NAFLD evaluation: Which is more appropriate,multislice computed tomography or ultrasound real-time shear wave elastography?

Abstract

OBJECTIVE:

This study aims to assess the value of ultrasound real-time shear wave elastography (US-SWE) for evaluation of nonalcoholic fatty liver disease (NAFLD) in a rabbit model compared with multislice computed tomography (MSCT).

MATERIAL AND METHODS:

Twenty-six rabbits were fed with high-fat, high-cholesterol diet and six rabbits were fed with a standard diet. All rabbits were performed with MSCT and US-SWE at various time points to measure changes in liver parenchyma. The diagnostic efficiency of US-SWE was analyzed using receiver operating characteristics (ROC) curves compared with MSCT based on the liver pathology.

RESULTS:

The statistically significant differences in the areas under the ROC curves between using MSCT and US-SWE modalities were detected to discriminate between normal vs. NAFLD or higher severity pathology. Similarly, for normal or NAFLD vs. borderline or NASH livers, statistically significant differences between using US-SWE and MSCT modalities were also detected for nonalcoholic steatohepatitis (NASH) vs. lower severity pathology.

CONCLUSIONS:

MSCT, but not US-SWE, had a better ability to differentiate normal or NAFLD livers from higher severity NAFLD livers. However, the diagnostic efficiency of US-SWE was superior to that of MSCT for differentiating NASH from normal or lower severity NAFLD.

Keywords

Shear wave elastography nonalcoholic fatty liver disease nonalcoholic steatohepatitis multislice computed tomography diagnostic efficiency

1 Introduction

NAFLD is defined as intracellular deposition of fat droplets exceeding 5% of triglycerides in the patients without clinically significant alcohol intake, infection or any other liver disease induced by specific cause [1–4].The earliest and most significant histological characteristic of NAFLD is the presence of intracellular triglyceride accumulation within hepatocytes. Oxidative stress is caused by hepatic intracellular triglycerides deposition, leading to simple steatosis, liver injury, inflammation, NASH, and eventually progresses to hepatocellular carcinoma and cirrhosis [5, 6].NAFLD has been confirmed as a risk factor for cardiovascular morbidity and metabolic disease (including type-2 diabetes) [7–9]. Specifically, NAFLD severity groups were classified into normal, NAFLD, borderline NASH, and NASH according to NAFLD activity score as following [18]: (1) Normal: activity score = 0; (2) NAFLD: activity score = 1, 2; (3) Borderline NASH: activity score = 3, 4; and (4) NASH: activity score ≥5. Therefore, early diagnosis of severity and degree of NAFLD is of great importance for clinical practice.

Currently, there are no specific serum biomarkers for hepatic fat content quantification. Aminotransferases are not sensitive and specific enough for detection of liver fat content. Liver biopsy is considered the gold standard for evaluation of NAFLD. However, the procedure is not widely available because of its invasiveness. It is crucial for clinical practice to accurately quantify the liver with NAFLD or higher severity livers noninvasively.

Non-invasive imaging modalities for diagnosis of NAFLD include ultrasound (US), computed tomography (CT) and magnetic resonance imaging (MRI) [10, 11]. However, conventional imaging modalities do not detect specific signs of NAFLD owing to many factors. The non-imaging technique of controlled attenuation parameters (CAP) is an objective assessment for ultrasound attenuation, which can be performed to measure the hepatic steatosis by using the appropriately selected FibroScan probe [12, 13]. This parameter can be used as a validated marker of steatosis. MR spectroscopy (MRS) is a technique that the relative strength signals of water and fat were determined by frequency-domain analysis of MRS data. Liver fat quantification can be evaluated by MRS, which requires physicist or technologist to set up the sequence and acquire data, the MRS analysis software is also needed to analyze the data. Liver fat content was assessed with MRS, however, MRS is expensive and not broadly available, the accuracy of magnetic resonance-derived methods may decrease with significant fibrosis [14]. In comparison to MRI, the advantages of US or MSCT are its relatively low cost and ease of access [11]. A recent study showed that a novel technique, quantitative US-SWE, may be useful to measure the severity and degree of liver cirrhosis [15–17].

