Clinical Study Of Prognosis And Esophageal Varices In Cirrhotic Patients

This research was composed of three parts:Ⅰ.Identification of theprognostic value of MELD and MELD-Na system in Chinese patients withdecompensated liver cirrhosis.Ⅱ.Prediction of esophageal varices andmedium to large esophageal varices in cirrhotic patients with Child-Pugh Ausing uninvasive method.Ⅲ. Assessment of early rebleeding afterendoscopic esophageal variceal ligation using a predictive model.Ⅰ.Identification of the prognostic value of MELD and MELD-Nasystem in Chinese patients with decompensated liver cirrhosis.Objectives To identify the prognostic value of MELD and MELD-Nasystem in Chinese patients with decompensated liver cirrhosis.Methods Acohort of 206 hospitalized patients with decompensated liver cirrhosis wereretrospectively studied with follow-up period of three months. TheChild-Pugh score,MELD score and MELD-Na score were respectivelycalculated according to the original formula for each patient to comparetheir ability in prognosis.Receiver operating characteristic curve (ROC) andthe area under the curve (AUC) were used to measure the prognostic abilityof the three system to predict the 3-month survival,and Z-test was used tocompare their predictive ability.ROC curves were also used to determine theChild-Pugh,MELD and MELD-Na score cut-offs with the best sensitivityand specificity in discriminating the high and low mortality patients.Results 39 patients died within 3 months.Child-Pugh score,MELD scoreand MELD-Na score for non-survivors(11.18±1.76,21.49±6.75,29.26±12.03) were higher than those for survivors(9.16±2.19,15.83±4.99,18.65±9.69) significantly (P＜0.001).The AUCvalues generated by the ROC curves for the Child-Pugh score,MELD scoreand MELD-Na score were 0.759, 0.760 and 0.777 respectively in predicting 3-month survial.The cut-off scores of the three system were 10, 18 and 23respectively, which could discriminate high and low mortality patientsaccurately.Compared with Child-Pugh classification and MELD, thepredicting accuracy of MELD-Na were significantly higher (P＜0.001),whilethe predicting accuracy between Child-Pugh and MELD was nosignificantly different (P=0.80).Conclusions 1.Child-Pugh score,MELDscore,and MELD-Na score can accurately predict the 3-month prognosis ofpatients with decompensated liver cirrhosis.2.MELD-Na score is superior toother systems and is worth using in clinic,while there was no difference inthe prognostic assessment between MELD score and Child-Pughscore.3.The correlation between MELD-Na,MELD,CS and CC was allsignificant(P＜0.001).Ⅱ.Prediction of esophageal varices and medium to largeesophageal varices in cirrhotic patients with Child-Pugh A usingnon-invasive methods.Objectives To analyze the independent risk factors which mightnon-invasively predicit the presence of esophageal varices and medium tolarge esophageal varices in first-treated cirrhotic patients with Child-PughA. Methods This study retrospectively included 225 cirrhotic patientsrefered to the Tianjin Third centred Hospital for hospitalizing betweenJanuary 2003 and December 2006.Endoscopy was performed in allpatients,and all of their liver function belong to Child-Pugh Aclassification.All of their clinical,biochemical,eikonic parameters andendoscopic appearances were collected.Firstly, according to theirendoscopic appearances, they were allocated to the group without varicesand the group with esophageal varices;meanwhile according to theirdegree of varices,they were allocated to the group with none to smallesophageal varices and the group with medium to large esophageal varices.All of their data were analyzed by univariate and multivariate analysis respectively.After the independent risk factors were obtained, ROCanalysis was used to evaluate their predictive efficacy.The cut-offs withthe best sensitivity and specificity were used to discriminate the high-riskgroup and low-risk group.Results In this study, 56 patients had largeesophageal varices, 96 patients had medium esophageal varices,34patients had small esophageal varices and 39 patients were absence ofoesophageal varices.When divided according to whether they hadvarices,univariate and multivariate analysis identified three independentvariables predictive of esophageal varices in cirrhotic patients withChild-Pugh A classification liver function:platelet count, portal venousflow velocity and spleen thickness.ROC curves were applied to determinethreshold values (cut-off) of platelet count (97×10~9/L) and spleenthickness (4.89cm).For platelet count,a threshold value＜97×10~9/L had asensitivity of 59% and a specificity of 79% with a positive predictivevalue (PPV) of 73.75% and a negative predictive value(NPV) of65.83%.For spleen thickness, a threshold value＞4.89cm had asensitivity of 79% and a specificity of 59% with a positive predictivevalue (PPV) of 65.83% and a negative predictive value(NPV) of73.75%.When divided according to their degree of varices,using the samemethod above-mentioned to analyze we found that platelet count, portalvenous flow velocity and spleen thickness were independent predictivevariables of presence of medium to large esophageal varices. ROC curveswere applied to determine threshold values (cut-off) of plateletcount(83×10~9/L),and it's corresponding sensitivity and specificity were63% and 77% with a positive predictive value (PPV) of 73.26% and anegative predictive value(NPV) of 67.54%.Conclusions Platelet count,portal venous flow velocity and spleen thickness were independent riskfactors in the prediction of esophageal varices and medium to largeesophageal varices in cirrhotic patients with Child-Pugh A classification.Platelet count and spleen thickness were better than portalvenous flow velocity in the prediction of the presence of esophagealvarices; While platelet count was better than portal venous flow velocityand spleen thickness in the prediction of the presence of medium to largeesophageal varices.Ⅲ. Assessment of early rebleeding after endoscopic esophagealvariceal ligation using a predictive modelObjectives To analyze the risk factors which can predict earlyrebleeding after endoscopic esophageal variceal ligation,and to develop apredictive model for the occurrence of early rebleeding.Methods Therecords of patients treated with EVL due to cirrhosis with or withouthepatocellular carcinoma were retrospectively analyzed.Totally 603proceduers of EVL had been performed in the period between January2001 and December 2005.They were randomly allocated to either anestimation group (423 cases) or a validation group (180 cases).Forty-threecommon clinical and serum markers were analyzed initially in theestimation group to derive a predictive model(EVLER Index) to predict theoccurrence of early bleeding after EVL.The model created was thenassessed with ROC analysis.It was also applied to the validation group totest its repeatability, and ROC analysis was used again to measure themodel's accuracy in order to further certain it's predictive efficacy.ResultsIn estimation group totally 38 patients generated early rebleeding afterEVL.α-fetoprotein level, Child-Pugh classification, bending loop and portalvenous flow velocity were indentifled by multivariate logistic regressionanalysis.The index model for predicting early rebleeding (EVLER index)was composed of by those four factors. In ROC analysis, AUC was 0.804for the estimation group and 0.843 for the validation group for predictingearly rebleeding after EVL. Using the optimal cut-off score 2.3, thesensitivity of the model was 75.1%, and the specificity was 78.9%. There was no significant differences in predictive efficacy between the estimationand the validation group (Z=0.95,P＞0.05). Conclusionsα-fetoprotein level,Child-Pugh classification, bending loop and portal venous flow velocitywere independent risk factors of early rebleeding after EVL;and a model forpredicting early rebleeding after EVL (EVLER index) was established withthe four easily accessible markers. It appears to be accurate, sensitive andreproducible,and it could be used to predict the early rebleeding after EVL.