A method for predicting the risk of cirrhotic esophagogastric varices with spontaneous portosystemic shunt

Abstract

The present application relates to the technical field of clinical decision-making of digestive diseases, and specifically discloses a risk prediction method for liver cirrhosis esophagogastric varices with spontaneous portosystemic shunt. The method comprises the following steps: obtaining clinical index data of a liver cirrhosis esophagogastric varices patient; taking the clinical index data as input data, and outputting the risk probability of spontaneous portosystemic shunt of the liver cirrhosis esophagogastric varices patient through multifactor logistic regression equation analysis; wherein the clinical index data comprises the widest inner diameter of the splenic vein, prothrombin time, ascites, and a history of esophagogastric varices rupture and hemorrhage. The risk prediction method provided by the present application has significantly better diagnostic efficiency than traditional single indicators, and has important clinical value for early screening, endoscopic scheme selection, complication prevention and control, and prognosis management of liver cirrhosis portal hypertension patients.

Description

Technical Field

[0001] This invention relates to the field of clinical decision-making technology for digestive diseases, specifically to a risk prediction method for esophageal and gastric varices with spontaneous portosystemic shunt in patients with cirrhosis. Background Technology

[0002] Cirrhosis is a common end-stage chronic liver disease in my country. Its core pathophysiological feature is portal hypertension, which can lead to serious complications such as esophageal and gastric varices, rupture and bleeding, ascites, hepatic encephalopathy, and spontaneous portosystemic shunt.

[0003] Spontaneous portosystemic shunt (SPSS) is an abnormal collateral pathway that forms between the portal venous system and the systemic circulation after portal venous pressure increases. It mainly includes three types: gastrorenal shunt, spleen-renal shunt, and umbilical vein opening. The incidence rate in patients with esophageal and gastric varices in cirrhosis is about 20% to 40%.

[0004] SPSS has a typical "double-edged sword" effect on disease progression: in patients with well-compensated liver function, SPSS can shunt portal vein blood flow, reduce variceal vein pressure, and reduce the risk of bleeding; in patients with decompensated liver function, SPSS can lead to insufficient liver perfusion, worsening of liver function, hepatic encephalopathy, and a significantly increased risk of rebleeding.

[0005] Current clinical methods for identifying and assessing SPSS primarily rely on CT portal vein imaging (CTPV). While this method can provide a definitive diagnosis, it has limitations such as high examination costs, radiation exposure, the need for specific contrast and post-processing procedures, and difficulty in using it as a routine screening method. These limitations hinder early, rapid, and repeatable risk identification in large populations with cirrhosis. Clinical practice often relies on physician experience and subjective judgment, lacking standardized and quantitative risk assessment strategies, resulting in insufficient accuracy in predicting the risk of SPSS. Summary of the Invention

[0006] To develop a standardized and quantifiable risk assessment strategy to improve the accuracy of predicting the risk of SPSS (Spontaneous Portosystemic Shunt), this invention provides a risk prediction method for esophageal and gastric varices with spontaneous portosystemic shunt in patients with cirrhosis. To achieve the above objective, this invention adopts the following technical solution.

[0007] This invention provides a risk prediction method for esophageal and gastric varices with spontaneous portosystemic shunt in patients with cirrhosis, comprising the following steps: S1. Clinical indicator data collection: The clinical indicator data of patients with esophageal and gastric varices due to cirrhosis are obtained as input data. Through multivariate logistic regression equation analysis, the risk probability of spontaneous portosystemic shunting in patients with esophageal and gastric varices due to cirrhosis is output.

[0008] The clinical indicators include the widest inner diameter of the splenic vein (measured by enhanced abdominal CT or ultrasound), prothrombin time (obtained by coagulation function test), ascites (determined by imaging or physical examination), and history of esophageal and gastric variceal rupture and bleeding (confirmed by medical history collection).

[0009] The acquired clinical indicator data were standardized and assigned values ​​as follows: ascites and history of esophageal and gastric variceal bleeding were binary variables, with 0 assigned to the absence of ascites and 1 assigned to the presence of ascites; 0 assigned to the absence of history of esophageal and gastric variceal bleeding and 1 assigned to the presence of history of esophageal and gastric variceal bleeding; the widest inner diameter of the splenic vein was a continuous measurement variable in seconds (s), and the original test values ​​were directly used as variable values ​​input into the model; prothrombin time was a continuous measurement variable in mm, and the original test values ​​were directly used as variable values ​​input into the model.

[0010] S3. Multivariate Logistic Regression Analysis and Risk Probability Calculation and Output: Standardized clinical indicator data are input into a multivariate logistic regression equation for analysis. The Logit transformed value of the risk of spontaneous portosystemic shunt is calculated, and the risk probability of spontaneous portosystemic shunt in this patient with esophageal and gastric varices due to cirrhosis is calculated and output. The multivariate logistic regression equation is constructed using SPSS 25.0 statistical software based on the clinical indicator data, as detailed below: Logit P )=β1×X1+β2×X2+β3×X3+β4×X4; P =1 / (1+e −Logit(P) ); In the above formula, P Logit( P X1 represents the Logit transformed value for the risk of spontaneous portosystemic shunt; X2 represents the variable assignment result for the widest internal diameter of the splenic vein, and β1 represents the regression coefficient corresponding to the widest internal diameter of the splenic vein; X3 represents the variable assignment result for ascites, and β3 represents the regression coefficient corresponding to ascites; X4 represents the variable assignment result for the history of ruptured esophageal and gastric varices, and β4 represents the regression coefficient corresponding to the history of ruptured esophageal and gastric varices.