As for the quantitative assessment of liver density by MSCT, which is based on X-ray attenuation value or density that corresponds to the “brightness” of the pixels. MSCT is less prone to anatomical constraints than US allowing complete visualization of the entire liver within a short breath-hold. With increasing fat accumulation, the CT value of liver parenchymal decreases. Therefore, the severity of hepatocellular steatosis may be represented by means of a continuous range of CT attenuation values, the CT attenuation values of fatty liver range from that of normal liver parenchyma to that of fat tissue [11]. According to this theory, so we studied the diagnostic efficiency of MSCT for assessment of NAFLD in the rabbit model in comparison to US-SWE in a rabbit NAFLD model.

2 Materials and methods

This study was approved by the animal care and use committee of Guizhou Medical University.

2.1 The study design for rabbit NAFLD model

Thirty-six New Zealand white rabbits (male = 18, female = 18), 8-weeks-old, were randomly classified into control group (n = 9) or experimental groups (9 rabbits/group). Randomization procedure: the randomization procedure was based on the random number and block randomization by using SPSS software (version 23.0). The control group was fed a standard diet. The experimental group was fed a high-fat and high-cholesterol diet (standard diet with an additional 2% cholesterol, 6% yolk powder, 10% lard, and 2% maltose dextrin). All rabbits were performed US-SWE and MSCT before experiment begun, and then underwent US and MSCT at various time points: 4 weeks (group 4W), 8 weeks (group 8W), or 12 weeks (group 12W) for different groups, respectively. The body weight also was also recorded. This process was conducted by the Laboratory Animal Research Center of Guizhou Medical University (Guiyang, China). The study design was as following (Fig. 1):

Fig.1

A flow chart showing our study design of NAFLD rabbit models. 36 rabbits were randomly classified into a control group (n = 9) or an experimental group (n = 27). Control group was fed with a standard diet, and all experimental groups were fed with a high-fat and high-cholesterol diet; All the rabbits underwent US-SWE and MSCT prior to being fed with standard diet or high-fat and high-cholesterol diet. Rabbits in the experimental groups were sacrificed either at 4 weeks (n = 9), 8 weeks (n = 9), or 12 weeks (n = 9). Control rabbits were sacrificed at 8 weeks. All liver tissue specimens were stained with H&E and Masson trichrome staining. Histological features were classified as follows: normal, NAFLD, borderline NASH, and NASH in our study.

2.2 MSCT protocols

After the US-SWE examination, all rabbits underwent 128-slice MSCT scanning (SOMATOM definition AS+, Siemens, Germany). The scan parameters were as follows: Tube voltage 120 kVp, tube current 250 mA, collimation 128×0.6 mm, slice thickness1 mm, FOV 500 mm.

2.3 Ultrasound examination

All rabbits underwent ultrasound imaging by the Supersonic Imaging Aixplorer with a linear array probe (L15-4) and a frequency of 4–15 Hz. Skin preparation was performed by the ultrasonographer. Firstly, we obtained a conventional B-mode gray-scale image for assessment of the entire liver parenchyma. Secondly, SWE imaging was performed on the liver parenchyma for assessment of liver stiffness.

2.4 Image data analysis

All the rabbits underwent MSCT and US-SWE on the same day. All CT images datasets were reconstructed at a workstation (Syngo CT Workplace VE40B). Three ROIs (diameter 0.7 cm–1.0 cm) were measured on the liver parenchyma of each rabbit by an experienced abdominal radiologist who was blinded to the rabbit’s diet. Care was taken to avoid large bile ducts and hepatic vessels on the CT image. One ROI was placed in the left lateral lobe, one in the left medial lobe, and the third in the right lobe, using the left hepatic vein and the gallbladder as anatomic landmarks. CT attenuation values of each ROI were recorded, average CT value of three ROIs was calculated for analysis. As for the ultrasound images of SWE, sample frames were full-color (≥95%) for stabilization and color homogeneity, three ROIs (diameter 0.7 cm–1.0 cm) were measured in the liver parenchyma by Q-Box software, avoiding intrahepatic vascular and bile ducts, the mean elastic modulus values were recorded and calculated for statistical analysis.

2.5 Histological analysis

Following all US-SWE and MSCT examinations, all rabbits were scarified under deep anesthesia with 5 ml of potassium chloride injected into the auricular veins. All liver tissue specimens were stained with Masson staining and hematoxylin and eosin staining, respectively. Pathological sections were reviewed by an experienced pathologist who was blinded to rabbit’s diet. The reference standards for pathological specimen were as follows [18]: NAFLD activity score was given for each pathological section: the unweighted sum of steatosis (score 0–3), lobular inflammation (score 0–3), and hepatocellular ballooning (score 0–2).

