Assessment of Liver Fibrosis by Transient Elastography in Children with Chronic Hepatitis B Virus Infection

Background: This study aimed to investigate the effectiveness of transient elastography (TE) by comparing liver biopsies to assess liver brosis in children with chronic hepatitis B (CHB). Methods: A total of 157 CHB children aged 0 - 6 years in China were enrolled in this single-center prospective study. All patients underwent liver stiffness measurement (LSM) by TE and liver biopsy at an interval of less than a week. Results: LSM, aspartate aminotransferase (AST)-platelet ratio index (APRI), and brosis-4 score (FIB-4) positively correlated with activity grade and brosis stage in children with CHB. The area under receiver operating characteristic curves (AUCs) of LSM for identifying signicant (F ≥ 2) and advanced brosis (F ≥ 3) were 0.732 and 0.94, the cut-off values were 5.6 kPa and 6.9 kPa, specicity of 75.7% and 91.5%, and sensitivity of 67.4% and 81.3%, respectively. Compared to LSM, the overall diagnostic performance of APRI and FIB-4 for signicant and advanced brosis was suboptimal with low AUCs and sensitivity. Since LSM, platelet, and Log 10 HBsAg were independent factors with the brosis stages (F < 2 and F ≥ 2) on the liver biopsy, the LPS index was formulated to predict F ≥ 2 by combining LSM, platelet, and Log 10 HBsAg. The AUC of LPS for F ≥ 2 was increased to 0.792, which was higher than that of LSM (0.732, p < 0.05), with an improved sensitivity (76.6% vs 67.4%). Conclusions: TE represents a promising technology for the diagnosis of advanced brosis in CHB children aged 0 6 years.


Introduction
Hepatitis B virus (HBV) infection is one of the most common causes of chronic liver disease (CLD) worldwide, especially in China where more than 80 million adults and 37,000 children are affected [1,2]. Although the natural history of chronic HBV-infected children remains poorly understood, limited studies have shown that 1% − 5% of hepatitis B e-antigen (HBeAg)-positive children developed cirrhosis before adulthood [3][4][5][6]. In addition, 25% of adult patients who acquire HBV infection in childhood will develop liver cancer or cirrhosis, which leads to the greatest burden of morbidity and mortality [7]. Thus, there is a critical need to decrease the risk of disease progression to cirrhosis and even achieve a functional cure for children with chronic hepatitis B (CHB) with antiviral treatment. Additionally, one of the most important indicators for antiviral treatment is histological evidence of necro-in ammation and brosis according to guidelines of the European Society of Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN), as well as the American Association for the Study of Liver Diseases [8]. Therefore, early diagnosis of the extent of liver in ammation and brosis is important for the treatment of CHB during childhood [8,9].
Currently, liver biopsy remains the gold standard for determining the degree of liver in ammation and brosis, and is pivotal for guiding antiviral treatment in children with CHB [8,9]. In addition, follow-up monitoring is required for these patients to evaluate the e cacy of antiviral treatment [10]; however, children may suffer from pain, additional expenses, and risks associated with post-procedure hospitalizations by undergoing a liver biopsy [11].
Moreover, a liver biopsy requires highly skilled physicians and medical devices. Thus, non-invasive diagnostic tests for CHB children to diagnose liver cirrhosis are necessary to avoid the risks and costs associated with liver biopsies.