[0011] This invention uses the widest internal diameter of the splenic vein, prothrombin time (PT), ascites, and esophageal variceal bleeding (EGVB) as core predictive factors for risk prediction. Through retrospective clinical studies, univariate and multivariate logistic regression analysis, ROC curves, calibration curves, and clinical decision curves, a risk prediction method with high diagnostic efficacy was constructed (the model involved has an AUC of 0.787). Specifically, the widest internal diameter of the splenic vein directly reflects the severity of portal hypertension and the tendency for collateral circulation to open, forming the anatomical basis for SPSS formation; PT reflects the degree of impairment of liver synthetic function, and decreased liver function can lead to increased portal vein resistance and promote the formation of abnormal shunt channels; ascites indicates a decompensated state of portal hypertension and shares a common hemodynamic driving mechanism with SPSS occurrence; a history of EGVB reflects the pressure balance state of varices, and SPSS formation can partially reduce pressure within varices and decrease the risk of rupture and bleeding. Therefore, these four factors cover the relevant aspects of SPSS occurrence from four levels: anatomical structure, liver function reserve, hemodynamic status, and clinical event outcome, possessing a scientifically reliable intrinsic correlation mechanism. The combined prediction model and risk prediction method constructed based on the aforementioned multi-dimensional factors can overcome the limitations of a single indicator reflecting only a local state, achieving integrated interpretation and quantitative judgment of the risk of SPSS occurrence, significantly improving prediction accuracy and stability. This model can complete risk calculation using routine non-invasive examination indicators, without relying on invasive, high-cost, and high-radiation examinations such as CT portal vein imaging. It can be widely used in primary healthcare institutions and large-scale population screening, effectively addressing the pain points of existing technologies that rely on special imaging examinations and are difficult to conduct early routine screening. Simultaneously, this risk prediction method can output continuous risk probabilities and achieve high- and low-risk stratification. Combined with Child-Pugh liver function classification, it further distinguishes the clinical significance of SPSS in different liver function states, clarifying that it is a protective factor in grade A and a risk factor in grades B / C. This guides the selection of endoscopic treatment plans, early warning of rebleeding risk, and short-term prognosis, providing clinically applicable individualized decision-making basis.

[0012] Preferably, the regression coefficient β1 for the widest inner diameter of the splenic vein is positive, the regression coefficient β2 for the prothrombin time is positive, the regression coefficient β3 for the ascites is positive, and the regression coefficient β4 for the history of esophageal and gastric variceal rupture and bleeding is positive.

[0013] Preferably, the output is the probability of spontaneous portosystemic shunt occurrence in this patient with cirrhotic esophageal and gastric varices. P Afterwards, the optimal threshold obtained by ROC curve verification was used as the risk stratification criterion. The risk probability of spontaneous portosystemic shunting in the patients with esophageal and gastric varices due to cirrhosis was compared with the optimal threshold to determine the risk stratification.

[0014] If the risk probability of spontaneous portosystemic shunt in a patient with esophageal and gastric varices due to cirrhosis is greater than or equal to the optimal threshold, the patient is considered to be in a high-risk group for spontaneous portosystemic shunt.

[0015] If the probability of spontaneous portosystemic shunt in a patient with esophageal and gastric varices due to cirrhosis is less than the optimal threshold, the patient is considered to be in a low-risk group for spontaneous portosystemic shunt.

[0016] Preferably, the optimal critical value is determined through ROC curve analysis and is the predicted probability value (cutoff value) corresponding to the maximum Youden's Index. The Youden's Index = Sensitivity + Specificity - 1, reflecting the model's comprehensive ability to distinguish between high-risk and low-risk groups.

[0017] Preferably, the optimal critical value (cutoff value) is 0.309. This cutoff value is the optimal estimated parameter for the current sample cohort. Affected by sampling error and population heterogeneity, this threshold may fluctuate slightly in different study populations and is not an absolutely fixed critical standard.

[0018] Preferably, the risk prediction model is evaluated for discrimination, calibration and clinical applicability, wherein the discrimination is evaluated by the area under the ROC curve, the calibration is evaluated by the Hosmer-Lemeshow test or calibration curve, and the clinical applicability is evaluated by clinical decision curve analysis.

[0019] The critical values ​​for the four independent predictors are set as follows: (1) The widest inner diameter of the splenic vein is ≥14.5mm.

[0020] (2) Prothrombin time ≥19.95s.

[0021] (3) Ascites is present.

[0022] (4) No history of esophageal or gastric variceal bleeding.

[0023] Preferably, the risk prediction method is deployed in the form of a computer program on a computer-readable storage medium, and the computer program, when executed by a processor, performs the following steps: Obtain detection data for the four independent predictors in patients with esophageal and gastric varices due to cirrhosis.

