Acoustic radiation force impulse-imaging and transient elastography for non-invasive assessment of liver fibrosis and steatosis in NAFLD

European Journal of Radiology 81 (2012) e325–e331

Contents lists available at SciVerse ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Acoustic radiation force impulse-imaging and transient elastography for non-invasive assessment of liver fibrosis and steatosis in NAFLD Mireen Friedrich-Rust a,∗ , Daniela Romen a , Johannes Vermehren a , Susanne Kriener b , Dilek Sadet a , Eva Herrmann c , Stefan Zeuzem a , Joerg Bojunga a,1 a

Department of Internal Medicine, J.W. Goethe-University Hospital, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany Institute of Pathology, J.W. Goethe-University Hospital, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany c Institute of Biostatistics and Mathematical Modelling, Faculty of Medicine, J.W. Goethe-University, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany b

a r t i c l e

i n f o

Article history: Received 23 July 2011 Received in revised form 29 October 2011 Accepted 31 October 2011 Keywords: Ultrasound ARFI CAP FibroScan Steatohepatitis Liver biopsy

a b s t r a c t Background: Transient elastography (TE) and acoustic radiation force impulse (ARFI)-imaging have shown promising results for the staging of liver fibrosis. Aim: The aim of the present study was to compare ARFI of the left and right liver lobe with TE using the standard and obese probes for the diagnosis of liver fibrosis in NAFL/NASH. In addition, liver steatosis is evaluated using the novel controlled attenuation parameter (CAP). Methods: Sixty-one patients with NAFLD/NASH were included in the study. All patients received TE with both probes, ARFI of both liver lobes and CAP. The results were compared with liver histology. Results: 57 patients were included in the final analysis. The diagnostic accuracy for TE measurements with the M-and XL-probe and for ARFI of the right and left liver lobe was 0.73, 0.84, 0.71 and 0.60 for the diagnosis of severe fibrosis, and 0.93, 0.93, 0.74 and 0.90 for the diagnosis of cirrhosis, respectively. No significant difference of results was observed between TE and ARFI in the subgroup of patients with reliable TE-measurement when taking into account the best results of both methods. However, while a significant correlation could be found for TE with histological liver fibrosis, the correlation of ARFI with liver fibrosis was not statistically significant. A significant correlation was found for CAP with histological steatosis (r = 0.49, p < 0.001). Conclusions: No significant difference in diagnostic accuracy for the non-invasive assessment of liver fibrosis was found for transient elastography and ARFI. Nevertheless TE significantly correlated with liver fibrosis while ARFI did not. CAP enables the non-invasive assessment of steatosis. © 2011 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Non-alcoholic fatty liver disease (NAFLD) is rapidly becoming a major health concern due to the increasing obesity epidemic and its potential to progress to liver fibrosis, cirrhosis and hepatocellular carcinoma [1]. At present, the prevalence of NAFLD in the United States is estimated to be as high as 30–50% [2]. Nonalcoholic steatohepatitis (NASH) represents a clinically important subset of NAFLD with increased risk for fibrosis progression and mortality [3]. Early detection of NASH-associated fibrosis is crucial for the prognosis of disease progression. However, identification of

∗ Corresponding author. Tel.: +49 069 6301 5297; fax: +49 069 6301 6247. E-mail addresses: [email protected] (M. Friedrich-Rust), [email protected] (J. Bojunga). 1 Tel.: +49 069 6301 87686; fax: +49 069 6301 6448. 0720-048X/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2011.10.029

patients with fibrosis has been difficult because liver biopsy was traditionally required for diagnosis. Liver biopsy is associated with substantial patient discomfort and carries a risk for complications [4]. In addition, liver biopsy is prone to sampling errors and intraand inter-observer variability [5]. Recently, advances have been made in the development of non-invasive methods for assessment and follow-up of patients with liver fibrosis. Transient elastography (TE; FibroScan® ; Echosens, Paris, France) has shown excellent diagnostic value in the detection of advanced fibrosis and cirrhosis in patients with NAFLD/NASH [6]. However, obesity, a major risk factor for NAFLD/NASH, was associated with TE failure in up to 25% of patients [6]. This limitation was recently overcome by the introduction of a new XL-probe with improved diagnostic utility in obese patients [7]. Acoustic radiation force impulse (ARFI) imaging (ACUSON S2000TM ; Siemens Medical Solutions, Mountain View, CA, USA) represents another promising ultrasound-based method for the

e326

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

assessment of liver stiffness. ARFI is integrated into a conventional ultrasound system and preliminary results have shown that even severe obesity is not a limitation for this technique [8]. However, it remains unclear how ARFI imaging compares to TE, including TE with the new XL-probe, in patients with NAFLD/NASH. The aim of the present study was to assess the diagnostic accuracy of ARFI imaging of the left and right liver lobe in comparison to TE with both standard (M) and obese (XL) probes for the diagnosis of fibrosis and cirrhosis in patients with NAFLD/NASH. Liver histology was used as the reference method. In addition, non-invasive measurement of liver steatosis was evaluated with a novel controlled attenuation parameter (CAP) that uses the same radio-frequency data as acquired by TE. 2. Materials and methods

Table 1 Patients’ characteristics. Characteristics

Patients (n = 57)

