the SSM and LSM. Consistent with our findings, one study reported that it seems possible to accurately estimate the HVPG by using a simple linear model that includes these two variables. This study revealed that the SSM and LSM can be evaluated in relation to each other during the management of individuals suffering from cirrhosis, and this joint evaluation approach is particularly suitable in the context of rapid changes in inflammatory activity in extensive liver tissue necrosis[32]. This finding deserves further verification and in-depth study.
The main advantage of this study is that all of our enrolled patients had viral cirrhosis, which minimizes the bias of etiology in the study results and provides strong reliability of our data. Second, the noninvasive predictive model we established has good predictive performance and clinical practicability, and the performance of the model has been verified in an external environment. From a technical point of view, the model is sufficiently safe and highly reproducible through TE and CT measurements. Our findings indicate that the newly developed model is both safe and effective, demonstrating promise in predicting the risk of decompensation and the severity of PH in individuals with viral cirrhosis. Nevertheless, certain limitations should be acknowledged. First, as a single-center retrospective study, our results might have been influenced by biases related to patient admission and selection. Second, the limited sample size could have impacted the precision of the outcomes. Additionally, the presence of substantial ascites in patients with decompensated cirrhosis may have interfered with the accuracy of the TE test results. Finally, this study exclusively recruited patients with hepatitis B virus/hepatitis C virus-related cirrhosis from Asia. Cirrhosis caused by different etiologies involves heterogeneous disease-driven mechanisms (e.g., the reversibility observed after cessation of alcohol consumption in alcoholic cirrhosis patients and the metabolic-inflammatory features of non-alcoholic steatohepatitis-related cirrhosis). Despite these differences, both the LSM and SSM continue to serve as objective indicators that effectively capture the core pathophysiological changes associated with end-stage liver disease. Their associations with liver decompensation, PH-related bleeding, and mortality risk have been validated through multietiological cohort studies[41-43]. Nevertheless, other etiology-specific variables, such as the insulin resistance index and duration of alcohol abstinence, were not incorporated into the model construction process. Given the racial variations in the etiological spectrum of liver cirrhosis and the impact of genetic and environmental factors on the progression of chronic liver diseases, the predictive performance of this model in non-Asian populations and patients with cirrhosis of other etiologies requires further clarification through multicenter, multiethnic external validation.
Our observations also prove that the LSM should be comprehensively considered with other parameters that can reflect dynamic changes in peripheral blood flow, namely, the SSM and SLD, to predict the presence of PH and CSPH. The HVPG continues to serve as the gold standard and is utilized for assessing the severity of PH; however, because the HVPG has limited application in the clinic, the introduction of a noninvasive predictive model represents a significant advance in clinical practice. Thus, accurate prediction of CSPH and HVPG from a simple formula may provide a way to quickly identify the various phases of disease development in individuals with cirrhosis. This approach will help cirrhosis patients identify their own risk level and guide patients on whether further HVPG and upper gastrointestinal endoscopy are needed. In conclusion, the noninvasive predictive model established in this research provides further insight into the potential role of TE in predicting clinical decompensation among individuals with viral cirrhosis. In particular, the noninvasive prediction models developed with the LSM, SSM and SLD are safe and effective and can help clinicians detect changes in PH and the possibility of decompensation, which has good clinical practical value. Our findings require additional confirmation through larger-scale prospective research.
CONCLUSION
The newly developed model effectively predicts clinical decompensation events in individuals with viral cirrhosis, thereby reducing the need for unnecessary HVPG tests.
ACKNOWLEDGEMENTS
We thank all the participants in this study.
Footnotes
Provenance and peer review: Unsolicited article; Externally peer reviewed.
Peer-review model: Single blind
Specialty type: Gastroenterology and hepatology
Country of origin: China
Peer-review report’s classification
Scientific Quality: Grade A, Grade B, Grade B
Novelty: Grade B, Grade B, Grade B
Creativity or Innovation: Grade B, Grade B, Grade B
Scientific Significance: Grade B, Grade B, Grade B
P-Reviewer: Jin S, MD, PhD, Chief Physician, Professor, China; Soldera J, MD, PhD, Associate Professor, Brazil S-Editor: Wu S L-Editor: A P-Editor: Zheng XM
Zhang X, Song J, Zhang Y, Wen B, Dai L, Xi R, Wu Q, Li Y, Luo X, Lan X, He Q, Luo W, Lai Q, Ji Y, Zhou L, Qi T, Liu M, Zhou F, Wen W, Li H, Liu Z, Chen Y, Zhu Y, Li J, Huang J, Cheng X, Tu M, Hou J, Wang H, Chen J. Baveno VII algorithm outperformed other models in ruling out high-risk varices in individuals with HBV-related cirrhosis.J Hepatol. 2023;78:574-583.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 19][Cited by in RCA: 23][Article Influence: 11.5][Reference Citation Analysis (0)]