Page 4/16
The aspartate aminotransferase (AST)-platelet ratio index (APRI) and brosis-4 score (FIB-4) obtained by calculating laboratory parameters have been used to identify the brosis stages in adult patients with CHB. The results showed that APRI and FIB-4 were insu cient due to high rates of misclassi cation [12], whereas the diagnostic performance of APRI and FIB-4 remained unknown in children with CHB. Additionally, transient elastography (TE) has been widely adopted as a novel noninvasive assessment tool to diagnose the stage of liver brosis and monitor the development of chronic liver disease (e.g., CHB and chronic hepatitis C [CHC]) due to its accuracy and reproducibility in adult patients [13]. Several studies have shown that liver stiffness measurement (LSM) by TE is useful to assess liver brosis in children with CLD [14][15]. Although LSM was evaluated for hepatitis B/C-related brosis in children across three studies, due to small sample sizes, limited patient populations (primarily adolescents and young adults) and multiple causes of liver disease [11,15,16] [17], underwent LSM and liver biopsy with an interval of less than a week, and written informed consent was obtained from the parent or legal guardian of the child subjects. The exclusion criteria consisted of: 1) patients with white blood cell (WBC) < 2.75 × 10 9 /L, PLT < 80 × 10 9 /L, total bilirubin > 51 μmol/L, alanine aminotransferase (ALT) ≥ 400 IU/L, serum creatinine > 133 μmol/L or international normalized ratio > 1.5; 2) patients with positive hepatitis A/C/delta virus, human immunode ciency virus, or a chronic liver disease other than CHB (e.g., autoimmune hepatitis, Wilson's disease, hepatolenticular degeneration, and hepatocellular carcinoma); 3) patients with evidence of decompensation (i.e., clinical ascites); or 4) any other serious physical and mental illnesses.

Clinical and Laboratory Parameters
Demographic data, including age, gender, body weight, and height (Body Mass Index [BMI] = body weight in kg/height in meters 2 ) were collected. Routine blood tests, liver function tests, plasma HBV DNA quanti cation, and serological HBV markers, including HBeAg and hepatitis B surface antigen (HBsAg) quanti cation, as well as abdominal ultrasound examination were performed. APRI and FIB-4 were calculated as previously reported [12].

Liver Histology and Liver Stiffness Measurements
After the laboratory examinations were performed, ultrasonic-guided liver biopsies were carried out in all subjects using a one-second needle biopsy. Liver specimens were prepared for histological evaluation by a senior pathologist who was blinded to the LSM results according to the meta-analysis of histological data for the viral hepatitis (METAVIR) scoring system [18]. LSM expressed in kilopascals (kPa) was measured using Fibroscan® with an S probe (Echosens, France) by a certi ed and experienced physician blinded to the liver biopsy results. In this study, only the LSM results were considered reliable when an interquartile range (IQR)/LSM of ≤ 0.3, up to 10 validated measurements, and a success rate of ≥ 60% were obtained.

Statistical Analysis
For the descriptive analysis, quantitative variables were expressed as the mean ± SD or medians (IQR), whereas categorical variables were expressed as the number of subjects (percentages). A comparison of quantitative variables was conducted using a Student's t-test/one-way ANOVA for normally distributed variables or Tamhane's T2 for anomalous distributed variables, whereas categorical variables were compared using a Chi-squared test.
Correlations were assessed using a Spearman's rank correlation coe cient, and factors associated with the degree of liver brosis were identi ed with a logistic regression analysis. The diagnostic value of LSM was evaluated based on sensitivity, speci city, positive, and negative predictive values (PPV and NPV), positive and negative likelihood ratio (PLR and NLR), and area under receiver operating characteristic (ROC) curves (AUC) using a Hanley-McNeil test. The LSM cut-off values for predicting different stages of liver brosis were determined at the highest sensitivity and speci city. All of the above statistical analyses were performed using SPSS 25.0 statistical software, and statistical signi cance was considered at p < 0.05.