Sex Age

30 male/27 female patients Mean ± SD: 45 ± 14 years; median: 45 years; range: 21–71 years Mean ± SD: 98 ± 14 mm; median: 97 mm; range: 65–140 mm Mean ± SD: 104 ± 11 mm; median: 102 mm; range: 80–140 mm Mean ± SD: 25 ± 7 mm; median: 24 mm; range: 15–45 mm Mean ± SD: 28 ± 5.5 kg/m2 ; median: 27.8 kg/m2 ; range 18–43 kg/m2 Mean ± SD: 50 ± 27 IU/L; median: 43 IU/L; range: 18–136 IU/L Mean ± SD: 72 ± 53 IU/L; median: 58 IU/L; range: 13–275 IU/L Mean ± SD: 161 ± 27 IU/L; median: 58 IU/L; range: 14–2580 IU/L Mean ± SD: 0.75 ± 0.38 mg/dL; median: 0.6 mg/dL; range: 0.3–2.0 mg/dL Mean ± SD: 235 ± 53 × 103 /mm3 ; median: 242 × 103 /mm3 ; range: 70–353 × 103 /mm3 Mean ± SD: 206 ± 52 mg/dL; median: 205 mg/dL; range: 94–363 mg/dL Mean ± SD: 185 ± 154 mg/dL; median: 131 mg/dL; range: 22–972 mg/dL Mean ± SD: 102 ± 33 mg/dL; median: 91 mg/dL; range: 69–266 mg/dL

Waist circumference Hip circumference Skin capsule distance BMI AST ALT GGT

2.1. Patients Total bilirubin

Sixty-one patients with NAFLD or NASH were enrolled consecutively between May 2009 and December 2010. All patients received acoustic radiation force impulse (ARFI)-imaging of the right and left liver lobe (Siemens, Mountain View, CA), as well as transient elastography (FibroScan® ) with the standard probe (Mprobe) and the obese probe (XL-probe) (Echosens, Paris, France) on the same day of presentation. The distance between skin and liver capsule at the site of ARFI and TE measurement was measured using conventional ultrasound. Diagnosis of NAFLD or NASH was made histologically by liver biopsy. As the mean progression rate of liver fibrosis is low, a time interval between liver biopsy and study inclusion of up to 18 months was accepted for enrolment in the present study. The time interval between liver biopsy and study inclusion ranged from 0 to 17 months (median 3.0 months, mean 4.3 ± 4.0 months). Men with alcohol consumption of more than 30 g of alcohol per week and women with more than 20 g of alcohol per week were excluded from the study. In addition, patients with other causes of liver disease (positive hepatitis B surface antigen or anti-hepatitis C virus antibody, positive autoantibodies) or histological evidence of other concomitant chronic liver diseases were excluded. Patient characteristics and biochemical values are shown in Table 1. The present study was performed in accordance with the ethical guidelines of the Helsinki Declaration and was approved by the local ethics committee. Written informed consent was obtained from all patients.

Platelet count Total cholesterol Triglycerides Fasting glucose Histological fibrosis stage F0 F1 F2 F3 F4

21 patients 20 patients 5 patients 9 patients 2 patients

NAS score NAS ≤ 4 NAS > 4

20 patients 37 patients

Histological steatosis grade S0 S1 S2 S3

1 patients 14 patients 18 patients 24 patients

SD, standard deviation; pat., patients; BMI, body mass index; AST, aspartate aminotransaminase; ALT, alanine aminotransaminase; GGT, gamma-glutamyl transpeptidase.

biopsy at least 1 cm or if the number of portal tracts was at least 6. The mean length of the included liver biopsies was 22.9 ± 9.5 mm (median 22 mm, range 10–60 mm).

2.2. Liver histology 2.3. Acoustic radiation force impulse (ARFI)-imaging Liver biopsy specimens were fixed in 4%-buffered formalin and embedded in paraffin. Two-micrometer-thick sections were stained with haematoxylin-eosin, Perls iron stain, dPAS (periodic acid Schiff after digestion with diastase) and Masson Trichrome. All biopsy specimens were analysed by an experienced pathologist who was blinded to the clinical results of the patients. Histological scoring was performed according to Kleiner et al. [9]. Steatosis was assessed according the number of hepatocytes with fatty degeneration: S0 = <5%, S1 = 5–33%, S2 = >33–66%, S3 = >66% of hepatocytes. Liver fibrosis was staged on a F0–F4 scale according to Kleiner: F0, no fibrosis; F1, perisinusoidal or periportal fibrosis; F2, perisinusoidal and portal or periportal fibrosis; F3, bridging fibrosis and F4, cirrhosis. The NAFLD Activity Score (NAS) was calculated according to Kleiner from the unweighted sum of the scores of steatosis (0–3), lobular inflammation (0–3) and ballooning (0–2). Using the NAS, the diagnosis of steatohepatitis is present if NAS is greater than 4. Biopsies were judged to be adequate if the length of liver

All patients received ARFI-imaging and TE by physicians blinded to the results of liver biopsy. Before ARFI-measurement the distance between the skin and the liver capsule at the site of the planned TE measurement and ARFI-measurement of the right lobe was measured using a 4.0-MHz curved ultrasound transducer. ARFI imaging (Virtual TouchTM Tissue Quantification, Siemens ACUSON S2000) involves targeting of an anatomic region to be interrogated for elastic properties with a region-of-interest (ROI) cursor while performing real time B-mode imaging. Tissue at the ROI is mechanically excited using short-duration acoustic pulses with a fixed transmit frequency of 2.67 MHz to generate localized tissue displacements in tissue. The displacements result in shear-wave propagation away from the region of excitation and are tracked using ultrasonic, correlation-based method. The maximum displacement is estimated for many ultrasound tracking beams laterally adjacent to the single push-beam, which are transmitted

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

e327

Fig. 2. Receiver-operating characteristic (ROC) curves for TE with the M probe and the XL probe and of ARFI of the left and right liver lobe for diagnosis of severe fibrosis (F ≥ 3) in 36 patients with valid measurements for all methods.