After the NAFLD activity score was recorded, each study subject was classified into various NAFLD severity groups [18]: Normal: activity score = 0; NAFLD: activity score = 1, 2; Borderline: activity score = 3, 4; and NASH: activity score ≥5.

The METAVIR score system was employed for evaluation of liver fibrosis [19]: F0 = no fibrosis; F1 = portal fibrosis without septa; F2 = portal fibrosis with few septa; F3 = a number of septa without cirrhosis; F4 = cirrhosis.

2.6 Data analysis

Statistical analysis was performed using SPSS software (version 23.0, SPSS Inc., Chicago, IL) and MedCalc software (MedCalc 15.2.2, Mariakerke, Belgium). Continuous variables were expressed as mean±SD. The normality assumption of variables was conducted according to the Kolmogorov-Smirnov Z test. The body weight, CT value, and elastic modulus value were tested by two-tailed paired-samples T Test before and after experiment for different groups. The CT value and elastic modulus value of liver parenchyma among the grades of different histological features were analyzed by one-way ANOVA or two-tailed independent-sample.

Diagnostic efficiency of MSCT was tested and compared with US-SWE; Corresponding sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the area under ROC curve (AUC) were also analyzed. 95% confidence interval (CI) was recorded. P < 0.05 was considered statistically significant difference.

3 Results

3.1 Characterization of NAFLD severity according to histologic features

Three rabbits died in control group before sacrifice date, and one rabbit died in group 12 W. At last, 32 rabbits (control group n = 6, experiment groups n = 26) were enrolled in our study. For the six animals in the control group, histological features were designated as normal liver.

For the nine rabbits in group 4W, all rabbits were fed 4 weeks on a high-fat, high-cholesterol diet, mean NAFLD activity score (NAS) was 2.56 (range, 2–4). Six rabbits were classified as NAFLD, three rabbits were borderline, but no rabbits were NASH or fibrotic.

Group 8W (n = 9) was maintained for 8 weeks on a high-fat, high-cholesterol diet. Mean NAS was 4.00 (range, 2–7). Two rabbits were classified as NAFLD, five as borderline, two as NASH. Two of five borderline and two NASH rabbits had the fibrosis of F1.

For group 12 W (n = 8), eight rabbits were fed 12 weeks of a high-fat, high-cholesterol diet. Mean NAS was 5.63 (range, 2–7). One rabbit was classified as NAFLD, one as borderline, six as NASH. One borderline and six NASH rabbits had pericellular fibrosis of F1.

In summary, according to histological feature scores, six rabbits were normal, nine had NAFLD, nine were borderline, eight had NASH, and eleven showed fibrosis of F1 (Table 1).

Table 1
The NAFLD severity livers at different time points

Sacrifice date Number of Rabbit Normal NAFLD Borderline NASH Fibrosis (F0/F1/F2/F3/F4)

Control (8 weeks) 6 6 0 0 0 6/0/0/0/0

Group 4 W (4 weeks) 9 0 6 3 0 9/0/0/0/0

Group 8 W (8 weeks) 9 0 2 5 2 5/4/0/0/0

Group 12 W (12 weeks) 8 0 1 1 6 1/7/0/0/0

Sacrifice date	Number of Rabbit	Normal	NAFLD	Borderline	NASH	Fibrosis (F0/F1/F2/F3/F4)
Control (8 weeks)	6	6	0	0	0	6/0/0/0/0
Group 4 W (4 weeks)	9	0	6	3	0	9/0/0/0/0
Group 8 W (8 weeks)	9	0	2	5	2	5/4/0/0/0
Group 12 W (12 weeks)	8	0	1	1	6	1/7/0/0/0

In addition, the body weight was increased in the control group, group 4 W, and group 12 W (all P < 0.05). As for the CT value of liver parenchymal, the CT value was decreased in all experiment groups (all P < 0.05). However, the elastic modulus value was increased in group 8 W and group 12 W (all P < 0.05). (Table 2 and Fig. 2).

Fig.2

Changes were observed in (A) body weight, (B) liver stiffness, and (C) liver density at various time points.

