Buscar en
Annals of Hepatology
Toda la web
Inicio Annals of Hepatology New model predicting gastroesophageal varices and variceal hemorrhage in patient...
Journal Information
Vol. 19. Issue 3.
Pages 287-294 (May - June 2020)
Share
Share
Download PDF
More article options
Visits
227
Vol. 19. Issue 3.
Pages 287-294 (May - June 2020)
Original article
DOI: 10.1016/j.aohep.2019.12.007
Open Access
New model predicting gastroesophageal varices and variceal hemorrhage in patients with chronic liver disease
Visits
...
Jia-li Maa, Ling-ling Hea, Yu Jianga, Jun-ru Yanga, Ping Lia, Yao Zangb, Hong-shan Weic,
Corresponding author
drwei@ccmu.edu.cn

Corresponding author at: Department of Gastroenterology, Beijing Ditan Hospital, Capital Medical University, Beijing, China.
a Capital Medical University Affiliated Beijing Ditan Hospital, Department of Gastroenterology, Beijing, China
b Capital Medical University Affiliated Beijing Ditan Hospital, Department of Medical Ultrasound, Beijing, China
c Department of Gastroenterology, Beijing Ditan Hospital, Capital Medical University, Beijing, China
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Figures (4)
Show moreShow less
Tables (3)
Table 1. The demographic data of patients.
Table 2. The etiology of patients in validation set without HBV infection.
Table 3. The result of multiple logistic regression.
Show moreShow less
Abstract
Introduction and objectives

The predictors for gastroesophageal varices (GOV) and hemorrhage development have not been well studied in different liver diseases or different population. This study aimed to evaluate whether a new algorithm focusing on chronic hepatitis B (CHB) patients is also applicable to other chronic liver diseases (CLDs) in Chinese population.

Patients or materials and methods

We retrospectively analyzed 659 CHB patients and 386 patients with other CLDs. A total of 439 CHB patients were included in training set, the other 220 CHB patients and other patients with CLDs were included in validation set. A new algorithm for diagnosing GOV was established and its sensitivity and specificity for predicting the varices was verified.

Results

Multivariable logistic regression revealed that the rough surface of the liver (p<0.001), splenic thickness (p<0.001), and liver stiffness (p=0.006) were independent predictors of GOV. The new algorithm was considered to be a reliable diagnostic model to evaluate the presence of varices. The AUROC was 0.94 (p<0.001) in CHB validation set and 0.90 (<0.001) in non-CHB validation set. When the cut-off value was chosen as −1.048, the sensitivity and specificity in diagnosing GOV in CHB population were 89.1% and 82.5%, respectively. Importantly, the new algorithm accurately predicted the variceal hemorrhage not only in CHB patients, but also in patients with other CLDs.

Conclusion

The new algorithm is regarded as a reliable model to prognosticate varices and variceal hemorrhage, and stratified not only the high-risk CHB patients, but also in patients with other CLDs for developing GOV and variceal bleeding.

Keywords:
Chronic hepatitis B
Gastroesophageal varices
Algorithm
Variceal hemorrhage
Full Text
1Background

Variceal hemorrhage is a major acute complication of cirrhotic patients and is a leading cause of death. Although multiple procedures were available to control bleeding, the mortality associated with it accounted for about 10%–20% currently [1,2]. Therefore, clinical guidelines recommend upper gastrointestinal endoscopic examination for all cirrhotic patients to identify esophageal and/or gastric varices [3,4]. The end-stage chronic liver disease (CLD) is the result of liver cirrhosis, although the complications are multiple, severe and potentially fatal in the compensated stage. Several patients with CLD in the early stage of liver cirrhosis or liver fibrosis were at increased risk of varices and hemorrhage. Apparently, early diagnosis and appropriate treatment of cirrhotic patients with clinically significant portal hypertension (CSPH) remains a critical step for improving the prognosis and quality of life of patients. Generally, CSPH is diagnosed based on hepatic venous pressure gradient of ≥10mmHg, and severe portal hypertension of ≥12mmHg [5,6]. However, direct measurement of hepatic venous pressure gradient is an invasive method, and so it is difficult to be used in clinical practice. Endoscopy is a technique to confirm gastroesophageal varices (GOV) in patients with CLDs, and it is difficult to be accepted by some patients due to its invasive nature. Endoscopic ultrasound significantly improves the diagnostic performance when compared with that of traditional endoscopy, but it is also an invasive method [7]. In fact, the above discussed methods are invasive, expensive and not available at every medical center.

