Volume 6 Issue 3, October 2023, pp. pp. 332-346


Individuals infected with hepatitis B (HBV), hepatitis C (HCV), and human immunodeficiency (HIV) viruses can experience compensated advanced chronic liver disease (cACLD) leading to esophageal varices (EV). In patients at low risk of esophageal varices needing treatment (EVNT), non-invasive criteria based on liver stiffness measurement (LSM) with platelets, or fibrosis biomarkers, may avoid unnecessary screening esophagogastroduodenoscopies (EGD). These approaches have not been compared among people infected with HIV, HBV, and HCV patients.

Patients with a diagnosis of cACLD (LSM ≥10 kPa) and EGD availability were included from two cohorts. Baveno VI and expanded Baveno VI criteria (based on LSM and platelets), fibrosis biomarkers Fibrosis-4 Index (FIB-4), AST-to-Platelets Ratio Index (APRI), AST-to-ALT ratio (AAR), and RESIST criteria (based on platelets and albumin) were applied to determine the proportion of spared EGD and of missed EVNT.

Three hundred fifty three patients (30.6% with HIV, 25.3% monoinfected with HBV, and 44.1% with HCV) were included. The prevalence of EVNT was 8.2%. Both Baveno VI and expanded Baveno VI criteria performed well in patients with virus-related cACLD, by sparing 26.1% and 51.6% EGD, respectively, while missing <2% EVNT. The proportion of spared EGD were 48.2%, 58%, and 24.3% by FIB-4 (<2.78), APRI (<1.1), and AAR (<0.75), respectively, while missing <3% EVNT. RESIST criteria spared 47.8% EGD while missing 1.9% EVNT.

Non-invasive criteria based on LSM can spare unnecessary EGD in virus-related cACLD. Simple fibrosis biomarkers can ameliorate resource utilization for EVNT screening in low resource settings.

Patients with advanced liver scarring due to viral hepatitis viruses and HIV are at risk for virus-related advanced chronic liver disease (cACLD) and for enlarged veins in the esophagus that can leak blood or even rupture, causing life-threatening bleeding. The usual test to diagnose enlarged veins in the esophagus is the endoscopy to view the interior of the esophagus and the stomach. Recently, non-invasive tests including a blood test to measure platelets and liver enzymes, and the measure of the stiffness of the liver through an ultrasound-based machine can reduce the number of invasive endoscopies to look for enlarged veins in the esophagus caused by advanced liver scarring. In 353 patients with liver scarring due to viral hepatitis or HIV from two Canadian studies, we found that these non-invasive tests can spare one-fourth to half invasive endoscopy procedures to diagnose the enlarged veins in the esophagus.

Chronic infections with hepatitis B (HBV) and hepatitis C (HCV) viruses represent a major cause of compensated advanced chronic liver disease (cACLD) leading to end-stage liver disease (1). People mono-infected with the human immunodeficiency virus (HIV) are also at high risk for cACLD due to emerging risk of nonalcoholic fatty liver disease (NAFLD) related to frequent metabolic imbalance and lifelong use of antiretroviral therapy (24). The development of clinically significant portal hypertension and esophageal varices (EV) impacts on prognosis of cACLD and is associated with increased mortality (1,5). The diagnosis of EV, and especially large varices (grade 2/3) requiring primary prophylaxis (EV needing treatment, EVNT), is of paramount prognostic importance in patients with ACLD (6,7). Esophagogastroduodenoscopy (EGD) remains the gold standard to diagnose EV in cACLD. However, EVNT are not frequent in patients with cACLD, and universal screening is therefore unnecessarily invasive given limited yield for a large proportion of patients. Additionally, patients with cACLD of viral etiology are more likely to be followed in lower resource settings than other etiologies of liver disease like nonalcoholic steatohepatitis or autoimmune diseases (8). Access to EGD can be problematic in the context of persons with poor linkage to care having cACLD, as well as in low resource countries that also experience the highest rates of HIV and HBV positivity (9,10). Strategies to reduce referral to hepato-gastroenterology units for unnecessary EGD screening are therefore crucial in this setting.

The Baveno VI criteria proposed that patients with cACLD with a liver stiffness measurement (LSM) by transient elastography <20 kPa and a platelet count >150,000/μL can avoid screening EGD. The Baveno VII guidelines also proposed to use the Baveno VI noninvasive criteria (1,11). Expanded Baveno VI criteria, obtained by optimizing LSM and platelets thresholds, have been proposed to spare an even higher proportion of unnecessary EGD when compared to the Baveno VI criteria (11,12). Simple fibrosis biomarkers have also been proposed for non-invasive diagnosis of EVNT, including the fibrosis-4 index (FIB-4), AST-to-Platelets Ratio Index (APRI), and AST-to-ALT ratio (AAR), but few studies compared their performance to LSM-based criteria in virus-related cACLD other than HCV (13,14). In settings where LSM is not available, the RESIST criteria were also proposed in HCV, based on platelets and albumin values (15). All these non-invasive criteria have not been compared to serum biomarkers and across various etiologies of virus-related cACLD, including chronic HBV and HCV infections and people with HIV at risk for NAFLD (1622). Moreover, the pathogenesis of liver fibrosis portending the development of cACLD and portal hypertension involves a number of discrete mechanisms unique to each etiology of virus-related cACLD (23).

We used data from two prospective/retrospective Canadian cohorts, the Canadian Hepatitis B Network and the LIVEr disease in HIV (LIVEHIV), to fulfill the following aims: (1) validate the Baveno VI and expanded Baveno VI criteria, based on platelets and LSM, to identify patients at low risk of EVNT in virus-related cACLD; (2) compare those criteria to simple fibrosis biomarkers; and (3) conduct a cost-analysis of these non-invasive strategies compared to the universal use of the gold standard EGD.

Study design and population

This was a cross-sectional study of two longitudinal cohorts. The Canadian HBV Network is an interprovincial network of 21 tertiary care-based clinical cohorts providing care for HBV-infected patients. Data were analyzed from six hepatology clinics or infectious disease clinics across Canada (British Columbia, Alberta, Ontario, Quebec, Nova Scotia). Each site conducted data collection by reviewing electronic and paper patient charts. Deidentified information was entered into a registry database housed at the University of Calgary, Canada. The LIVEHIV cohort is a prospective screening program for liver disease running at the McGill University Health Centre, Montreal, Canada. Patients are regularly followed up by their treating physician for their HIV care every 6 months. Moreover, they are consecutively screened for HCV, HBV virus serology, and they undergo annual transient elastography measurement using the FibroScan® (EchoSens, Paris, France). For the present study, we included patients between January 2013 and March 2022 who met the following criteria: (1) age ≥18 years; (2) diagnosis of cACLD (LSM ≥10 kPa) (1); (3) availability of EGD within 1 year of LSM and of relevant biochemical parameters to compute the simple fibrosis biomarkers. Exclusion criteria included: patients coinfected with multiple hepatitis viruses, alcohol intake >20 g/day during the previous year (evaluated by interview of patients on amount, frequency and type), previous decompensating events from liver cirrhosis (Child-Pugh B and C), hepatocellular carcinoma, liver transplantation, previous prophylactic EV banding, portal or splenic vein thrombosis, non-cirrhotic portal hypertension, and splenectomy. The combined study cohort included: 109 patients from the Canadian Hepatitis B Network cohort and 244 patients from the LIVEHIV cohort (104 with HIV and 140 with HCV).

