Shear-Wave Elastography Using Commercially Available Ultrasound in a Mouse Model of Chronic Liver Disease

: Elastography is currently used clinically to diagnose the degree of liver stiffness. We sought to develop a shear-wave elastography (SWE) measurement method using ultrasound in mice and to compare its results with those of other noninvasive tests for liver ﬁbrosis. We divided male mice into three groups (normal (G1), liver ﬁbrosis (G2), and fatty liver (G3)). We measured mouse liver SWE values and compared them with T 1rho and T 2 values from magnetic resonance imaging results. We also compared the SWE values with the expression levels of a serum liver ﬁbrosis biomarker (Mac-2-binding protein (M2BP)) and hepatic genes. SWE values signiﬁcantly increased over time in G2 but did not change in G3. T 1rho values in G2 and G3 were signiﬁcantly increased compared with those in G1. T 2 values in G2 did not increase compared with those in group 1. T 2 values in G3 signiﬁcantly increased compared with those in groups 1 and 2. In G2, SWE values signiﬁcantly and positively correlated with T 1rho values. SWE values signiﬁcantly correlated with serum M2BP levels in G2 but did not correlate with inﬂammatory gene expression. We could measure SWE values to assess the degree of liver ﬁbrosis in mouse models of liver disease.


Introduction
Liver fibrosis occurs as a reparative action after chronic liver injury and often progresses to cirrhosis [1]. The degree of progression determines the prognosis of patients with chronic liver disease, including those with virus-associated hepatitis and nonalcoholic fatty liver disease (NAFLD) [2,3]. Generally, the stage of liver fibrosis is determined by examining a liver biopsy specimen. However, an invasive liver biopsy is unsuitable as a diagnostic test for large numbers of patients with chronic liver disease. Therefore, the development of a reproducible, noninvasive test (NIT) that can be used to accurately diagnose the stage of liver fibrosis is urgently needed. Currently, biomarkers of blood liver fibrosis and imaging methods are representative NITs. Elastography (using ultrasound or magnetic resonance imaging (MRI)) for measuring liver stiffness has been used to diagnose liver fibrosis. For quantitative measurements of liver stiffness, shear-wave elastography (SWE) is used clinically. This method can be integrated with ultrasound imaging to latitudinally generate propagating shear waves using acoustic radiation forces and to track their velocity. If SWE can be implemented for rodents, it is expected to have a wide range of applications in various animal experiments. Studies using SWE measurements in rat models of liver disease have been previously reported [4,5]. However, there are few reports on studies using SWE measurements in mice [6].
We developed a serum biomarker of liver fibrosis, Mac-2-binding protein (M2BP) [7][8][9][10][11]. Serum M2BP levels increase with the progression of liver fibrosis in humans and mice, and measurements of its levels are useful to assess the stage of liver fibrosis. Our previous study demonstrated that mouse serum M2BP levels increased with the stage of liver fibrosis in a carbon tetrachloride (CCl 4 )-induced liver fibrosis model and an NAFLD model [7]. In addition, serum M2BP levels positively correlated with hepatic fibrosisrelated gene expression.
Certain imaging modalities can be used to evaluate liver fibrosis noninvasively. While ultrasound elastography is commonly used in hospitals, MRI is superior to those methods in terms of objectivity. To develop a biomarker for imaging liver fibrosis with higher quality, we investigated liver injury assessment methods using 7T-MRI. Our recent study used T 1rho and T 2 values to evaluate acute liver injury after CCl 4 administration; we found that T 1rho mapping could be used to diagnose acute liver injury more accurately than T 2 mapping [12]. Another group reported that T 1rho quantification could also be used to assess the degree of liver fibrosis in a rat model of fibrosis after chronic CCl 4 administration [13].
The purpose of our study was to develop an SWE measurement method using a commercially available ultrasound system in mice. In addition, we compared the results of SWE values in two different mouse models of liver disease with other liver fibrosis NITs (blood fibrosis biomarkers and MRI), as well as with hepatic gene expression levels. The determination of the utility of SWE measurements via commercially available US in mouse liver disease models is important to both verify the effectiveness of therapeutic drugs and for experiments using genetically modified mice.