Correlation between LSM, APRI or FIB-4 and histological features in children with CHB
The activity grades in our study were divided into two groups: A < 2 and A ≥ 2, and the liver brosis stages were classi ed into three groups: F0-F1, F2, and F3-F4 in accordance with previous studies [14]. The distribution of LSM, APRI, and FIB-4 according to activity grade and stages of liver brosis are displayed in Figure 1.  . 1c and d). In addition, the FIB-4 levels were higher in the A ≥ 2 group compared with those in the A < 2 group (0.1104 vs 0.0814, p < 0.01); however, there was no signi cant differences in the brosis stages among the three groups (F0-F1, 0.0896, F2, 0.1321 and F3-F4, 0.1337, all p > 0.05) (Fig. 1e and f). We next estimated the correlation between LSM, APRI or FIB-4, and activity grades (A < 2 and A ≥ 2) or liver brosis stages (F0-F1, F2, and F ≥ 3). The results revealed that LSM (r = 0.275, P < 0.001), APRI (r = 0.478, P < 0.001), and FIB-4 (r = 0.249, P< 0 .01) were positively correlated with the degree of activity. We also found the three parameters were positively correlated with the brosis degree (LSM, r = 0.414, P < 0.001; APRI, r = 0.357, P < 0.001 and FIB-4, r = 0.277, P < 0.001). Overall, these data suggest that LSM, APRI and FIB-4 are positively associated with the severity of liver in ammation and brosis in CHB children.
Performance of LSM, APRI, and FIB-4 for liver brosis stages To further evaluate the performance of LSM, APRI, and FIB-4 for the liver brosis stages, an ROC curve analysis was performed for all patients. The AUCs of LSM identifying brosis stages F ≥ 2 and F ≥ 3 among children with CHB were 0.732 (95% con dence interval, 0.639 -0.826) and 0.941 (0.897 -0.985), respectively ( Table 2). The  Table 2 and Fig. 2a and b). Additionally, compared to LSM, although the speci cities of APRI predicting F ≥ 2 and F ≥ 3 were moderately higher, both the AUCs and sensitivities of APRI and FIB-4 for F ≥ 2 and F ≥ 3 were lower, especially for F ≥ 3 ( Table 2 and Fig. 2a and b). Overall, these data suggest that LSM is reliable for assessing advanced liver brosis, which is superior to that of APRI and FIB-4, whereas all of these parameters were suboptimal for identifying signi cant liver brosis.
Independent parameters associated with the brosis stage F ≥ 2 according to the liver biopsy We next performed a univariate analysis of the parameters associated with brosis stages of the liver biopsies (Table S1). We observed that ALT, AST, gamma-glutamyl transpeptidase (γ-GT), cholinesterase, PLT, HBeAg, and HBsAg quanti cation, Log 10 HBsAg, Log 10 HBV DNA, A ≥ 2, and LSM were signi cantly associated with the brosis stage (F ≥ 2) of the liver biopsy (all p < 0.05) (Table S3). Based on these results, further multivariate analyses showed that LSM, PLT, and Log 10 HBsAg were independent factors associated with the brosis stages of the liver biopsy (all p < 0.05) ( Table 3).
Combination of LSM, PLT and Log 10 HBsAg to determine liver brosis stage F ≥ 2 Since LSM was associated with a relatively poor diagnostic accuracy for F ≥ 2 as shown in Table 2 and Fig. 2, LSM, PLT, and Log 10 HBsAg were further combined as independent factors associated with brosis stages to create an algorithm that could predict the presence of F ≥ 2 in our patients. This algorithm was the LPS index (LSM, PLT and Log 10 HBsAg) = 0.511 × LSM -0.006 × PLT -0.682 × Log 10 HBsAg + 0.769. The data revealed that the AUC increased to 0.792 (0.720 -0.852), which was higher than that of LSM (0.792 vs 0.732, p < 0.05) ( Table 4 and Fig. S1). More importantly, the sensitivity increased by almost 10 percent (76.7% vs 67.4%) ( Table 5 and Fig.   S1). Taken together, these ndings demonstrate that compared to LSM, the combination of LSM, PLT, and Log 10 HBsAg could better predict liver brosis of F ≥ 2 with a higher AUC and greater sensitivity.

