Aim: to evaluate the frequency of portal vein thrombosis (PVT) and build predictive models of the development of PVT for patients with liver cirrhosis (LC) of A and B/C classes by Child-Pugh.Materials and methods. Research design is a case-control. The Case group included 130 patients with newly diagnosed PVT not caused by invasive hepatocellular carcinoma (HCC); 29 patients were assigned to class A, 101 patients were assigned to class B/C. From the database of cirrhotic patients without PVT 60 Controls for class A and 205 for B/C were selected using sratified randomization by sex, age and etiology of cirrhosis. The Mann-Whitney U-test and Pearson's chi-squared test were used to compare the groups. Odds ratios (OR) and 95 % confidence intervals (95 % CI) were calculated. Logistic regression models are constructed with the separation of the sample into training and test (0.7; 0.3). The operational characteristics of the models were calculated on the test sample; ROC analysis was carried out, the area under the ROC curve (AUC) was calculated.Results. The overall frequency of PVT was 4.1 % (95 % CI 2.7-5.8 %) in class A and 10.4 % (95 % CI 8.5-12.5 %) class B/C. Patients with class A and B/C PVT differed from the corresponding controls by more severe portal hypertension: the frequency of bleeding / number of interventions on varices compared with the control were 41/45 % vs. 7/8 % (p < 0.001) for class A and 25.7/30.7 % vs. 16.1/16.1 % (p < 0.05) for class B/C, ascites frequency was 24 % vs. 8 % (p < 0.05) for class A and 89.1 % vs. 68.3 % (p < 0.001) for class B/C. The cutoff by the portal vein diameter was the same for both classes — 13.4 mm; the spleen length was similar and amounted 17.5 mm for class A, 17.1 mm for class B/C. Patients with PVT differed from the corresponding controls by neutrophil-to-lymphocyte ratio: class A 2.33 (1.82; 3.61) vs. 1.76 (1.37; 2.20), p < 0.01, class B/C 2.49 (1.93; 3.34) vs. 2.15 (1.49; 3.26), p < 0.05. Patients of class B/C had a higher incidence of newly diagnosed malignant tumors - 23.8% (primarily HCC that does not invade the portal vein), compared with control and cases of class A - 6.3 % and 3 % (p < 0.05), respectively. The best model for class A included variceal bleeding, ascites, portal vein diameter, absolute number of neutrophils, for class B — ascites, spleen length, portal vein diameter, malignant tumors / local factors; sensitivity, specificity, accuracy and AUC were 79.3 %, 90 %, 86.5 %, 0.897 and 73.3 %, 68.3 %, 69.9 %, 0.789, respectively.Conclusion. Independently of the Child-Pugh class of LC, the main risk factor for PVT is severe portal hypertension.
Read full abstract