331jum_online_Layout 1 12/19/13 1:55 PM Page 73
ORIGINAL RESEARCH
Indirect Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness Using the FibroScan System Xiangdong Hu, MD, Xiaoluan Xu, MD, Qing Zhang, MD, Hong Zhang, MD, Jing Liu, MD, Linxue Qian, MD Objectives—To evaluate quantitative measurement of spleen stiffness for indirect assessment of liver fibrosis in patients with chronic hepatitis B and to correlate spleen stiffness with liver stiffness using pathologic examination as a reference standard. Methods—Sixty patients with clinically confirmed chronic hepatitis B (n = 54) and liver cirrhosis (n = 6) were enrolled. Quantitative stiffness measurements (kilopascals) were obtained from spleen and liver parenchyma with the FibroScan system (Echosens, Paris, France). Correlation analyses were performed between spleen and liver stiffness and between spleen stiffness and liver fibrosis stages. The diagnostic performance of spleen stiffness for indirect prediction of liver fibrosis was estimated by receiver operating characteristic curves.
Received May 16, 2013, from the Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, Beijing, China. Revision requested May 29, 2013. Revised manuscript accepted for publication October 7, 2013. This study was funded by Beijing Training Support of Talents (grant 2011-D003034000006) and Capital Health Development Research Projects (grant 2011-1002-02). Address correspondence to Linxue Qian, MD, Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, 95 Yongan Rd, 100050 Beijing, China. E-mail: [email protected]
Abbreviations
ARFI, acoustic radiation force impulse imaging; AUROC, area under the receiver operating characteristic curve; CI, confidence interval; MR, magnetic resonance doi:10.7863/ultra.33.1.73
Results—Both spleen and liver stiffness increased as liver fibrosis progressed. Spleen stiffness values had a positive correlation with liver stiffness values in all patients (Pearson r = 0.810; P < .001). The correlation between spleen stiffness and fibrosis stages was statistically significant (Spearman r = 0.833; P < .001). The areas under the receiver operating characteristic curves for spleen stiffness were 0.902 (95% confidence interval [CI], 0.825–0.978) for stage S1 (fibrous portal expansion and limited perisinusoidal or lobular fibrosis) or higher, 0.927 (95% CI, 0.852–1.0) for S2 (periportal fibrosis and few fibrous septa but intact architecture) or higher, 0.962 (95% CI, 0.918–1.0) for S3 (numerous fibrous septa with architectural distortion but no obvious cirrhosis) or higher, and 0.983 (95% CI, 0.957–1.0) for S4 (cirrhosis) (all P < .001). The differences between the areas under the curves for spleen and liver stiffness in liver fibrosis staging were not statistically significant (P = .115–.756). Conclusions—Quantitative measurement of spleen stiffness is a feasible and promising technique for estimating liver fibrosis. The overall diagnostic performance of spleen stiffness for liver fibrosis staging is comparable with that of liver stiffness. Key Words—FibroScan; gastrointestinal ultrasound; liver fibrosis; liver stiffness; spleen stiffness
L
iver fibrosis is an important histopathologic feature of chronic liver disease, which is related to the development of liver cirrhosis, portal hypertension, hepatocellular carcinoma, and other conditions. Studies have shown that liver fibrosis can be healed or reversed if the patient is treated with appropriate therapy for different stages of fibrosis.1,2 Thus, it is imperative to assess the progression of liver fibrosis in deciding both treatment strategies and follow-up.
©2014 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2014; 33:73–81 | 0278-4297 | www.aium.org
331jum_online_Layout 1 12/19/13 1:55 PM Page 74
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
To date, liver biopsy has long been considered the reference standard for evaluating liver fibrosis. However, because it is an invasive procedure with a mortality rate of 0.1% to 0.01%, sampling errors due to sampling of only 1/50,000 of the liver parenchyma, and interobserver and intraobserver diagnostic variability, liver biopsy cannot be widely used in clinical practice.3,4 Thus, investigators have conducted a great deal of research on noninvasive diagnosis of liver fibrosis.5,6 Over the past decade, a method to quantitatively evaluate liver fibrosis with elasticity information from liver parenchyma has become an area of great interest.7–17 Yeh et al7 investigated the elastic modulus of liver parenchyma in 19 fresh human liver samples. The Pearson correlation coefficients between the pathologic grading of liver fibrosis and the Young modulus value under 5%,10%, and 15% preload strain conditions were 0.702, 0.669, and 0.626, respectively (P < .01). Commonly used imaging techniques include transient elastography, acoustic radiation force impulse imaging (ARFI), supersonic shear imaging, and magnetic resonance (MR) elastography. Acoustic radiation force impulse imaging uses high-intensity ultrasound beams to induce displacements inside the liver and provides tissue stiffness as the shear wave velocity. Toshima et al8 and Lupsor et al9 reported that the area under the receiver operating characteristic curve (AUROC) for the diagnosis of fibrosis (grade ≥3) with ARFI in patients with chronic liver disease was superior to serum markers and comparable with transient elastography. Supersonic shear imaging relies on the use of radiation force and a high–frame rate ultrasound scanner. The AUROC of supersonic shear imaging for mild and intermediate hepatic fibrosis was confirmed by Bavu et al10 to be higher than that of transient elastography. Magnetic resonance elastography11–13 has demonstrated the ability to detect advanced hepatic fibrosis, including occult cirrhosis, based on measurements of liver stiffness. Ultrasound-based transient elastography has been studied in both research and clinical settings. Transient elastography uses an ultrasound probe mounted on the axis of a mechanical vibrator, which generates a 50-Hz shear wave with a small amplitude. The displacements induced by the shear wave in the medium are measured, and the shear wave velocities are calculated by pulse-echo ultrasound as a reflection of liver stiffness. A meta-analysis aimed at the diagnostic accuracy of transient elastography for liver fibrosis showed that the average AUROC was 0.84 (95% confidence interval [CI] 0.82–0.86) for substantial fibrosis, and the average AUROC was 0.94 (95% CI, 0.93– 0.95) for cirrhosis.14 However, despite high AUROC values, several groups also reported that liver stiffness significantly
