This study received approval from our institutional ethics committee, and informed consent was obtained from patients. Between January 2014 and September 2018, 1932 patients with a history of inactive HBV carrier status, CHB, or hepatitis B-related cirrhosis referred to our tertiary hospital for follow-up and US evaluation of the liver were prospectively recruited (patients referred several times were counted as a single participant). Fig. 1 presents a flowchart of sample selection. Because CHB arises from HBV infection in the normal liver, and further evolves to liver fibrosis or cirrhosis, the liver stiffness of CHB patients may be different from that of healthy people, inactive HBV carriers, and patients with decompensated cirrhosis. Therefore, to conduct an integrated study, we compared the liver stiffness of CHB patients with that of patients with different pathologies as controls (to determine lower and upper limit parameters of liver stiffness). Patients with decompensated cirrhosis could be diagnosed through clinical data (including laboratory tests), but patients with compensated cirrhosis could not be diagnosed accurately in the absence of biopsy and histopathology findings; therefore, patients with compensated cirrhosis were excluded. The exclusion criteria were patients with obesity (body mass index >30 kg/m²) (n=7), marked fatty liver (n=10), pregnancy (n=3), cholestasis (n=2), splenectomy (n=2), diabetes mellitus (n=1), liver tumors (n=3), liver nodular lesions (diameter >7 mm, with a space-occupying effect) (n=81), liver congestion (n=1), mitral stenosis (n=1), heart failure (n=1), severe chronic obstructive pulmonary disease (n=2), renal failure (n=1), compensated liver cirrhosis based on US criteria and biochemical tests (n=227) [20]; patients without comprehensive laboratory tests results for the evaluation of the liver (n=238), HBV DNA>2,000 IU/mL, alanine transaminase (ALT) levels>60 U/L (>40 U/L for inactive HBV carriers); and patients who did not agree to undergo P-SWE evaluation (n=37). The inclusion criteria were referral according to the American Association for the Study of Liver Diseases guideline for the diagnosis of liver disease [1], a history of hepatitis B infection, inactive HBV, or CHB for at least 3 years, and regular follow-up since diagnosis and all inactive HBV carriers and CHB patients were inactive in recent 3 years. The diagnosis of decompensated cirrhosis was based on the presence of previous clinical data indicating a Child-Pugh class of B and C. The study population was selected consecutively from 1,215 patients who met the inclusion criteria, including 100 CHB patients without patchy echogenicity of the liver and 100 CHB patients with patchy echogenicity of the liver. Among the CHB patients (with and without patchy echogenicity of the liver), 32 consecutive patients underwent percutaneous liver biopsy. One hundred inactive HBV carriers without abnormal echogenicity of the liver (HBV DNA<1,000 IU/ mL, ALT<40 U/L) and 100 patients with decompensated liver cirrhosis were selected as control groups. Eighty participants with a healthy liver who had been hospitalized for various operations not involving the liver and did not show abnormalities on laboratory tests were recruited as a blank control group. For all patients and healthy participants, anthropometric tests were performed, including determinations of body weight, height, and body mass index. Biochemical tests of ALT, aspartate aminotransferase, γ -glutamyltransferase, alkaline phosphatase, total bilirubin, and serum albumin levels, platelet count, and prothrombin time were performed within 7 days of the US examination. The baseline demographic and clinical characteristics of the healthy participants and patients are listed in Table 1.