Fig. 1. Bland–Altmann regression of TE with the M probe and the XL probe (A) and ARFI quantifications in the left and right lobe (B) in 36 patients with valid measurements for all methods.

at nominal center-frequency of 3.08 MHz and pulse-repetitionfrequency ranging from 4500 to 9000 Hz. By measurement of the time to peak displacement at each lateral location, the shear wave speed of the tissue can be reconstructed [10]. The shear velocity is estimated in the central window of 5 mm axial by 4 mm width within a region of interest (ROI) graphically displayed with a size of 1 cm axial by 6 mm width. The shear wave propagation velocity is proportional to the square root of tissue elasticity. Results are expressed in m/s (range: 0.5–4.4 m/s with ±20% accuracy over the range). ARFI-imaging is integrated in a conventional high-end ultrasound machine (Siemens S2000). The transmitted frequency of 2.67 MHz is integrated as fixed frequency in the ultrasound machine and cannot be changed by the operator. ARFI-imaging was performed with a curved array at 4 MHz for B-mode imaging. The examination was performed on the right and left lobe of the liver. An area was chosen where the liver tissue was at least 6 cm thick and free of large blood vessels. A measurement depth of 1–2 cm below the liver capsule was chosen to standardize the examination. Ten acquisitions at different locations within the right liver lobe and ten acquisitions within the left liver lobe were performed on each patient and documented separately. The median of 10 valid measurements for each liver lobe was calculated and included in the final analysis. ARFI-failure was defined as no successful ARFI-measurement after 10 attempts. 2.4. Transient elastography (TE) FibroScan® (Echosens, Paris, France) is a medical device based on transient elastography (TE). It is equipped with a probe including an ultrasonic transducer mounted on the axis of a vibrator. A vibration transmitted from the vibrator towards the tissue induces an elastic shear wave that propagates through the tissue. These

propagations are followed by pulse-echo ultrasound acquisitions and their velocity is measured which is directly related to tissue stiffness. Results are expressed in kilopascal. Details have been described in previous studies [11]. The XL-probe used in this study was a prototype version of the probe now commercialised and was loaned by the device’s manufacturer. The standard probe (M-probe) and the XL-prototype probe differ in the tip diameter (9 mm vs. 13 mm), the ultrasound central frequency (3.5 MHz vs. 1.75 MHz), the vibration amplitude (2 mm vs. 3 mm) and the measurement depth (2.5–6.5 cm for the M-probe vs. 3.5–7.5 cm for the XL-probe). The examination was performed on the right lobe of the liver through the intercostal space in all patients. After the area of measurement was located, the examiner pressed the button of the probe to start the acquisition. Ten successful acquisitions were performed on each patient with both probes. The success rate was automatically calculated by the machine as the ratio of the number of successful acquisitions over the total number of acquisitions. Only TE results obtained with 10 valid measurements, with a success rate of at least 60% and an interquartile range (IQR) ≤ 30% of the median were considered reliable and were therefore used for the final analysis only. TE failure is defined when less than ten valid measurements are obtained. Time duration of TE measurement with each probe was documented. 2.5. Controlled attenuation parameter (CAP) Ultrasonic signals of the radiofrequency backpropagated signals were acquired during the TE examination using the standard probe (M-probe, 3.5 MHz). The measured parameter is called Controlled Attenuation Parameter (CAP). It is a gain controlled ultrasonic attenuation coefficient estimating ultrasound attenuation at 3.5 MHz using a novel proprietary algorithm and is expressed in dB/m. Details have been published recently [12]. During TE examination, the raw ultrasonic radiofrequency signals were stored in the TE examination file. CAP was then computed off-line retrospectively. Ultrasound attenuation is only calculated when the TE measurement is valid ensuring to get an attenuation value of the liver. The median value of 10 measurements was calculated and used for the

e328

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

further analysis. CAP values were extracted for each patient after the end of the study by Echosens (Paris, France) staff. The company was blinded to patient data and liver histology results. 2.6. The fatty liver index The fatty liver index was calculated using the published algorithm [13]: FLI =

M-probe, 4 patients examined with the XL-probe, and 5 patients evaluated with the TE-combination-score. The IQR/median quotient was greater 30% in additional 9 patients examined with the M-probe, 6 patients examined with the XL-probe, and no additional patient evaluated with the TE-combination-score. Therefore, reliable and valid TE-measurement was only available for 37 patients measured with the M-probe, 43 patients measured with the XLprobe, and 44 patients evaluated with the TE-combination-score.