74
increases during inflammatory episodes and mechanical cholestasis,15–17 which will weaken the value of liver stiffness measurement for liver fibrosis staging. Importantly, these pathologic processes have less influence on the spleen. Thus, measurement of spleen stiffness may provide additional parameters for indirectly reflecting the status of the liver stiffness. It is well known that the splenic vein is a tributary of the portal system. The hemodynamics and morphologic characteristics of the spleen may change during the development of chronic liver disease,18–20 which may provide helpful information for diagnosis of cirrhosis. However, the changing trend of splenic elasticity during the development of liver fibrosis and its diagnostic value for liver fibrosis staging are currently unknown. Recently, some studies have used MR elastography, ARFI, and transient elastography to investigate spleen stiffness in patients with chronic liver diseases, especially for assessment of portal hypertension and esophageal varices.21–24 The aim of this study was to evaluate quantitative measurement of spleen stiffness with the FibroScan system (Echosens, Paris, France) for indirect assessment of liver fibrosis in patients with chronic liver disease and to correlate spleen stiffness with liver stiffness using pathologic examination as a reference standard.
Materials and Methods Patients From May 2011 to December 2011, 60 patients including 54 with clinically confirmed chronic hepatitis B and 6 with a clinical diagnosis of Child-Pugh class A hepatitis B virus– related cirrhosis were enrolled in this prospective study. All patients had a history of hepatitis B virus infection for at least 6 months before enrollment. The patients with liver cirrhosis were diagnosed by clinical history with serologic examination and computed tomographic (CT) or MR imaging confirmation. Patients were excluded if they were infected with other hepatitis viruses or the human immunodeficiency virus or if they were pregnant. Also, patients with fatty liver, drug-induced liver injury, alcoholic liver disease, autoimmune disease, biliary tract disease, ascites, and liver tumors were excluded. This study was approved by the University Ethics Committee, and written informed consent was obtained from all patients before the study. Measurement of Stiffness Quantitative stiffness values were obtained by the FibroScan system from parenchyma of the spleen and liver in all 60 patients, who fasted overnight. The scanning operator was a well-trained radiologist with at least 5 years of
J Ultrasound Med 2014; 33:73–81
331jum_online_Layout 1 12/19/13 1:55 PM Page 75
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
experience in sonography and special training in FibroScan examinations. Patients’ clinical and laboratory information were blinded to the operator. The scanning technique and examination procedure for measurement of liver stiffness were adapted from a previously published technique.25 All patients were examined with a medium-sized probe with a frequency of 3.5 MHz. First, conventional a ultrasound system equipped with a broadband 2.5–3.5-MHz convex probe (MyLab 70Xvision; Esaote Biomedica, Genoa, Italy) was used to image and localize the spleen and liver via an intercostal space. Once the appropriate areas of the spleen and liver were identified, the probe was manually held in a perpendicular position to the skin surface for measurements of spleen and liver stiffness at a depth range from 25 to 65 mm below the skin surface. The sample volume of the measurement on the FibroScan system is approximately 1 cm wide
and 4 cm long in a cylindrical shape. For spleen stiffness measurements, the sample volume was targeted at the region where the splenic parenchymal thickness was more than 4 cm and free of large vessels. These conditions were met in all 60 patients. The quality of data acquisition was further verified by TM-mode and A-mode imaging on the FibroScan system (Figure 1). The TM-mode represents the ultrasonic amplitude in a log scale as a function of depth and time, whereas the A-mode represents a real-time ultrasonic amplitude in a log scale as a function of depth. If the sample volume includes large vessels or dilated bile ducts, the A-mode line will display nonlinearity and fluctuation. When the TM-mode image was homogeneous and the A-mode line appeared linear with less fluctuation, it implied that the measured spleen/liver tissue was free of large vessels, which was appropriate for the examination.
Figure 1. Monitor image from FibroScan system: A, TM-mode; B, A-mode; C, elastographic image; D, tissue stiffness; E, success rate. TM-mode represents the ultrasonic amplitude in a log scale as a function of depth and time. A-mode represents the real-time ultrasonic line amplitude in a log scale as a function of depth. The operator uses TM-mode and A-mode images to select and localize the sample area. CS indicates current stiffness; and IQR, interquartile range.