Imaging methods such as X-ray, ultrasound, computed tomography (CT) scan, and magnetic resonance imaging were also generally used to diagnose CSPH [7,8]. Barium swallow is considered as a quick and reliable method for assessing esophageal varices till 1990s [9]. The major disadvantage of this method is patients are exposed to radiation, and few centers still use it currently. CT is used to screen GOV instead of endoscopy, but its suboptimal specificity should be taken into consideration [10]. Transient elastography (TE) is a relatively new noninvasive technique for diagnosing portal hypertension in patients with CLDs [7]. TE is highly sensitive in estimating the absence of liver fibrosis or the presence of high-degree liver fibrosis. Therefore, the American Gastroenterological Association and Baveno VI Consensus have recommended the use of TE to evaluate GOV. TE is a rapid and painless method, and is well accepted by most of the patients [2,11]. However, acute hepatocytes injury, fasting status, and extra-hepatic cholestasis influence TE performance in evaluating GOV [12]. Consequently, some new algorithms were reported to diagnose varices in population with CLDs [13,14]. The algorithm was based on the results of capsule endoscopy, which was relatively an expensive method. Hence, it is difficult to be used in clinical practice, especially in developing countries.

For creating a noninvasive model to predict the presence of GOV in chronic HBV infected patients and other CLDs, 1045 patients with and without chronic HBV infection were retrospectively analyzed and provided a new algorithm for diagnosing GOV, with a relatively high sensitivity and specificity.

2Methods2.1Patients

This was a retrospective analysis conducted according to the ethical guidelines of Declaration of Helsinki. The protocol was approved by the Ethics Committee of Beijing Ditan Hospital, Capital Medical University. All patients provided written informed consent before receiving endoscopy or endotherapy. CLD was defined as an inclusive category for all patients that presented with a history and/or evidence of chronic liver dysfunction as detected by ultrasound to prove hepatomegaly and/or persistent abnormality of liver function (more than 6 months duration). For chronic HCV infection patients, all variables were recorded before they receive direct antivirals that produce sustained clearance of HCV. From October, 2015 to December, 2017, a total of 1133 patients with CLDs or suspected cirrhosis have received elastography, endoscopy, ultrasound, and systematical laboratory test during the same period in our hospital. For 1133 patients with CLDs, 88 patients were excluded from the group as 11 patients were without TE data and 77 patients have ALT levels ≥60U/L after analyzing the TE data (the ALT threshold is selected according to the suggestions of AASLD 2018 Hepatitis B Guidelines [15]). CHB patients were randomly divided into two groups (439 into training set and 220 into validation set), and CLD patients with other etiologies were also chosen as validation set. The criteria for patients in training group included HBsAg-positive for over 6 months and with nearly normal ALT levels (≤1.5ULN), and those with or without receiving antiviral therapy. For validation group, the diagnosis of patients was based on clinical guidelines, with nearly normal ALT levels (≤1.5ULN).

A total of 659 CHB patients and 386 patients with other CLDs were included. CHB is usually acquired perinatally or during childhood in China, or infected perinatally. There is a common period with normal serum ALT levels, positive sera for hepatitis B e-antigen, high HBV DNA levels, and minimal or no liver inflammation [16], and the shape of the liver might also be approximately normal (Fig. 1).

Fig. 1.

Study design. A total of 1133 CHD patients received endoscopy from October, 2015 to December, 2017 in our medical center. Eleven patients without TE and 77 patients with ALT60U/L, were excluded from analysis, and a total of 1045 patients were left (CLD, chronic hepatitis disease; CHB, chronic hepatitis B; non-CHB, CLD patients without hepatitis B virus infection; TE, transient elastography; ALT, alanine transaminase).

(0.27MB).
2.2LS, ultrasonographic measurement and endoscopy

LS measurements for each patient were performed using FibroScan (FibroScan®, France) by one of the three experienced sonographers (at least 10 years experiences). Patients who have undergone Fibroscan were all in the fasting state. The method for measuring spleen thickness involves locating ultrasonic probe in the left intercostal skew, the maximum distance between the spleen and the opposite convex surface of the spleen. Liver surface nodularity was defined as rough surface, which was assessed by using a transducer to examine the whole liver surface and the outer 2–3cm of the liver parenchyma. Multiple scans were used to search the signs in both the left and the right lobes. The GOV and bleeding were diagnosed by endoscopic examination (Olympus GIF-260/H260, Japan). The gastroenterologist should have at least 10 years of experience.

2.3Sample size calculation

To achieve 80% power for a 5% significance level, the incidence of cirrhosis in CHB patients should be approximately 10% [17,18], and nearly 50% of patients with newly diagnosed liver cirrhosis should accompany the varices [19]. By considering a 20% patient dropout rate, at least 113 patients should be allocated to each group.

2.4Statistics

Statistical analysis was performed by SPSS 19.0 (IBM Corp., Armonk, NY, United States) and GraphPad Prism 7.0 (GraphPad Prism Software Inc., San Diego, CA, USA). The quantitative data were presented as means±SD. The predictive index of GOV was analyzed by multivariable logistic regression analysis. A total of 659 CHB patients were included in the training set, and 386 patients with non-HBV related to other CLDs were included in the validation set. The differences in measurement data between the training set and validation set were evaluated by univariate regression analysis. Predictive markers with significant relation to GOV were identified (p<0.05) and then were subjected to multivariate logistic regression analysis. Using the new algorithm, a risk score was calculated for each patient in the training set. Student's t test was used to evaluate the differences in risk score for diagnosing varices between training set and validation set. Receiver operating characteristic curve (ROC) analysis was used to define a plot of diagnostic sensitivity and specificity regarding the present algorithm for predicting GOV. For evaluating the predictive performance of variceal hemorrhage, the new algorithm was applied for all 1045 patients.