e0.953∗log e(triglycerides)+0.139∗BMI+0.718∗log e(GGT)+0.053∗waist circumference−15.745 1 + e0.953∗log e(triglycerides)+0.139∗BMI+0.718∗log e(GGT)+0.053∗waist circumference−15.745

2.7. Statistical analysis Statistical analysis was performed using SigmaPlot and SigmaStat for Windows (version 11.0, Systat Software, Inc., Germany) and BiAS for Windows (version 9.08, epsilon 2010, Frankfurt, Germany). Correlations were assessed by Spearman’s correlation coefficient. Clinical and laboratory characteristics of patients were expressed as mean ± SD, median and range. Characteristics of TE-measurements with both probes, as well as of ARFI-measurement in the right and left liver lobe were compared using the Wilcoxon–Mann–Whitney paired t-test and assessed with Bland–Altmann regression. A p value less than 0.05 was judged to be statistically significant. McNemar’s test was used to compare the number of patients with 0, ≥5 and ≥10 valid measurements for both probes, as well as for ARFI in both liver lobes. The diagnostic performance of TE, ARFI and CAP was assessed by receiver-operating-characteristic (ROC) curves. The ROC curve represents sensitivity versus 1-specificity for all possible cut-off values for the prediction of the different fibrosis stages, respectively. The areas under the ROC curves (AUROC) as well as 95% CI of AUROC were calculated including all patients. AUROC values for different diagnostic criteria for the same data set were compared with the non-parametric DeLong test. Note that AUROC values for the different methods are correlated and that this test accounts for such correlations. Therefore, it may find significant differences in diagnostic accuracy even when confidence intervals of the single AUROC values, which ignore these correlations, are overlapping. In addition, the diagnostic accuracy was measured for a combination of TE-measurements with the M-probe and XL-probe (TE-combination-score). For this purpose, the median TE value of the M-probe was used for patients with a skin-to-liver capsule distance of less than 2.5 cm, and the median TE value of the XL-probe was used for patients with a skin-to-liver capsule distance of 2.5 cm and greater.

∗ 100

The Spearman correlation coefficient between TE with the M-probe and TE with the XL-probe on the one hand and the different histological fibrosis stages on the other hand was 0.36 (p = 0.049) and 0.53 (p = 0.00089), respectively. The correlation coefficient of the median measurements with both probes was 0.64 (p < 0.000048). Nevertheless, Bland–Altmann regression indicates that median measurements with the XL probe are significantly smaller than those with the M probe (p = 0.0020, Fig. 1). The Spearman correlation coefficient between TE with the Mprobe and TE with the XL-probe on the one hand and the NAFLD activity score (NAS) on the other hand was 0.37 (p = 0.023) and 0.25 (p = 0.10), respectively. The correlation coefficient between the TEcombination-score score with histological fibrosis stage and NAS was 0.29 (p = 0.049) and 0.35 (p = 0.020), respectively. No significant correlation was found for histological steatosis with TE measurement with both probes (p = 0.62 and p = 0.84). The diagnostic accuracy of the M-probe, XL-probe and TEcombination-score are shown in Table 2. No significant difference of AUROC was found between the M-probe and XL-probe for the diagnosis of significant fibrosis (p = 0.90), for the diagnosis of severe fibrosis (p = 0.44), for the diagnosis of cirrhosis (p = 0.62), and for the diagnosis of steatohepatitis (NAS > 4, p = 0.61). When using the M-probe for patients with a skin-to-liver distance of less than 2.5 cm and the XL-probe for patients with skin-to-liver distance of 2.5 cm and greater only three patients did not have 10 valid measurements. When using the quality criteria (10 valid measurements, success rate > 60%, IQR < 0.3) 44/57 (77%) patients could be analysed when using the TE-combinationscore, as compared to 43/57 (75%) when using the XL-probe and 37/57 (65%) when using the M-probe. The difference between the XL-probe and TE-combination-score was not significant (p = 0.32), while significantly more patients could be evaluated with the XL-probe and TE-combination-score as compared to the M-probe (p = 0.031 and p = 0.016). Details are shown in Table 3. 3.2. Fibrosis assessment: ARFI of the left and right liver lobe

3. Results Sixty-one patients with NALFD/NASH were included in the study. Four patients were excluded due to poor quality of liver biopsy (<10 mm length or <6 portal tracts), therefore, 57 patients were included in the final analysis. Patients’ characteristics at the time of study inclusion are shown in Table 1. 3.1. Fibrosis assessment: TE with the M and XL-probe and TE-combination-score TE failure (<10 valid measurements) was observed in 8 patients measured with the M-probe, 3 patients measured with the XLprobe, and 3 patients, when using the combination of M-probe and XL-probe (TE-combination-score) as described in the statistics section. In one patient the XL-probe had a technical defect; therefore, an examination was only possible with the M-probe. The success rate was below 60% in additional 3 patients examined with the

All patients could be successfully examined with ARFI in both lobes of the liver. The Spearman correlation coefficient between ARFI of the right liver lobe and ARFI of the left liver lobe on the one hand and the different histological fibrosis stages on the other hand was 0.20 (p = 0.10) and 0.22 (p = 0.10), respectively. The correlation coefficient of the median measurements in both lobes was 0.34 (p = 0.0095). The Spearman correlation coefficient between ARFI of the right liver lobe and ARFI of the left liver lobe on the one hand and the NAFLD activity score (NAS) on the other hand was 0.21 (p = 0.11) and 0.05 (p = 0.69), respectively. No significant correlation was found for histological steatosis with ARFI measurement in both liver lobes (p = 0.99 and p = 0.51). The diagnostic accuracy of ARFI of the left and right liver lobe is shown in Table 2. No significant difference in AUROC was found between both liver lobes for the diagnosis of significant fibrosis (p = 0.76), for

