Diagnostic Performance of Two-dimensional Shear Wave Elastography in Advanced Liver Fibrosis: A Prospective Pathology-based Study

Studies for evaluating the diagnostic performance of two-dimensional Shear-wave Elastography (2D-SWE) in a patient cohort including various liver disorders, remain limited. We aimed to evaluate the validity of 2D-SWE in the diagnosis of advanced liver brosis amongst patients with various liver disorders. In this pathology-based study, patients who underwent a liver biopsy for various benign liver diseases were prospectively recruited during the period between February, 2017 and September, 2020. Data of 2D-SWE, Fibrosis-4 Index (FIB-4), and Aspartate Aminotransferase to Platelet Ratio Index (APRI) were simultaneously collected. The cut-off values for predicting advanced brosis, i.e. Metavir brosis stage ≥ F3, were determined using Receiver Operating Characteristic (ROC) analysis. The diagnostic performance was evaluated and then compared by Area Under the ROC (AUROC). In total, 95 patients were recruited for study analysis. The diagnostic performance of 2D-SWE was signicantly superior to that of both FIB-4 (AUROC: 0.88, 95% condence interval [CI]: 0.80-0.94; vs 0.72, 95%CI: 0.62-0.81; p=0.001) and APRI (AUROC: 0.88, 95%CI: 80-0.94; vs 0.76, 95%CI: 0.66-0.84; p=0.007). With an optimal cutoff value of 9.3 kPa, the sensitivity and specicity were 90.91% and 76.47%, respectively. In subgroup analysis, the AUROC of 2D-SWE was the highest when compared to that of FIB-4 and APRI in patients with chronic hepatitis B, chronic hepatitis C, fatty liver, and concurrent hepatitis. 2D-SWE can therefore be a valid non-invasive method in the detection of advanced liver brosis in various liver diseases. was performed using the Kruskall-Wallis test. Statistical analysis was performed using SPSS software version 22.0.


Introduction
Chronic hepatitis has been estimated to affect at least 325 million people worldwide 1 , and can eventually result in advanced brosis, liver malignancies, and death. Accurate evaluation regarding the degree of liver brosis is therefore essential in the diagnosis and treatment of chronic liver diseases 2,3 . Early diagnosis of advanced liver brosis can be life-saving in patients with chronic liver diseases 4,5 . However, although a liver biopsy remains the gold standard in the evaluation of liver brosis, it is limited by its invasive nature and the possibility of severe complications and/or discomfort 6 . Thus, an effective noninvasive test is highly helpful in clinical practices. However, the diagnostic performance of most commonly used non-invasive tests, such as Aspartate Aminotransferase (AST) to Platelet Ratio Index (APRI) or the Fibrosis-4 Index (FIB-4), are not satisfactory and should be improved.
Although several novel non-invasive tests have been developed in recent years; for example, using serum biomarkers which are directly related to the molecular pathogenesis of liver brosis; they are usually only available in certain health care units 7 . In addition, in recent years several ultrasound-based methods have been used for measuring liver stiffness, including Transient Elastography (TE), Acoustic Radiation Force Impulse (ARFI), and two-dimensional Shear Wave Elastography (2D-SWE); however, some disadvantages in each may limit their applications in clinical practice or research. For example, TE, i.e. Fibroscan®, has been proven to offer strong diagnostic performance with an Area Under the Receiver Operating characteristic Curve (AUROC) of 0.8 to 0.9 for the diagnosis of advanced brosis in patients with chronic hepatitis C 8 ; however, TE requires separate equipment in addition to the sonography probe. Moreover, the TE measurement may be affected by non-parenchymal structures, such as blood vessels, ligaments, or tumor lesions 9 . ARFI and 2D-SWE are both ultrasound-based methods which could be incorporated into a conventional ultrasound device without additional equipment. However, ARFI has been proved to have lower diagnostic performance in a liver with steatosis 10 .
2D-SWE is a novel technique for non-invasive evaluation of liver brosis with several advantages. For example, multiple measurements can be performed over a relatively larger eld of view, compared to ARFI 11 . In addition, operators can choose a proper area by avoiding non-parenchymal structures 12 , and by checking the shear-wave propagation displayed by parallel lines in the selected region. This way, the operator can ensure a higher reliability of the measurement 13 . Although 2D-SWE has also been proved to have a good diagnostic performance for liver diseases 14, 15,16 , only a few studies have been validated by the reference standard of liver brosis, i.e. liver pathology. In addition, data from 2D-SWE applied in a patient cohort with various liver disorders remains limited. The aim of this study was to evaluate the diagnostic performance of 2D-SWE amongst patients with various disorders.

