European Review for Medical and Pharmacological Sciences 2020; 24: 3876-3881 A study on two kinds of scoring models in predicting the degree of esophageal varices and bleeding X.-K. WANG1,2, P. WANG2,3, Y. ZHANG2, S.-L. QI2, W. LIU2, G.-C. WANG2 1First Clinical College of Liaoning University of Traditional Chinese Medicine, Shenyang, China 2Department of Liver Disease, Dalian Sixth People Hospital, Dalian, China 3Department of Gastroenterology, Yantaishan Hospital, Yantai, China Xie-Kui Wang and Ping Wang contributed equally to this work Abstract. – OBJECTIVE: This study aims to in- Introduction vestigate the value and determine the accuracy of two kinds of scoring models in predicting the In liver cirrhosis (LC) patients, portal hy- degree of esophageal varices (EV) and esopha- pertension mainly manifests as esophagogas- geal variceal bleeding (EVB) in patients with liv- er cirrhosis (LC). tric varices (EGVs), hypersplenism, and ascites. PATIENTS AND METHODS: A total of 189 Among these, EGVs are the main manifesta- patients with LC, who underwent esophagogas- tion of portal hypertension, and the main cause troduodenoscopy (EGD), color Doppler ultra- of death. According to statistics, EGVs can be sound (CDU), and computed tomography (CT), observed in approximately 50% of LC patients. were retrospectively analyzed. Then, the routine Patients without varicose veins develop varicose blood examination, liver function test, M-index veins at a rate of 5% a year. The varicose vein of of the spleen in CT, EGD, and CDU results were recorded. According to the EGD result, these the venule (≤5 mm in diameter of the varicose patients were divided into five groups: varicose vein) also develops from 5% to 12% per year to bleeding group, severe varices group, moderate the middle or varicose veins of the vena cava varices group, mild varices group, and no var- (the diameter of the varicose vein is >5 mm). ices group. Then, the receiver operating char- The annual incidence of varicose veins is 5-15%. acteristic curves of all predicting parameters Esophageal variceal bleeding (EVB) can spon- studied were respectively drawn, the area un- der the receiver operating characteristic curves taneously stop in 40% of patients. Although the were calculated, and the predictive value of EV treatment of EGV bleeding and the prevention of and EVB was evaluated. EGV hemorrhage have significantly improved in RESULTS: The area under the receiver op- recent years, the mortality rate in six weeks can erating characteristic curve of the VAP score still reach up to 20%, and the re-bleeding rate model and Plt/S-D score model was 0.901 of untreated patients remains at approximately and 0.835, respectively. The VAP score model 60%1,2. Although esophagogastroduodenoscopy cut-off value of 461.5 for predicting moderate esophageal varices (MoEV), severe esophageal (EGD) is the gold standard for diagnosing EGVs, varices (SEV), and EVB has a specificity and in recent years, non-invasive methods to diag- sensitivity of 100% and 68.7%, respectively, nose EGVs can reduce the cost and discomfort while the Plt/S-D score model cut-off value of of some patients, and many scholars have begun 835.5 for predicting MoEV, SEV, and EVB has a to research on more accurate EGV prediction specificity and sensitivity of 95.1% and 58.2%, non-invasive methods. Therefore, it is of great respectively. CONCLUSIONS: These two kinds of scoring clinical value to develop an accurate, practical, models can predict the degree of esophageal and repeatable non-invasive examination meth- varices and bleeding in liver cirrhosis patients od for predicting the degrees of EGVs and bleed- and has good predictive accuracy. ing in LC that allow for the good compliance of patients. The present study mainly uses two Key Words: kinds of scoring models to predict the degree of Scoring models, Esophageal varices, Varicose esophageal varices (EV) and risk of hemorrhage bleeding, Cirrhosis. and to evaluate the accuracy of the prediction. 3876 Corresponding Author: Gong-Chen Wang, MD; e-mail: [email protected] Scoring models of esophageal varices and bleeding Patients and Methods tion within 48 hours of bleeding; patients whose esophageal varicose hemorrhage (blood seepage) Main Materials could be observed through an endoscope, patients The materials used were: EGD purchased who had a white thrombus in the hemorrhage of from Olympus 260 (Olympus, Tokyo, Japan); EV, or patients with who had blood clots on the computed tomography (CT); Brilliant 16 (Philips, surface, or thrombosis or scab formation. Pa- Amsterdam, The Netherlands); Color Doppler tients with other potential bleeding sites and other Ultrasound (CDU); HI Vision Preirus (Hitachi, bleeding lesions were excluded. Tokyo, Japan). Exclusion Criteria Subject of the Study Patients with the following symptoms were There were 189 patients with LC. Among these excluded: patients who were previously admin- patients, 151 patients were male, and 38 patients istered with