Patient consent statement

The study was conducted in accordance with the principles of the Declaration of Helsinki. For the Canadian Hepatitis B Network cohort, eligible participants either provided consent or were included with a waiver of consent, based on each site's local REB approval. All data received were anonymous and collected under an approved University of Calgary Conjoint Ethics Research Board (CHREB) approved protocol (Ethics ID# REB16-0041) and analyzed under CHREB Ethics ID# REB17-2329. The LIVEHIV cohort was approved by the Research Ethics Board of the Research Institute of the McGill University Health Centre (code 14-182-BMD). Participants in the LIVEHIV cohort provided written informed consent before enrollment.

Clinical and biological parameters

Data collected at the time of enrollment and within 6 months from LSM included demographic information, medication history, liver biochemistries, hematological, and virological parameters. EGD was performed by a small number of experienced operators at each Hospital. At endoscopy, high-risk EV warranting primary prophylaxis against EV bleeding (EVNT) were defined by medium or large size (grade 2/3) or the presence of high-risk stigmata findings (red wale marks, cherry red spots) (11). Small varices were defined as small size varices (grade 1) without high-risk stigmata.

Transient elastography and serum fibrosis biomarkers

Transient elastography was performed with the FibroScan® medical device, using the M probe. When available, the XL probe was used in cases of failure of transient elastography with the M probe. In each center, LSM was assessed after at least 4 hours fasting, by a trained operator who had previously performed at least 300 examinations. The following criteria were applied to define the result of transient elastography as reliable: at least 10 valid LSM, and an interquartile range (IQR) <30% of the median LSM (24). Criteria based on LSM and platelets (11,12) and the simple fibrosis biomarkers FIB-4 (25,26), APRI (27) and AAR (28) were applied to identify patients not requiring screening EGD. The recently proposed RESIST criteria based on platelets and albumin were also evaluated (15).

Outcome measures

As primary outcome, we validated the diagnostic accuracy of non-invasive tests to identify low risk patients not requiring screening EGD as per Baveno VI (favorable status: LSM <20 kPa and platelets >150,000) (11), expanded Baveno VI (favorable status: LSM <25 kPa and platelets >110,000) (11,12), FIB-4 (favorable status: <2.78), APRI (favorable status: <1.0), AAR (favorable status: <0.75) (13,29), and RESIST criteria (favorable status: albumin >36 g/L and platelets >120,000) (15).

Cost analysis

Potential direct cost savings were estimated using Canadian data. As previously reported (30,31), costs of non-invasive tests in Canadian dollars were as follows: FIB-4 and APRI: $17; AST:ALT ratio: $10; RESIST criteria: $12; TE: $125. As for the EGD, costs were as follows: direct cost: $281.95 (including nursing equipment $151.08 in 2015, inflated to $167.96 (32); and cost of EGD $113.99); physician minor consult ($105.10); indirect costs: $31.58/h average wage; 8 hours for patient, 4 hours for caregiver ($378.95), for a total cost of $766 per EGD.

Statistical analysis

Performance evaluations were made in terms of percentage of spared endoscopies, percentage of missed EVNT, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive and negative likelihood ratios (LR). Confidence intervals (CIs) for sensitivity, specificity were “exact” Clopper-Pearson CI (33). CI for predictive values were the standard logit CI given by Mercaldo et al. (34). Optimized cut-offs for the simple fibrosis biomarkers were established by using the area under the curve (AUC) analysis. The modelling maximized the absolute number of spared endoscopies while keeping the risk of missed EVNT below 5% (or, equivalently, while constraining the NPV to be at least 95%). The 5% false negative rate of undetected EVNT was agreed as a reasonable criterion by experts in the Baveno VII consensus conference and adopted by the American Gastroenterological Association and several other authors (1,11,35). Correlation coefficients among LSM, FIB-4, APRI, and AAR were calculated using the Pearson correlation analysis. A concordance analysis between each of the non-invasive criteria and EGD was carried out using the kappa score, with results interpreted as follows: less than 0, less than chance agreement; 0.01–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement and 0.81–0.99, almost perfect agreement (36). A two-sided level of significance of 5% was used for all statistical inferences. Potential direct cost savings were estimated using Canadian data. Statistical analysis was performed using STATA 17 (STATA Corp. LP, College Station, Texas, USA).

Patient characteristics

The baseline characteristics of the 353 patients with virus-related cACLD by etiology category are shown in Table 1. In the whole cohort, mean age was 55.3 years (SD 11.2), 28.0% were females, mean model for end-stage liver disease (MELD) score was 7.9 (SD 3.9), mean LSM was 21.0 kPa (SD 13.8), mean platelets were 150.7 × 109/L (SD 66.5). All patients were Child-Pugh Class A. The M probe was used in most of the cases (97.5%). Overall, EVNT and small varices were present in 8.2% and 21.2% of cases, respectively. LSM was significantly higher in patients with HBV and HCV compared to those with HIV. In the HBV subgroup, 36.8% of patients were HBeAg positive, 32% had undetected or very low detectable HBV viral load (<20 IU), and 82% of patients were receiving antiviral therapy for HBV. In the HIV subgroup, mean CD4 cell count was 578 (SD 355), 68% had undetectable HIV viral load (≤50 copies), all patients were on antiretroviral therapy (82% on nucleoside reverse transcriptase inhibitors, 32% on non-nucleoside reverse transcriptase inhibitors, 47% on protease inhibitors, 33% on integrase inhibitors). In the HCV subgroup, most patients (85%) had received antiviral therapy with direct antiviral agents and achieved sustained virological response (undetectable HCV RNA). In our cohort, 85 patients (24.1%) had LSM >25, which would meet the Baveno VII diagnostic criteria to rule-in clinically significant portal hypertension. Among these patients, 19 (22.3%) had EVNT.