J Ultrasound Med 2014; 33:73–81
75
331jum_online_Layout 1 12/19/13 1:55 PM Page 76
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
The measurement values were expressed in kilopascals. Ten valid measurements were performed on each patient. The success rate was calculated as the number of validated measurements divided by the number of total measurements. Measurements with an interquartile range of less than 30% of the median value and a success rate of greater than 60% were considered reliable data sets. According to the manufacturer’s recommendations, the median value of 10 measurements was considered representative of tissue stiffness (liver or spleen). Sonographically Guided Liver Biopsy Percutaneous sonographically guided liver biopsy was performed by 2 experienced radiologists using a Magnum automatic biopsy gun (C. R. Bard, Inc, Tempe, AZ) and a 16-gauge core biopsy needle after the stiffness measurements were obtained from the 54 patients with chronic hepatitis B within the same day of the study and the other 6 patients with cirrhosis 3 to 6 months before the study. The probes for both FibroScan scanning and biopsy guidance were manually held in a position perpendicular to the skin surface. The liver biopsy attempted to target the same area as the region of stiffness measurements by marking the location of the probe on the skin and sampling the tissue at the same depth as the transient elastographic examination. All specimens were at least 1.5 cm in length, fixed in 4% formalin, and embedded in paraffin. The specimen slices were stained with hematoxylin-eosin and Masson trichrome for light microscopic analyses. Each specimen was examined under microscopy by an experienced pathologist, who was blinded to both the clinical and imaging data of the patients. Liver fibrosis was scored in 5 grades based on a published standard26 as follows: stage S0, no fibrosis; S1, fibrous portal expansion and limited perisinusoidal or lobular fibrosis; S2, periportal fibrosis and few fibrous septa but intact architecture; S3, numerous fibrous septa with architectural distortion but no obvious cirrhosis; and S4, cirrhosis.
fibrosis stage) and the Pearson correlation test (for spleen stiffness and liver stiffness). The F test followed by the Games-Howell test was used to compare the liver and spleen stiffness values between different pathologic fibrosis stages. The diagnostic performance was evaluated by the AUROC. Optimal cutoff values were chosen to maximize the sum of sensitivity and specificity.
Results The 60 patients enrolled in this study include 46 male and 14 female patients with a mean age of 37 ± 15 years. The clinical findings of the patients are summarized in Table 1. Of the 60 patients who underwent liver biopsy, 17 had S0 fibrosis, 18 had S1, 9 had S2, 8 had S3, and 8 had S4. Spleen and Liver Stiffness for Staging Liver Fibrosis Spleen and liver stiffness values were successfully measured in all patients. The elastic moduli of spleen parenchyma in patients with S0, S1, S2, S3, and S4 fibrosis were 13.83 ± 2.60, 17.96 ± 3.85, 22.93 ± 5.97, 33.64 ± 12.38, and 57.53 ± 16.40 kPa, respectively. The elastic moduli of liver parenchyma in patients with S0, S1, S2, S3, and S4 fibrosis were 4.64 ± 0.70, 6.06 ± 1.29, 6.34 ± 1.10, 12.89 ± 3.49, and 17.28 ± 4.27 kPa. The spleen and liver stiffness values as related to fibrosis stages are shown as box plots in Figure 2. On the whole, there was a dramatic increase in spleen and liver stiffness in parallel with progression of liver fibrosis. A significant correlation was found between spleen stiffness and fibrosis stages (Spearman r = 0.833; P < .001). At the same time, the spleen stiffness values had a positive correlation with the liver stiffness values in all patients (Pearson r = 0.810; P < .001; Figure 3). P values for multiple comparisons of spleen and liver stiffness between different liver fibrosis stages are shown in Table 2. Table 1. Clinical Characteristics Characteristic
Statistical Analyses Statistical analyses were performed with SPSS version 13.0 software (SPSS Inc, Chicago, IL) and MedCalc version 12.3 software (MedCalc Software, Mariakerke, Belgium). P < .05 was considered statistically significant. The liver and spleen stiffness values and variables were expressed as mean ± standard deviation. Box plots were used to illustrate the distributions of the different parameters. The Levene test was used to investigate the homogeneity of variance. Correlations between different parameters were analyzed by the Spearman correlation test (for spleen stiffness and
76
Male, n (%) Age, y BMI, kg/m2 AST, U/L ALT, U/L ALP, U/L GGT, U/L T-BIL, μM
Value 46 (76.7) 37 ± 15 (20–55) 19.36 ± 2.46 (16.65–25.61) 56.23 ± 22.27 (27–84) 48.33 ± 18.75 (23–76) 62.28 ± 20.15 (36–115) 28.16 ± 13.40 (11–45) 10.25 ± 3.58 (4.26–16.49)
Data are presented as mean ± SD (range) where applicable. ALP indicates alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; GGT, γ-glutamyltranspeptidase; and T-BIL, total bilirubin.