Study design and patient selection
In this prospective trans-sectional study, patients who were scheduled to receive a liver biopsy for the evaluation of benign liver diseases, were recruited in Taichung Veterans General Hospital, a tertiary medical center in central Taiwan, during the period between February, 2017 and February, 2020. Patients with malignant liver tumors were excluded. All patients received a liver biopsy after a 2D-SWE study had been performed on the same day. Liver function-related blood tests, such as bilirubin, albumin, AST, Alanine Aminotransferases (ALT), complete blood count, and Prothrombin Time (PT), were collected within 2 weeks prior to the liver biopsy. This study has been approved by the Institutional Review Board of Taichung Veterans General Hospital (IRB No. CE18315A). All methods of this study were performed in accordance with the Declaration of Helsinki and the relevant guidelines and regulations. The informed consent has been obtained from all the study participants.

Liver biopsy and histopathology
After a 2D-SWE examination, real-time sonography-guided percutaneous liver biopsies were performed at the same site of the 2D-SWE examination, using a 16-gauge core needle. At least one liver tissue fragment at a length ≥ 15 mm was obtained through the right intercostal space. All the liver specimens were reviewed by two independent pathologists, and a consensus for liver brosis staging was made by the two pathologists if the brosis stages were different. The degree of liver brosis was classi ed according to the Metavir system 17 .

2D-shear Wave Elastography
The 2D-SWE measurements were performed using the Aplio 500 Platinum Series ultrasound machine and 6C1 (PVT-375BT) curved array ultrasound transducer (Canon Medical Systems Corporation). All patients fasted for at least 6 hours prior to the examination. The examinations were performed using the right intercostal approach, with patients holding their breath during expiration. Under the view of a sonography, a 2 × 4 cm acquisition box was selected for SWE detection, with a 10-mm diameter circle used to select the Region Of Interest (ROI). The parallel propagation of shear wave contours was observed in the ROIs.
The shear wave conduction speed was measured in the selected region and expressed in meters per second (m/s). The Liver Stiffness Measurement (LSM), also known as the elastic modulus, was calculated by the program incorporated in the sonography machine, which was expressed in kilopascal (kPa) 18 . At least 10 measurements were obtained for each patient, and the median of the measurements was calculated 19 .

Subgroup analysis
We further analyzed the diagnostic performance of 2D-SWE in subgroups of patients with chronic hepatitis B, chronic hepatitis C, fatty liver disease, and concurrent ALT elevation (de ned as ALT > 35 U/L in male patients and ALT > 25 U/L in female patients) 20 . Fatty liver disease was de ned as steatosis of more than 5% hepatocytes in the histopathology specimen 21 .

Statistical analysis
We presented continuous variables as median values (25-75% Interquartile Range [IQR]) and categorical variables as numbers (percentages). FIB-4 and APRI were calculated according to the formula presented in previous literature 22,23 . The upper normal limit of AST for calculating APRI was 35 U/L according to the reference range of our hospital. The diagnostic performances of FIB-4, APRI and 2D-SWE were determined according to the results of liver pathology. Metavir brosis stage F3 or above was de ned as advanced brosis. Data were expressed with a Receiver Operating Characteristic (ROC) curve and AUROC.
Optimal cutoff values were calculated by the measurement of the highest Youden index. Paired comparison of the AUROC of each method to predict advanced liver brosis was performed using the Kruskall-Wallis test. Statistical analysis was performed using SPSS software version 22.0.

Study subjects
As shown in Table 1, 95 patients were recruited in total. Most of the patients were middle-aged, with a median age of 59 years. In the etiology of liver diseases, 53% patients had chronic Hepatitis B Virus (HBV) infection, 38% patients had chronic Hepatitis C Virus (HCV) infection, and 3.2% patients had HBV and HCV coinfection. Approximately 30% of the patients had diabetes mellitus, hypertension, or dyslipidemia. The median levels of AST and ALT in the blood were 47 U/L and 53 U/L, respectively. Around 45% of patients suffered from ALT elevation. The median value of calculated FIB-4 and APRI was The measurements of 2D-SWE The diagnostic performance of 2D-SWE  (Fig. 2).