Table 1: Baseline demographic, metabolic, laboratory, and instrumental features of patients with virus-related cACLD by etiology category (n = 353)

Table 1: Baseline demographic, metabolic, laboratory, and instrumental features of patients with virus-related cACLD by etiology category (n = 353)

HBV (n = 109) HIV (n = 104) HCV (n = 140)
Age (years) 57.2 (12.7) 51.9 (7.4) 56.5 (11.7)
Female (%) 33 (30.3) 25 (24.0) 41 (29.3)
Ethnicity (%)
 White 13 (11.9) 83 (79.9) 96 (68.6)
 Black 18 (16.5) 12 (11.5) 18 (12.8)
 South East Asian 31 (28.5) 1 (1.0) 4 (2.9)
 Other Asian 43 (39.4) 4 (3.8) 0
 Spanish 0 2 (1.9) 14 (10.0)
 Middle Eastern 4 (3.7) 0 8 (5.7)
 Other 0 2 (1.9) 0
Transmission (%)
 IDU 2 (1.8) 42 (40.4) 35 (25)
 Vertical 9 (8.3) 0 3 (2.2)
 Sexual 2 (1.8) 52 (59.6) 5 (3.6)
 Iatrogenic 5 (4.6) 0 24 (17.1)
 Unknown 91 (83.5) 0 73 (52.1)
Duration of viral infection (years) 17.2 (15.1) 20.5 (9.9) 14.3 (8.1)
INR 1.10 (0.17) 1.08 (0.18) 1.10 (0.17)
Platelets (109/L) 137.9 (62.4) 162.1 (69.9) 152.4 (65.7)
ALT (IU/L) 57.9 (131.9) 70.4 (60.8) 91.3 (51.9)
AST (IU/L) 52.3 (74.5) 64.1 (42.4) 83.1 (49.9)
Total bilirubin (μmol/L) 18.5 (25.9) 16.7 (16.3) 16.7 (9.9)
Albumin (g/L) 39.4 (5.3) 40.5 (6.8) 39.1 (4.2)
MELD sodium 9.8 (4.1) 6.4 (3.2) 7.3 (3.6)
LSM (kPa) 22.7 (16.5) 16.3 (8.6) 22.1 (12.3)
APRI 1.63 (3.62) 1.54 (1.60) 2.3 (2.4)
FIB-4 4.26 (6.79) 3.50 (4.41) 4.71 (4.84)
AAR 1.10 (0.46) 1.11 (0.60) 0.96 (0.34)
EVNT (%) 11 (10.1) 5 (4.8) 13 (9.3)
Small EV (%) 26 (23.9) 17 (16.3) 32 (22.9)

Note: Continuous variables are expressed as mean (standard deviation) and categorical variables are expressed as frequencies (%).

AAR = AST-to-ALT ratio; ALT = alanine aminotransferase; APRI = AST-to-Platelet ratio index; AST = aspartate aminotransferase; BMI = body mass index; CAP = controlled attenuation, parameter; EV = esophageal varices; EVNT = esophageal varices needing treatment; FIB-4 = fibrosis-4 index; HBV = hepatitis B virus; HCV = hepatitis C virus; HIV = human immunodeficiency virus; IDU = injection drug use; INR = international normalized ratio; IU = international units; LSM = liver stiffness measurement; MELD = model for end-stage liver disease.

Diagnostic accuracy of non-invasive criteria to diagnose EVNT

Figure 1 summarizes the proportion of spared EGD and of missed EVNT according to the various non-invasive criteria in the whole cohort of patients with virus-related cACLD. Across etiologies, the AUC of FIB-4, APRI, and AAR for the diagnosis of EVNT were 0.784 (95% CI 0.691–0.862), 0.818 (95% CI 0.728–0.889), and 0.518 (95% CI 0.413–0.617), respectively. AAR had a significant lower overall performance by AUC compared to FIB-4 and APRI (p < 0.001). The optimized cut-offs for the simple fibrosis biomarkers based on the AUC analysis were 2.78 for FIB-4, 1.1 for APRI, and 0.75 for AAR. The diagnostic performance of the non-invasive criteria to diagnose EVNT by etiology category of cACLD is shown in Table 2. In the HBV subgroup, LSM based criteria had favorable status and thus spared EGD in 27.5–50.5% of cases. No EVNTs were missed with the application of Baveno VI and 1.9% were missed with expanded Baveno VI, while NPV was 100% and 96.3%, respectively. Among the serum biomarkers, FIB-4, APRI, and RESIST showed similar performance, with 47.7–59.6% spared EGD. No EVNTs were missed with the application of the RESIST criteria, while 0.9% were missed with FIB-4 and APRI. The NPV was 100–98.1% for these fibrosis biomarkers. AAR had lower performance compared to the other biomarkers. The LR of the various non-invasive criteria to diagnose EVNT in HBV are reported in Figure 2a. In the HIV subgroup, LSM based criteria had favorable status and thus spared EGD in 41.3–64.4% of cases. Missed EVNT were 1% for both Baveno VI and expanded Baveno VI, while NPV was 97.5% and 98.5%, respectively. Among the serum biomarkers, FIB-4, APRI, and RESIST showed similar performance, with 47.1–64.4% spared EGD. Also, the proportion of EVNTs missed by FIB-4, APRI, and the RESIST criteria were similar and lower than 2%, while the NPV range was 96.4–98.5%. AAR had lower performance compared to the other biomarkers. The LR of the various non-invasive criteria to diagnose EVNT in HIV is reported in Figure 2b. In the HCV subgroup, LSM based criteria had favorable status and thus spared EGD in 18.6–42.9% of cases. Missed EVNT were 1.2% and 2.5% for Baveno VI and expanded Baveno VI, while NPV was 92.3% and 98.4%, respectively. Among the serum biomarkers, FIB-4, APRI, and RESIST showed similar performance, with 37.1–60.0% spared EGD. Also, the proportion of EVNT missed by FIB-4 and APRI was the same at 0.9%, while the RESIST criteria and AAR missed 3.5% and 4% EVNT, respectively. The NPV for the simple fibrosis biomarkers ranged between 86.7% and 92.4%. LR of the various non-invasive criteria to diagnose EVNT in HCV are reported in Figure 2c. We also conducted a sensitivity analysis in the subgroup of 46 patients (32.9%) with untreated HCV and found no difference compared to patients with treated HCV (data not shown). Overall, the rate of missed EVNT was less than 5% in all subgroup etiologies of virus-related cACLD. The concordance analysis showed slight to fair agreement between non-invasive criteria and EGD (Table 2). Correlation analysis was significant among APRI, FIB-4, and AAR across the three virus-related etiologies of cACLD. Interestingly, the correlation between LSM and serum fibrosis biomarkers was significant only in HCV (Figure 3).