J Ultrasound Med 2014; 33:73–81
331jum_online_Layout 1 12/19/13 1:55 PM Page 77
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
Diagnostic Value of Liver and Spleen Stiffness Evaluated by the AUROC The receiver operating characteristic curve analyses showed that spleen stiffness and liver stiffness had high predictive values for S1 or higher to S4 (AUROC = 0.899–0.999; P < .001, Table 3). On the other hand, the sensitivities decreased when using spleen or liver stiffness to estimate mild hepatic fibrosis (S1). When compared with the diagnostic value of liver and spleen stiffness, no statistical significance was found between the AUROC for estimating liver fibrosis stages (Z = 0.310–1.577; P = .115–0.756; Figure 4).
Discussion In the literature, several methods, such as MR elastography, transient elastography, and ARFI, have been used to measure spleen stiffness.21–24,27 The FibroScan system used in this study is an ultrasound-based modality for quantitative assessment of tissue elasticity. The probe generates elastic shear waves that propagate through tissues. The propagation velocity, which is correlated with tissue stiffness, is measured by pulse-echo ultrasound. In this study, the mean spleen stiffness value in patients with S0 liver fibrosis was 13.83 kPa, which was similar to that (17.8 kPa) in 17 control patients in a previous study that used the FibroScan system23 but greater than that (2.35– 5.58 kPa) measured in 16 healthy participants reported by
Nedredal et al21 using MR elastography. Because of the small sample sizes of these studies, further investigation is needed to establish the normal range of spleen stiffness. It is well known that histologic modifications lead to changes in the mechanical behavior of organs. Mechanical and histologic analyses have been combined in a number of studies of the human liver.7,28,29 By means of aspiration experiments, Mazza et al28 reported that the stiffness values (0.3 and 0.44 bar/mm) of 2 fibrotic liver samples among a total of 17 samples were 3 times greater than those of normal tissue (0.07–0.18 bar/mm). A study reported by DeWall et al29 showed F4-graded fibrotic liver tissue (n = 2) was significantly stiffer than F0-graded tissue (n = 14), with a modulus contrast of 4:1. In our study, the mean liver stiffness was similar to the Young modulus of liver simples under a 15% preload strain reported by Yeh et al.7 The S4 to S0 liver stiffness ratio was also comparable with the ratios of elastic parameters in previous studies.28,29 Several clinical studies have shown that liver stiffness is correlated with the severity of liver fibrosis and that liver stiffness is a good indicator for estimating liver fibrosis stages.30,31 This study demonstrated that there was an excellent correlation (Pearson r = 0.810; P < .001) between spleen and liver stiffness in all patients from S0 to S4, which implied that measurement of spleen stiffness may be suitable for indirectly assessing liver fibrosis. Theoretically, the related correlation between spleen and liver stiffness is
Figure 2.Spleen stiffness measurement (SSM) and liver stiffness measurement (LSM) values for different stages of liver fibrosis. The circles display all observations. The ends of the boxes are the upper and lower quartiles, such that the boxes span the interquartile ranges. The medians are marked by horizontal lines inside the boxes. The whiskers extend to the highest and lowest observations. Spleen and liver stiffness increased as the liver fibrosis stage increased, especially in S3 and S4.
J Ultrasound Med 2014; 33:73–81
77
331jum_online_Layout 1 12/19/13 1:55 PM Page 78
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
based on the pathophysiologic changes of chronic liver disease, including hyperdynamic splenic blood flow, splenic parenchymal hyperplasia, and splenomegaly,18,19,32 especially in late stages of fibrosis. The results of this study showed that spleen stiffness values were greater than liver stiffness values in the same patient, which is consistent with other studies.21,23,33 Stefanescu et al23 reported that the mean spleen and liver stiff-
ness in values in patients with cirrhosis without esophageal varices were 46.05 (range, 23.7–75.0) and 26.3 (range, 9.9–40.3) kPa, respectively, and these values were similar to the results obtained in this study. The gaps in spleen stiffness measurements between adjacent stages were also greater than those of liver stiffness, in which later stages may be helpful identifying the adjacent liver fibrosis stage, and the standard deviations of spleen stiffness were also greater than those of liver stiffness; thus, the differences in spleen stiffness between different liver fibrosis stages were not always evident and distinguishable. Spleen stiffness measurements in this study had no significant differences for S1 versus S2 and S2 versus S3 but had statistically significant differences between the other groups. Liver stiffness measurements were not significantly different for S1 versus S2 and S3 versus S4 but were statistically significantly different between the rest of the groups. Therefore, spleen stiffness measurement was better than liver stiffness measurement for identification of S3 versus S4, whereas spleen stiffness measurement was inferior to liver stiffness measurement in terms of S2 versus S3 differentiation.
Figure 3. Spleen stiffness measurement (SSM) and liver stiffness measurement (LSM) values in different patients (Pearson r = 0.810; P < .001).