[0024] The detection data were substituted into the multivariate logistic regression equation to calculate the risk probability of spontaneous portosystemic shunt in the patient with esophageal and gastric varices due to liver cirrhosis.

[0025] Output the risk probability of spontaneous portosystemic shunt and the corresponding risk stratification results for this patient with esophageal and gastric varices due to liver cirrhosis.

[0026] Preferably, the risk prediction method further incorporates Child-Pugh liver function grading to achieve stratified prediction: Child-Pugh Grade A: SPSS indicates a protective factor, reducing the risk of EGVB.

[0027] Child-Pugh Category B / C: SPSS is a risk factor, increasing the risk of rebleeding after endoscopic treatment.

[0028] Preferably, the clinical indicator data are obtained through medical imaging examinations, laboratory tests, and medical record reviews.

[0029] Preferably, the standardized assignment rules are pre-stored in the database and are automatically invoked by the system during assignment processing.

[0030] Preferably, β1, β2, β3 and β4 are determined by fitting a multivariate logistic regression based on training samples.

[0031] Preferably, the multifactor logistic regression equation is predetermined using the following model construction method, which includes the following steps: S1. Data Collection: Collect the following clinical indicators from patients with esophageal and gastric varices due to cirrhosis: white blood cell count, red blood cell count, gamma-glutamyl transferase level, albumin level, prothrombin time (PT), prothrombin activity (PT%), thrombin time (TT), D-dimer level, blood ammonia level, widest diameter of portal vein, widest diameter of splenic vein, widest diameter of superior mesenteric vein, esophageal and gastric variceal bleeding (EGVB), ascites, hepatic encephalopathy (HE), portal hypertensive gastropathy (PHG), and portal vein thrombosis (PVT).

[0032] S2. Candidate Variable Screening: A computer system is used to perform univariate logistic regression analysis on the clinical indicator data to... P A value <0.05 is the statistically significant screening criterion, and candidate variables that are statistically associated with spontaneous gate shunting are selected.

[0033] S3. Determination of Independent Predictors: The candidate variables were included in a multivariate logistic regression analysis using a computer system. Stepwise regression (forward selection) was adopted, with the inclusion criterion (Entry) = 0.05 and the exclusion criterion (Removal) = 0.10 as the variable selection criteria. The widest internal diameter of the splenic vein, prothrombin time, ascites, and history of esophageal and gastric variceal rupture and bleeding were identified as four independent predictors. The regression coefficients and constant terms of each independent predictor were obtained, and a multivariate logistic regression equation was established.

[0034] S4. Model Validation: Calculate the predicted probability based on the multivariate logistic regression equation using a computer system, plot the ROC curve to evaluate the model's discriminative ability, determine the optimal critical value (cutoff value), plot the calibration curve to evaluate the model's calibration degree, and plot the clinical decision curve to evaluate the model's clinical net benefit.

[0035] S5. Model Construction: The model is constructed from the multivariate logistic regression equation, the optimal critical value, and the validation results. It is used to predict the risk of spontaneous portosystemic shunt in patients with esophageal and gastric varices due to cirrhosis.

[0036] More preferably, the method for constructing the model includes the following steps: Data Collection: Clinical data of patients with esophageal and gastric varices due to cirrhosis were collected. The clinical data included the widest diameter of the splenic vein measured by enhanced abdominal CT or ultrasound, prothrombin time obtained by coagulation function tests, ascites determined by imaging or physical examination, history of esophageal and gastric variceal rupture and bleeding confirmed retrospectively by medical records, and the status of spontaneous portosystemic shunt (SPSS) confirmed by imaging as outcome variables.

[0037] Univariate analysis: Univariate logistic regression analysis was performed on the clinical indicator data to screen candidate predictors associated with the occurrence of SPSS.

[0038] Multivariate analysis: The candidate predictors were included in multivariate logistic regression analysis to determine the widest internal diameter of the splenic vein, prothrombin time, ascites, and history of esophageal and gastric variceal rupture and bleeding as independent predictors, and the regression coefficients and model intercepts of each predictor were obtained.

[0039] Model construction and validation: A risk prediction method was constructed based on the independent predictors and their regression coefficients. The model was internally validated using ROC curves, calibration curves, and clinical decision curves, and the area under the ROC curve (AUC) was found to be 0.787.

[0040] Critical value determination: The optimal critical value of the model was determined to be 0.309 based on the ROC curve, corresponding to a sensitivity of 76% and a specificity of 67%, and risk stratification criteria were set accordingly.

[0041] Preferably, the collection of the above-mentioned clinical indicator data of patients with esophageal and gastric varices due to cirrhosis also includes Child-Pugh liver function classification data for subsequent stratified predictive analysis.

[0042] The univariate analysis also includes normality tests and collinearity diagnoses for continuous variables.

[0043] The multivariate logistic regression analysis employed either stepwise regression or forced entry methods, with inclusion criteria as follows: P <0.05, exclusion criteria are P ≥0.10.

[0044] The present invention also provides a spontaneous gate diversion risk prediction system based on the aforementioned risk prediction method, the risk prediction system comprising the following functional modules: The indicator input and standardization module is used to input the test data of four independent predictors of esophageal and gastric varices in patients with cirrhosis and to standardize the test data.