3Results3.1Demographic data

A total of 1045 patients with complete data were included in the final analysis. The mean age of patients with CHB in the training set was 47.7±11.6 years, and 284 (64.7%) of these were men. In total, 139 patients had GOV, and 46 patients had history of GOV bleeding. The mean age of CHB patients in the validation set was 48.3±11.4 years, and 136 (61.8%) of these were men. In total, 71 patients had GOV, and 23 patients had history of GOV bleeding. The mean age of 386 non-CHB patients included in the validation set was 54.6±11.9 years, and 199 (51.6%) of these were men. In total, 164 patients had GOV, 53 patients had history of GOV bleeding. Most of the patients in CLD validation set had HCV infection, primary biliary cirrhosis and cryptogenic cirrhosis. The demographic and clinical characteristics and flow chart of subjects are shown in Tables 1 and 2 and Fig. 1.

Table 1.

The demographic data of patients.

Characteristics  Training cohort(CHB N=439)  Validation cohort(CHB N=220)  p  CHB cohort(N=659)  CLD cohort(N=386)  p 
Rough surface of liver  165 (38.0%)  88 (40.0%)  0.529  254 (39.3%)  165 (42.7%)  0.726 
Portal vein diameter (mm)  11 (11–12)  11 (11–12)  0.766  11 (11–12)  11 (11–12)  0.719 
Spleen thickness (mm)  35 (30–45)  34 (30–46)  0.659  35 (30–45)  38 (31–48)  0.002 
Ascites  59 (13.4%)  29 (13.2%)  0.935  88 (13.6%)  71 (18.4%)  0.096 
Gastroesophageal varices  139 (32.0%)  71 (32.3%)  0.859  210 (32.3%)  164 (42.5%)  0.017 
Varices hemorrhage history  46 (10.5%)  23 (10.5%)  1.000  69 (10.6%)  53 (13.7%)  0.284 
Endotherapy  44 (10.0%)  18 (8.2%)  0.450  62 (9.5%)  46 (11.9%)  0.418 
LS (kPa)  7.9 (5.3–15.1)  7.3 (5.3–17.3)  0.946  7.7 (5.3–15.7)  13.3 (6.7–26.6)  ≤0.001 
Age (yr)  47.7±11.6  48.3±11.4  0.530  47.8±11.5  54.6±11.9  ≤0.001 
Male sex  284 (64.7%)  136 (61.8%)  0.487  425 (64.5%)  199 (51.6%)  ≤0.001 
WBC (×109/L)  5.0±1.9  4.7±1.7  0.143  4.8 (3.7–6.0)  4.6 (3.4–6.0)  0.223 
PNG (%)  57.7±10.7  57.6±10.5  0.930  57.6±10.7  56.2±11.8  0.051 
HGB (g/L)  142 (128–157)  145 (125–157)  0.984  143 (127–157)  132 (115–145)  ≤0.001 
RBC (×1012/L)  4.7 (4.2–5.1)  4.7 (4.2–5.1)  0.899  4.7 (4.2–5.1)  4.3 (3.7–4.7)  ≤0.001 
PLT (×109/L)  154.9 (83.7–210.0)  166.0 (86.7–218.0)  0.325  159.7 (84.6–213.0)  135.7 (79.0–193.3)  0.004 
ALT (U/L)  23.1 (16.0–34.1)  25.0 (17.5–33.1)  0.266  23.5 (16.6–33.9)  24.8 (15.9–44.6)  0.009 
AST (U/L)  23.8 (18.8–30.1)  25.2 (19.5–33.3)  0.089  24.1 (19.0–31.7)  29.6 (21.1–51.0)  ≤0.001 
TBIL (μmol/L)  13.3 (9.9–18.7)  13.5 (9.7–18.6)  0.788  13.4 (9.9–18.6)  14.3 (10.1–21.4)  0.023 
ALP (U/L)  72.6 (58.9–90.1)  75.6 (59.6–94.2)  0.697  73.0 (59.1–91.9)  85.7 (65.2–124.1)  ≤0.001 
ALB (g/L)  45.6 (41.0–48.3)  45.4 (40.7–48.7)  0.865  45.6 (41.0–48.4)  41.0 (36.7–45.2)  ≤0.001 

HBV, hepatitis B virus; WBC, white blood cell; PNG, percent of neutrophile granulocyte; HGB, hemoglobin; RBC, red blood cell; PLT, platelet count; ALT, alanine transaminase; AST, aspartate aminotransferase; TBIL, total bilirubin; ALP, alkaline phosphatase; ALB, albumin.