Table 2

Dynamic changes of body weight, US-SWE, and MSCT for evaluation of liver parenchyma at different time points for various groups

Subjects	Body Weight (kg)	US-SWE (kPa)	MSCT (HU)
Group 4 W
1 W	2.01±0.15	6.29±1.21	58.11±4.48
4 W	2.68±0.39	5.73±0.86	41.85±9.73
P value^*	<0.001	0.161	0.002
Group 8 W
1 W	2.14±0.29	6.88±1.08	60.56±7.74
8 W	2.59±0.49	12.27±3.97	28.74±4.03
P value^*	0.055	0.004	<0.001
Group 12 W
1 W	2.23±0.16	7.15±0.85	65.00±6.30
12 W	3.13±0.30	12.75±3.39	28.71±5.44
P value^*	<0.001	0.001	<0.001
Normal
1 W	2.32±0.15	7.11±0.73	58.50±4.73
8 W	3.48±0.36	7.23±0.70	56.54±5.50
P value^*	0.001	0.122	0.110

“*” All the data were tested with two-tailed paired-samples T Test.

3.2 The characteristic between liver density or stiffness and histological features

MSCT and US-SWE measurements for evaluation of histological features were shown in Fig. 3. There was statistically significant difference in steatosis grades for liver density and stiffness (P < 0.001 and P = 0.007, respectively). Statistically significant difference was observed in fibrosis grades (P = 0.002 and P = 0.026, respectively). Statistically significant difference was also existed in inflammation grades for MSCT and US-SWE (P = 0.001 and P < 0.001, respectively).

Fig.3

Box-plots of MSCT and SWE measurements for evaluation of histological features, respectively, including steatosis grade (A and B), Fibrosis grade (C and D), and inflammation grade (E and F). Boundaries of the boxes indicate the lower and upper percentiles; lines within the boxes indicate medians; and error bars indicate ranges for each grade of histological feature.

3.3 Comparison of diagnostic efficiency between US-SWE and MSCT

The statistically significant difference was observed to discriminate normal liver parenchyma from NAFLD livers between MSCT and US-SWE (Z = 4.599, P < 0.001). The difference of AUC between MSCT (AUC 0.949) and US-SWE (AUC 0.551) was 0.398 (95% CI 0.228, 0.567). For discrimination between normal liver or NAFLD versus borderline or NASH, there were also significant differences between US-SWE and MSCT (Z = 2.631, P = 0.009). The difference of AUC between MSCT (AUC 0.710) and US-SWE (AUC 0.941) was 0.231 (95% CI 0.059, 0.404); For discrimination of livers with NASH versus those with borderline or lower severity liver pathology, we also saw a statistically significant difference (Z = 2.003, P < 0.045). The difference of AUC between US-SWE (AUC 0.953) and MSCT (AUC 0.773) was 0.180 (95% CI 0.004, 0.355). The AUC, cutoff values for MSCT and US-SWE parameters, sensitivities, specificities, PPV, and NPV are displayed in Table 3, Figs. 4, and Figs. 5.

Fig.4

Box-plots show elastic modulus and density of liver parenchyma for normal, NAFLD, borderline, and NASH. (A) US-SWE for normal, NAFLD or higher severity pathology. The lowest line indicates the minimum value, the highest line indicates the maximum value, lines in the box indicate median, boundary of box indicate lower and upper quartile. (B) MSCT for normal, NAFLD or higher severity pathology.

Fig.5

ROC curve analysis for US-SWE and MSCT. (A) ROC curves show that there was significant statistical difference in the area under the ROC curve between MSCT and US-SWE to evaluate for normal vs. ≥NAFLD severity livers. (B) there was significant statistical difference between MSCT and US-SWE with respect to discrimination of normal or NAFLD vs. borderline or NASH livers. (C) Significant statistical difference was also existed between US-SWE and MSCT for NASH vs. lower severity livers.

Table 3

Receive operating characteristic curve analysis for NAFLD severity livers by US-SWE and MSCT