Experimental Protocol
All experimental protocols described in this study were approved by the Ethics Review Committee for Animal Experimentation of Osaka University Graduate School of Medicine. C57BL/6J mice were purchased from Oriental Yeast (Suita, Osaka, Japan). The animals were provided with unrestricted amounts of food and water, housed in temperature-and humidity-controlled rooms, and maintained on a 12/12 h light/dark cycle.
Male mice (n = 18) were divided into three groups that received different experimental protocols ( Figure 1A). Both high-fat/high-cholesterol diet (HFHC; 15% cocoa butter fat, 1.25% cholesterol, and 0.5% cholic acid) and normal diet (ND) feed were purchased from Oriental Yeast (Suita, Osaka, Japan). The ND group (group 1; n = 6) was fed an MF diet for 4 weeks. The CCl 4 group (group 2; n = 6) was injected with CCl 4 (500 µL/kg body weight) intraperitoneally twice per week for 4 weeks to induce persistent liver damage and liver fibrosis. The HFHC group (group 3; n = 6) was fed an HFHC diet for 4 weeks. Mice were killed 3 days after the final CCl 4 injection. For each experiment, mice were fasted for 5 h with free access to water and then were weighed and killed. Blood was collected aseptically from the inferior vena cava and centrifuged (13,000× g, 5 min, 4 • C) to collect the serum, and the samples were frozen at -80 • C until measurements were performed. Livers were harvested and fixed with 10% buffered paraformaldehyde or were immediately frozen in liquid nitrogen for mRNA extraction and lipid analysis, as previously described [14].
to collect the serum, and the samples were frozen at -80 °C until measurements were formed. Livers were harvested and fixed with 10% buffered paraformaldehyde or w immediately frozen in liquid nitrogen for mRNA extraction and lipid analysis, as pr ously described [14].

SWE Measurements
We measured mouse liver SWE values using a diagnostic ultrasound scanner (A 500) equipped with a probe (PLT-1005BT; Canon Medical Systems, Otawara, Tochigi pan), as described in a previous study [5] ( Figure 1B).
Before SWE measurements, mice were fasted for 4 h and anesthetized by an intra itoneal injection of medetomidine (0.75 mg/kg), midazolam (4.0 mg/kg), and butorpha (5.0 mg/kg). Mice then had their abdominal hair shaved; residual abdominal hair wa moved using a chemical depilation cream (Veets, Reckitt Benckiser Japan, Tokyo, Jap (Figure 1Ba-c). A probe was placed in the median fossa region for imaging with tak
Before SWE measurements, mice were fasted for 4 h and anesthetized by an intraperitoneal injection of medetomidine (0.75 mg/kg), midazolam (4.0 mg/kg), and butorphanol (5.0 mg/kg). Mice then had their abdominal hair shaved; residual abdominal hair was removed using a chemical depilation cream (Veets, Reckitt Benckiser Japan, Tokyo, Japan) (Figure 1Ba-c). A probe was placed in the median fossa region for imaging with taking care not to put pressure on the mouse liver (Figure 1Bd), and the SWE measurement was performed. Ten consecutive and distinct SWE images of mouse liver parenchyma were acquired in the same image plane. For each image, an SW speed map and a propagation map were displayed, and two circular regions of interest (ROI; 3 mm in diameter) were identified in the liver parenchyma. After a few seconds of immobilization to allow the SWE image to stabilize, the images were captured and saved. An average of nine acquisitions was documented for each mouse. After the SWE measurements, we administered an anesthetic antagonist (0.75 mg/kg) to awaken the mice.

Statistical Analysis
Statistical analyses were conducted using JMP Pro 16.1 software (SAS Institute, Inc., Cary, NC, USA). Results are presented as the mean ± standard deviation. Statistical analy-ses included analysis of variance, the Wilcoxon and Kruskal-Wallis tests, and Spearman R correlations. The diagnostic performance of the SWE value was assessed by analyzing receiver operating characteristic (ROC) curves. The probabilities of true-positive (sensitivity) and true-negative (specificity) assessments were determined for selected cutoff values, and the area under the curve (AUC) was calculated. The Youden index was used to identify the optimal cutoff point of the SWE value. Differences were considered statistically significant at P values less than 0.05.