[0045] Risk probability calculation module: used to substitute the standardized detection data into the multivariate logistic regression equation to calculate the risk probability of spontaneous portosystemic shunt in patients with esophageal and gastric varices due to liver cirrhosis.

[0046] Risk stratification module: used to compare the risk probability of spontaneous portosystemic shunt in the patients with cirrhosis and esophageal varices with the optimal threshold value to perform risk stratification; and to determine whether the patients with cirrhosis and esophageal varices are high-risk or low-risk for spontaneous portosystemic shunt.

[0047] Prognostic assessment and follow-up suggestion module: This module is used to output prognostic assessment information and follow-up suggestions based on the risk stratification results.

[0048] The present invention also provides an electronic device comprising one or more processors and a memory for storing executable instructions, wherein the one or more processors are configured to invoke the executable instructions stored in the memory to implement the function of the spontaneous gate diversion risk prediction system.

[0049] The present invention also provides a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, implement the function of the spontaneous gate diversion risk prediction system.

[0050] Compared with the prior art, the beneficial effects of the present invention are as follows: 11 1. This invention aims to provide a stable, reliable, and quantitatively operable risk prediction method to improve the accuracy of early risk identification for SPSS. Based on non-invasive and readily available routine clinical indicators, this invention constructs a stable, reliable, and quantitatively operable risk prediction method to improve the accuracy of early risk identification for SPSS and achieve individualized prognostic stratification and clinical decision support. By integrating non-invasive and easily accessible routine clinical and imaging indicators, this invention can achieve accurate, quantitative, and early prediction of SPSS risk and risk stratification, which is superior to the limitations of existing technologies that rely on single-indicator prediction and invasive imaging. It achieves early prediction of SPSS risk and individualized diagnosis and treatment support. The risk prediction method provided by this invention can serve as a standardized and quantitative risk assessment strategy, with advantages of stability, reliability, efficiency, and ease of implementation.

[0051] 2. This invention also establishes a non-invasive, rapid, accurate, and quantifiable SPSS risk prediction model, capable of achieving the following functions: 1) Identify high-risk individuals for SPSS early.

[0052] 2) Clarify the clinical significance of SPSS under different liver function states.

[0053] 3) Reduce medical costs and improve accessibility at the grassroots level. Attached Figure Description

[0054] Figure 1 The figures show the relationship between the two groups of patients and the widest inner diameters of the portal vein, splenic vein, and superior mesenteric vein in this invention; the left figure shows the widest inner diameter of the portal vein, the middle figure shows the widest inner diameter of the splenic vein, and the right figure shows the widest inner diameter of the superior mesenteric vein.

[0055] Figure 2 This is a heatmap analysis of variable correlation in this invention.

[0056] Figure 3 This is an analysis of SPSS prediction nomograms in this invention.

[0057] Figure 4 This is the regression forest diagram in this invention.

[0058] Figure 5 This is for the ROC curve analysis in this invention.

[0059] Figure 6 This is for the calibration curve analysis in this invention.

[0060] Figure 7 This is for the analysis of clinical decision curves in this invention. Detailed Implementation

[0061] The specific embodiments of the present invention are described in detail below, but it should be understood that the scope of protection of the present invention is not limited to the specific embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise specified, the experimental methods described in the embodiments of the present invention are conventional methods, and the materials and reagents used in the following embodiments are commercially available unless otherwise specified.

[0062] Example 1 1. Data collection of research subjects This study retrospectively collected data from patients diagnosed with esophageal and gastric varices due to cirrhosis who visited the First Affiliated Hospital of Xi'an Medical University between January 2018 and December 2023, underwent endoscopy, and completed CTPV examination. The protocol of this invention was approved by the Ethics Committee of the First Affiliated Hospital of Xi'an Medical University, ethics approval number: XYYFY2024LSKY-029. Inclusion criteria were:

[0063] (1) Patients with a clear diagnosis of esophageal and gastric varices due to cirrhosis.

[0064] (2) Patients with complete imaging data and clinical laboratory records. Exclusion criteria: 1) Patients with malignant tumors; 2) Patients with a history of transjugular intrahepatic portosystemic shunt (TIPS), surgical shunt, or hepatectomy; 3) Patients with spontaneous portosystemic shunts (SPSS) other than gastrorenal shunt (GRS), splenorenal shunt (SRS), and umbilical vein opening, such as retroperitoneal venous shunts and cardiac angle venous shunts; 4) Patients with incomplete medical records. Exclusion criterion: Patients with unclear diagnosis of SPSS by CTPV.

[0065] A total of 306 patients meeting the inclusion criteria were collected, of whom 26 were excluded, including 7 with unclear diagnosis and triage, 3 patients who had previously undergone TIPS, 5 patients with malignant tumors, and 11 patients with missing necessary clinical information. Ultimately, 280 patients were included in this study. Clinical data collected included: age, sex, smoking history, alcohol consumption history, cause of cirrhosis, white blood cell count, red blood cell count, hemoglobin, platelet count, total bilirubin, indirect bilirubin, total bile acids, cholinesterase, AST, ALT, ALP, gamma-glutamyl transferase, total protein, albumin, globulin, creatinine, uric acid, PT, PT%, INR, APTT, TT, FIB, FDP, D-dimer, serum ammonia, widest diameter of portal vein, widest diameter of splenic vein, widest diameter of superior mesenteric vein, Child-Pugh classification, and occurrence of cirrhosis complications (ascites, PVT, HE, PHG, EGVB).

