Background
Non-alcoholic fatty liver disease (NAFLD) is a chronic disease characterized by hepatic fat accumulation combined with underlying metabolic dysregulation, mainly encompassing hepatic steatosis, non-alcoholic steatohepatitis (NASH), non-alcoholic cirrhosis, and even hepatocellular carcinoma [
1,
2]. It is rapidly becoming one of the most common liver diseases, with a 20–30% incidence in Western countries [
3].
Liver biopsy (LB) is regarded as the gold standard for assessing disease severity [
4,
5]. However, due to tissue invasion, sampling error, and the associated complications [
6], imaging methods are increasingly being favored for the non-invasive evaluation of fatty liver. B-Mode Ultrasound (US) based on grayscale images is the most common method for evaluating fatty liver due to its low cost, safety, and availability. Hepatic steatosis is usually graded by various US features, including liver brightness, the hepatorenal ratio, and vessel blurring [
7,
8]. However, the sensitivity of US B-mode imaging for detecting mild steatosis (fat content > 5%) is reportedly between 60.9% and 65% [
9], and it does not allow quantitative evaluation. Although the controlled attenuation parameter (CAP) based on Fibroscan can accurately detect and quantify liver steatosis (> 10%), a meta-analysis showed that overlapping boundaries limit CAP’s clinical value in distinguishing between mild and moderate steatosis [
10,
11]. Liver ultrasound attenuation (LiSA) is a new technique similar to CAP, but it may not accurately reflect the actual performance of LiSA tested in LB-validated patients [
12]. In contrast, Sound Speed Estimation could detect and grade hepatic steatosis with a sensitivity of 80% and specificity of 85.7% in a small pilot study, whereas further research in a larger sample of patients with NAFLD is needed [
13].
Although CT is more effective in assessing fatty liver, it is less accurate for mild to moderate hepatic steatosis, and involves radiation exposure [
14]. Magnetic resonance spectroscopy is a non-invasive method for the quantification of liver fat with high sensitivity and specificity; nevertheless, at present it is mainly used as a research tool [
15].
Ultrasound elastography techniques, like transient elastography (TE), shear wave elastography (SWE), or acoustic radiation force impulse(ARFI), based on shear wave generation, aim to measure elasticity [
16]. TE showed a great performance when identifying and staging fibrosis in NAFLD, but steatosis or inflammatory activity may influence the accuracy of liver stiffness measurements (LSMs) in predicting fibrosis [
17‐
19]. A large-scale meta-analysis has shown that 2D-SWE is superior to TE in diagnosing significant fibrosis and cirrhosis [
20]. However, whether steatosis affects the elasticity values measured by SWE is still controversial [
21]. Some studies found no correlation between steatosis and SWE measurements based on multivariate analysis [
22,
23], whereas others have shown increasing liver viscoelasticity with the presence of steatosis [
24].
Hence, we performed an experimental study in rats that had undergone SWE examinations, using liver biopsy as reference standard. This study aimed to investigate the correlation between hepatic steatosis and elastic modulus values measured by SWE and discuss feasibility of SWE for grading steatosis in the absence of fibrosis in NAFLD.
Discussion
This study aims to demonstrate the influence of steatosis on liver stiffness measured by SWE on a rat model in the absence of fibrosis, indicating that steatosis, but not inflammation (B = − 0.031, P = 0.920) and ballooning (B = 0.216, P = 0.458), was an independent factor affecting the mean elastic modules (B = 1.558, P < 0.001). In terms of predictive efficacy, the AUROC of the mean elastic modulus model for differentiating steatosis was > 0.90. After adjusting for inflammation and ballooning, the AUROC of the mean elastic modulus value for distinguishing the steatosis degree was also ≥ 0.92, which means SWE has excellent predictive validity for grading the steatosis stages regardless of the presence or absence of inflammation and ballooning.
SWE can quantitatively reflect tissue stiffness according to shear wave propagation in the tissue expressed as Young's modulus. Zhaoke Pi et al. [
28] showed that the elasticity values µ attained by SWE in vivo had a significant correlation with liver steatosis in NAFLD in a mouse model. Grimal et al. [
29] found a steadily increase of median liver stiffness in rats with NASH, measured by SWE, with exacerbation of steatosis grade. Similarly, our results also suggested that steatosis is highly associated with liver mean elastic modules. This could be explained because the presence of fat droplets within hepatocytes affects the liver structure and may alter the propagation time of vibration waves in the liver, a crucial principle of 2D-SWE.