Evaluation of Mouse Liver Histology
To compare the histological changes in each group, we stained mouse livers with H&E and Picrosirius red (Figure 2). Compared with normal liver (group 1), injured hepatocytes and bridging liver fibrosis were observed in group 2. In group 3 mouse livers, significant inflammatory cell infiltration, abundant steatosis, and mild liver fibrosis were observed.

Statistical Analysis
Statistical analyses were conducted using JMP Pro 16.1 software (SAS Institute, Cary, NC). Results are presented as the mean ± standard deviation. Statistical ana included analysis of variance, the Wilcoxon and Kruskal-Wallis tests, and Spearm correlations. The diagnostic performance of the SWE value was assessed by analyzin ceiver operating characteristic (ROC) curves. The probabilities of true-positive (sen ity) and true-negative (specificity) assessments were determined for selected cutoff ues, and the area under the curve (AUC) was calculated. The Youden index was us identify the optimal cutoff point of the SWE value. Differences were considered sta cally significant at P values less than 0.05.

Evaluation of Mouse Liver Histology
To compare the histological changes in each group, we stained mouse livers H&E and Picrosirius red (Figure 2). Compared with normal liver (group 1), inj hepatocytes and bridging liver fibrosis were observed in group 2. In group 3 mouse li significant inflammatory cell infiltration, abundant steatosis, and mild liver fibrosis observed.

Changes in SWE Values
In group 2 mice, liver SWE values significantly increased over time ( Figure 3A,B week 4, the SWE values of group 2 mice were significantly higher than those of mi groups 1 and 3. In group 3 mice, liver SWE values did not change during the experim time course. Using ROC analyses, we set the cutoff values of SWE values for the stif of the liver in a mouse liver fibrosis model treated by CCL4 for 4 weeks using group 1

Changes in SWE Values
In group 2 mice, liver SWE values significantly increased over time ( Figure 3A,B). At week 4, the SWE values of group 2 mice were significantly higher than those of mice in groups 1 and 3. In group 3 mice, liver SWE values did not change during the experimental time course. Using ROC analyses, we set the cutoff values of SWE values for the stiffness of the liver in a mouse liver fibrosis model treated by CCL 4 for 4 weeks using group 1 and group 2 ( Figure 3C). The cutoff value was 1.649 m/s, and the AUC, sensitivity, and specificity of this cutoff value were 0.967, 86.7%, and 100%, respectively. Figure 4A shows the color-coded T 1rho maps and T 2 maps in mouse livers at weeks 2 and 4. Figure 4B shows T 1rho and T 2 relaxation times in mouse livers. T 1rho relaxation times in groups 2 and 3 were significantly longer at week 4 than those in group 1. T 2 relaxation times in group 3 were significantly longer at week 4 than those in groups 1 and 2. T 2 relaxation times in group 2 were not longer than those in group 1.

MRI (T 1rho , T 2 ) Maps
Gastrointest. Disord. 2022, 4, FOR PEER REVIEW 6 group 2 ( Figure 3C). The cutoff value was 1.649 m/s, and the AUC, sensitivity, and specificity of this cutoff value were 0.967, 86.7%, and 100%, respectively.   Figure 4A shows the color-coded T1rho maps and T2 maps in mouse livers at weeks 2 and 4. Figure 4B shows T1rho and T2 relaxation times in mouse livers. T1rho relaxation times in groups 2 and 3 were significantly longer at week 4 than those in group 1. T2 relaxation times in group 3 were significantly longer at week 4 than those in groups 1 and 2. T2 relaxation times in group 2 were not longer than those in group 1.

Relationship between SWE Values and MRI Parameters
We investigated the relationship between SWE values and MRI parameters in this study ( Figure 5). In group 2 mice, SWE values significantly and positively correlated with T 1rho relaxation times but not with T 2 relaxation times ( Figure 5A). In group 3 mice, SWE values did not correlate with either T 1rho or T 2 relaxation times ( Figure 5B).