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

e329

Table 2 Area under the ROC curve (95% confidence interval) for transient elastography with the M-probe and the XL-probe and for ARFI of the right and left lobe of the liver according to Kleiner’s fibrosis stage and NAS. Method

F≥2

F≥3

F=4

NAS > 4

TE with M-probe (n = 37) p-Value TE with XL-probe (n = 43) p-Value TE combination score (n = 44) p-Value ARFI right lobe (n = 57) p-Value ARFI left lobe (n = 57) p-Value

0.80 (0.63–0.97) 0.0067 0.82 (0.68–0.95) 0.0014 0.77 (0.61–0.93) 0.011 0.66 (0.50–0.81) 0.073 0.62 (0.44–0.80) 0.15

0.73 (0.52–0.94) 0.065 0.84 (0.69–0.99) 0.0018 0.76 (0.57–0.95) 0.022 0.71 (0.55–0.87) 0.034 0.60 (0.34–0.85) 0.34

0.93 (0.78–1.00) 0.044 0.93 (0.77–1.00) 0.043 0.95 (0.85–1.00) 0.032 0.74 (0.28–1.00) 0.26 0.90 (0.70–1.00) 0.056

0.62 (0.43–0.81) 0.21 0.65 (0.49–0.82) 0.086 0.71 (0.55–0.86) 0.019 0.58 (0.43–0.73) 0.31 0.56 (0.41–0.72) 0.45

Table 3 Comparison of the performance of the M-probe with the XL-probe. Characteristics

M-probe

XL-probe

p Value

Mean ± SD of median liver stiffness Mean ± SD of number of valid measurements Mean ± SD of number of invalid measurements Mean ± SD of success rate (%) Mean ± SD of IQR/median liver stiffness Mean ± SD of measurement time (min) No valid measurements (Nr. of patients) Less than 5 valid measurements (Nr. of patients) Less than 10 valid measurements (Nr. of patients)

7.7 ± 4.7 9.2 ± 2.6 3.0 ± 4.3 80 ± 27 23 ± 15 3.5 ± 1.4 5 6 8

7.0 ± 4.3 9.6 ± 1.9 2.3 ± 4.0 86 ± 22 20 ± 12 3.1 ± 2.1 3 3 3

0 < 0.001 0.31 0.14 0.09 0.43 <0.005 0.50 0.25 0.06

SD, standard deviation; Nr., number.

the diagnosis of severe fibrosis (p = 0.26), and for the diagnosis of cirrhosis (p = 0.22). While 10 valid measurements were achieved in both liver lobes in all patients, significant more invalid measurements occurred when examining the left liver lobe (Table 4). 3.3. Comparison of ARFI and TE 10 valid measurements could be obtained in all patients measured with ARFI of both lobes. This was non-significantly more than with the TE XL-probe (57 vs. 53 patients, p = 0.13), but significantly more than with the TE M-probe (57 vs. 49 patients, p = 0.0078). The comparisons of AUROC were performed only in the group of patients with valid and reliable TE values with both probes (n = 36). The comparison of ARFI and TE was performed for the different fibrosis stages while using the best method of TE and ARFI. No

Table 4 Comparison of the performance of ARFI of the left and right lobe of the liver. Characteristics

ARFI right liver lobe

ARFI left liver lobe

p Value

Mean ± SD of median ARFI Mean ± SD of number of valid measurements Mean ± SD of number of invalid measurements Mean ± SD of success rate (%) No valid measurements (Nr. of patients) Less than 5 valid measurements (Nr. of patients) Less than 10 valid measurements (Nr. of patients)

1.4 ± 0.7

1.5 ± 0.5

0.05

10 ± 0

10 ± 0

na

0.7 ± 1.4

1.6 ± 2.5

<0.01

94 ± 10

89 ± 14

<0.01

0

0

na

0

0

na

0

0

na

SD, standard deviation; Nr., number.