Table 2. P Values for Multiple Comparisons of Spleen and Liver Stiffness Measurements Between Different Liver Fibrosis Stages Fibrosis Stage S0 S1 S2 S3 S4
S0
S1
Spleen Stiffness S2 S3
.007 .007 .01 .016 .001
.221 .049 .001
.01 .221 .247 .003
.016 .049 .247
S4 .001 .001 .003 .039
.039
S0
S1 .003
.003 .009 .002
ORIGINAL RESEARCH
Indirect Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness Using the FibroScan System Xiangdong Hu, MD, Xiaoluan Xu, MD, Qing Zhang, MD, Hong Zhang, MD, Jing Liu, MD, Linxue Qian, MD Objectives—To evaluate quantitative measurement of spleen stiffness for indirect assessment of liver fibrosis in patients with chronic hepatitis B and to correlate spleen stiffness with liver stiffness using pathologic examination as a reference standard. Methods—Sixty patients with clinically confirmed chronic hepatitis B (n = 54) and liver cirrhosis (n = 6) were enrolled. Quantitative stiffness measurements (kilopascals) were obtained from spleen and liver parenchyma with the FibroScan system (Echosens, Paris, France). Correlation analyses were performed between spleen and liver stiffness and between spleen stiffness and liver fibrosis stages. The diagnostic performance of spleen stiffness for indirect prediction of liver fibrosis was estimated by receiver operating characteristic curves.
Received May 16, 2013, from the Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, Beijing, China. Revision requested May 29, 2013. Revised manuscript accepted for publication October 7, 2013. This study was funded by Beijing Training Support of Talents (grant 2011-D003034000006) and Capital Health Development Research Projects (grant 2011-1002-02). Address correspondence to Linxue Qian, MD, Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, 95 Yongan Rd, 100050 Beijing, China. E-mail: [email protected]
Abbreviations
ARFI, acoustic radiation force impulse imaging; AUROC, area under the receiver operating characteristic curve; CI, confidence interval; MR, magnetic resonance doi:10.7863/ultra.33.1.73
Results—Both spleen and liver stiffness increased as liver fibrosis progressed. Spleen stiffness values had a positive correlation with liver stiffness values in all patients (Pearson r = 0.810; P < .001). The correlation between spleen stiffness and fibrosis stages was statistically significant (Spearman r = 0.833; P < .001). The areas under the receiver operating characteristic curves for spleen stiffness were 0.902 (95% confidence interval [CI], 0.825–0.978) for stage S1 (fibrous portal expansion and limited perisinusoidal or lobular fibrosis) or higher, 0.927 (95% CI, 0.852–1.0) for S2 (periportal fibrosis and few fibrous septa but intact architecture) or higher, 0.962 (95% CI, 0.918–1.0) for S3 (numerous fibrous septa with architectural distortion but no obvious cirrhosis) or higher, and 0.983 (95% CI, 0.957–1.0) for S4 (cirrhosis) (all P < .001). The differences between the areas under the curves for spleen and liver stiffness in liver fibrosis staging were not statistically significant (P = .115–.756). Conclusions—Quantitative measurement of spleen stiffness is a feasible and promising technique for estimating liver fibrosis. The overall diagnostic performance of spleen stiffness for liver fibrosis staging is comparable with that of liver stiffness. Key Words—FibroScan; gastrointestinal ultrasound; liver fibrosis; liver stiffness; spleen stiffness
L
iver fibrosis is an important histopathologic feature of chronic liver disease, which is related to the development of liver cirrhosis, portal hypertension, hepatocellular carcinoma, and other conditions. Studies have shown that liver fibrosis can be healed or reversed if the patient is treated with appropriate therapy for different stages of fibrosis.1,2 Thus, it is imperative to assess the progression of liver fibrosis in deciding both treatment strategies and follow-up.
©2014 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2014; 33:73–81 | 0278-4297 | www.aium.org
331jum_online_Layout 1 12/19/13 1:55 PM Page 74
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
To date, liver biopsy has long been considered the reference standard for evaluating liver fibrosis. However, because it is an invasive procedure with a mortality rate of 0.1% to 0.01%, sampling errors due to sampling of only 1/50,000 of the liver parenchyma, and interobserver and intraobserver diagnostic variability, liver biopsy cannot be widely used in clinical practice.3,4 Thus, investigators have conducted a great deal of research on noninvasive diagnosis of liver fibrosis.5,6 Over the past decade, a method to quantitatively evaluate liver fibrosis with elasticity information from liver parenchyma has become an area of great interest.7–17 Yeh et al7 investigated the elastic modulus of liver parenchyma in 19 fresh human liver samples. The Pearson correlation coefficients between the pathologic grading of liver fibrosis and the Young modulus value under 5%,10%, and 15% preload strain conditions were 0.702, 0.669, and 0.626, respectively (P < .01). Commonly used imaging techniques include transient elastography, acoustic radiation force impulse imaging (ARFI), supersonic shear imaging, and magnetic resonance (MR) elastography. Acoustic radiation force impulse imaging uses high-intensity ultrasound beams to induce displacements inside the liver and provides tissue stiffness as the shear wave velocity. Toshima et al8 and Lupsor et al9 reported that the area under the receiver operating characteristic curve (AUROC) for the diagnosis of fibrosis (grade ≥3) with ARFI in patients with chronic liver disease was superior to serum markers and comparable with transient elastography. Supersonic shear imaging relies on the use of radiation force and a high–frame rate ultrasound scanner. The AUROC of supersonic shear imaging for mild and intermediate hepatic fibrosis was confirmed by Bavu et al10 to be higher than that of transient elastography. Magnetic