Relationship between SWE Values and MRI Parameters
We investigated the relationship between SWE values and MRI parameters in this study ( Figure 5). In group 2 mice, SWE values significantly and positively correlated with T1rho relaxation times but not with T2 relaxation times ( Figure 5A). In group 3 mice, SWE values did not correlate with either T1rho or T2 relaxation times ( Figure 5B).

Comparison of SWE Results with Levels of a Liver Fibrosis Biomarker and Hepatic Genes
Next, we measured the serum levels of the liver fibrosis biomarker, M2BP, and liver gene expression levels ( Figure 6). Serum M2BP levels were significantly higher in groups 2 and 3 than those in group 1 (group 1, 65.8 ± 16.8 ng/mL; group 2 175.9 ± 41.8 ng/mL, and group 3, 202.8 ± 136.4 ng/mL). Hepatic gene expression levels of Tgf-β, Tnf-α, and iNos were significantly higher in group 3 than the levels in groups 1 and 2. These levels were not increased in group 2 compared with group 1.
Next, we investigated the relationship between M2BP and liver gene expression levels with imaging values (SWE, T1rho, T2; Table 1). In group 2, we found there were significant, positive relationships between serum M2BP levels and SWE values and T1rho relaxation times (Table 1A). In addition, hepatic iNos gene expression was positively correlated with T1rho and T2 relaxation times. In group 3, serum M2BP levels significantly correlated with T1rho and T2 relaxation times (Table 1B). Hepatic gene expression levels (Tgf-β, Tnf-α, and iNos) also significantly correlated with T1rho and T2 relaxation times. There was no significant correlation with SWE values in group 3 mice.

Comparison of SWE Results with Levels of a Liver Fibrosis Biomarker and Hepatic Genes
Next, we measured the serum levels of the liver fibrosis biomarker, M2BP, and liver gene expression levels ( Figure 6). Serum M2BP levels were significantly higher in groups 2 and 3 than those in group 1 (group 1, 65.8 ± 16.8 ng/mL; group 2 175.9 ± 41.8 ng/mL, and group 3, 202.8 ± 136.4 ng/mL). Hepatic gene expression levels of Tgf-β, Tnf-α, and iNos were significantly higher in group 3 than the levels in groups 1 and 2. These levels were not increased in group 2 compared with group 1.   Next, we investigated the relationship between M2BP and liver gene expression levels with imaging values (SWE, T 1rho , T 2 ; Table 1). In group 2, we found there were significant, positive relationships between serum M2BP levels and SWE values and T 1rho relaxation times (Table 1A). In addition, hepatic iNos gene expression was positively correlated with T 1rho and T 2 relaxation times. In group 3, serum M2BP levels significantly correlated with T 1rho and T 2 relaxation times (Table 1B). Hepatic gene expression levels (Tgf-β, Tnf-α, and iNos) also significantly correlated with T 1rho and T 2 relaxation times. There was no significant correlation with SWE values in group 3 mice. Table 1. Relationships between imaging results and liver-injury-related factors in each group.
A. Comparison of factors of mice in groups 1 and 2.