Table 2.

The etiology of patients in validation set without HBV infection.

Clinical diagnosis  N (%) 
Chronic HCV infectionPrimary biliary cirrhosis  10263 
Cryptogenic hepatitis  60 
Alcoholic liver disease  56 
Non-alcoholic fatty liver disease  38 
Autoimmune hepatitis  33 
Drug-induced liver disease  30 
Hepatitis E 
Primary sclerotic cholangitis 
Wilson disease 
Total  386 

HCV, hepatitis C virus.

3.2TE Performance to predict varices in patients with and without chronic HBV infection

Patients with high liver stiffness were associated with high risk of varices, and the value of TE to predict varices was validated by ROC curve. In all 659 CHB patients, the results showed that TE had an area under the ROC curve of 0.84 (95% CI: 0.81–0.87, p<0.0001). The cut-off value was set as 12.5kPa, the sensitivity and the specificity were 72.90% (95% CI: 66.42–78.73%) and 80.04% (95% CI: 76.33–83.4%), respectively. In the 386 non-CHB patients, the area under the ROC curve was 0.78 (95% CI: 0.74–0.83, p<0.0001). The cut-off value was set as 16.2kPa, the sensitivity and the specificity were 70.20% (95% CI: 62.22–77.36%) and 74.03% (95% CI: 68.23–79.27%), respectively (Fig. 2).

Fig. 2.

Liver stiffness for diagnosing GOV in population with or without chronic HBV infection. (A) Liver stiffness was different between GOV positive group and GOV negative group in patients with chronic HBV infection. (B) ROC curve of liver stiffness for diagnosing GOV in population with HBV infection. (C) Liver stiffness was different between GOV positive and GOV negative in CLD patients without HBV infection. (D) ROC curve of liver stiffness for diagnosing GOV in CLD patients without HBV infection.

(0.15MB).
3.3Predictors of GOV from the training set and the predictive performance of new algorithm to GOV

Variables associated with GOV were assessed by multivariate analysis. Table 3 shows that the rough surface of liver (p<0.001), splenic thickness (p<0.001), liver stiffness (p=0.006) were independent predictors of GOV. Based on the three related variables, the final prediction algorithm was developed, which was as follows: Index=−6.56+1.948*rough surface of liver (rough 1, smooth 0)+0.106*splenic thickness (millimeter, mm)+0.028*LS (kPa). The cut-off point was selected according to the ROC curve analyses based on the presence of GOV.

Table 3.

The result of multiple logistic regression.

Variables  Coefficient  S.E.  p value  OR 95% CI 
Rough surface of liver  1.948  0.338  0.000  7.016 (3.618–13.606) 
Splenic thickness (mm)  0.106  0.016  0.000  1.111 (1.077–1.147) 
Liver stiffness (kPa)  0.028  0.01  0.006  1.028 (1.008–1.049) 
Constants  −6.56  0.692  0.000   

Algorithm=Index=−6.56+1.948*rough surface of liver (rough 1, smooth 0)+0.106*splenic thickness (millimeter, mm)+0.028*LS (kPa).

ROC curve analyses were used to evaluate the predictive effect of the algorithm in 220 CHB validation patients. The results showed that the new algorithm demonstrated a high area under the ROC curve of 0.94 (95% CI: 0.899–0.975, p<0.001). When the cut-off point was −1.048, the sensibility to predict GOV was 89.1%, and the specificity was 82.5%.

For further evaluating the predictive performance of new algorithm in non-CHB population, a new algorithm was put forwarded in 386 non-CHB patients. The results showed that the new algorithm had a similar high area under the ROC curves of 0.90 (95% CI: 0.856–0.922, p<0.001; Fig. 3).

Fig. 3.

The predictive performance of the new algorithm for GOV. (A) The model score was different between GOV positive group and GOV negative group in CHB training cohort. (B) ROC curve of the model score for diagnosing GOV in CHB training cohort. (C) The model score was different between GOV positive and GOV negative in CHB validation cohort. (D) ROC curve of the model score for diagnosing GOV in CHB validation cohort. (E) The model score was different between GOV positive and GOV negative in non-CHB validation cohort. (F) The model score was different between GOV positive and GOV negative in non-CHB validation cohort.

(0.34MB).
3.4The predictive performance of new algorithm to variceal hemorrhage

The predictive performance of new algorithm to variceal hemorrhage was evaluated by ROC curve. The results revealed that the algorithm showed a surprisingly high area under the ROC curve of 0.89 (95% CI: 0.84–0.93, p<0.001) in the CHB training set, 0.92 (95% CI: 0.880–0.957, p<0.001) in the CHB validation set and 0.84 (95% CI: 0.796–0.892, p<0.001) in the non-CHB validation set. The cut-off point of 0.093 was chosen, the sensibility to predict variceal hemorrhage was 93.0%, and the specificity was 77.9%, respectively (Fig. 4).