A Siemens Acuson S2000 ultrasound system (Siemens Healthcare, Ultrasound Business Unit, Mountain View, CA, USA) equipped with a curvilinear transducer (4C1, 2-4 MHz, mean push pulse, 2.67 MHz), the Virtual Touch quantification software package (version 2.0), and a 14 L5 linear transducer (6-14 MHz) was used to acquire US images and SWVs. A radiologist specializing in US (S.W.) with 17 years of US experience and 2 years of ARFI P-SWE experience (as of 2014) who was blinded to the clinical data performed US and P-SWE. Each US examination was first performed using the conventional B-mode. The linear transducer was used to assess the liver surface and superficial parenchyma. After the US examination of the liver parenchyma, blood vessels, and color Doppler flow imaging, P-SWE was performed. The SWV measurements were made by the same operator. P-SWE was performed with the patient in the supine position and the right arm at maximal abduction. SWV measurements of the liver were made at the S5 and S6 segments of the right liver lobe, with a depth of 1-3 cm under the liver capsule. During the procedure, patients were instructed to suspend breathing, and the acoustic beam was directed perpendicularly to the liver parenchyma free of large vessels via the intercostal space (Fig. 2). A rectangular region of interest (ROI) was set with fixed dimensions of 1.0 cm×0.5 cm with the convex transducer (4C1) when performing SWV. An adequate amount of gel was placed onto the skin overlying the liver, the transducer was placed gently onto the gel without compression, and the SWV was measured in several different ROIs for each patient. Ten valid measurements were obtained at four different regions of the right lobe for each patient, and we followed the guideline that at least 60% of the measurements should be "good"; if not, a value should not be reported [21]. A "good" measurement was defined as one where a numerical result was obtained, not an "x.xx." The interquartile range (IQR), also called the "middle 50," is a measure of the statistical dispersion defined as the difference between the upper and lower quartiles. An IQR/ median value of less than 0.30 suggests that a data set is good [21]. SWV failure was defined as a success rate of 60% or less or an IQR of 30% or more of the entire range. The median SWV was expressed in meters per second (m/sec). The mean value of all the acquisitions was calculated as the SWV result.

US, color Doppler flow imaging, and P-SWE images of the liver surface, liver parenchyma, hepatic veins, portal vein, patent paraumbilical vein (if present), and the spleen were saved in the picture archiving and communications system.

Taking normal liver and spleen parenchyma as reference areas of homogeneous moderate (not fine or coarse) parenchyma, the parenchymal texture of the participants with a healthy liver and patients with CHB was categorized as homogeneous moderate parenchyma, heterogeneous coarse liver parenchyma (Fig. 3), liver parenchyma without patchy echogenicity, and liver parenchyma with scattered patchy echogenicity (Figs. 4-6).

A US-guided percutaneous liver biopsy was performed within 3 days following P-SWE and US examination using a 16-gauge Magnum needle (MG1522/BARD Magnum, C.R. Bard Inc., Murray Hill, NJ, USA), and the biopsy sites were predominantly in parenchymal areas of the right lobe of the liver with scattered patchy echogenicity, and in other places in a few cases. The biopsy specimens (15-22 mm long) were fixed in formalin and embedded in paraffin.

All biopsy specimens were analyzed by two experienced liver pathologists, who were blinded to the results of P-SWE and US, but not to the clinical and biochemical data of the patient. Liver fibrosis was evaluated using the METAVIR scoring system, as follows: score of F0, no fibrosis; F1, portal fibrosis without septa; F2, portal fibrosis and few septa; F3, numerous septa without cirrhosis; and F4, cirrhosis. Significant fibrosis was defined as a score of F2 or greater [22]. Hepatic steatosis was defined as at least 30% of hepatocytes containing lipid droplets.

When the quantitative variables were normally distributed, the results were expressed as mean values and standard deviations; otherwise, medians and IQR (25th-75th percentile) were reported. Qualitative variables were summarized as counts and percentages. The fibrosis score (METAVIR) of CHB patients with and without patchy echogenicity was compared using the chi-square test. The SWVs of the P-SWE results were not normally distributed across different groups and different fibrosis stages; therefore, the Wilcoxon rank-sum test was used to test the equality of the median SWV measurements. The level of significance was determined using a 2-sided test with an α level of 0.05. SPSS version 20 (IBM Corp., Armonk, NY, USA) was used for all statistical analyses.