Boursier J, Zarski JP, de Ledinghen V, Rousselet MC, Sturm N, Lebail B, Fouchard-Hubert I, Gallois Y, Oberti F, Bertrais S, Calès P; Multicentric Group from ANRS/HC/EP23 FIBROSTAR Studies. Determination of reliability criteria for liver stiffness evaluation by transient elastography.Hepatology. 2013;57:1182-1191.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 392][Cited by in RCA: 497][Article Influence: 41.4][Reference Citation Analysis (0)]
Yoo JJ, Maeng SA, Chang Y, Lee SH, Jeong SW, Jang JY, Cheon GJ, Kim YS, Kim HS, Kim SG. Enhancing liver cirrhosis varices and CSPH risk prediction with spleen stiffness measurement using 100-Hz probe.Sci Rep. 2024;14:13674.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 4][Reference Citation Analysis (0)]
Colecchia A, Montrone L, Scaioli E, Bacchi-Reggiani ML, Colli A, Casazza G, Schiumerini R, Turco L, Di Biase AR, Mazzella G, Marzi L, Arena U, Pinzani M, Festi D. Measurement of spleen stiffness to evaluate portal hypertension and the presence of esophageal varices in patients with HCV-related cirrhosis.Gastroenterology. 2012;143:646-654.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 359][Cited by in RCA: 375][Article Influence: 28.8][Reference Citation Analysis (0)]
Pons M, Augustin S, Scheiner B, Guillaume M, Rosselli M, Rodrigues SG, Stefanescu H, Ma MM, Mandorfer M, Mergeay-Fabre M, Procopet B, Schwabl P, Ferlitsch A, Semmler G, Berzigotti A, Tsochatzis E, Bureau C, Reiberger T, Bosch J, Abraldes JG, Genescà J. Noninvasive Diagnosis of Portal Hypertension in Patients With Compensated Advanced Chronic Liver Disease.Am J Gastroenterol. 2021;116:723-732.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 163][Cited by in RCA: 164][Article Influence: 41.0][Reference Citation Analysis (0)]
Giannini E, Botta F, Borro P, Risso D, Romagnoli P, Fasoli A, Mele MR, Testa E, Mansi C, Savarino V, Testa R. Platelet count/spleen diameter ratio: proposal and validation of a non-invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis.Gut. 2003;52:1200-1205.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 306][Cited by in RCA: 299][Article Influence: 13.6][Reference Citation Analysis (0)]
Giannini EG, Zaman A, Kreil A, Floreani A, Dulbecco P, Testa E, Sohaey R, Verhey P, Peck-Radosavljevic M, Mansi C, Savarino V, Testa R. Platelet count/spleen diameter ratio for the noninvasive diagnosis of esophageal varices: results of a multicenter, prospective, validation study.Am J Gastroenterol. 2006;101:2511-2519.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 164][Cited by in RCA: 172][Article Influence: 9.1][Reference Citation Analysis (0)]
Dajti E, Ravaioli F, Zykus R, Rautou PE, Elkrief L, Grgurevic I, Stefanescu H, Hirooka M, Fraquelli M, Rosselli M, Chang PEJ, Piscaglia F, Reiberger T, Llop E, Mueller S, Marasco G, Berzigotti A, Colli A, Festi D, Colecchia A; Spleen Stiffness—IPD-MA Study Group. Accuracy of spleen stiffness measurement for the diagnosis of clinically significant portal hypertension in patients with compensated advanced chronic liver disease: a systematic review and individual patient data meta-analysis.Lancet Gastroenterol Hepatol. 2023;8:816-828.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 42][Cited by in RCA: 48][Article Influence: 24.0][Reference Citation Analysis (0)]
Dajti E, Ravaioli F, Marasco G, Alemanni LV, Colecchia L, Ferrarese A, Cusumano C, Gemini S, Vestito A, Renzulli M, Golfieri R, Festi D, Colecchia A. A Combined Baveno VII and Spleen Stiffness Algorithm to Improve the Noninvasive Diagnosis of Clinically Significant Portal Hypertension in Patients With Compensated Advanced Chronic Liver Disease.Am J Gastroenterol. 2022;117:1825-1833.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 67][Cited by in RCA: 62][Article Influence: 20.7][Reference Citation Analysis (0)]
Grgurević I, Bokun T, Mustapić S, Trkulja V, Heinzl R, Banić M, Puljiz Ž, Lukšić B, Kujundžić M. Real-time two-dimensional shear wave ultrasound elastography of the liver is a reliable predictor of clinical outcomes and the presence of esophageal varices in patients with compensated liver cirrhosis.Croat Med J. 2015;56:470-481.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 47][Cited by in RCA: 45][Article Influence: 4.5][Reference Citation Analysis (0)]
Janik MK, Kruk B, Szczepankiewicz B, Kostrzewa K, Raszeja-Wyszomirska J, Górnicka B, Lammert F, Milkiewicz P, Krawczyk M. Measurement of liver and spleen stiffness as complementary methods for assessment of liver fibrosis in autoimmune hepatitis.Liver Int. 2021;41:348-356.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 8][Cited by in RCA: 22][Article Influence: 5.5][Reference Citation Analysis (0)]
Thorhauge KH, Semmler G, Johansen S, Lindvig KP, Kjærgaard M, Hansen JK, Torp N, Hansen CD, Andersen P, Hofer BS, Gu W, Israelsen M, Mandorfer M, Reiberger T, Trebicka J, Thiele M, Krag A; Microb-Predict, Galaxy and MicrobLiver consortia. Using liver stiffness to predict and monitor the risk of decompensation and mortality in patients with alcohol-related liver disease.J Hepatol. 2024;81:23-32.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 3][Cited by in RCA: 18][Article Influence: 18.0][Reference Citation Analysis (0)]