[0066] 2. Statistical methods All data were statistically analyzed using SPSS 25.0 statistical software. P A value <0.05 was considered statistically significant, and missing data were filled using the median. Normality was tested for continuous data. Normally distributed continuous data were expressed as mean ± standard deviation (x ± s), and independent samples t-tests were used for comparisons between groups. Non-normally distributed continuous data were expressed as median (P50) and interquartile ranges (P25, P75), and Mann-Whitney U tests were used for comparisons between groups. Count data were expressed using chi-square (χ²) data. 2 ) Test probability method, between-group analysis P Factors with a value <0.1 were included in the univariate logistic regression analysis to screen out factors in the univariate analysis. P For factors with a correlation coefficient <0.05, multiple linear and variable correlation diagnostics are performed. After confirming no linear correlation, the included factors are subjected to binary logistic analysis again to screen out risk factors. A nomogram is created using R4.4.2 software for visualization analysis, and receiver operating characteristic (ROC) curves, calibration curves, and clinical decision curves are used to evaluate the model efficacy.

[0067] 3. Clinical Feature Analysis The age, sex, smoking history, alcohol consumption history, and etiology of cirrhosis (hepatitis B, hepatitis C, alcoholic hepatitis, and others including cholestatic hepatitis, autoimmune hepatitis, etc.) of patients in the SPSS and non-SPSS groups were statistically analyzed, as shown in Table 1.

[0068] Table 1 Basic Data Analysis Note: P<0.05* P <0.01** P <0.001***.

[0069] As shown in Table 1, there were no statistically significant differences among the above indicators. P >0.05), patients without SPSS were older (57.16±12.79 vs. 54.06±12.73), but the difference was not statistically significant.

[0070] Peripheral venous blood was collected from all patients on the morning of the second day after admission in an empty stomach. After standing at room temperature for 2 hours, the blood was centrifuged at 3000 r / min for 10 minutes, the supernatant serum was separated and stored at -80℃ for later testing. A fully automated biochemical analyzer was used to detect complete blood count and liver and kidney function related indicators, including white blood cell count, red blood cell count, hemoglobin, platelet count, total bilirubin, direct bilirubin, indirect bilirubin, total bile acids, cholinesterase, aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), gamma-glutamyl transferase, total protein, albumin, globulin, creatinine, and uric acid. A fully automated coagulation analyzer was used to detect prothrombin time (PT), prothrombin activity (PT%), international normalized ratio (INR), activated partial thromboplastin time (APTT), thrombin time (TT), fibrinogen (FIB), fibrin degradation products (FDP), and D-dimer. Blood ammonia concentration was detected using enzymatic kinetics or ion exchange methods. All testing procedures, reagent usage, and result interpretation were strictly performed according to the instrument and reagent kit instructions, and the testing process conformed to standardized operating procedures for clinical laboratories to ensure accurate and reliable results.

[0071] The results are shown in Table 2.

[0072] Table 2 Laboratory Indicator Analysis Table 2 shows the results of white blood cell, red blood cell, hemoglobin, platelet count, total bilirubin, indirect bilirubin, total bile acids, cholinesterase, AST, ALT, ALP, gamma-glutamyl transferase, total protein, albumin, globulin, creatinine, uric acid, PT, PT%, INR, APTT, TT, FIB, FDP, D-dimer, and blood ammonia levels between the two groups. The statistically significant differences between the two groups were in white blood cell count (P=0.009) and PT. P =0.006), the red blood cell count was slightly higher in the SPSS group, but the difference was not statistically significant. P >0.05).

[0073] The internationally recognized Child-Pugh liver function grading system was used. Based on five core indicators at admission—total bilirubin, albumin, ascites, hepatic encephalopathy, and prothrombin time—patients were stratified into three groups: Child-Pugh A, B, and C. The distribution of different Child-Pugh grades was compared between the two groups with and without SPSS (Patients with and without SPSS).

[0074] The results are shown in Table 3.

[0075] Table 3 Liver function grading analysis As shown in Table 3, the SPSS group consisted of 82 patients, of whom 32 (39.02%) were Child-Pugh grade A, 36 (43.90%) were grade B, and 14 (17.07%) were grade C. The non-SPSS group consisted of 198 patients, of whom 84 (42.42%) were Child-Pugh grade A, 73 (36.87%) were grade B, and 41 (20.71%) were grade C. The χ² test showed no statistically significant difference in Child-Pugh classification between the two groups (χ²=1.29). P >0.05).

[0076] Based on the aforementioned Child-Pugh liver function classification results, Child-Pugh grade A patients were selected as the study subgroup, further divided into an SPSS-positive group (n=32) and a non-SPSS-positive group (n=84). Standardized assessments were performed for four portal hypertension-related complications of cirrhosis: ascites, PHG, PVT, and EGVB.