resonance elastography11–13 has demonstrated the ability to detect advanced hepatic fibrosis, including occult cirrhosis, based on measurements of liver stiffness. Ultrasound-based transient elastography has been studied in both research and clinical settings. Transient elastography uses an ultrasound probe mounted on the axis of a mechanical vibrator, which generates a 50-Hz shear wave with a small amplitude. The displacements induced by the shear wave in the medium are measured, and the shear wave velocities are calculated by pulse-echo ultrasound as a reflection of liver stiffness. A meta-analysis aimed at the diagnostic accuracy of transient elastography for liver fibrosis showed that the average AUROC was 0.84 (95% confidence interval [CI] 0.82–0.86) for substantial fibrosis, and the average AUROC was 0.94 (95% CI, 0.93– 0.95) for cirrhosis.14 However, despite high AUROC values, several groups also reported that liver stiffness significantly
74
increases during inflammatory episodes and mechanical cholestasis,15–17 which will weaken the value of liver stiffness measurement for liver fibrosis staging. Importantly, these pathologic processes have less influence on the spleen. Thus, measurement of spleen stiffness may provide additional parameters for indirectly reflecting the status of the liver stiffness. It is well known that the splenic vein is a tributary of the portal system. The hemodynamics and morphologic characteristics of the spleen may change during the development of chronic liver disease,18–20 which may provide helpful information for diagnosis of cirrhosis. However, the changing trend of splenic elasticity during the development of liver fibrosis and its diagnostic value for liver fibrosis staging are currently unknown. Recently, some studies have used MR elastography, ARFI, and transient elastography to investigate spleen stiffness in patients with chronic liver diseases, especially for assessment of portal hypertension and esophageal varices.21–24 The aim of this study was to evaluate quantitative measurement of spleen stiffness with the FibroScan system (Echosens, Paris, France) for indirect assessment of liver fibrosis in patients with chronic liver disease and to correlate spleen stiffness with liver stiffness using pathologic examination as a reference standard.
Materials and Methods Patients From May 2011 to December 2011, 60 patients including 54 with clinically confirmed chronic hepatitis B and 6 with a clinical diagnosis of Child-Pugh class A hepatitis B virus– related cirrhosis were enrolled in this prospective study. All patients had a history of hepatitis B virus infection for at least 6 months before enrollment. The patients with liver cirrhosis were diagnosed by clinical history with serologic examination and computed tomographic (CT) or MR imaging confirmation. Patients were excluded if they were infected with other hepatitis viruses or the human immunodeficiency virus or if they were pregnant. Also, patients with fatty liver, drug-induced liver injury, alcoholic liver disease, autoimmune disease, biliary tract disease, ascites, and liver tumors were excluded. This study was approved by the University Ethics Committee, and written informed consent was obtained from all patients before the study. Measurement of Stiffness Quantitative stiffness values were obtained by the FibroScan system from parenchyma of the spleen and liver in all 60 patients, who fasted overnight. The scanning operator was a well-trained radiologist with at least 5 years of
J Ultrasound Med 2014; 33:73–81
331jum_online_Layout 1 12/19/13 1:55 PM Page 75
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
experience in sonography and special training in FibroScan examinations. Patients’ clinical and laboratory information were blinded to the operator. The scanning technique and examination procedure for measurement of liver stiffness were adapted from a previously published technique.25 All patients were examined with a medium-sized probe with a frequency of 3.5 MHz. First, conventional a ultrasound system equipped with a broadband 2.5–3.5-MHz convex probe (MyLab 70Xvision; Esaote Biomedica, Genoa, Italy) was used to image and localize the spleen and liver via an intercostal space. Once the appropriate areas of the spleen and liver were identified, the probe was manually held in a perpendicular position to the skin surface for measurements of spleen and liver stiffness at a depth range from 25 to 65 mm below the skin surface. The sample volume of the measurement on the FibroScan system is approximately 1 cm wide
and 4 cm long in a cylindrical shape. For spleen stiffness measurements, the sample volume was targeted at the region where the splenic parenchymal thickness was more than 4 cm and free of large vessels. These conditions were met in all 60 patients. The quality of data acquisition was further verified by TM-mode and A-mode imaging on the FibroScan system (Figure 1). The TM-mode represents the ultrasonic amplitude in a log scale as a function of depth and time, whereas the A-mode represents a real-time ultrasonic amplitude in a log scale as a function of depth. If the sample volume includes large vessels or dilated bile ducts, the A-mode line will display nonlinearity and fluctuation. When the TM-mode image was homogeneous and the A-mode line appeared linear with less fluctuation, it implied that the measured spleen/liver tissue was free of large vessels, which was appropriate for the examination.
Figure 1. Monitor image from FibroScan system: A, TM-mode; B, A-mode; C, elastographic image; D, tissue stiffness; E, success rate. TM-mode represents the ultrasonic amplitude in a log scale as a function of depth and time. A-mode represents the real-time ultrasonic line amplitude in a log scale as a function of depth. The operator uses TM-mode and A-mode images to select and localize the sample area. CS indicates current stiffness; and IQR, interquartile range.