However, several studies using SWE found no correlation between hepatic steatosis and elasticity [
30‐
32]. Possible factors accounting for this difference could be experimental operators, steatosis type, unadjusted confounding factor, and different diets animal models. In rats fed with MCD, mainly macrovesicular steatosis developed, whereas mainly microvesicular steatosis was observed in the rats fed with a choline deficient diet [
33]. Previous study had suggested that microvesicular steatosis was associated significantly with advanced fibrosis, but macrovesicular steatosis was not [
34]. Therefore, different types of steatosis may affect the LSM through SWE due to the presence or absence of fibrosis. Besides, most research investigated the relationship between fibrosis with steatosis and the elastic value by subgroup analysis. A bidirectional relationship between steatosis and fibrosis is found in patients with NAFLD. Specifically, hepatic steatosis can promote fibrosis in the early stages of NAFLD, while advanced fibrosis or cirrhosis reduces the steatosis degree [
35]. Consequently, it is difficult to distinguish their separate effects on LSM without adjusting for confounding factors.
In addition, multivariate regression analysis demonstrated that neither inflammation nor ballooning was independently related to liver stiffness, in accordance with previous studies [
36,
37]. However, our findings differed from Sugimoto et al.’s [
38] and Takeuchi et al.’s [
39]. This may be due to differences in experimental subjects, viscoelasticity parameters, and the inherent limitations of non-invasive imaging methods. Interestingly, some studies in NAFLD patients pointed out that inflammation affected LSM accuracy with TE rather than with SWE [
17,
31]. Accordingly, inflammation may lead to an increase in LSM with TE but not with 2D-SWE; a combination of TE and 2D-SWE could help identify suitable individuals with inflammation for participation in clinical trials.
Hepatic steatosis and lobular inflammation are strongly associated with NASH progression [
40]. To diagnose NASH as early as possible, more attention should be drawn for diagnosis of steatosis. In our study, a rat model was used to limit the confounding effects of fibrosis; our results demonstrated that steatosis was an independent factor affecting elastic modules, and the AUROC of mean elastic modulus value is > 0.90. Our results indicated that LSM can provide useful information on the status of hepatic steatosis obtained by SWE with or without inflammation and ballooning. Compared to other elastography techniques, SWE can detect tissue elasticity in real time and steadily when observing two-dimensional morphology, allowing to easily select the ideal ROI for measurement of liver elasticity, thus offering more reliable estimates of elastic values.
Some advantages of our study are worth recapitulating. Firstly, at present the “gold standard” for diagnosing NAFLD is liver biopsy. However, it is not suitable for wide application due to its invasiveness, sampling error, and associated complications. Therefore, we established an animal NAFLD model to obtain imaging information and pathological data and accurately assess NAFLD histological features in SWE. Secondly, we demonstrated the influence of steatosis on liver stiffness measured by SWE by adjusting for inflammation and ballooning, and excluding fibrosis. We obtained the estimated effect through multiple regression and determined an association between elasticity and steatosis stages. Thirdly, we evaluated the predictive performance of the mean elastic modulus value for steatosis by ROC analysis, as well as the AUC value adjusted by inflammation and ballooning, and DCA and AUDCA were calculated in this study.
Several limitations should be acknowledged. Firstly, an NAFLD animal model could develop a single pathological feature like steatosis, then evaluating its impact on liver stiffness measurement, which would be preferred. With prolonged feeding of MCD, rats develop other pathological features, such as inflammation and fibrosis, that may interfere with the purpose of this study, even though we utilized the statistical approach to adjust. Secondly, to explore the association between steatosis and elastic modules, we excluded rats with fibrosis, as it has been associated with liver stiffness [
16], which leads to a relatively small sample size and may induce sampling error. In addition, the proportion of rats with different degrees of steatosis was uneven, especially in the severe steatosis group. Thirdly, the animal model in our study were Sprague–Dawley rats. Rodents and humans have markedly different metabolic rates, affecting liver homeostasis [
30]. Consequently, NAFLD pathogenesis may not exactly replicate in humans; so further research is required to validate and improve SWE accuracy.
In conclusion, mean elasticity was significantly associated with hepatic steatosis rather than ballooning or inflammation. The elasticity measured by SWE can reflect the grades of hepatic steatosis to a certain extent. Despite the usefulness of liver biopsy, SWE may have the feasibility to be introduced as an assistive technology for patients with NAFLD in grading steatosis. Further studies with prospective, more scientific research modalities and larger sample sizes are needed to clinically validate our results.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.