[0077] The results are shown in Table 4.

[0078] Table 4. Analysis of complications of cirrhosis in Child-Pugh A stage. As shown in Table 4, the incidence of EGVB in patients with SPSS in Child-Pugh A status was significantly lower than that in patients without SPSS, and the difference was statistically significant. P =0.005), but the incidence of ascites was higher in patients with SPSS than in those without SPSS, and the difference was close to statistical significance.

[0079] Based on the aforementioned Child-Pugh liver function classification results, Child-Pugh grade B and C patients were selected and divided into an SPSS group (n=50) and a non-SPSS group (n=114). Standardized assessments were performed on four complications related to portal hypertension in cirrhosis: ascites, PHG, PVT, and EGVB.

[0080] The results are shown in Table 5.

[0081] Table 5. Analysis of complications of cirrhosis in Child B / C stage. Table 5 shows that, in Child-Pugh B / C status, the probability of ascites, PHG, PVT, and EGVB was significantly higher in the SPSS group than in the non-SPSS group, but the difference was not statistically significant. This suggests that SPSS may promote the occurrence of clinical decompensated events as liver function deteriorates.

[0082] 4. Prediction Model Establishment With the presence or absence of SPSS as the dependent variable, binary variables were assigned the following values: SPSS=1, no SPSS=0; no EGVB=0, EGVB=1; no ascites=0, ascites=1; no HE=0, HE=1; no PHG=0, PHG=1; no PHG=0, PVT=1, PVT=0. Continuous variables included white blood cells, red blood cells, gamma-glutamyl transferase, albumin, PT, PT%, TT, D2-mer, blood ammonia, maximum diameter of portal vein, maximum diameter of splenic vein, and maximum diameter of superior mesenteric vein. A binary univariate logistic regression model was constructed using SPSS 25.0 statistical software. Each of the above clinical indicators was individually included in the model, and the independent association between each indicator and the risk of SPSS occurrence was analyzed.

[0083] The results are shown in Table 6.

[0084] Table 6 Results of Univariate Logistic Analysis As shown in Table 6, univariate logistic regression analysis revealed that six indicators—white blood cell count, gamma-glutamyl transferase (GGT), total PT, widest diameter of splenic vein, EGVB, and ascites—were significantly correlated with the occurrence of SPSS in patients with cirrhosis. P All <0.05).

[0085] Furthermore, compared with patients without SPSS, patients with cirrhosis and SPSS had a significantly wider splenic vein diameter. P <0.001, suggesting that splenic vein dilation may be closely related to the formation of SPSS; while no significant difference was observed in the widest inner diameter of the portal vein and superior mesenteric vein, suggesting that the correlation between SPSS and different portal vein tributaries varies. Figure 1 ).

[0086] Collinearity diagnosis and correlation heatmap analysis were performed on the above 6 pre-included candidate input variables.

[0087] The results are shown in Table 7.

[0088] Table 7. Diagnoses of multicollinearity in variables pre-included in multivariate analysis Combined with Table 7 and Figure 2 As a result, no significant linear correlation was found among the independent variables.

[0089] Based on the results of the aforementioned univariate logistic regression analysis, screening... P Clinically significant indicators with values ​​<0.05 were included in the multivariate analysis, including: white blood cell count, gamma-glutamyl transferase (GGT), total thromboembolic rate (PT), widest splenic vein diameter (GVB), and ascites. SPSS 25.0 statistical software was used to construct a binary multivariate logistic regression model to identify the core predictors of SPSS incidence independently. The model outputs regression coefficient β, standard error SE, Z-score, and... P Value, odds ratio (OR), and 95% confidence interval (95% CI); P <0.05 is used as the criterion for determining an independent risk factor.

[0090] The results are shown in Table 8.

[0091] Table 8 Results of Multivariate Logistic Analysis Multivariate logistic regression analysis showed that PT, the widest diameter of the splenic vein, EGVB, and ascites were identified as independent predictors of portosystemic shunt. P <0.05), among which PT, the widest inner diameter of the splenic vein, and ascites were positively correlated with the occurrence of SPSS, while EGVB was negatively correlated with the occurrence of SPSS.

[0092] These factors were visualized and analyzed by plotting nomograms using R4.4.2 software. Figure 3 The results showed that ascites, the widest diameter of the splenic vein, and PT were positively correlated with the occurrence of SPSS, while EGVB was negatively correlated with the occurrence of SPSS. Figure 4 ).

[0093] Furthermore, to further explore the model efficacy, ROC curve analysis was constructed: Among the single-indicator predictive efficacy, the widest internal diameter of the splenic vein (AUC=0.747, 95%CI, 0.680~0.814) had a Youden index of 0.399 and an optimal cut-off value of 14.5 (sensitivity 63%, specificity 76%). The predictive efficacy for PT (AUC=0.610, 95%CI: 0.531~0.689) had a Youden index of 0.331 and an optimal cut-off value of 19.95 (sensitivity 58%, specificity 75%), ascites (AUC=0.555, 95%CI: 0.477~0.633), and EGVB (AUC=0.583, 95%CI: 0.506~0.660) was lower than that of the multi-indicator joint prediction model (AUC=0.787, 95%CI, 0.731~0.843). P The value was <0.001, the Youden index was 0.428, and the optimal cut-off value was 0.309 (sensitivity 76%, specificity 67%), suggesting that the comprehensive assessment strategy has better clinical discriminative ability. Figure 5 ).