Parameters	Normal vs. ≥NAFLD	≤NAFLD vs. ≥ Borderline	≤Borderline vs. NASH
CT value cutoff (HU)	46.33	33.33	33
Sensitivity (%)	88.46 (69.8, 97.6)	88.24 (63.6, 98.5)	100 (63.1, 100.0)
Specificity (%)	100.00 (54.1, 100)	93.33 (68.1, 99.8)	70.83 (48.9, 87.4)
PPV (%)	100.00 (85.2, 100)	93.70 (69.8, 99.8)	53.3 (26.6, 78.7)
NPV (%)	66.70 (29.9, 92.5)	87.50 (61.7, 98.4)	100 (80.5, 100.0)
AUC	0.949 (0.808, 0.996)^*	0.941 (0.797, 0.993)^*	0.773 (0.592, 0.902)^*
E value cutoff (kPa)	8.04	8.04	8.04
Sensitivity (%)	46.15 (26.6, 66.6)	58.82 (32.9, 81.6)	100 (63.1, 100.0)
Specificity (%)	100 (54.1, 100)	86.67 (59.5, 98.3)	83.33 (62.6, 95.3)
PPV (%)	100 (73.5, 100)	83.30 (51.6, 97.9)	66.7 (34.9, 90.1)
NPV (%)	30 (11.9, 54.3)	65.0 (40.8, 84.6)	100 (83.2, 100.0)
AUC	0.551 (0.336, 0.727)	0.71 (0.523, 0.856)	0.953 (0.814, 0.997)

“*” In comparison to US-SWE, there was significant statistical difference in the area under ROC curve, P < 0.05.

4 Discussion

The results we got were that we detected a significant difference between US-SWE and MSCT to evaluate the NAFLD severity in a rabbit model. For the ability to discriminate 1) normal liver versus NAFLD, and 2) normal liver or NAFLD versus borderline liver or NASH, the diagnostic efficiency of MSCT was superior to that of US-SWE. However, for discrimination of livers with NASH versus borderline or lower pathological severity livers, the diagnostic performance of US-SWE was superior to that of MSCT. This suggests that MSCT and US-SWE provide different diagnostic efficiencies in diagnosis of the severity of NAFLD.

We chose the rabbit as NAFLD model in our study due to its size. The body weight of rabbit was observed in our study because we needed to know the volume of pentobarbital sodium for anesthesia before we performed US-SWE and MSCT for assessing the liver stiffness and density dynamically. In our study, the body weight was decreased at 8 weeks compared with control group. Histological results shown that the rabbit NAFLD livers had pericellular fibrosis by Masson staining after 8 weeks. It is hepatic fibrosis and damage that previous studies suggested may be associated with body weight; In addition, liver fibrosis presented with anorexia and emaciation were observed in animal study [20, 21]. In our study, the liver had fibrosis in the borderline NASH or NASH rabbit model after 8 weeks (Fig. 6).

Fig.6

Ultrasound SWE image for assessment of liver parenchyma. The normal liver of (A) MSCT, (B) US-SWE, (C) hematoxylin and eosin staining, and (D) Masson staining; the NAFLD liver of (E) MSCT, (F) US-SWE, (G) hematoxylin and eosin staining, and (H) Masson staining.

Histological features for NAFLD livers were classified as normal, NAFLD, borderline NASH, and NASH according to the unweighted sum of steatosis, lobular inflammation, and hepatocellular ballooning. In our study, the CT value of liver parenchyma was decreased when increasing the steatosis grades, but the stiffness of liver parenchyma was increased. Previous human or animal studies have shown that the liver stiffness increased when increasing the fibrosis stages or inflammation grades [22–24]. In our study, the elastic modulus value of F1 was higher than that of F0, and the elastic modulus value was increased when increasing the inflammation grades. However, the histological features of steatosis, fibrosis, and inflammation coexisted in the NAFLD liver specimen in our study (11/26). Therefore, it is very difficult to evaluate the difference of liver density or stiffness induced only by steatosis, fibrosis or inflammation, respectively.

The main advantages of conventional ultrasound are ease of access and low cost. However, many factors affect the quality and accuracy of ultrasound, such as type of equipment, the experience of operator, and anatomical constraints [25]. In addition, sonographic images are confounded in advanced NAFLD because of coexisting inflammation and fibrosis, therefore, conventional ultrasound is not recommended for diagnosis of mild steatosis or for assessing the severity of NAFLD [26–28]. For mild steatosis, the diagnostic sensitivity and specificity of sonographic features were 62.2–82.1% and 76.2% –90.1%, respectively [10]. To overcome the subjectivity of conventional ultrasound, several quantitative ultrasound techniques for assessment of liver lesions have been utilized [29–31]. Real-time SWE is a quantitative technique with excellent severity grading performance and reproducibility [32]. Several studies have shown that SWE may be used in quantitative ultrasound elastography to assess liver stiffness associated with cirrhotic features [33–35].