Discussion
To the best of our knowledge, our study is the first approach for the measurement of mouse liver stiffness to assess the progression of mouse liver disease in vivo using a commercially available US machine. In this study, we compared the SWE results with the MRI results, serum liver fibrosis biomarker levels, and hepatic gene expression levels. In a CCl 4 -induced liver fibrosis model (group 2), SWE values significantly increased with liver disease progression. In the HFHC diet induced-NAFLD model (group 3), SWE values did not change during the experiment. One of the reasons for the lack of change in the SWE values in group 3 mice may be because there was very mild hepatic fibrosis but ample hepatic steatosis (soften mouse liver stiffness) in mice fed an HFHC diet for 4 weeks.
In group 2 mice, SWE values positively correlated with T 1rho values and serum M2BP levels but did not correlate with T 2 values. In a previous report, T 1rho and T 2 values were associated with the severity of liver fibrosis in experimental models of rat liver fibrosis (4 weeks in a bile duct ligation model, 16 weeks in a CCl 4 -induced liver fibrosis model) [16]. In a study evaluating disease severity in a rat model of CCl 4 -induced liver fibrosis, T 1rho values were shown to be more useful than T 2 values [17]. However, Xie et al. reported that edema and inflammation had a greater impact on liver T 1rho values than liver fibrosis [18]. T 2 values are reported to increase with the grade of steatosis and inflammation [19,20]. Serum M2BP levels are increased not only by the progression of liver fibrosis but also by increased steatosis and inflammation [7]. Considering these findings together, SWE values appear to be less susceptible to edema and inflammation and would be superior to other NITs used in this study for evaluations of disease progression in a mouse model of CCl 4 -induced liver fibrosis.
In group 3 mice, the SWE values did not change during our experiment. Four weeks of an HFHC diet induced very mild liver fibrosis and fatty changes in mice. According to our study results, we believe the SWE measurements were insufficient for the detection of such mild fibrotic changes in group 3 mice. In addition, the grade of liver steatosis is known to affect liver stiffness as measured with ultrasound-based elastography [21][22][23].
The effects of steatosis could thus offset elevations in liver stiffness induced after 4 weeks of an HFHC diet. The SWE values did not correlate with either serum M2BP levels or hepatic gene expression levels in group 3 mice. These results also indicate that SWE values are less susceptible to factors other than liver stiffness. On the other hand, T 1rho and T 2 values are well-correlated with serum M2BP levels and hepatic gene expression levels. These results indicate that T 1rho and T 2 values are affected by inflammation and/or steatosis.
In addition to SWE, there are other methods (e.g., MR elastography (MRE), transient elastography) for measuring liver stiffness in mice, as described in preclinical studies [6,[24][25][26]. MRE requires expensive equipment, and US elastography measurement methods for mice reported to date have required special equipment [6]. Yeh et al. firstly demonstrated the quantitative determination of the mouse liver stiffness in vivo using two single-element high-frequency transducers. In this study, we could measure hepatic SWE in mice in vivo using a commercially available US machine and probe. We could measure liver stiffness using SWE over time in the same individual mouse. Our findings could promote the utility of SWE measurements in mouse liver disease models to both verify the effectiveness of therapeutic drugs and for experiments using genetically modified mice. From an animal welfare perspective, the SWE measurement using commercially available US in mice could contribute to reducing the number of mice sacrificed.
Our study has several limitations. First, the experimental duration was relatively short (4 weeks) for the detection of obvious changes in SWE values in mouse liver. To detect changes in SWE values in an NAFLD model, a longer duration of diet feeding is needed to achieve obvious progression of liver fibrosis. Second, mouse liver histological data at baseline and 2 weeks were not available in this study. Therefore, we could not investigate the precise data at 2 weeks. However, our study is significant in that we took imaging data over time on the same mouse. Third, our study did not investigate the effects of NITs in a regression model (e.g., one group of mice receives continuous CCl 4 administration during the experimental protocol (progressive disease model), while another group of mice receives a limited duration of CCl 4 administration and are then monitored without CCl 4 administration (regression model)). The measurements of each NIT in these mice could then be used to assess the strength of each test for the detection of therapeutic effects.

Conclusions
In conclusion, we could measure SWE values using a commercially available US machine to evaluate the degree of liver fibrosis in a mouse liver disease model. After comparing SWE values with other test values (the MRI results and levels of a serum biomarker of liver fibrosis and hepatic genes), SWE was found to be a more specific tool for assessing the degree of liver fibrosis in mouse liver disease than other NITs. Our findings promote the utility of SWE measurements by commercially available US in mouse liver disease models to both verify the effectiveness of therapeutic drugs and for experiments using genetically modified mice. Funding: This study was supported by JSPS KAKENHI grants (grant numbers 20K08383 (to Y.K.)) from the Japan Society for the Promotion of Science.
Institutional Review Board Statement: All experimental protocols described in this study were approved by the Ethics Review Committee for Animal Experimentation of Osaka University Graduate School of Medicine (no. R03-01-0). Informed Consent Statement: All experimental protocols described in this study were approved by the Ethics Review Committee for Animal Experimentation of Osaka University Graduate School of Medicine.