[0094] To further verify the goodness of fit of the logistic regression model, a calibration curve was plotted. The calibration curve shows that the predicted probability is basically consistent with the observed probability in the range of 0.2 to 0.6, while slightly overestimating the actual risk in the high-risk range (>0.6). Overall, the Hosmer-Lemeshow test results showed no statistically significant difference. P =0.955, indicating that the model calibration is good ( Figure 6 ).

[0095] To quantify the net benefit of the model and assess its clinical application value, a clinical decision curve was plotted to provide a reference for clinical decision-making. The results showed that when the risk threshold probability was within the range of 0.2–0.7, the net benefit of the predictive model was significantly higher than that of the "full intervention" and "no intervention" strategies, suggesting that applying this model within this range can optimize clinical decision-making. The net benefit peaked (approximately 0.25) when the threshold probability was between 0.3 and 0.5, indicating that the model had the best classification efficacy for intermediate-risk patients, consistent with the cut-off value of the ROC curve (0.309). Figure 7 ).

[0096] 5. Application module of risk prediction model The SPSS risk prediction model constructed in this application, when used in clinical practice, follows four steps: standardized indicator collection, risk probability calculation, risk stratification determination, and individualized follow-up recommendations. This forms a risk prediction method for esophageal and gastric varices with spontaneous portosystemic shunt in patients with cirrhosis. The specific steps are as follows: 1) Indicator collection and standardization steps (corresponding to the original: Indicator entry and standardization module) The core indicators collected by clinicians included four core predictive factors and liver function stratification indicators for patients with esophageal and gastric varices in the target cirrhosis: splenic vein maximum diameter, prothrombin time (PT), ascites, esophageal and gastric variceal bleeding (EGVB), and Child-Pugh classification. Specifically: the splenic vein maximum diameter was measured via abdominal ultrasound; prothrombin time (PT) was obtained from fasting peripheral venous blood collected upon admission and analyzed using a coagulation analyzer; ascites and EGVB were classified into two categories based on abdominal ultrasound, gastroscopy, and clinical history, with 1 indicating presence and 0 indicating absence; the Child-Pugh classification was divided into grades A, B, and C based on the patient's liver function indicators and clinical symptoms.

[0097] The standardization and processing of indicators involves standardizing the collected indicators: continuous indicators retain their original measured values, and binary indicators are uniformly coded as 0 / 1; the integrity of the indicators is checked, and abnormal and missing data are removed to obtain standardized parameters that can be used for model calculation.

[0098] 2) Risk probability calculation steps (corresponding to the original: Risk probability calculation module) The SPSS risk prediction model formula (multifactor logistic regression equation) constructed using this application through multifactor logistic regression is as follows: Logit P )=β1×X1+β2×X2+β3×X3+β4×X4; P =1 / (1+e −Logit(P) ); In the above formula, P X1 represents the risk probability of spontaneous portosystemic shunt; X2 represents the widest internal diameter of the splenic vein, and β1 is its regression coefficient; X3 represents the prothrombin time, and β2 is its regression coefficient; X4 represents the ascites, and β3 is its regression coefficient; X4 represents the history of esophageal and gastric variceal rupture and bleeding, and β4 is its regression coefficient.

[0099] Substitute the standardized four core indicator parameters into the above model formula (multivariate logistic regression equation), and calculate manually or with the help of a regular calculator to obtain the probability of SPSS occurrence in the 0-1 range for this patient.

[0100] 3) Risk stratification determination steps (corresponding to the original: Risk stratification determination module) The optimal critical value of the model obtained by ROC curve verification in this application was used as the stratification criterion; and the calculated SPSS risk probability was compared with the optimal critical value (0.309): a. If the predicted probability is greater than or equal to the optimal critical value, the individual is identified as a high-risk group for SPSS.

[0101] b. If the predicted probability is less than the optimal threshold, the individual is considered to be in a low-risk group for SPSS.

[0102] 4) Individualized follow-up and intervention recommendations (corresponding to the original: follow-up recommendation module) Based on the patient's risk stratification results and Child-Pugh liver function classification, clinicians will develop corresponding treatment and follow-up plans. High-risk groups: It is recommended to prioritize CT portal vein imaging (CTPV) to clarify SPSS anatomical structures; for Child-Pugh B / C decompensated patients, enhanced endoscopic monitoring and early hemostasis intervention should be adopted to reduce the risk of bleeding.

[0103] Low-risk groups: Closely monitor portal hypertension-related indicators, conduct portal venous system imaging examinations at scheduled times, routinely perform endoscopic screening, and dynamically assess the risk of SPSS occurrence and progression.

[0104] 6. Conclusion This invention is the first to construct and validate a risk prediction and calculation tool for esophageal and gastric varices in liver cirrhosis based on splenic vein diameter, PT, ascites, and EGVB, enabling non-invasive, rapid, and accurate early risk assessment and endoscopic treatment decision support.