In general, fatty liver disease develops from a simple fatty liver to steatohepatitis, and eventually to fatty cirrhosis [36, 37]. However, there are exceptions, for example, fatty liver disease may not exhibit clear steatohepatitis, but directly evolve to cirrhosis [36]. In group fed with high-fat and high-cholesterol diet for 12 weeks, we suspect that higher elastic modulus value may be associated with the histological feature of fibrosis.

MSCT is based on the attenuation values which are standardized by means of Housfield Units (HU). On the unenhanced CT image, the attenuation value of liver parenchyma is approximately 60±10 HU. Subcutaneous adipose tissue is approximately –90 HU. Therefore, the severity of hepatocellular steatosis may be represented by means of a continuous range of CT attenuation values from normal liver parenchyma to fat tissue [11]. However, CT attenuation value can be confounded by many factors for evaluation of hepatic steatosis [38]. In addition, accumulation of water, iron, collagen, and glycogen often occurs in the diffuse liver disease [39]. All these factors may affect the CT attenuation value of liver parenchyma and may confound the evaluation of hepatic steatosis. In our study, the CT attenuation value of rabbit liver parenchyma decreased gradually in all experimental groups.

The mean elastic modulus value increased as the severity of NAFLD increased (P < 0.05). There was a statistically significant difference in discrimination capability for NASH from borderline or lower severity pathology, with an AUC value of 0.953 (0.814, 0.997). In comparison to MSCT, the diagnostic efficiency of SWE was superior to that of MSCT (P < 0.05). Using MSCT, the CT number for liver was decreased compared with normal liver as the severity of NAFLD increased (P < 0.05). There was a statistically significant difference between modalities for discrimination between normal liver and liver with various severities of pathology. However, the diagnostic efficiency of MSCT was superior to that of US-SWE (P < 0.05) for differentiating 1) normal liver from NAFLD, and 2) normal or NAFLD from borderline or NASH; AUC value 0.949 (0.808, 0.996) and 0.940 (0.797, 0.993), respectively.

There are some limitations in our study. First, the sample size was small, including the number of subjects in the control group and experimental group. Second, we used a semiquantitative grading system for evaluation of histological features, including steatosis, hepatocellular ballooning, lobular inflammation, and fibrosis. Our study showed that the histological feature of rabbit liver specimens with NAFLD was micro-vesicular steatosis predominately (24/26). Previous studies demonstrated a correlation between histological features of micro-vesicular steatosis and advanced histological results of NAFLD [40]. Third, the information about fat content of liver tissue was not measured directly in our study, which may better reveal the steatosis of NAFLD livers. Fourth, ROIs were not a consistent one-to-one match for measuring the density and stiffness of the liver parenchyma by MSCT and US-SWE, respectively. Last, the repeatability and reproducibility of US-SWE and MSCT were not performed in this study.

In conclusion, the results we got are promising. As for these two quantitative techniques of US-SWE and MSCT, different modalities with different diagnostic efficiencies for assessing the different stages of NAFLD livers. MSCT, but not US-SWE, had a better ability to differentiate normal or early NAFLD livers from higher severity NAFLD livers, which suggests that MSCT could be recommended to evaluate or monitor the early stage of NAFLD livers. However, the diagnostic efficiency of US-SWE was superior to that of MSCT for differentiating NASH from normal or lower severity NAFLD, which suggests that US-SWE could be recommended to assess or follow up the dynamic changes for the advanced stage of NAFLD livers. It means that the diagnostic efficiencies of these two techniques are mutually supplementary for assessment of NAFLD.

Conflict of interest

The authors do not have any disclosures to report.

Footnotes

Acknowledgments

We are thankful to Dr. Lujun Dai and Dr. Yining Xiang who supported the techniques of H&E staining and Masson trichrome staining.

This study was supported by Joint Fund of Guizhou Province Department of Science and Technology (No. LH[2017]7208); National Natural Science Foundation of China (No. 81760312); Guizhou Province Clinical Discipline Subject Fund (No. QWBF2011[52]-03); Doctoral Research Initiation Fund of the Affiliated Hospital of Guizhou Medical University.

References

Blachier

Leleu

Peck-Radosavljevic

, et al.,The burden of liver disease in Europe: A review of available epidemiological data, J Hepatol 58(3) (2013), 593–608.

Bedossa

, Current histological classification of NAFLD: Strength and limitations, Hepatol Int 7(Suppl 2) (2013), 765–770.