Fig. 4.

The predictive performance of the new algorithm for variceal hemorrhage. (A) The new model score was significantly different between hemorrhage group and non-hemorrhage group in CHB training cohort. (B) ROC curve of the model score for diagnosing variceal hemorrhage in CHB training cohort. (C) The new model score was significantly different between hemorrhage group and non-hemorrhage group in CHB validation cohort. (D) ROC curve of the model score for diagnosing variceal hemorrhage in CHB validation cohort. (E) The new model score was significantly different between hemorrhage group and non-hemorrhage group in non-CHB validation cohort. (F) ROC curve of the model score for diagnosing variceal hemorrhage in non-CHB validation cohort.

(0.33MB).
3.5The predictive performance of new algorithm for predicting varices in patients with compensated cirrhosis

Patients in the compensated stage of cirrhosis are completely asymptomatic, but require endoscopic screening of GOV. Since patients with compensated liver cirrhosis have a low probability of having large esophageal varices, noninvasive predictors are considered more critical. The new algorithm was applied on 204 patients with compensated cirrhosis. The results showed GOV in 104/204 patients, and were small (G1) in 66 and medium/large (G2, G3) in 38 patients. The algorithm showed the area under the ROC curve as 0.76 (95% CI: 0.69–0.83, p<0.001).

3.6The predictive value of other surrogate markers in patients with compensated cirrhosis

In Baveno VI consensus, TS20–25kPa can be used to diagnose CSPH in virus-related CLD, and the diagnostic value of TE in other etiologies should be confirmed [2]. In our study, 110 non-virus related CLDs with TS20kPa were observed, in which 70/110 (63.64%) had GOV (small to large), and 40/110 (36.36%) had no varices at all. In cases with misdiagnosis, 18 patients were considered to have cryptogenic hepatitis, 12 patients were with alcoholic liver disease, 8 patients were with autoimmune liver disease, 1 had non-alcoholic fatty liver disease and the remaining 1 patient had drug-induced liver disease. Four established algorithms, including the evaluation of AAR, FIB-4, APRI and platelet count/spleen diameter ratio in patients with severe varices in compensated liver cirrhosis [20,21], were performed in the compensated liver cirrhosis group. The algorithms showed that the area under the ROC curve as 0.60 (95% CI: 0.52–0.88, p<0.001), 0.54 (95% CI: 0.49–0.70, p=0.082), 0.53 (95% CI: 0.44–0.63, p=0.538) and 0.72 (95% CI: 0.63–0.82, p=0.000), respectively.

4Discussion

GOV occurs as a serious consequence of portal hypertension in patients with CLDs. Portal hypertension mostly occurs as a late manifestation of liver disease. However, some patients with early liver disease have non-cirrhotic hypertension [22]. However, single biomarkers are usually insufficient due to missing high-risk esophageal varices rate, and this exceeds the acceptable limit of 5% as indicated according to the Baveno VI consensus. In our study, we attempted to develop a single algorithm for routinely available results to predict GOV in CHB and non-CHB patients. We found that the rough surface of the liver, splenic thickness and liver stiffness (LS) as independent predictors of GOV. We hypothesized that the new algorithm that consisted of the above 3 variables was better in comparison to other predictors.

According to the previous report, the liver stiffness-spleen size-to-platelet ratio score (LSPS), which is a combination of 3 simple examination methods (LS, spleen size, and platelet count), predicted esophageal varices (EV) and high risk EV in patients with compensated cirrhosis. The area under ROC value of LSPS was 0.918, and more than 80% of the patients were accurately classified [23]. The diagnostic and prognostic abilities of LSPS were assessed in another study, and the results revealed that LSPS showed significant correlation with EV grade (p<0.001). It was independently associated with the presence of EV (p<0.001) and elevated EV risk (p=0.013) by multivariate logistic regression analysis [24]. Our new algorithm has similar high predictive value with an excellent area under the ROC curve of 0.94 (95% CI: 0.899–0.975, p<0.001). When applying the best cut-off point of the curve, the sensibility to predict GOV was 89.1%, and the specificity was 82.5%. This was used in CHB validation, non-CHB validation and compensated liver cirrhosis patients, and the algorithm still predicted GOV with a high area under ROC. Other previous studies used noninvasive markers such as serum albumin [7], platelets count [21,25], PVD [21], spleen size [21,26], platelet/spleen diameter ratio [27], and bilirubin [28] for investigation. However, the results showed that a low area under ROC value of 0.65–0.82 when compared with our study of 0.94. Meanwhile, there are few research studies that focused on chronic HBV infected patients only.