Figure 1: Proportion of patients with no varices and EVNT in the study cohort (n = 353) according to favorable status of non-invasive criteria

Figure 2: Likelihood ratios of non-invasive criteria: a) in HBV subgroup; b) in HIV subgroup; c) in HCV subgroup

Figure 3: Correlation matrix with heatmap based on correlation coefficient analysis


Table 2: Performance of non-invasive criteria for prediction of EVNT by etiology category (n = 353)

Table 2: Performance of non-invasive criteria for prediction of EVNT by etiology category (n = 353)

Sensitivity (%) Specificity (%) NPV (%) PPV (%) Spared EGD (%) EVNT missed (%) Kappa coefficient
HBV (n = 109)
Baveno VI 100 (71.5–100) 30.5 (21.5–40.8) 100 14.3 (12.7-16.0) 27.5 0 0.084
Extended Baveno VI 81.8 (48.2–97.7) 54.7 (44.2–65.0) 96.3 (88.0–98.9) 17.3 (12.8–23.0) 50.5 1.9 0.138
FIB-4 90.9 (58.7–99.8) 53.7 (43.2–64.0) 98.1 (88.6–99.7) 18.5 (14.6–23.2) 48.6 0.9 0.185
APRI 90.9 (58.7–99.8) 66.3 (55.9–75.7) 98.4 (90.6–99.8) 23.8 (18.2–30.5) 59.6 0.9 0.264
AAR 81.2 (48.2–97.7) 15.8 (9.1–24.7) 88.2 (66.3–96.6) 10.1 (7.8–13.1) 16.5 1.9 0.021
RESIST 100 (71.5–100) 53.2 (42.6–63.6) 100 20 (16.8–23.7) 47.7 0 0.192
HIV (n = 104)
Baveno VI 80.0 (23.4–99.5) 42.4 (32.6–52.8) 97.7 (87.8–99.6) 6.6 (4.2–10.1) 41.3 1.0 0.035
Extended Baveno VI 80.0 (23.4–99.5) 66.7 (56.5–75.8) 98.5 (91.9–99.7) 10.8 (6.7–16.9) 64.4 1.0 0.116
FIB-4 80.0 (23.4–99.5) 48.5 (38.3–58.8) 98.0 (89.2–99.6) 7.3 (4.6–11.2) 47.1 1.0 0.071
APRI 60.0 (14.7–94.7) 53.5 (43.2–63.6) 96.4 (89.9–98.8) 6.1 (3.0–12.1) 52.9 1.9 0.071
AAR 60.0 (14.7–94.7) 20.2 (12.8–29.5) 90.9 (76.1–96.9) 3.7 (1.8–7.3) 22.1 1.9 -0.050
RESIST 80.0 (28.4–99.5) 67.4 (57.0–76.6) 98.5 (91.7–99.7) 11.4 (7.1–17.9) 64.4 1 0.123
HCV (n = 140)
Baveno VI 96.2 (80.4–99.9) 21.8 (11.8–35.0) 92.3 (72.2–98.9) 36.7 (33.1–40.5) 18.6 1.2 0.126
Extended Baveno VI 87.0 (77.1–94.3) 46.2 (32.3–59.1) 98.4 (88.3–99.8) 15.1 (7.9–23.4) 42.9 2.5 0.226
FIB-4 82.1 (68.7–99.8) 51.3 (43.2–64.0) 89.4 (84.6–95.7) 21.5 (15.6–24.2) 48.6 0.9 0.199
APRI 65.1 (48.7–83.8) 81.4 (65.9–95.7) 92.4 (80.6–99.8) 74.2 (48.2–94.5) 60.0 0.9 0.156
AAR 75.2 (42.2-93.7) 49.0 (19.1–75.7) 88.2 (66.3–96.6) 12.1 (9.7–15.2) 32.9 3.5 0.112
RESIST 84.6 (65.1–95.6) 47.3 (33.7–61.2) 86.7 (71.7–94.4) 43.1 (36.0–50.6) 37.1 4 0.254

Note: Ranges in parentheses indicate the 95% confidence interval. The following cut-offs of the non-invasive criteria were applied: Baveno VI, LSM <20 kPa and platelets >150,000; expanded Baveno VI, LSM <25 kPa and platelets > 110,000; FIB-4, <2.78; APRI, <1.1; AAR, <0.75; RESIST, albumin >36 g/L and platelets > 120,000.

AAR = AST-to-ALT ratio; APRI = AST-to-platelet ratio index; EVNT = esophageal varices needing treatment; FIB-4 = fibrosis-4 index; HBV = hepatitis B virus; HCV = hepatitis C virus; HIV = human immunodeficiency virus; NPV = negative predictive value; PPV = positive predictive value.

Cost analysis

The universal endoscopic screening for EVNT in 100 patients with virus-related cACLD would cost $76,600. Considering the proportion of patients in Figure 1 where EGD would be spared according to LSM-based non-invasive criteria, the total direct costs saved per 100 patients would be $7,416 and $27,332 with Baveno VI and extended Baveno VI criteria, respectively. As for simple serum biomarkers, total direct cost saved compared to universal endoscopic screening would be as follows: (1) $35,068 with FIB-4; (2) $42,728 with APRI; (3) $17,384 with AAR; and (4) $35,568 with RESIST criteria.

In this multicenter clinical cohort of patients with virus-related cACLD, we found that non-invasive criteria based on LSM perform well in sparing a significant number of EGD. Furthermore, non-invasive criteria based on simple fibrosis biomarkers, namely FIB-4, APRI, and RESIST, have good performance and could thus be used in remote primary care and general medicine clinics, as well as those located in low resource countries. Finally, our data show that the performance of both LSM-based criteria and simple fibrosis biomarkers were consistent and comparable across the individual etiologies of virus-related cACLD, with a high yield to predict the absence of EVNT.