significant difference of AUROC was found between ARFI and TE for the diagnosis of significant fibrosis (0.84 for TE with the XLprobe vs. 0.71 for ARFI of the right lobe of the liver, p = 0.11), for the diagnosis of severe fibrosis (0.83 for TE with the XL-probe vs. 0.75 for ARFI of the right lobe, p = 0.21, Fig. 2), for the diagnosis of cirrhosis (0.96 for TE combination score vs. 0.94 for ARFI of the left lobe of the liver, p = 0.67), and for the diagnosis of steatohepatitis (0.71 for TE combination score vs. 0.59 for ARFI of the left liver lobe, p = 0.31). 3.4. Steatosis assessment using the fatty liver index and the controlled attenuation parameter (CAP) Valid CAP measurements were available in 46 patients measured with the M-probe. These were 1 patient with steatosis grade S0, 11 patients with S1, 13 patients with S2 and 21 patients with S3. The median CAP value was 220 for steatosis grade S0/S1 (mean ± SD: 241 ± 71; range: 130–374), 299 for steatosis grade S2 (mean ± SD: 298 ± 30; range: 251–356), and 319 for steatosis grade S3 (mean ± SD: 314 ± 39; range: 230–372). While a significant correlation of histological steatosis was found with CAP (0.49, p = 0.00069), NAS (0.74, p = 0.00000020), BMI (0.30, p = 0.043), and waist circumference (0.34, p = 0.021), no significant correlation was found with the fatty liver index (0.25, p = 0.10), triglycerides (0.02, p = 0.90), and GGT (0.07, p = 0.66). In addition a significant correlation was found for CAP with the fatty liver index (0.49, p = 0.00060), NAS (0.42, p = 0.0043) BMI (0.51, p = 0.00029), and waist circumference (0.38, p = 0.0087). The diagnostic accuracy of CAP was 0.78 (95%-CI: 0.58–0.99) for the diagnosis of steatosis grade ≥S2 vs. S0/1, 0.72 (95%-CI: 0.57–0.86) for the diagnosis of steatosis grade S3 vs. S0/1/2, and 0.65 (95%-CI: 0.49–0.82) for the diagnosis of NAS > 4, respectively. The optimal cut-off was 245 dB/m for the diagnosis of steatosis grade ≥S2 with a sensitivity of 97% and a specificity of 67%, 301 dB/m for the diagnosis of steatosis grade S3 with a sensitivity of 76% and a specificity of 68%, and 303 dB/m for the diagnosis of NAS > 4 with a sensitivity of 74% and a specificity of 67%, respectively.

e330

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

4. Discussion Non-invasive tests are safe, easy-to-perform and are thus potentially useful for large-scale screening of liver fibrosis in patients with NAFLD/NASH [6,14,15]. To our knowledge, this is the first study evaluating the diagnostic accuracy of TE performed with two different probes (standard and obese probe) in comparison to ARFI of the right and left lobe of the liver in patients with NAFLD. Our study showed comparable results for both methods for the diagnosis of severe liver fibrosis (F ≥ 3). In addition, CAP is a novel quantitative method to measure steatosis non-invasively and this is to our knowledge the first study evaluating CAP in a population of only NAFLD/NASH with promising results. In the present study, TE performed with the new XL-probe for obese patients had comparable diagnostic accuracy to that of the standard M-probe for the detection of significant fibrosis (F ≥ 2) and cirrhosis (F = 4) in patients with NASH. Significantly more patients could be reliably examined with the XL-probe (75%) or the TE-combination-score (77%) as compared to the M-probe (65%). These findings are consistent with those observed in a previous study in a comparable patient cohort [7]. The TE-combinationscore was based on M-probe measurements in patients with a skin-to-liver capsule distance of less than 2.5 cm, and XL-probe measurements in patients with a skin-to-liver capsule distance of at least 2.5 cm. However, no significant difference was observed between the XL-probe and TE-combination-score, indicating that switching between different probes may not improve diagnostic quality in this setting and this should be clarified in larger patient cohorts. In line with a previous study, median liver stiffness measured with the XL-probe was significantly lower than that measured with the M-probe (7.0 vs. 7.7 kPa) [7]. A possible explanation may be increased liver stiffness with the M-probe due to measurement of subcutaneous fatty tissue in patients with a skin-to-liver capsule distance greater than 2.5 cm. However, in a recent study, high BMI had no effect on liver stiffness measurement [6]. Therefore, until new cut-off values for the XL-probe are defined, a mean difference of 1–2 kPa between the two probes must be taken into account. Unlike in TE, there is currently no standardized procedure available for ARFI imaging. ARFI imaging offers the advantage of shear wave velocity measurements within a ROI that can be freely placed at different depths in different parts of the liver. Thus, large, representative parts of the liver could be evaluated within one examination. This may be of particular interest, as histological studies have shown that lesions of NASH including perisinusoidal fibrosis are unevenly distributed throughout the liver parenchyma [16]. Additional advantages of ARFI imaging include its superior feasibility in obese patients and patients with ascites [17]. However, standardization of measurement site and depth, required numbers of measurements as well as precise definitions of ARFI failure has yet to be determined. In a recent study, ARFI was comparable to TE for the diagnosis of liver fibrosis in NAFLD. However, only ARFI of the right lobe and only TE with the M-probe were available in this study [18]. In our study, ARFI imaging could be successfully performed in both lobes of the liver, even in the most obese patients (with BMI up to 43 kg/m2 ). ARFI imaging of the right and left lobe of the liver was significantly correlated (p = 0.0095). However, shear wave velocity largely varied between the two lobes with higher AUROC values in the right lobe for F ≥ 2 and F ≥ 3 whereas the AUROC for F = 4 was higher in the left lobe. Furthermore, the number of invalid measurements was significantly higher in the left lobe (p < 0.01). These findings may well represent the heterogeneity of fibrosis progression in different parts of the liver. Therefore, evaluation of an ARFI combination score that encompasses both right and left