Baveno VI consensus recommended that combination of LS (<20kPa) and platelet count (>150×109/L) accurately identified patients who were at very low risk of having clinically significant varices and avoiding screening endoscopy [2]. The LS was considered to be a good single noninvasive variable for identifying patients with CSPH. For predicting the presence of GOV, the new algorithm demonstrated a better diagnostic performance, and the combination of parameters showed superior performance of LS alone for predicting GOV. Our study included a sufficient number of patients and 2 validation sets. The variables selected in the algorithm are easily collected and applicable, and accurately predicted GOV in CHB and non-CHB patients, making it a novel way for patients who need endoscopy. Meanwhile, our new algorithm had a fine diagnostic performance for predicting varices in compensated liver cirrhosis patients, and was better than AAR, FIB-4, APRI and platelet count/spleen diameter ratio that are used to evaluate severe varices in patients with compensated liver cirrhosis.

One study used non-invasive markers to assess the predictive value of esophageal varices and variceal bleeding in Albanian patients diagnosed with liver cirrhosis. A total of 139 newly diagnosed cirrhotic patients without variceal bleeding were included, and the major end point was the first esophageal variceal bleeding event. The results revealed that FIB-4 was the strong and significant predictor of esophageal varices (OR=1.57; 95% CI: 1.15–2.14), but was a poor predictor of EVB (area under the ROC only 0.51) [29]. In contrast, none of the non-invasive markers, such as AST/ALT, APRI, PC/SD, FI, MELD and King's score were considered as useful predictors of EVB in this study. Surprisingly, we found that the new algorithm had a satisfactory value to predict variceal hemorrhage, in which the area under the ROC was 0.92 in CHB patients and 0.84 in other CLDs. Xavier et al. have assessed the prognostic performance of Albumin–Bilirubin, Child–Pugh, and model for end-stage liver disease in patients with liver cirrhosis complicated with acute upper gastrointestinal bleeding. They found that Albumin–Bilirubin score has accurately predicted both in-stay and 30-day mortality, with an area under the curve of 0.80, suggesting that the Albumin–Bilirubin score was particularly useful in assessing short-term outcomes inpatients with liver cirrhosis complicated with acute upper gastrointestinal bleeding [30]. The new algorithm and the Albumin–Bilirubin score demonstrated similar area under the ROC. However, further evaluation on validation set should be performed to confirm the value.

LS is considered as a useful marker for evaluating whether a patient has progressed into GOV based on the Baveno VI guidelines. However, LS is only applicable to CLD patients with HBV or HCV infection. For CLD patients with other etiologies, the clinical performance of LS still requires elucidation. Consequently, patients with CHB were chosen for the training set, and other CLD patients for the validation set. Our present data indicated that the diagnostic algorithm of varices for CHB is also applicable to other CLD patients.

However, this study has some potential limitations. Firstly, CHB patients were chosen for the training set due to their high prevalence in China. More importantly, there was more evidence regarding the predictive value of TE in CHB patients according to the Baveno Consensus. Secondly, the severity of GOV was not classified, and only one cut-off point was chosen to distinguish GOV positive or not, lacking the accurate prediction of whether a patient belongs to high-risk group. Thirdly, the use of TE has several limitations, including high BMI and/or central obesity, necro-inflammatory activity, and extrahepatic cholestasis, and so a low threshold of ALT<60U/L was set, impacting the results toward the more compensated population, and the algorithm might not be suitable in all the patients [31,32].

5Conclusion

The new algorithm based on LS is a reliable model to prognosticate GOV and variceal hemorrhage, and used to stratify not only the high-risk CHB patients, but also to patients with other CLDs for developing GOV and variceal bleeding.AbbreviationsLS

liver stiffness

CHB

chronic hepatitis B

HBV

hepatitis B virus

GOV

gastroesophageal varices

PVD

portal vein diameter

ST

splenic thickness

PLT

platelet count

PNG

percent of neutrophil granulocyte

CSPH

clinically significant portal hypertension

CT

computed tomography

TE

transient elastography

ROC

receiver operating characteristic curve

ALT

alanine transaminase

LSPS

spleen size-to-platelet ratio score

EV

esophageal varices

PHT

portal hypertension

EVB

esophageal varices bleeding

AST

aspartate transaminase

CHD

chronic liver disease

Ethical statement

Ethics approval and consent to participate: Our retrospective study was approved by the Clinical Ethics Committees of Beijing Ditan Hospital, Capital Medical University (JDLKZ2018-021-01). All patients provided written informed consents in accordance with the Declaration of Helsinki. Availability of data and material: The datasets generated and/or analyzed during the current study are available in the Springer Nature (https://figshare.com/articles/raw_data_xls/6168803).

Funding

Grants to conduct this study were provided by the Science Foundation of Capital Medicine Development (2014-2-2171); Capital Foundation for Clinical Characteristic Applied Research Projects (Z141107002514132) and Capital Foundation for Clinical Characteristic Applied Research Projects (Z181100001718084); Beijing Natural Science Foundation (7202071); The Digestive Medical Coordinated Development Center of Beijing Municipal Administration of Hospitals (XXZ0404) to Prof. Hongshan Wei.