Bedossa

, Histological Assessment of NAFLD, Dig Dis Sci 61(5) (2016), 1348–1355.

Petäjä

E.M.

and Yki-Järvinen

, Definitions of Normal Liver Fat and the Association of Insulin Sensitivity with Acquired and Genetic NAFLD-A Systematic Review, Int J Mol Sci 17(5) (2016), 633.

Hagström

Nasr

Ekstedt

, et al., Fibrosis stage but not NASH predicts mortality and time to development of severe liver disease in biopsy-proven NAFLD, J Hepatol 67(6) (2017), 1265–1273.

Jennings

Faselis

and Yao

M.D.

, NAFLD-NASH: An under-recognized epidemic, Curr Vasc Pharmacol 16 (2018)209–213.

Mikolasevic

Orlic

Milic

, et al., Nonalcoholic fatty liver disease (NAFLD) proven by transient elastography in patients with coronary heart disease, Wien Klin Wochenschr 126(15-16) (2014), 474–479.

Patel

P.J.

Hayward

K.L.

Rudra

, et al., Multimorbidity and polypharmacy in diabetic patients with NAFLD: Implications for disease severity and management, Medicine (Baltimore) 96(26) (2017), e6761.

Munteanu

M.A.

and Mircea

P.A.

, From NAFLD to Cardiovascular Disease. Is it (Still) the Metabolic Syndrome, Clujul Med 87(2) (2014), 80–86.

10.

Bohte

A.E.

van Werven

J.R.

Bipat

and Stoker

, The diagnostic accuracy of US, CT, MRI and 1H-MRS for the evaluation of hepatic steatosis compared with liver biopsy: A meta-analysis, Eur Radiol 21(1) (2011), 87–97.

11.

Kinner

Reeder

S.B.

and Yokoo

, Quantitative Imaging Biomarkers of NAFLD, Dig Dis Sci 61(5) (2016), 1337–1347.

12.

Pembroke

Deschenes

Lebouché

, et al., Hepatic steatosis progresses faster in HIV mono-infected than HIV/HCV co-infected patients and is associated with liver fibrosis, J Hepatol 67 (2017), 801–808.

13.

Caussy

Alquiraish

M.H.

Nguyen

, et al., Optimal threshold of controlled attenuation parameter with MRI-PDFF as the gold standard for the detection of hepatic steatosis, Hepatology 6 (2017), doi:10.1002/hep.29639

14.

Machado

M.V.

and Cortez-Pinto

, Non-invasive diagnosis of non-alcoholic fatty liver disease, A critical appraisal, J Hepatol 58 (2013), 1007–1019.

15.

Pawluś

Inglot

Chabowski

, et al., Shear wave elastography (SWE) of the spleen in patients with hepatitis B and C but without significant liver fibrosis, Br J Radiol 89 (2016), 20160423.

16.

Sporea

Bota

Gradinaru-Taşcău

Sirli

Popescu

and Jurchiş

, Which are the cut-off values of 2D-Shear Wave Elastography (2D-SWE) liver stiffness measurements predicting different stages of liver fibrosis, considering Transient Elastography (TE) as the reference method, Eur J Radiol 83(3) (2014), e118–122.

17.

Jeong

W.K.

Lim

H.K.

Lee

H.K.

J.M.

and Kim

, Principles and clinical application of ultrasound elastography for diffuse liver disease, Ultrasonography 33(3) (2014), 149–160.

18.

Kleiner

D.E.

Brunt

E.M.

Van Natta

, et al., Design and validation of a histological scoring system for nonalcoholic fatty liver disease, Hepatology 41(2005), 1313–1321.

19.

Poynard

Bedossa

and Opolon

, Natural history of liver fibrosis progression in patients with chronic hepatitis C. The OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups, Lancet 349(9055) (1997), 825–832.

20.

Peretz

Link

Pappo

Bruck

and Ackerman

, Effect of hepatic iron concentration reduction on hepatic fibrosis and damage in rats with cholestatic liver disease, World J Gastroenterol 12(6) (2006), 240–245.

21.

Riet-Correa

Carvalho

K.S.

Riet-Correa

, et al., Periacinar liver fibrosis caused by Tephrosia cinerea in sheep, Res Vet Sci 95(1) (2013), 200–203.

22.