Discussion
This study is the rst to report that LSM represents a better noninvasive index for predicting HBV-related brosis stages from liver biopsies than APRI and FIB-4 in a large sample size of children aged 0 -6 years. In addition, LSM could better distinguish the patients with F0-F2 vs F3-F4 (AUC 0.941) compared to the patients with F0-F1 vs F2-F4 (AUC 0.732), suggesting a promising index to diagnose liver brosis of F ≥ 3.
To date, liver biopsies remain the most common test used to assess HBV-related brosis in children; however, its invasiveness restricts the ability to perform the repeated assessments required for dynamic monitoring of CHB development and the effects of antiviral treatment [8][9]. Recently, a pediatric nonalcoholic steatohepatitis study presented AUCs of TE for brosis of F ≥ 2 and F ≥ 3 to be 0.992 and 1, respectively, and the cut-off values were 7 kPa and 9 kPa, respectively for predicting the corresponding brosis stages [14]. Another study found that the 8.6 kPa cutoff point could be used to discriminate between stages F0-F2 and F3-F4 for children and young adults with multiple causes of liver disease [11]. In our study, we found that the AUCs were 0.732 and 0.941 and the cut-off values were 5.6 kPa and 6.9 kPa for brosis stages F ≥ 2 and F ≥ 3, respectively. The discrepancies between the ndings of these studies may be due to differences in the age of the participants at the time of enrolment and causes of the disease [19]. Consistent with our ndings, the study by Anna et al. reported an LSM of 5.4 (4.0, 7.1) kPa for the F2 stage in children with CHC (20). Moreover, a previous study demonstrated that LSM was able to adequately predict the liver brosis stage in adult patients with CHB, and the ROC curves were 0.81 for F0-F1 vs F2-F4 and 0.93 for F0-F2 vs F3-F4 [21], which is consistent with the children with CHB in our study. However, the cutoff values in adult patients with CHB were 7.2 kPa and 8.1 kPa for brosis stages F ≥ 2 and F ≥ 3, respectively [21]. This difference in the cut-off values was also affected by the age of subjects in two studies [22]. Overall, this is the rst study to suggest that TE represents a highly effective methodology for identifying children aged 0 -6 years with CHB who exhibit advanced brosis (F ≥ 3). Moreover, TE is vital for outpatient monitoring and clinical decision-making for patients with advanced brosis, similar to CHB in adults [23].
We additionally found that APRI and FIB-4 provided no advantages over LSM in the discriminated hepatic brosis stages of F ≥ 2 and F ≥ 3. In agreement with some studies that have focused on adults with CHB [24,25], we found that APRI and FIB-4 were not suitable for predicting the HBV-related brosis stages of F ≥ 2 and F ≥ 3 in CHB children. These data also indicate that TE, APRI, and FIB-4 were suboptimal for the diagnosis of the F ≥ 2 stage. Previous studies have shown that PLT count, log 10 HBsAg, alkaline phosphatase, ALT, AST, BMI, and in ammation were correlated with HBV/HCV-related brosis [26][27][28]. Similarly, in our study, we found that LSM, PLT, and Log 10 HBsAg were independent factors associated with the brosis stage F ≥ 2. Based on these independent factors, this study is the rst to apply the LPS index to improve the diagnostic performance of F ≥ 2 with a higher AUC (0.792) and sensibility (76.6%).
This study had several limitations. First, due to the low incidence of children with HBV-related advanced brosis, the sample size in subjects with F3-F4 was small, which also limited our ability to validate the cutoff points for identifying advanced brosis. Second, we only evaluated the children aged 0 -6 years and our study was performed at a single center. To address these drawbacks, future studies are required to enlarge the sample size to validate the cutoff points and evaluate the performance of TE for children aged 7 -18 years with CHB by conducting a multi-center study.
In conclusion, LSM rather than APRI and FIB-4 offers excellent performance for children aged 0 -6 years with HBV-related advanced brosis in China, whereas TE, APRI and FIB-4 are suboptimal for the diagnosis of stage F ≥ 2. Thus, the combination of LSM, PLT, and log 10 HBsAg could signi cantly enhance the diagnostic performance for stage F ≥ 2.

Declarations Acknowledgments
We thank the staff from the Department of liver disease and Ming-yuan Ji from the National day school for assisting in data collection. We also thank all of the patients for their participation in this study.