Authors’ contributions

WH designed the present study; HL, YJ, YX and YJ collected the clinical data; MJ, LP, HJ, ZY and LX performed statistical analysis; MJ was a major contributor in writing the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

We acknowledge Dr. Zhenglin Ai for helping statistics data, and other colleagues for clinical data collection.

References
[1]
N.D. Grace.
Prevention of initial variceal hemorrhage.
Gastroenterol Clin North Am, 21 (1992), pp. 149-161
[2]
R. de Franchis.
Baveno VI Faculty Expanding consensus in portal hypertension: report of the Baveno VI Consensus Workshop: stratifying risk and individualizing care for portal hypertension.
J Hepatol, 63 (2015), pp. 743-752
[3]
J.A. Flemming, V. Saxena, H. Shen, N.A. Terrault, C. Rongey.
Facility- and patient-level factors associated with esophageal variceal screening in the USA.
Dig Dis Sci, 61 (2016), pp. 62-69
[4]
C.A. Mayorga, D.C. Rockey.
Clinical utility of a standardized electronic order set for the management of acute upper gastrointestinal hemorrhage in patients with cirrhosis.
Clin Gastroenterol Hepatol, 11 (2013), pp. 1342-1348
[5]
T.Y. Kim, W.K. Jeong, J.H. Sohn, J. Kim, M.Y. Kim, Y. Kim.
Evaluation of portal hypertension by real-time shear wave elastography in cirrhotic patients.
Liver Int, 35 (2015), pp. 2416-2424
[6]
A.K. Pillai, B. Andring, A. Patel, C. Trimmer, S.P. Kalva.
Portal hypertension: a review of portosystemic collateral pathways and endovascular interventions.
Clin Radiol, 70 (2015), pp. 1047-1059
[7]
S. Bhattarai, K.R. Dewan, G. Shrestha, B.S. Patowary.
Non-invasive predictors of gastro-oesophageal varices.
JNMA J Nepal Med Assoc, 56 (2017), pp. 298-303
[8]
A. Kapoor, N. Dharel, A.J. Sanyal.
Endoscopic diagnosis and therapy in gastroesophageal variceal bleeding.
Gastrointest Endosc Clin N Am, 25 (2015), pp. 491-507
[9]
A.Z. Ginai, H.R. van Buuren, W.C. Hop, S.W. Schalm.
Oesophageal varices: how reliable is a barium swallow?.
Br J Radiol, 66 (1993), pp. 322-326
[10]
Y.J. Tseng, X.Q. Zeng, J. Chen, N. Li, O.J. Xu, S.Y. Chen.
Computed tomography in evaluating gastroesophageal varices in patients with portal hypertension: a meta-analysis.
Dig Liver Dis, 48 (2016), pp. 695-702
[11]
S. Singh, A.J. Muir, D.T. Dieterich, Y.T. Falck-Ytter.
American Gastroenterological Association Institute technical review on the role of elastography in chronic liver diseases.
Gastroenterology, 152 (2017), pp. 1544-1577
[12]
H. Perazzo, F.F. Fernandes, E.C. Castro Filho, R.M. Perez.
Points to be considered when using transient elastography for diagnosis of portal hypertension according to the Baveno's VI consensus.
J Hepatol, 63 (2015), pp. 1048-1049
[13]
P. Calès, S. Sacher-Huvelin, D. Valla, C. Bureau, A. Olivier, F. Oberti, et al.
Large oesophageal varice screening by a sequential algorithm using a cirrhosis blood test and optionally capsule endoscopy.
Liver Int, 38 (2018), pp. 84-93
[14]
C. Jansen, C. Bogs, W. Verlinden, M. Thiele, P. Mōller, J. Gōrtzen, et al.
Algorithm to rule out clinically significant portal hypertension combining Shear-wave elastography of liver and spleen: a prospective multicentre study.
[15]
N.A. Terrault, A.S.F. Lok, B.J. McMahon, K.M. Chang, J.P. Hwang, M.M. Jonas, et al.
Update on prevention, diagnosis, and treatment of chronic hepatitis B: AASLD 2018 hepatitis B guidance.
Hepatology, 67 (2018), pp. 1560-1599
[16]
B.J. McMahon.
Epidemiology and natural history of hepatitis B.
Semin Liver Dis, 25 (2005), pp. 3-8
[17]
G. Fattovich, F. Bortolotti, F. Donato.
Natural history of chronic hepatitis B: special emphasis on disease progression and prognostic factors.
J Hepatol, 48 (2008), pp. 335-352
[18]
Chinese Society of Hepatology, Chinese Medical Association, Chinese Society of Infectious Diseases, Chinese Medical Association, J.L. Hou, W. Lai.
The guideline of prevention and treatment for chronic hepatitis B: a 2015 update.
Zhonghua Gan Zang Bing Za Zhi, 23 (2015), pp. 888-905
[19]
Y.D. Kim.
Management of acute variceal bleeding.
Clin Endosc, 47 (2014), pp. 308-314
[20]