The development of portal hypertension remains a crucial event in the natural history of ACLD as it underlines most of the clinical end-stage hepatic complications (1,5). The risk of variceal bleeding from EV is 5% to 15% per year in patients with ACLD and the mortality rate is at least 20% at 6 weeks after a bleeding episode (1,37). Given the prognostic burden of a bleeding episode, screening is recommended as early intervention can be provided to patients with EVNT to prevent bleeding. The gold standard procedure for the diagnosis of EVNT is EGD. Traditionally, it is recommended that all patients with cACLD should undergo EGD surveillance for varices at the time of diagnosis and then yearly, with identification rates of EVNT ranging between 6% and 12% (38). However, EGD is an invasive procedure which carries the risk of uncommon but significant complications, such as pain, bleeding, and perforation (39). Moreover, it needs to be performed by well-trained healthcare providers in high-resources centers. Despite the advancement in antiviral therapy for HCV and HBV, virus etiology remains the third most common cause of ACLD in Western countries (40,41). HIV can also cause ACLD independently of viral hepatitis coinfections due to high frequency of NAFLD caused by metabolic imbalance and lifelong use of antiretroviral therapy (24). Access to EGD can be problematic in the context of primary care settings, especially in low resource countries. Strategies to reduce referral to hepato-gastroenterology units for unnecessary EGD screening may be essential in this clinical setting. The Baveno VI guidelines proposed that patients cACLD with LSM <20 kPa and platelets >150,000/μL can avoid screening EGD. Expanded Baveno VI criteria, obtained by optimizing LSM and platelets thresholds, have been proposed to spare a even higher proportion of unnecessary EGD (12). These LSM-based non-invasive criteria have been validated in patients with cACLD of various etiologies and in a meta-analysis (35,4247). However, these criteria have not been sufficiently compared to serum biomarkers and across various etiologies of virus-related cACLD. Ad hoc validation is relevant considering that most studies have been conducted in chronic HCV, that patients with HIV may have higher frequency of thrombocytopenia unrelated to liver disease, and that LSM presents with specific cut-offs and diagnostic accuracy in HBV (48,49). Moreover, the pathogenesis of liver fibrosis portending the development of cirrhosis is complex and involves a number of discrete mechanisms unique to the HIV and HBV infections (23,50). For instance, the fibrogenetic process in HBV includes a combination of processes including inflammation, cytokine production, viral components, and HBV DNA integration (50). Finally, since EV development results from liver fibrosis and is due to increased intrahepatic resistance, simple fibrosis biomarkers can be potentially useful in detecting EV. However, most available studies have been conducted in patients with HCV (13,51).

In the present study, we found that the LSM-based criteria Baveno VI and expanded Baveno VI would have spared 26.1% and 51.6% EGD while missing 0.6% and 2% EVNT, respectively. The number of spared EGD is somewhat similar to that reported in previous studies (1622). Of note, there was no difference in proportions of spared EGD and missed EVNT across subgroup etiologies of virus-related cACLD. For HBV subgroup, the NPV of Baveno VI and expanded Baveno VI was 100% and 96.3%, respectively. In the HIV subgroup, the NPV of Baveno VI and expanded Baveno VI was 97.7% and 98.5%, respectively. These figures were no different from the HCV subgroup, where the NPV of Baveno VI and expanded Baveno VI was 92.3% and 98.4%, respectively. As such, our data support the use of these LSM-based criteria across these different viral etiologies of cACLD. Interestingly, we found that serum fibrosis biomarkers also performed well for the diagnosis of EVNT in virus-related cACLD. FIB-4 and APRI would have spared 48.2% and 58% EGD while missing 0.8% and 1.2% EVNT, respectively. We also applied the RESIST criteria, recently proposed in HCV, and we found that they perform well in virus-related cACLD, with 47.8% spared EGD and 1.9% missed EVNT. Conversely, AAR had suboptimal performance compared to other serum fibrosis biomarkers, with an AUC of 0.518 compared to 0.784 for FIB-4 and 0.818 for APRI. Despite having been proposed in the past for the non-invasive detection of EVNT, our data show that it is imperative to adopt non-invasive criteria incorporating platelets (13,52). The performance of FIB-4, APRI, and RESIST was similar across HBV, HIV, and HCV subgroups. Overall, all non-invasive criteria but AAR had NPV >95% and missed EVNT in less than 5% of cases. The LR analysis, which provides information of the diagnostic test independently of disease prevalence, confirmed the good performance of the LSM-based non-invasive criteria, as well as of the simple fibrosis biomarkers, except AAR. In a cost analysis based on Canadian data, we found that simple fibrosis biomarkers would save more direct costs compared to LSM-based criteria.

Our study has several strengths. The population was included from two prospective and ethnically diverse cohorts. We validated both LSM-based non-invasive criteria and simple fibrosis biomarkers to detect EVNT. We included only patients mono-infected with HBV, HIV, and HCV in order to validate the accuracy of non-invasive criteria for each individual viral etiology of cACLD. A cost analysis was also proposed. We also wish to acknowledge several limitations. First, some data collection and analysis of data were retrospective and the evaluation of EV size was performed by multiple endoscopists, thus there is a potential inter-observer variability due to intrinsic subjectivity of the endoscopic examination. Second, we did not have follow-up data to evaluate the performance of non-invasive criteria to predict bleeding events. Third, referral bias may have occurred as included patients were part of tertiary care centers, and not from resource limited areas. Fourth, the cost analysis did not account for missed EVNT, for cost of acquisition and maintenance of both transient elastography and endoscope, and for variations across different Canadian provinces of wages, billing practices, and employability. Lastly, we wish to underline that, beyond detection of EVNT, LSM may have superior diagnostic accuracy for liver fibrosis (53). Moreover, our study specifically addresses the use of endoscopy for EVNT but there are other aspects of the clinical management of cACLD and portal hypertension, including screening for hepatocellular carcinoma and the use of carvedilol (1).

In conclusion, in a multicenter Canadian cohort of patients with virus-related cACLD non-invasive criteria based on LSM, platelets, albumin, and liver transaminases can safely spare 26.1–58% EGD while missing 0.6–2% EVNT. In HBV and HCV, APRI seems to have a better potential for EGD sparing with the lowest missed EVNT and the highest total direct cost saved. In HIV, the RESIST criteria may have a better potential for EGD sparing with the lowest missed EVNT and the highest total direct cost saved. EGD remains the gold standard in the identification of EV and has the important advantage of being therapeutic. However, it is invasive, associated with potential complications and has significantly high costs. Non-invasive criteria could help risk stratification and allow for triaging patients, thereby benefitting patients at higher risk in obtaining EGD in a prompter fashion. Our study will inform judicious use of scarce resources in healthcare jurisdictions burdened by increased costs and inflation. Further validation studies are needed to confirm our findings, particularly for the prediction of clinical outcomes such as EV bleeding and in etiologies where validation of the Baveno VI criteria is suboptimal, such as autoimmune hepatitis, primary biliary and sclerosing cholangitis, and alcoholic liver disease.

The study was supported by the Canadian HBV Network. GS is supported by a Senior Salary Award from Fonds de Recherche du Quebec – Sante (FRQS) (#296306). Part of this work has been presented at the Conference on Retroviruses and Opportunistic Infections (CROI) (US; February 2022).

