| ObjectiveEsophageal variceal bleeding is a severe complication in cirrhotic patients. The diagnosis of esophageal varices(EV) early will be of important significant to prevent esophageal variceal bleeding. Currently, there are several non-invasive diagnostic tools for EV proposed, such as hematological index, traditional ultrasound examination and ultrasound elastography examination. However, few studies compared the diagnostic accuracy of those parameters. Thus, the first part of the present study was aim to compare the diagnostic efficacy of real-time tissue elastography(RTE) with laboratory tests in detecting EV in cirrhotic patients. The second part of our study intend to compare the diagnostic efficacy of the predicting model for moderate and severe EV with LS?PSR?LSPS?EV risk score and Lo K score, and validated internally.Material and Method1.1 Ultrasonography, laboratory tests and gastroscope were performed in 111 liver cancer parents with cirrhosis respectively. The level of aspartate aminotransferase(AST), alanine aminotransferase(ALT), platelet(PLT) and the results of gastroscope were recorded. The ratio of AST and ALT(AAR), and the ratio of AST and PLT(APRI) were calculated.1.2 Comparing the non-invasive parameters mentioned above between EV 0 and EV?1. ROC curve analysis is performed at the parameters different statistically significant and then compare their diagnostic efficacy.1.3 Comparing the non-invasive parameters mentioned above between EV?1and EV?2. ROC curve analysis is performed at the parameters different statistically significant and then compare their diagnostic efficacy.2.1 Transient Elastography, laboratory tests and gastroscope were performed in 167 liver cancer parents with cirrhosis respectively. The liver stiffness(LS), standard serum index and the base characteristics of patients were recorded. The PSR?LSPS?EV risk score and Lo K score were calculated.2.2 Comparing the non-invasive parameters mentioned above between EV?2 and EV<2 in training set, Thereafter, variables with P < 0.05 in the univariate analysis were included in subsequent multivariate analysis, where multiple logistic regression analysis was used to select variables to be maintained in the final model and fit EV index.2.3 Taking the results of gastroscope examination as golden standard, to compare the diagnostic accuracy of EV index with LS, PSR, LSPS, EV risk score and Lo K score in training set.2.4 validating the diagnostic efficacy of EV index for moderate and severe EV and comparing to the LS, PSR, LSPS, EV risk score and Lo K score in validation set. Result1 Comparing the diagnostic efficacy between LFI and laboratory indexes1.1 There were significant difference in LFI, AST, PLT and APRI between EV0 and EV?1 patients(all P<0.05), but no significant differences in age, sex ratio, ALT and AAR(all P>0.05). The area under ROC curve(AUC) of LFI, AST, APRI and PLT for EV?1 is 0.78(95%CI0.69-0.85),0.60(95%CI0.50-0.69),0.77(95%CI0.68-0.85)and 0.77(95%CI0.68-0.84), respectively. There was statistic different in AUC of LFI and AST for EV?1.1.2 There were significant difference in LFI, PLT, AAR and APRI between EV?1 and EV?2 patients(all P<0.05), but no significant difference in age, sex ratio, AST and ALT(all P>0.05). The AUC of LFI, PLT, AAR and APRI for EV?2 is 0.77(95%CI0.68-0.84),0.71(95%CI0.61-0.79),0.84( 95%CI0.76-0.90)and 0.84(95%CI0.76-0.90),respectively. There is no significant different in AUC between any variables(all P?0.05).2 Comparing the diagnostic efficacy between the predicting model for moderate and severe EV with LS?PSR?LSPS?EV risk score and Lo K score2.1 There were significant difference in LS, WBC, HB, ALB, PAB, INR, APTT and PSR between EV?2 and EV < 2 patients(all P < 0.05). These univariate predictors were entered into a stepwise logistic regression model. On the basis of the above multivariate analysis, we derived a multiple fractional equation(EV index=4.27+0.191*LS-0.054*HB-1.914*PSR) for prediction of moderate and severe EVs that included LS, HB and PSR.2.2 The AUC of LS?PSR?LSPS?EV risk score?Lo K score and EV index in training set for EV?2 is 0.793( 95%CI0.705-0.864), 0.812(95%CI0.723-0.883),0.860(95%CI0.777-0.921), 0.841(95%CI0.755-0.907),0.748(95%CI0.655-0.826)and 0.909(95%CI 0.857-0.957),respectively. The AUC of EV index for EV?2 is significant higher than LS?PSR?Lo K score(all P<0.05).2.3 The AUC of LS?PSR?LSPS?EV risk score?Lo K score and EV index in training set for EV?2 is 0.901(95%CI0.793-0.964),0.879(95%CI0.765-0.950),0.877(95%CI0.763-0.949),0.926(95%CI0.822-0.928),0.844(95%CI0.724-0.927)? 0.950(95%CI0.856-0.991),respectively. There is no significant different in AUC between any variables(all P?0.05).ConclusionAll above, the result of present study shows that the diagnostic efficiency with LFI was not significantly better than those with serum indexes. LFI, PLT, APRI and AAR have certain value in diagnosis of EV?1 and EV?2. The diagnostic efficiency with EV index was significantly better than those with LS, PSR and Lo K score. EV index has an excellent diagnostic accuracy of moderate and severe EV, and can serve as a reference index for high EV bleeding risk. However, it can't replace the examination of grastroscope. |
PDF Full Text Request