Discussion
This prospective study clearly presents that 2D-SWE is a useful non-invasive tool for evaluating the degree of liver brosis in clinical practices, as well as being more effective in determining advanced brosis than either FIB-4 or APRI. In this prospective pathology-based study, all study subjects underwent the liver biopsy and 2D-SWE examination on the same day, so the bias in previous studies of 2D-SWE, such as lacking a pathology standard reference 24 or an up to 24-week interval between liver biopsy and 2D-SWE examination 25 , could have been well overcome. Although there are fewer studies of 2D-SWE than there are of other modalities in evaluating liver brosis, the results of this study provide solid data that 2D-SWE can be implemented as another non-invasive alternative in clinical practices.
Several non-invasive tests have been used for the evaluation of liver brosis; however some disadvantages may limit their applications. Although FIB-4 and APRI have been widely used to predict advanced liver brosis because of being non-invasive, cheap, and convenient, the sensitivity and speci city of FIB-4 and APRI have not been satisfactory. Both FIB-4 and APRI were initially established to predict advanced brosis in patients with chronic hepatitis C; however in a previous study, the brosis status could not be determined in a signi cant proportion of patients (35.4% in FIB-4 and 49% in APRI) 22,23 . In a recently published study examining the diagnostic performance of FIB-4 and APRI in cirrhotic patients with chronic hepatitis B, a lower AUROC in FIB-4 (0.75) and APRI (0.65) was also reported 26 . In addition, although TE has been proven to be superior to APRI in the evaluation of liver brosis, its measurements were proved to be less reproducible in patients with fatty liver disease, obesity, and ALT elevation 27 . Furthermore, the diagnostic performance of TE has been found to be inferior to that of 2D-SWE in predicting advanced brosis in a meta-analysis study 28 . Although ARFI has fewer limitations than TE 29 , its accuracy also decreases in patients with fatty liver disease 29,30 . In this study, the diagnostic performance of 2D-SWE was validated not only within the whole patient cohort experiencing various liver disorders, but also in the individual patient subgroups. 2D-SWE may well be an easy-to-use tool for effectively accessing patients with various liver conditions. Although there have been several studies in recent years involving 2D-SWE for the purpose of evaluating liver brosis, most of them were retrospective studies. The measurement of 2D-SWE and the performing of a liver biopsy were usually not conducted on the same day, with the time interval possibly being up to 12 weeks 31,32 . In this situation, the severity of liver brosis may have changed during this period. Importantly, the liver area selected for a 2D-SWE measurement might be different from that chosen for a liver biopsy, with the data taken from the 2D-SWE possibly not being representative of the pathology ndings. In this pathology-based prospective study, the data regarding the 2D-SWE measurements and blood tests were collected right before the liver biopsy, with the liver specimens acquired from the area selected for 2D-SWE measurement. The possible bias which occurred in previous 2D-SWE studies has been minimized in this study. In addition, data regarding 2D-SWE in a patient cohort, including various liver disorders, remains limited. Most previous 2D-SWE studies were conducted in East Asian countries, and the reported data basically investigated from patients with chronic HBV infection 14, 33 . On the other hand, studies which have been carried out in Western countries mainly focused on patients with chronic HCV infection 34 . In this study which included patients with various liver disorders, 2D-SWE presented a good diagnostic performance not only in the whole patient cohort, but also in patients with various liver disorders.
Several limitations should be acknowledged in this study. First, the sample size of our study was not large enough to con dently evaluate the accuracy of 2D-SWE in all patient subgroups, such as patients with alcoholic liver disease or autoimmune liver disease. A larger study for investigating other patient subgroups should be conducted in the future. Second, most patients recruited for this study had been infected with HBV and/or HCV, therefore patients without chronic viral hepatitis should be interpreted with caution. For example, although this study demonstrated that the diagnostic performance of 2D-SWE was not interfered in patients with concurrent fatty liver disease, subgroup analysis for patients experiencing purely fatty liver disease could not be conducted due to the small sample size. Although the strong diagnostic performance of 2D-SWE in non-alcoholic liver disease has been reported in a recent small study 35 , further validation studies for patients without chronic viral hepatitis should be encouraged. Third, this has been a study conducted in Taiwan, and the data of this study would need to be validated in Western countries.
In conclusion, 2D-SWE can be widely used as an effective non-invasive tool for the diagnosis of advanced liver brosis in patients with various liver disorders.

Data Availability
All relevant data has been reported within the manuscript. Further supplementary datasets can be obtained upon written request addressed to the corresponding author.