J Ultrasound Med 2014; 33:73–81
75
331jum_online_Layout 1 12/19/13 1:55 PM Page 76
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
The measurement values were expressed in kilopascals. Ten valid measurements were performed on each patient. The success rate was calculated as the number of validated measurements divided by the number of total measurements. Measurements with an interquartile range of less than 30% of the median value and a success rate of greater than 60% were considered reliable data sets. According to the manufacturer’s recommendations, the median value of 10 measurements was considered representative of tissue stiffness (liver or spleen). Sonographically Guided Liver Biopsy Percutaneous sonographically guided liver biopsy was performed by 2 experienced radiologists using a Magnum automatic biopsy gun (C. R. Bard, Inc, Tempe, AZ) and a 16-gauge core biopsy needle after the stiffness measurements were obtained from the 54 patients with chronic hepatitis B within the same day of the study and the other 6 patients with cirrhosis 3 to 6 months before the study. The probes for both FibroScan scanning and biopsy guidance were manually held in a position perpendicular to the skin surface. The liver biopsy attempted to target the same area as the region of stiffness measurements by marking the location of the probe on the skin and sampling the tissue at the same depth as the transient elastographic examination. All specimens were at least 1.5 cm in length, fixed in 4% formalin, and embedded in paraffin. The specimen slices were stained with hematoxylin-eosin and Masson trichrome for light microscopic analyses. Each specimen was examined under microscopy by an experienced pathologist, who was blinded to both the clinical and imaging data of the patients. Liver fibrosis was scored in 5 grades based on a published standard26 as follows: stage S0, no fibrosis; S1, fibrous portal expansion and limited perisinusoidal or lobular fibrosis; S2, periportal fibrosis and few fibrous septa but intact architecture; S3, numerous fibrous septa with architectural distortion but no obvious cirrhosis; and S4, cirrhosis.
fibrosis stage) and the Pearson correlation test (for spleen stiffness and liver stiffness). The F test followed by the Games-Howell test was used to compare the liver and spleen stiffness values between different pathologic fibrosis stages. The diagnostic performance was evaluated by the AUROC. Optimal cutoff values were chosen to maximize the sum of sensitivity and specificity.
Results The 60 patients enrolled in this study include 46 male and 14 female patients with a mean age of 37 ± 15 years. The clinical findings of the patients are summarized in Table 1. Of the 60 patients who underwent liver biopsy, 17 had S0 fibrosis, 18 had S1, 9 had S2, 8 had S3, and 8 had S4. Spleen and Liver Stiffness for Staging Liver Fibrosis Spleen and liver stiffness values were successfully measured in all patients. The elastic moduli of spleen parenchyma in patients with S0, S1, S2, S3, and S4 fibrosis were 13.83 ± 2.60, 17.96 ± 3.85, 22.93 ± 5.97, 33.64 ± 12.38, and 57.53 ± 16.40 kPa, respectively. The elastic moduli of liver parenchyma in patients with S0, S1, S2, S3, and S4 fibrosis were 4.64 ± 0.70, 6.06 ± 1.29, 6.34 ± 1.10, 12.89 ± 3.49, and 17.28 ± 4.27 kPa. The spleen and liver stiffness values as related to fibrosis stages are shown as box plots in Figure 2. On the whole, there was a dramatic increase in spleen and liver stiffness in parallel with progression of liver fibrosis. A significant correlation was found between spleen stiffness and fibrosis stages (Spearman r = 0.833; P < .001). At the same time, the spleen stiffness values had a positive correlation with the liver stiffness values in all patients (Pearson r = 0.810; P < .001; Figure 3). P values for multiple comparisons of spleen and liver stiffness between different liver fibrosis stages are shown in Table 2. Table 1. Clinical Characteristics Characteristic
Statistical Analyses Statistical analyses were performed with SPSS version 13.0 software (SPSS Inc, Chicago, IL) and MedCalc version 12.3 software (MedCalc Software, Mariakerke, Belgium). P < .05 was considered statistically significant. The liver and spleen stiffness values and variables were expressed as mean ± standard deviation. Box plots were used to illustrate the distributions of the different parameters. The Levene test was used to investigate the homogeneity of variance. Correlations between different parameters were analyzed by the Spearman correlation test (for spleen stiffness and
76
Male, n (%) Age, y BMI, kg/m2 AST, U/L ALT, U/L ALP, U/L GGT, U/L T-BIL, μM
Value 46 (76.7) 37 ± 15 (20–55) 19.36 ± 2.46 (16.65–25.61) 56.23 ± 22.27 (27–84) 48.33 ± 18.75 (23–76) 62.28 ± 20.15 (36–115) 28.16 ± 13.40 (11–45) 10.25 ± 3.58 (4.26–16.49)
Data are presented as mean ± SD (range) where applicable. ALP indicates alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; GGT, γ-glutamyltranspeptidase; and T-BIL, total bilirubin.