beta-blockers medication or blood were female. Furthermore, among these patients, transfusion, non-cirrhosis patients who had EV 25 patients had EVB, 50 patients had severe rupture hemorrhage, patients with isolated gastric esophageal varices (SEV), 47 patients had mod- varices and hemorrhage, patients with liver can- erate esophageal varices (MoEV), 28 patients had cer and other malignant tumors, patients with oth- mild esophageal varices (MiEV), and 39 patients er non-liver diseases that can cause platelet abnor- had no esophageal varices (NEV). A total of 189 malities, patients who underwent a liver or spleen patients with LC underwent EGD and received intervention operation, and patients who received CDU and CT examinations during their hospi- a trans jugular intrahepatic portosystemic shunt talization in Dalian Sixth People’s Hospital from (TIPS). January 1, 2012 to November 1, 2017. The age, gender, degree of EV, level of albumina, blood Laboratory Observation Parameters platelet count (BPC), m-index of the spleen, m-in- The diagnosis of patients, and the relevant re- dex, and EGD EV diagnosis were determined. sults of the blood routine test, liver function, EGD, The present study was approved by the Ethics CDU, and CT were recorded. On the basis of the Committee of Dalian Sixth People’s Hospital. A above laboratory examinations and special exam- written informed consent was obtained from each inations, the following indexes were calculated: participant. varices and portal hypertensive gastropathy scor- ing system: VAP=[A(g/dl)×BPC(/mm3)]/[m-in- Inclusion Criteria dex(cm 3)]; M-index: the splenic multidimension- The inclusion criteria were as follows: patients al index, or spleen volume, which is the length, diagnosed with LC, that is, patients with a his- width and thickness of the spleen measured under tory that led to LC, such as viral hepatitis and CT scan; platelet count and spleen diameter ratio: long-term heavy drinking; patients with a clin- (Plt/S-D)=BPC(/mm3)/spleen diameter (cm); the ical manifestation of liver dysfunction or portal diameter of the spleen: the ratio of the maximum hypertension, and had a chemical index of liver length of the spleen to the two levels measured by function decompensation, such as decreased se- abdominal CDU. rum albumin, increased bilirubin, and prolonged prothrombin time; patients with LC according to Statistical Analysis CDU or CT, and EV revealed by EGD, excluding Categorical variables were compared using pre-hepatic portal hypertension and post-hepatic X2-test. The man variables between two groups portal hypertension (the gold standard for diag- were compared using Student’s t-test, the mean nosing this disease is liver biopsy examination, variables among multi-groups were compared us- which presents as a form of pseudo lobule); pa- ing one-way analysis of variance, and the post- tients who were admitted to the hospital within hoc test of Bonferroni; Duncan test was used to 48 hours after onset, underwent EGD, and were validate ANOVA. The area under the receiver op- evaluated using the endoscopic EV diagnostic cri- erating characteristic curve (AUROC) was used teria based on the endoscopic diagnosis and treat- to select the parameter that revealed a good dis- ment for gastrointestinal varices and bleeding3; criminative power for predicting the presence of patients diagnosed according to the diagnostic EV. In addition, the receiver operating character- criteria of EGV rupture hemorrhage; patients who istic (ROC) curve was used to determine the cut- underwent an emergency endoscopic examina- off value with the best sensitivity and specificity. 3877 X.-K. Wang, P. Wang, Y. Zhang, S.-L. Qi, W. Liu, G.-C. Wang Data were presented as mean ± standard deviation group and SEV group (p=0.766) in m-index vari- (SD), and each odds ratio (OR) and AUROC curve ables. However, there was a significant difference were presented together with its 95% confidence between the other groups (p<0.05). There was no interval (95% CI). A two-sided p<0.05 was con- statistically significant difference among the EVB sidered statistically significant for all analyses. group, MoEV group, and SEV group (p>0.05), When AUC<0.5 was not in accordance with the but there was a significant difference between the real situation, it was rarely observed in the actual other groups (p<0.05). The VAP scoring model situation4, and the AUC was compared with the had no statistically significant difference among z-test. p<0.05 was considered statistically signif- each group in the EVB group, MoEV group, and icant. The data were analyzed using Statistical SEV group (p>0.05). However, there was a sta- Product and Service Solution (SPSS 13.0, Chica- tistically significant difference when compared to go, IL, USA). the other groups (p<0.05) (Table I). Comparison of Prediction Models Results Among the Recombined Groups There were no significant statistical differenc- Comparison of all Parameters es among the five groups above.