[0105] This invention has high predictive efficacy, is easy to operate, and is highly scalable. It can significantly improve the early warning, precision treatment, and prognosis management of patients with portal hypertension due to cirrhosis, and has important clinical application value and scientific significance.

[0106] As can be seen from the above, the predictive model constructed in this invention has stable performance, is easy to operate, and is easy to promote and apply in clinical practice. It can effectively improve the identification efficiency of high-risk groups in SPSS, optimize the risk stratification management and individualized treatment strategies for patients with cirrhosis and portal hypertension, and has important clinical application value for improving patient prognosis and reducing the risk of complications.

[0107] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0108] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A risk prediction method for esophageal and gastric varices with spontaneous portosystemic shunt in patients with cirrhosis, characterized in that, Includes the following steps: Acquire clinical indicator data of patients with esophageal and gastric varices due to cirrhosis; wherein, the clinical indicator data includes the widest internal diameter of the splenic vein, prothrombin time, ascites, and history of esophageal and gastric variceal rupture and bleeding; The acquired clinical indicator data were standardized and assigned values. The standardized clinical indicator data were input into a multivariate logistic regression equation to calculate and output the risk probability of spontaneous portosystemic shunt in this patient with esophageal and gastric varices due to liver cirrhosis; wherein the multivariate logistic regression equation is as follows: Logit( P )=β1×X1+β2×X2+ β3×X3+β4×X4; P =1 / （1+e −Logit(P) ）； In the above formula, P The probability of spontaneous gate shunting occurring; Logit( P X1 represents the Logit transformed value for the risk of spontaneous portosystemic shunt; X2 represents the variable assignment result for the widest internal diameter of the splenic vein, and β1 is its corresponding regression coefficient; X3 represents the variable assignment result for ascites, and β3 is its corresponding regression coefficient; X4 represents the variable assignment result for the history of esophageal and gastric variceal rupture and bleeding, and β4 is its corresponding regression coefficient.

2. The risk prediction method according to claim 1, characterized in that, After outputting the risk probability of spontaneous portosystemic shunt in the patient with cirrhosis and esophageal varices, the optimal threshold value obtained by ROC curve verification was used as the risk stratification criterion. The risk probability of spontaneous portosystemic shunt in the patient with cirrhosis and esophageal varices was compared with the optimal threshold value to determine the risk stratification. If the risk probability of spontaneous portosystemic shunting in the cirrhotic esophageal and gastric varices patient is greater than or equal to the optimal threshold, the cirrhotic esophageal and gastric varices patient is determined to be a high-risk group for spontaneous portosystemic shunting. If the probability of spontaneous portosystemic shunt in a patient with esophageal and gastric varices due to cirrhosis is less than the optimal threshold, the patient is considered to be in a low-risk group for spontaneous portosystemic shunt.

3. The risk prediction method according to claim 2, characterized in that, The optimal critical value is determined through ROC curve analysis and is the predicted probability value corresponding to the maximum Yoden index.

4. The risk prediction method according to claim 3, characterized in that, The optimal critical value is 0.

309.

5. The risk prediction method according to claim 1, characterized in that, The clinical indicator data were obtained through medical imaging examinations, laboratory tests, and medical record reviews.

6. The risk prediction method according to claim 1, characterized in that, The standardized assignment rules are pre-stored in the database and are automatically invoked by the system during assignment processing.

7. The risk prediction method according to claim 1, characterized in that, β1, β2, β3, and β4 were determined by fitting a multivariate logistic regression based on the training samples.

8. The risk prediction method according to claim 7, characterized in that, The multifactor logistic regression equation is predetermined using the following model construction method, which includes the following steps: The following clinical data were collected from patients with esophageal and gastric varices due to cirrhosis: white blood cell count, red blood cell count, gamma-glutamyl transferase level, albumin level, prothrombin time, prothrombin activity, thrombin time, D-dimer level, blood ammonia level, widest diameter of portal vein, widest diameter of splenic vein, widest diameter of superior mesenteric vein, esophageal and gastric variceal rupture and bleeding, ascites, hepatic encephalopathy, portal hypertensive gastropathy, and portal vein thrombosis. Univariate logistic regression analysis was performed on the clinical indicator data to... P <0.05 is the screening criterion for statistical significance, and candidate variables that are statistically associated with the occurrence of spontaneous gate shunting are selected; The candidate variables were included in a multivariate logistic regression analysis. The stepwise regression method was used, with the inclusion criterion = 0.05 and the exclusion criterion = 0.10 as the variable selection criteria. The widest internal diameter of the splenic vein, prothrombin time, ascites and history of esophageal and gastric variceal rupture and bleeding were identified as four independent predictors. The regression coefficients of each independent predictor were obtained, and a multivariate logistic regression equation was established. Based on the multivariate logistic regression equation, the predicted probability is calculated, the ROC curve is plotted to evaluate the model's discriminative ability, and the optimal critical value is determined; the calibration curve is plotted to evaluate the model's calibration degree; and the clinical decision curve is plotted to evaluate the model's clinical net benefit. The model is composed of the multifactor logistic regression equation, the optimal critical value, and the validation results.

Images