J Ultrasound Med 2014; 33:73–81
331jum_online_Layout 1 12/19/13 1:55 PM Page 77
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
Diagnostic Value of Liver and Spleen Stiffness Evaluated by the AUROC The receiver operating characteristic curve analyses showed that spleen stiffness and liver stiffness had high predictive values for S1 or higher to S4 (AUROC = 0.899–0.999; P < .001, Table 3). On the other hand, the sensitivities decreased when using spleen or liver stiffness to estimate mild hepatic fibrosis (S1). When compared with the diagnostic value of liver and spleen stiffness, no statistical significance was found between the AUROC for estimating liver fibrosis stages (Z = 0.310–1.577; P = .115–0.756; Figure 4).
Discussion In the literature, several methods, such as MR elastography, transient elastography, and ARFI, have been used to measure spleen stiffness.21–24,27 The FibroScan system used in this study is an ultrasound-based modality for quantitative assessment of tissue elasticity. The probe generates elastic shear waves that propagate through tissues. The propagation velocity, which is correlated with tissue stiffness, is measured by pulse-echo ultrasound. In this study, the mean spleen stiffness value in patients with S0 liver fibrosis was 13.83 kPa, which was similar to that (17.8 kPa) in 17 control patients in a previous study that used the FibroScan system23 but greater than that (2.35– 5.58 kPa) measured in 16 healthy participants reported by
Nedredal et al21 using MR elastography. Because of the small sample sizes of these studies, further investigation is needed to establish the normal range of spleen stiffness. It is well known that histologic modifications lead to changes in the mechanical behavior of organs. Mechanical and histologic analyses have been combined in a number of studies of the human liver.7,28,29 By means of aspiration experiments, Mazza et al28 reported that the stiffness values (0.3 and 0.44 bar/mm) of 2 fibrotic liver samples among a total of 17 samples were 3 times greater than those of normal tissue (0.07–0.18 bar/mm). A study reported by DeWall et al29 showed F4-graded fibrotic liver tissue (n = 2) was significantly stiffer than F0-graded tissue (n = 14), with a modulus contrast of 4:1. In our study, the mean liver stiffness was similar to the Young modulus of liver simples under a 15% preload strain reported by Yeh et al.7 The S4 to S0 liver stiffness ratio was also comparable with the ratios of elastic parameters in previous studies.28,29 Several clinical studies have shown that liver stiffness is correlated with the severity of liver fibrosis and that liver stiffness is a good indicator for estimating liver fibrosis stages.30,31 This study demonstrated that there was an excellent correlation (Pearson r = 0.810; P < .001) between spleen and liver stiffness in all patients from S0 to S4, which implied that measurement of spleen stiffness may be suitable for indirectly assessing liver fibrosis. Theoretically, the related correlation between spleen and liver stiffness is
Figure 2.Spleen stiffness measurement (SSM) and liver stiffness measurement (LSM) values for different stages of liver fibrosis. The circles display all observations. The ends of the boxes are the upper and lower quartiles, such that the boxes span the interquartile ranges. The medians are marked by horizontal lines inside the boxes. The whiskers extend to the highest and lowest observations. Spleen and liver stiffness increased as the liver fibrosis stage increased, especially in S3 and S4.
J Ultrasound Med 2014; 33:73–81
77
331jum_online_Layout 1 12/19/13 1:55 PM Page 78
Hu et al—Prediction of Liver Fibrosis by Quantitative Measurement of Spleen Stiffness
based on the pathophysiologic changes of chronic liver disease, including hyperdynamic splenic blood flow, splenic parenchymal hyperplasia, and splenomegaly,18,19,32 especially in late stages of fibrosis. The results of this study showed that spleen stiffness values were greater than liver stiffness values in the same patient, which is consistent with other studies.21,23,33 Stefanescu et al23 reported that the mean spleen and liver stiff-
ness in values in patients with cirrhosis without esophageal varices were 46.05 (range, 23.7–75.0) and 26.3 (range, 9.9–40.3) kPa, respectively, and these values were similar to the results obtained in this study. The gaps in spleen stiffness measurements between adjacent stages were also greater than those of liver stiffness, in which later stages may be helpful identifying the adjacent liver fibrosis stage, and the standard deviations of spleen stiffness were also greater than those of liver stiffness; thus, the differences in spleen stiffness between different liver fibrosis stages were not always evident and distinguishable. Spleen stiffness measurements in this study had no significant differences for S1 versus S2 and S2 versus S3 but had statistically significant differences between the other groups. Liver stiffness measurements were not significantly different for S1 versus S2 and S3 versus S4 but were statistically significantly different between the rest of the groups. Therefore, spleen stiffness measurement was better than liver stiffness measurement for identification of S3 versus S4, whereas spleen stiffness measurement was inferior to liver stiffness measurement in terms of S2 versus S3 differentiation.
Figure 3. Spleen stiffness measurement (SSM) and liver stiffness measurement (LSM) values in different patients (Pearson r = 0.810; P < .001).
Table 2. P Values for Multiple Comparisons of Spleen and Liver Stiffness Measurements Between Different Liver Fibrosis Stages Fibrosis Stage S0 S1 S2 S3 S4
S0
S1
Spleen Stiffness S2 S3
.007 .007 .01 .016 .001
.221 .049 .001
.01 .221 .247 .003
.016 .049 .247
S4 .001 .001 .003 .039
.039
S0
S1 .003
.003 .009 .002