Verlinden

Bourgeois

Gigase

, et al., Liver Fibrosis Evaluation Using Real-time Shear Wave Elastography in Hepatitis C-Monoinfected and Human Immunodeficiency Virus/Hepatitis C-Coinfected Patients, J Ultrasound Med 35(6) (2016), 1299–1308.

23.

Dhyani

Gee

M.S.

Misdraji

Israel

E.J.

Shah

and Samir

A.E.

, Feasibility study for assessing liver fibrosis in paediatric and adolescent patients using real-time shear wave elastography, J Med Imaging Radiat Oncol 59(6) (2015), 687–694.

24.

Kang

B.K.

Lee

S.S.

Cheong

Hong

S.M.

Jang

and Lee

M.G.

, Shear Wave Elastography for Assessment of Steatohepatitis and Hepatic Fibrosis in Rat Models of Non-Alcoholic Fatty Liver Disease, Ultrasound Med Biol 41 (2015), 3205–3215.

25.

Bohte

A.E.

Koot

B.G.

van der Baan-Slootweg

O.H.

, et al., US cannot be used to predict the presence or severity of hepatic steatosis in severely obese adolescents, Radiology 262(1) (2012), 327–334.

26.

Dasarathy

Khiyami

Joseph

Lopez

and McCullough

A.J.

, Validity of real time ultrasound in the diagnosis of hepatic steatosis: A prospective study, J Hepatol 51(6) (2009), 1061–1067.

27.

Tobari

Hashimoto

Yatsuji

Torii

and Shiratori

, Imaging of nonalcoholic steatohepatitis: Advantages and pitfalls of ultrasonography and computed tomography, Intern Med 48(10) (2009), 739–746.

28.

Lee

S.S.

and Park

S.H.

, Radiologic evaluation of nonalcoholic fatty liver disease, World J Gastroenterol 20(23) (2014), 7392–7402.

29.

Z.F.

Zagzebski

J.A.

and Lee

F.T.

, Ultrasound backscatter and attenuation in human liver with diffuse disease, Ultrasound Med Biol 25(7) (1999), 1047–1054.

30.

Xia

M.F.

Yan

H.M.

W.Y.

, et al., Standardized ultrasound hepatic/renal ratio and hepatic attenuation rate to quantify liver fat content: An improvement method, Obesity (Silver Spring) 20(2) (2012), 444–452.

31.

Webb

Yeshua

Zelber-Sagi

, et al., Diagnostic value of a computerized hepatorenal index for sonographic quantification of liver steatosis, AJR Am J Roentgenol 192(4) (2009), 909–914.

32.

Yoon

H.M.

Cho

Y.A.

Kim

J.R.

, et al., Real-time two-dimensional Shear-wave elastography for liver stiffness in children: Interobserver variation and effect of breathing technique, Eur J Radiol 97 (2017), 53–58.

33.

Wei

Tang

, et al., Evaluation of fatty liver fibrosis in rabbits using real-time shear wave elastography, Exp Ther Med 8(2) (2014), 355–362.

34.

Wang

Zhang

, et al., Comparison of Two-Dimensional Shear Wave Elastography and Real-Time Tissue Elastography for Assessing Liver Fibrosis in Chronic Hepatitis B, Dig Dis 34 (2016), 640–649.

35.

Poynard

Pham

Perazzo

, et al., Real-Time Shear Wave versus Transient Elastography for Predicting Fibrosis: Applicability, and Impact of Inflammation and Steatosis. A Non-Invasive Comparison, PLoS One 11(10) (2016), e0163276.

36.

Diehl

A.M.

, Liver disease in alcohol abusers: Clinical perspective, Alcohol 27(1) (2002), 7–11.

37.

Brunt

E.M.

, Nonalcoholic steatohepatitis, Semin Liver Dis 24(1) (2004), 3–20.

38.

Birnbaum

B.A.

Hindman

Lee

and Babb

J.S.

, Multi-detector row CT attenuation measurements: Assessment of intra- and interscanner variability with an anthropomorphic body CT phantom, Radiology 242(1) (2007), 109–119.

39.

Fritz

G.A.

Schoellnast

Deutschmann

H.A.

, et al., Density histogram analysis of unenhanced hepatic computed tomography in patients with diffuse liver diseases, J Comput Assist Tomogr 30(2) (2006), 201–205.

40.

Tandra

Yeh

M.M.

Brunt

E.M.

, et al., Presence and significance of microvesicular steatosis in nonalcoholic fatty liver disease, J Hepatol 55(3) (2011), 654–659.