Conceptualization, A Zoughlami, J Serero, G Sebastiani (lead); Methodology, A Zoughlami, J Serero, SE Congly, G Sebastiani (lead); Validation, SE Congly, I Zhao, A Ramji, C Cooper, P Wong, R Bailey ; Formal Analysis, SE Congly, G Sebastiani; Resources: G Sebastiani (lead); Data Curation, A Zoughlami, J Serero, SE Congly, I Zhao, A Ramji, C Cooper, P Wong, R Bailey, G Sebastiani (lead); Writing —Original Draft: G Sebastiani (lead); Writing — Review & Editing, A Zoughlami, J Serero, SE Congly, I Zhao, A Ramji, C Cooper, P Wong, R Bailey; Supervision: G Sebastiani (lead); Project Administration, A Zoughlami, J Serero ; Funding Acquisition, G Sebastiani

This study was approved by the University of Calgary Conjoint Ethics Research Board (CHREB; Calgary, AB, Canada, Ethics ID: REB16-0041), and the Research Ethics Board of the Research Institute of the McGill University Health Centre (MUHC-RI REB; Montreal, Quebec, Canada, Code 14-182-BMD).



Funding for this study was received by the Canadian HBV Network and the Fonds de Recherche du Quebec – Sante (FRQS) – 296306.

S Congly reports grants from Gilead Sciences, Genfit, Allergan, Axcella Health, Inc, and Sequana Medical and personal fees from Intercept Pharmaceuticals, AstraZeneca, Paladin Labs, and Eisai. A Ramji served on advisory board and speaker bureau for Gilead, Intercept, AbbVie, Celgene, Merck, Novartis; clinical trial funding AbbVie, Assembly, Gilead, Intercept, Galmed, Janssen, Springbanks, Allergan, Novartis. C Cooper has served as a consultant and has participated in sponsored lectures and received research funding from Merck, Gilead, and AbbVie Canada. P Wong has acted as a consultant for BMS, Gilead, Merck, and Novartis. CS Coffin reports investigator- initiated research grant and research materials from GSK, Gilead Sciences, and Arbutus Biopharma; consulting fees with Gilead Sciences and Advisory board with Altimmune (paid to University of Calgary); and local-site principal investigator participation in clinical trials with Arbutus, Gilead Sciences, and Janssen Inc. G Sebastiani has acted as speaker for Merck, Gilead, AbbVie, Novonordisk, Novartis, Pfizer, served as an advisory board member for Pfizer, Merck, Novartis, Gilead, Allergan and Intercept and has received unrestricted research funding from Theratec.

This manuscript has been peer reviewed.

Data can be accessed upon request to the Canadian HBV Network.