lobe measurements is deemed desirable and should be evaluated in larger patient cohorts. While a significant correlation was found for TE with histological liver fibrosis, no significant correlation could be found for ARFI with liver fibrosis. Also the diagnostic accuracies of TE showed a trend to higher values than the diagnostic accuracies of ARFI. Nevertheless, the paired comparison of diagnostic accuracies between TE and ARFI for the diagnosis of significant fibrosis, severe fibrosis and liver cirrhosis did not show statistical significance. For the staging of patients with NALFD the diagnosis or exclusion of severe fibrosis F ≥ 3 is of great importance, since severe fibrosis is a risk factor for liver-related mortality [19]. No significant difference was found between the diagnostic accuracy of ARFI performed in the right liver lobe and TE performed with the XL-probe (75% vs. 83%, p = 0.21). It must be noted however, that comparisons of AUROC were only performed in patients in whom valid TE measurements with both probes were available, whereas ARFI AUROC values tended to be lower when all 57 patients were taken into account as shown in Table 2. Therefore, quality criteria as defined for TE might also apply for ARFI and need evaluation in future studies. The AUROC values in the present study are lower than in a previously published study reporting an AUROC of 97% for ARFI performed in the right liver lobe for the diagnosis of severe fibrosis [18]. TE and ARFI were comparable in that study [18]. Larger studies are awaited. Estimation of liver stiffness using the Fibroscan is derived from the speed of the shear wave by calculating the Young’s modulus expressed as: E = 3PVs 2 where P is the mass density and Vs is the shear wave velocity. This calculation is automatically performed in the Fibroscan machine and the results are expressed in kPa [11]. However, this calculation can induce an error since the mass density P of the liver is considered to be a constant while in reality it differs between normal, congestive, fatty, and fibrous liver tissue. Therefore, ARFI results expressed in m/s seem to be more relevant, since it corresponds to a direct measure of a physical phenomenon whereas Fibroscan results correspond to a calculated and thus indirect measure [20]. The values of measurement of both methods (FibroScan and ARFI) cannot be directly compared and the present study therefore compares to different elastography methods with different normal and pathological values, the cut-offs developed for FibroScan cannot be used for ARFI. Finally, the two methods were not significantly different for the diagnosis of steatohepatitis according to NAS. The histological NAFLD Activity Score (NAS) was proposed by Kleiner and co-workers to specifically include only features of active injury that are potentially reversible (e.g. after therapeutic interventions) [9] and non-invasive methods may be useful to monitor such histological changes. Therapeutic interventions for NAFLD such as lifestyle modifications are most likely to be successful at an early stage when hepatic steatosis is barely present. Several non-invasive techniques have been evaluated for the diagnosis of steatosis. B-mode ultrasound imaging is the most commonly used and easily accessible technique to detect the presence of steatosis. However, ultrasound is highly operator- and machine-dependent and can only detect steatosis beyond 30% fatty infiltration [21]. Computed tomography (CT) and magnetic resonance imaging (MRI) techniques have shown promising results in a number of studies [21,22]. However, lack of standardization, availability, and high costs have given rise to the need for alternative non-invasive methods to detect steatosis. In this study, we evaluated a new ultrasonic controlled attenuation parameter (CAP) that uses a proprietary algorithm based on TE to detect and quantify hepatic steatosis [12]. CAP can be measured along with TE examinations and is operator- and

M. Friedrich-Rust et al. / European Journal of Radiology 81 (2012) e325–e331

machine-independent. CAP is currently only available with the M-probe. In a previous study of patients 115 with chronic liver disease (only 15 patients with NAFLD), CAP was significantly correlated to steatosis and showed very good to excellent diagnostic accuracy (AUROC) for the detection of steatosis grade 1 (here defined as >10% fatty infiltration; 0.91), grade 2 (>33%; 0.95) and grade 3 (>66%; 0.89) [12]. In this study, we confirmed a significant correlation with histological steatosis in a study population of NAFLD/NASH. The AUROC for steatosis grade ≥2 vs. <2 and ≥3 vs. <3 were lower compared to the study by Sasso [12] with mixed liver diseases (0.78 and 0.72). An explanation might be the higher BMI of our study population (28 vs. 25 kg/m2 ) and the present availability of CAP measurement with the M-probe only. In addition, our study was limited by the small sample size and, possibly, long time interval of up to 17 months between liver biopsy and CAP measurement and further studies are certainly necessary including the measurement of CAP with the XL-probe in obese patients. The long time interval between liver biopsy and non-invasive evaluation may have also affected fibrosis assessment in this study. However, as the mean progression rate of liver fibrosis in untreated patients was estimated to be 0.085–0.120 fibrosis stages per year [23], changes over 17 months were expected to be minimal only. In addition, biopsies that are shorter than the usual standard of 15 mm were included in the present study, if at least 6 portal tracts were present. Nevertheless, this was a comparative study between ARFI imaging and different TE probes, where the limitations of liver biopsy affected all methods equally. 5. Conclusions Taken together, ARFI imaging can be performed with comparable diagnostic accuracy to that of transient elastography for the diagnosis of severe fibrosis. However a trend toward higher AUROCs was observed for TE. Quality criteria for ARFI imaging have yet to be established and incorporation of left lobe measurements should be considered. ARFI imaging and the TE XL-probe showed superior feasibility in obese patients and should be used as the preferred methods in patients with NASH and a high BMI. However, larger, prospective trials are necessary to develop cut-off values for the staging of liver fibrosis using these new methods. CAP represents a promising, easy-to-perform method for steatosis detection and quantification, though larger clinical trials have to be awaited. Funding None. Conflict of interest No conflict of interest exists for any author. Contributions All authors have participated in the study concept and design; acquisition of data; analysis and interpretation of data; critical revision of the manuscript for important intellectual content; in addition the authors Friedrich-Rust and Bojunga have performed the drafting of the manuscript; and Friedrich-Rust, Bojunga and Herrmann performed the statistical analysis.