Y. Iwata, H. Enomoto, Y. Sakai, N. Aizawa, H. Tanaka, N. Ikeda, et al.
Evaluation of the AST to ALT ratio in association with the severity of esophageal varices in patients with HCV-related compensated liver cirrhosis.
Hepato Gastroenterol, 60 (2013), pp. 149-152
[21]
E. Giannini, F. Botta, P. Borro, D. Risso, P. Romagnoli, A. Fasoli, et al.
Platelet count/spleen diameter ratio: proposal and validation of a non-invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis.
Gut, 52 (2003), pp. 1200-1205
[22]
R. Khanna, S.K. Sarin.
Non-cirrhotic portal hypertension – diagnosis and management.
J Hepatol, 60 (2014), pp. 421-441
[23]
A. Berzigotti, S. Seijo, U. Arena, J.G. Abraldes, F. Vizzutti, J.C. García-Pagán, et al.
Elastography, spleen size, and platelet count identify portal hypertension in patients with compensated cirrhosis.
Gastroenterology, 144 (2013), pp. 102-111
[24]
S. Shibata, S. Joshita, T. Umemura, T. Yamaaki, N. Fujimori, Y. Ichikawa, et al.
Liver stiffness-spleen size-to-platelet ratio risk score detects esophageal varices in chronic liver disease.
Spingerplus, 5 (2016), pp. 998
[25]
M.R. Adami, C.O. Kieling, F.P. Schwengber, V.N. Hirakata, S.M.G. Vieira.
Noninvasive method of predicting large esophageal varices in children with intrahepatic portal hypertension.
World J Gastroenterol, 19 (2013), pp. 2053-2059
[26]
P. Witters, D. Hughes, P. Karthikeyan, S. Ramakrishna, M. Davenport, A. Dhawan, et al.
King's variceal prediction score: a novel noninvasive marker of portal hypertension in pediatric chronic liver disease.
J Pediatr Gastroenterol Nutr, 64 (2017), pp. 518-523
[27]
H.Y. Zhou, T.W. Chen, X.M. Zhang, L.Y. Wang, L. Zhou, et al.
The diameter of the originating vein determines esophageal and gastric fundic varices in portal hypertension secondary to posthepatitic cirrhosis.
Clinics (Sao Paulo), 67 (2012), pp. 609-614
[28]
B. Bressler, R. Pinto, D. Ei-Ashry, E.J. Heathcote.
Which patients with primary biliary cirrhosis or primary sclerosing cholangitis should undergo endoscopic screening for oesophageal varices detection.
[29]
B. Kraja, I. Mone, I. Akshija, A. Kocollari, S. Prifti, G. Burazeri.
Predictors of esophageal varices and first variceal bleeding in liver cirrhosis patients.
World J Gastroenterol, 23 (2017), pp. 4806-4814
[30]
S.A. Xavier, R. Vilas-Boas, P. Boal Carvalho, J.T. Magalhаes, C.M. Marinho, J.B. Cotter.
Assessment of prognostic performance of Albumin-Bilirubin Child-Pugh, and Model for End-stage Liver Disease score in patients with liver cirrhosis complicated with acute upper gastrointestinal bleeding.
Eur J Gastroenterol Hepatol, 30 (2018), pp. 652-658
[31]
P. Stål.
Liver fibrosis in non-alcoholic fatty liver disease-diagnostic challenge with prognostic significance.
World J Gastroenterol, 21 (2016), pp. 11077-11087
[32]
R. Huang, N. Jiang, R. Yang, X. Geng, J. Lin, G. Xu, et al.
Fibroscan improves the diagnosis sensitivity of liver fibrosis in patients with chronic hepatitis B.
Exp Ther Med, 11 (2016), pp. 1673-1677
Copyright © 2020. Fundación Clínica Médica Sur, A.C.
Article options
Tools
es en pt

¿Es usted profesional sanitario apto para prescribir o dispensar medicamentos?

Are you a health professional able to prescribe or dispense drugs?

Você é um profissional de saúde habilitado a prescrever ou dispensar medicamentos

es en pt
Política de cookies Cookies policy Política de cookies
Utilizamos cookies propias y de terceros para mejorar nuestros servicios y mostrarle publicidad relacionada con sus preferencias mediante el análisis de sus hábitos de navegación. Si continua navegando, consideramos que acepta su uso. Puede cambiar la configuración u obtener más información aquí. To improve our services and products, we use "cookies" (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here. Utilizamos cookies próprios e de terceiros para melhorar nossos serviços e mostrar publicidade relacionada às suas preferências, analisando seus hábitos de navegação. Se continuar a navegar, consideramos que aceita o seu uso. Você pode alterar a configuração ou obter mais informações aqui.