1. de Franchis R, Bosch J, Garcia-Tsao G, Reiberger T, Ripoll C, Baveno VIIF. Baveno VII - renewing consensus in portal hypertension. J Hepatol. 2022;76(4):95974. https://doi.org/http://dx.doi./10.1016/j.jhep.2021.12.022. PMID: 35120736 MedlineGoogle Scholar
2. Castellares C, Barreiro P, Martin-Carbonero L, et al. Liver cirrhosis in HIV-infected patients: prevalence, aetiology and clinical outcome. J Viral Hepat. 2008;15(3):16572. https://doi.org/http://dx.doi./10.1111/j.1365-2893.2007.00903.x. PMID: 18233989 MedlineGoogle Scholar
3. Benmassaoud A, Nitulescu R, Pembroke T, et al. Liver-related events in human immunodeficiency virus-infected persons with occult cirrhosis. Clin Infect Dis. 2019;69(8):142230. https://doi.org/http://dx.doi./10.1093/cid/ciy1082. PMID: 30561558 MedlineGoogle Scholar
4. Cervo A, Shengir M, Patel K, Sebastiani G. NASH in HIV. Curr HIV/AIDS Rep. 2020;17(6):60114. https://doi.org/http://dx.doi./10.1007/s11904-020-00531-0. PMID: 3298492 MedlineGoogle Scholar
5. Abraldes JG, Bureau C, Stefanescu H, et al. Noninvasive tools and risk of clinically significant portal hypertension and varices in compensated cirrhosis: the “Anticipate” study. Hepatology. 2016;64(6):217384. https://doi.org/http://dx.doi./10.1002/hep.28824. PMID: 27639071 MedlineGoogle Scholar
6. D’Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol. 2006;44(1):21731. https://doi.org/http://dx.doi./10.1016/j.jhep.2005.10.013. PMID: 16298014 MedlineGoogle Scholar
7. Merchante N, Rivero-Juarez A, Tellez F, et al. Liver stiffness predicts variceal bleeding in HIV/HCV-coinfected patients with compensated cirrhosis. AIDS. 2017;31(4):493500. https://doi.org/http://dx.doi./10.1097/QAD.0000000000001358. PMID: 27922855 MedlineGoogle Scholar
8. Lemoine M, Nayagam S, Thursz M. Viral hepatitis in resource-limited countries and access to antiviral therapies: current and future challenges. Future Virol. 2013;8(4):37180. https://doi.org/http://dx.doi./10.2217/fvl.13.11. PMID: 23662157 MedlineGoogle Scholar
9. Mwachiro M, Topazian HM, Kayamba V, et al. Gastrointestinal endoscopy capacity in Eastern Africa. Endosc Int Open. 2021;9:E182736. https://doi.org/http://dx.doi./10.1055/a-1551-3343. PMID: 34790551 MedlineGoogle Scholar
10. Sankaranarayanan R, Ramadas K, Qiao YL. Managing the changing burden of cancer in Asia. BMC Med. 2014;12:3. https://doi.org/10.1186/1741-7015-12-3 Google Scholar
11. de Franchis R, Baveno VIF. Expanding consensus in portal hypertension: report of the Baveno VI Consensus Workshop: stratifying risk and individualizing care for portal hypertension. J Hepatol. 2015;63(3):74352. https://doi.org/http://dx.doi./10.1016/j.jhep.2015.05.022. PMID: 26047908 MedlineGoogle Scholar
12. Augustin S, Pons M, Maurice JB, et al. Expanding the Baveno VI criteria for the screening of varices in patients with compensated advanced chronic liver disease. Hepatology. 2017;66(6):19808. https://doi.org/http://dx.doi./10.1002/hep.29363. PMID: 28696510 MedlineGoogle Scholar
13. Sebastiani G, Tempesta D, Fattovich G, et al. Prediction of oesophageal varices in hepatic cirrhosis by simple serum non-invasive markers: results of a multicenter, large-scale study. J Hepatol. 2010;53(4):6308. https://doi.org/http://dx.doi./10.1016/j.jhep.2010.04.019. PMID: 20615567 MedlineGoogle Scholar
14. Ishida K, Namisaki T, Murata K, et al. Accuracy of Fibrosis-4 index in identification of patients with cirrhosis who could potentially avoid variceal screening endoscopy. J Clin Med. 2020;9(11):3510. https://doi.org/http://dx.doi./10.3390/jcm9113510. PMID: 33138335 MedlineGoogle Scholar
15. Calvaruso V, Celsam C, D’Ambrosio R, et al. RESIST-HCV criteria to monitor progression of low-risk esophageal varices in patients with compensated cirrhosis after HCV eradication. The SIMPLE study: SIMPLE: scoring index to monitor progression of low-risk esophageal varices. Am J Gastroenterol. 2022;117(11):181624. https://doi.org/http://dx.doi./10.14309/ajg.0000000000001878. PMID: 35973181 MedlineGoogle Scholar
16. Franchini M, Veneri D, Lippi G. Thrombocytopenia and infections. Expert Rev Hematol. 2017;10(1):99106. https://doi.org/http://dx.doi./10.1080/17474086.2017.1271319. PMID: 27936979 MedlineGoogle Scholar
17. Morse CG, McLaughlin M, Proschan M, et al. Transient elastography for the detection of hepatic fibrosis in HIV-monoinfected adults with elevated aminotransferases on antiretroviral therapy. AIDS. 2015;29(17):22972302. https://doi.org/http://dx.doi./10.1097/QAD.0000000000000841. PMID: 26544701 MedlineGoogle Scholar
18. Prat LI, Roccarina D, Lever R, et al. Aetiology and severity of liver disease in HIV positive patients with suspected NAFLD: lessons from a cohort with available liver biopsies. J Acquir Immune Defic Syndr. 2019;80(4):47480. https://doi.org/http://dx.doi./10.1097/QAI.0000000000001942. PMID: 30807482 MedlineGoogle Scholar
19. Macias J, Recio E, Vispo E, et al. Application of transient elastometry to differentiate mild from moderate to severe liver fibrosis in HIV/HCV co-infected patients. J Hepatol. 2008;49(6):91622. https://doi.org/http://dx.doi./10.1016/j.jhep.2008.07.031. PMID: 18929426 MedlineGoogle Scholar
20. de Ledinghen V, Douvin C, Kettaneh A, et al. Diagnosis of hepatic fibrosis and cirrhosis by transient elastography in HIV/hepatitis C virus-coinfected patients. J Acquir Immune Defic Syndr. 2006;41(2):1759. https://doi.org/http://dx.doi./10.1097/01.qai.0000194238.15831.c7. PMID: 16394849 MedlineGoogle Scholar
21. Merchante N, Rivero-Juarez A, Tellez F, et al. Liver stiffness predicts clinical outcome in human immunodeficiency virus/hepatitis C virus-coinfected patients with compensated liver cirrhosis. Hepatology. 2012;56(1):22838. https://doi.org/http://dx.doi./10.1002/hep.25616. PMID: 22278746 MedlineGoogle Scholar
22. Wang H, Wen B, Chang X, et al. Baveno VI criteria and spleen stiffness measurement rule out high-risk varices in virally suppressed HBV-related cirrhosis. J Hepatol. 2021;74(3):58492. https://doi.org/http://dx.doi./10.1016/j.jhep.2020.09.034. PMID: 33039403 MedlineGoogle Scholar
23. Kaspar MB, Sterling RK. Mechanisms of liver disease in patients infected with HIV. BMJ Open Gastroenterol. 2017;4(1):e000166. https://doi.org/http://dx.doi./10.1136/bmjgast-2017-000166. PMID: 29119002 MedlineGoogle Scholar
24. Castera L. Noninvasive methods to assess liver disease in patients with hepatitis B or C. Gastroenterology. 2012;142(6):12931302.e4. https://doi.org/http://dx.doi./10.1053/j.gastro.2012.02.017. PMID: 22537436 MedlineGoogle Scholar
25. McPherson S, Stewart SF, Henderson E, Burt AD, Day CP. Simple non-invasive fibrosis scoring systems can reliably exclude advanced fibrosis in patients with non-alcoholic fatty liver disease. Gut. 2010;59(9):12659. https://doi.org/http://dx.doi./10.1136/gut.2010.216077. PMID: 20801772 MedlineGoogle Scholar
26. Vallet-Pichard A, Mallet V, Nalpas B, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology. 2007;46(1):326. https://doi.org/http://dx.doi./10.1002/hep.21669. PMID: 17567829 MedlineGoogle Scholar
27. Wai CT, Greenson JK, Fontana RJ, et al. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology. 2003;38(2):51826. https://doi.org/http://dx.doi./10.1053/jhep.2003.50346. PMID: 1288349 MedlineGoogle Scholar