e331

Acknowledgements The authors thank Céline Fournier and other staff members of Echosens International (Paris, France) for the loan of the XLprototype probe for the present study and for the calculation of CAP. No financial support was obtained for the present study. References [1] Starley BQ, Calcagno CJ, Harrison SA. Nonalcoholic fatty liver disease and hepatocellular carcinoma: a weighty connection. Hepatology 2010;51:1820–32. [2] Williams CD, Stengel J, Asike MI, et al. Prevalence of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis among a largely middle-aged population utilizing ultrasound and liver biopsy: a prospective study. Gastroenterology 2011;140:124–31. [3] Ekstedt M, Franzen LE, Mathiesen UL, et al. Long-term follow-up of patients with NAFLD and elevated liver enzymes. Hepatology 2006;44:865–73. [4] Cadranel JF, Rufat P, Degos F. Practices of liver biopsy in France: results of a prospective nationwide survey. For the Group of Epidemiology of the French Association for the Study of the Liver (AFEF). Hepatology 2000;32:477–81. [5] Bedossa P, Dargere D, Paradise V. Sampling variability of liver fibrosis in chronic hepatitis C. Hepatology 2003;38:1449–57. [6] Wong VW, Vergniol J, Wong GL, et al. Diagnosis of fibrosis and cirrhosis using liver stiffness measurement in nonalcoholic fatty liver disease. Hepatology 2010;51:454–62. [7] Friedrich-Rust M, Hadji-Hosseini H, Kriener S, et al. Transient elastography with a new probe for obese patients for non-invasive staging of non-alcoholic steatohepatitis. Eur Radiol 2010;20:2390–6. [8] Palmeri ML, Wang MH, Rouze NC, et al. Noninvasive evaluation of hepatic fibrosis using acoustic radiation force-based shear stiffness in patients with nonalcoholic fatty liver disease. J Hepatol 2011. [9] Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005;41:1313–21. [10] Nightingale KR, Zhai L, Dahl JJ, Frinkley KD, Palmeri ML. Shear wave velocity estimation using acoustic radiation force impulsive excitation in liver in vivo. In: Proceedings of the 2006 IEEE Ultrasonics Symposium (IEEE). 2006. p. 1156–60. [11] Sandrin L, Fourquet B, Hasquenoph JM, et al. Transient elastography: a new non-invasive method for assessment of hepatic fibrosis. Ultrasound Med Biol 2003;29:1705–13. [12] Sasso M, Beaugrand M, de Ledinghen V, et al. Controlled attenuation parameter (CAP): a novel VCTE guided ultrasonic attenuation measurement for the evaluation of hepatic steatosis: preliminary study and validation in a cohort of patients with chronic liver disease from various causes. Ultrasound Med Biol 2010;36:1825–35. [13] Bedogni G, Bellentani S, Miglioli L, et al. The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol 2006;6:33. [14] Yoneda M, Yoneda M, Mawatari H, et al. Noninvasive assessment of liver fibrosis by measurement of stiffness in patients with nonalcoholic fatty liver disease (NAFLD). Dig Liver Dis 2008;40:371–8. [15] Poynard T, Ratziu V, Charlotte F, et al. Diagnostic value of biochemical markers (NashTest) for the prediction of non alcoholo steato hepatitis in patients with non-alcoholic fatty liver disease. BMC Gastroenterol 2006;6:34. [16] Ratziu V, Charlotte F, Heurtier A, et al. Sampling variability of liver biopsy in nonalcoholic fatty liver disease. Gastroenterology 2005;128:1898–906. [17] Friedrich-Rust M, Wunder K, Sotoudeh F, et al. Acoustic radiation force Elastografie versus FibroScan zur nicht-invasiven Beurteilung des Leberfibrosestadiums bei viraler hepatitis. Endoskopie heute 2009;1:62A. [18] Yoneda M, Suzuki K, Kato S, et al. Nonalcoholic fatty liver disease: US-based acoustic radiation force impulse elastography. Radiology 2010;256:640–7. [19] Musso G, Gambino R, Cassader M, Pagano G. Meta-analysis: natural history of non-alcoholic fatty liver disease (NAFLD) and diagnostic accuracy of noninvasive tests for liver disease severity. Ann Med 2010. [20] Boursier J, Isselin G, Fouchard-Hubert I, et al. Acoustic radiation force impulse: a new ultrasonographic technology for the widespread noninvasive diagnosis of liver fibrosis. Eur J Gastroenterol Hepatol 2010;22:1074–84. [21] Schwenzer NF, Springer F, Schraml C, Stefan N, Machann J, Schick F. Noninvasive assessment and quantification of liver steatosis by ultrasound, computed tomography and magnetic resonance. J Hepatol 2009;51:433–45. [22] Bohte AE, van Werven JR, Bipat S, Stoker J. 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 2011;21:87–97. [23] Thein HH, Yi Q, Dore GJ, Krahn MD. Estimation of stage-specific fibrosis progression rates in chronic hepatitis C virus infection: a meta-analysis and meta-regression. Hepatology 2008;48:418–31.