28. Williams AL, Hoofnagle JH. Ratio of serum aspartate to alanine aminotransferase in chronic hepatitis. Relationship to cirrhosis. Gastroenterology. 1988;95(3):7349. https://doi.org/http://dx.doi./10.1016/s0016-5085(88)80022-2. PMID: 3135226 MedlineGoogle Scholar
29. Hassan EM, Omran DA, El Beshlawey ML, Abdo M, El Askary A. Can transient elastography, Fib-4, Forns Index, and Lok Score predict esophageal varices in HCV-related cirrhotic patients? Gastroenterol Hepatol. 2014;37(2):5865. https://doi.org/http://dx.doi./10.1016/j.gastrohep.2013.09.008. PMID: 24365388 MedlineGoogle Scholar
30. Congly SE, Shaheen AA, Swain MG. Modelling the cost effectiveness of non-alcoholic fatty liver disease risk stratification strategies in the community setting. PLoS One. 2021;16(5):e0251741. https://doi.org/http://dx.doi./10.1371/journal.pone.0251741. PMID: 34019560 MedlineGoogle Scholar
31. Ma I, Lau CK, Ramdas Z, Jackson R, Naugler C. Estimated costs of 51 commonly ordered laboratory tests in Canada. Clin Biochem. 2019;65:5860. https://doi.org/http://dx.doi./10.1016/j.clinbiochem.2018.12.013. PMID: 30615855 MedlineGoogle Scholar
32. Monakova J, Wong J, Blais I, et al. Establishing funding rates for colonoscopy and gastroscopy procedures in Ontario. Curr Oncol. 2019;26(2):98101. https://doi.org/http://dx.doi./10.3747/co.26.4405. PMID: 31043810 MedlineGoogle Scholar
33. Altman DG MD, Bryant TN, Gardner MJ. Statistics with Confidence. 2nd ed. BMJ Books; 2000. Google Scholar
34. Mercaldo ND, Lau KF, Zhou XH. Confidence intervals for predictive values with an emphasis to case-control studies. Stat Med. 2007;26(10):217083. https://doi.org/http://dx.doi./10.1002/sim.2677. PMID: 16927452 MedlineGoogle Scholar
35. Petta S, Sebastiani G, Bugianesi E, et al. Non-invasive prediction of esophageal varices by stiffness and platelet in non-alcoholic fatty liver disease cirrhosis. J Hepatol. 2018;69(4):87885. https://doi.org/http://dx.doi./10.1016/j.jhep.2018.05.019. PMID: 29802949 MedlineGoogle Scholar
36. Cohen J. Weighted kappa: nominal scale agreement with provision for scaled disagreement or partial credit. Psychol Bull. 1968;70(4):21320. https://doi.org/http://dx.doi./10.1037/h0026256. PMID: 19673146 MedlineGoogle Scholar
37. D’Amico G, De Franchis R, Cooperative Study G. Upper digestive bleeding in cirrhosis. Post-therapeutic outcome and prognostic indicators. Hepatology. 2003;38(3):599612. https://doi.org/http://dx.doi./10.1053/jhep.2003.50385. PMID: 12939586 MedlineGoogle Scholar
38. Garcia-Tsao G, Sanyal AJ, Grace ND, Carey W. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis. Hepatology. 2007;46(3):92238. https://doi.org/http://dx.doi./10.1002/hep.21907. PMID: 17879356 MedlineGoogle Scholar
39. Levy I, Gralnek IM. Complications of diagnostic colonoscopy, upper endoscopy, and enteroscopy. Best Pract Res Clin Gastroenterol. 2016;30(5):70518. https://doi.org/http://dx.doi./10.1016/j.bpg.2016.09.005. PMID: 27931631 MedlineGoogle Scholar
40. Flemming JA, Djerboua M, Groome PA, Booth CM, Terrault NA. NAFLD and alcohol-associated liver disease will be responsible for almost all new diagnoses of cirrhosis in Canada by 2040. Hepatology. 2021;74(6):333044. https://doi.org/http://dx.doi./10.1002/hep.32032. PMID: 34174003 MedlineGoogle Scholar
41. Scaglione S, Kliethermes S, Cao G, et al. The Epidemiology of Cirrhosis in the United States: a population-based study. J Clin Gastroenterol. 2015;49(8):6906. https://doi.org/http://dx.doi./10.1097/MCG.0000000000000208. PMID: 25291348 MedlineGoogle Scholar
42. Thabut D, Bureau C, Layese R, et al. Validation of Baveno VI criteria for screening and surveillance of esophageal varices in patients with compensated cirrhosis and a sustained response to antiviral therapy. Gastroenterology. 2019;156(4):9971009.e5. https://doi.org/http://dx.doi./10.1053/j.gastro.2018.11.053. PMID: 30768988 MedlineGoogle Scholar
43. Stafylidou M, Paschos P, Katsoula A, et al. Performance of Baveno VI and expanded baveno vi criteria for excluding high-risk varices in patients with chronic liver diseases: a systematic review and meta-analysis. Clin Gastroenterol Hepatol. 2019;17(9):17441755.e11. https://doi.org/http://dx.doi./10.1016/j.cgh.2019.04.062. PMID: 31077823 MedlineGoogle Scholar
44. Maurice JB, Brodkin E, Arnold F, et al. Validation of the Baveno VI criteria to identify low risk cirrhotic patients not requiring endoscopic surveillance for varices. J Hepatol. 2016;65(5):899905. https://doi.org/http://dx.doi./10.1016/j.jhep.2016.06.021. PMID: 27388923 MedlineGoogle Scholar
45. Augustin S, Pons M, Genesca J. Validating the Baveno VI recommendations for screening varices. J Hepatol. 2017;66(2):45960. https://doi.org/http://dx.doi./10.1016/j.jhep.2016.09.027. PMID: 27826055 MedlineGoogle Scholar
46. Moctezuma-Velazquez C, Saffioti F, Tasayco-Huaman S, et al. Non-invasive prediction of high-risk varices in patients with primary biliary cholangitis and primary sclerosing cholangitis. Am J Gastroenterol. 2019;114(3):44652. https://doi.org/http://dx.doi./10.1038/s41395-018-0265-7. PMID: 30315285 MedlineGoogle Scholar
47. Marot A, Trepo E, Doerig C, Schoepfer A, Moreno C, Deltenre P. Liver stiffness and platelet count for identifying patients with compensated liver disease at low risk of variceal bleeding. Liver Int. 2017;37(5):70716. https://doi.org/http://dx.doi./10.1111/liv.13318. PMID: 27862856 MedlineGoogle Scholar
48. Merchante N, Saroli Palumbo C, Mazzola G, et al. Prediction of esophageal varices by liver stiffness and platelets in persons with human immunodeficiency virus infection and compensated advanced chronic liver disease. Clin Infect Dis. 2020;71(11):28107. https://doi.org/http://dx.doi./10.1093/cid/ciz1181. PMID: 31813962 MedlineGoogle Scholar
49. Li Q, Chen L, Zhou Y. Diagnostic accuracy of liver stiffness measurement in chronic hepatitis B patients with normal or mildly elevated alanine transaminase levels. Sci Rep. 2018;8(1):5224. https://doi.org/http://dx.doi./10.1038/s41598-018-23646-2. PMID: 29588489 MedlineGoogle Scholar
50. Suhail M, Abdel-Hafiz H, Ali A, et al. Potential mechanisms of hepatitis B virus induced liver injury. World J Gastroenterol. 2014;20(35):1246272. https://doi.org/http://dx.doi./10.3748/wjg.v20.i35.12462. PMID: 25253946 MedlineGoogle Scholar
51. Cifci S, Ekmen N. Evaluation of non-invasive fibrosis markers in predicting esophageal variceal bleeding. Clin Endosc. 2021;54(6):85763. https://doi.org/http://dx.doi./10.5946/ce.2021.028. PMID: 34034454 MedlineGoogle Scholar
52. Iwata Y, Enomoto H, Sakai Y, et al. Elevation of the AST to ALT ratio in association with the severity of esophageal varices in patients with HCV-related compensated liver cirrhosis. Hepatogastroenterology. 2013;60(121):14952. https://doi.org/http://dx.doi./10.5754/hge12578. PMID: 22975583 MedlineGoogle Scholar
53. Cheng PN, Chiu HC, Chiu YC, Chen SC, Chen Y. Comparison of FIB-4 and transient elastography in evaluating liver fibrosis of chronic hepatitis C subjects in community. PLoS One. 2018;13(11):e0206947. https://doi.org/http://dx.doi./10.1371/journal.pone.0206947. PMID: 30403744 MedlineGoogle Scholar