Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Jee-Young Son, MD Jae Young Lee, MD Nam-Joon Yi, MD Kwang-Woong Lee, MD Kyung-Suk Suh, MD Kwang Gi Kim, PhD Jeong Min Lee, MD Joon Koo Han, MD Byung Ihn Choi, MD
Purpose:
To investigate the diagnostic performance of acoustic structure quantification (ASQ) for the assessment of hepatic steatosis by using hydrogen 1 (1H) magnetic resonance (MR) spectroscopy as the reference standard and to compare ASQ with hepatorenal ratio.
Materials and Methods:
This prospective study was approved by an institutional review board, and informed written consent was obtained from all participants. ASQ and MR spectroscopy were performed in 89 participants (mean age, 41.48 years 6 14.16; 35 men, 54 women) without history of chronic liver disease. Obtained were focal disturbance (FD) ratio by using ASQ, hepatic fat fraction (HFF) by using MR spectroscopy, and hepatorenal ratio by using a histogram. Correlation coefficient, intraclass correlation coefficient, and receiver operating curve analyses were performed.
Results:
FD ratio measured with ASQ had a strong linear correlation with HFF measured with MR spectroscopy after logarithmic transformation of both variables (r = 20.87; P , .001). By using HFF of 5.79% as a cutoff value of 10% hepatic steatosis, 29 of 89 participants (32.6%) were categorized into the group with hepatic steatosis of 10% or greater (mean HFF, 13.18% 6 4.89). The area under curve of the FD ratio for diagnosing hepatic steatosis 10% or greater was 0.959 (95% confidence interval: 0.895, 0.990) with sensitivity of 86.2% (95% confidence interval: 68.3%, 96.0%) and specificity of 100% (95% confidence interval: 94.0%, 100.0%) by using a cutoff value of 0.1; the area under curve and specificity of the FD ratio were significantly higher than those of the hepatorenal ratio (respectively, 0.772 and 73.3%; respective P values, .001 and ,.001).
Conclusion:
This pilot study in a cohort of patients with hepatic steatosis without other parenchymal disease suggested ASQ may be valuable for the quantification of hepatic steatosis and detection of hepatic steatosis 10% or greater in living liver donors.
1
From the Department of Radiology, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Korea (J.Y.S.); Department of Radiology, Institute of Radiation Medicine (J.Y.L., J.M.L., J.K.H., B.I.C.), and Department of Surgery (N.J.Y., K.W.L., K.S.S.), Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Republic of Korea; and Biomedical Engineering Branch, Division of Convergence Technology, National Cancer Center, Goyang, Gyeonggi-do, Korea (K.G.K.). Received August 18, 2014; revision requested October 14; revision received April 1, 2015; accepted April 21; final version accepted May 1. Supported by grant no. 04-2012-0590 of the SNUH Research Fund and by the Ministry of Science, ICT and Future Planning (MSIP), Korea, under the Convergence Information Technology Research Center (C-ITRC) support program (NIPA-2014-H0401-14-1002) supervised by the National IT Industry Promotion Agency (NIPA). Address correspondence to J.Y.L. (e-mail: [email protected]).
RSNA, 2015
q
RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
1
Original Research n Ultrasonography
Hepatic Steatosis: Assessment with Acoustic Structure Quantification of US Imaging1
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
H
epatic steatosis, which indicates the accumulation of fat in hepatocytes, is common and has a broad disease spectrum according to its pathogenesis and its severity (1). Clinically, the occurrence of nonalcoholic fatty liver disease, the most common type of hepatic steatosis, is strongly correlated with metabolic disease, including type 2 diabetes mellitus and atherosclerotic cardiovascular disease (2). Simple hepatic steatosis can evolve into a more severe stage, such as steatohepatitis or cirrhosis. Furthermore, hepatic steatosis is an important risk factor for postoperative complications after major liver resection and living donor liver transplantation (3–7). Hepatic steatosis of 10% or greater is known to be critical in patients who undergo living donor liver transplantation because of the risks of initial graft dysfunction, poor graft survival, and other complications (8–10). The widespread use of liver biopsy (which is a reference standard for quantification of liver fat) as a screening tool and to monitor treatment response is limited mainly because of invasiveness. Therefore, noninvasive imaging methods that can quantitatively measure liver fat were actively investigated. Among them, hydrogen 1 (1H) magnetic resonance (MR) spectroscopy gained ground in the past decade as an alternative noninvasive reference
Advances in Knowledge nn In a population of living donors for liver transplantation, acoustic structure quantification (ASQ) provides an excellent quantitative tool for hepatic steatosis, and there is a strong linear correlation (r = 20.87; P , .001) between focal disturbance ratio measured with ASQ and hepatic fat fraction measured with MR spectroscopy. nn ASQ provides a tool with a diagnostic accuracy (area under curve, 0.959) by using MR spectroscopy as the reference standard for diagnosing hepatic steatosis that is 10% or greater for living donor liver transplantation. 2
standard to evaluate liver fat content because of its high diagnostic accuracy and reproducibility (1,2,11–15). However, MR spectroscopy is expensive to perform as a screening and monitoring tool for hepatic steatosis. Ultrasonographic (US) imaging is a modality that is simple and inexpensive to perform and is safe for the patient. To overcome the intrinsic subjectivity associated with the use of US imaging to grade fatty liver, several methods that use US image data were proposed for the quantification of hepatic steatosis (14–20). However, these methods are not widely used in clinical practice mainly because of their complexity or their use of noncommercial software. A commercially available quantification method, acoustic structure quantification (ASQ), was introduced into clinical practice for the evaluation of diffuse liver disease. The method is based on statistical analysis of the difference between theoretical and real echo amplitude distribution (21,22). Theoretical echo amplitude distribution indicates that a speckle pattern in a certain liver region is approximated by a function of a Reyleigh distribution based on the assumption that the speckle pattern is generated only by ultrasonic interference of very small scattering objects that are located closer than the wavelength of US. However, real echo amplitude distribution of normal liver does not fit a Reyleigh distribution mainly because of the presence of small vessel walls. As hepatic steatosis progresses, the real echo amplitude distribution more likely approaches the theoretical echo amplitude distribution with masking of the small structures, such as small vessel walls (21). In an animal model of hepatic steatosis, a significant correlation was observed between focal disturbance (FD) ratio, calculated by using ASQ, and either fat droplet area or fat droplet size (21). Therefore, it was suggested that the FD ratio could serve as a quantitative
Implication for Patient Care nn ASQ with a statistical model of echo amplitudes distribution may be beneficial to detect substantial hepatic steatosis in liver donors.
Son et al
biomarker of hepatic steatosis. However, to our knowledge, there has been no clinical study that investigated the value of ASQ for the quantification of hepatic steatosis. Therefore, the purpose of our study was to investigate the diagnostic performance of ASQ for the assessment of hepatic steatosis by using 1H MR spectroscopy as the reference standard and to compare ASQ with hepatorenal ratio.
Materials and Methods Study Population This prospective study was approved by our institutional review board and all participants gave written informed consent. Between September 2012 and September 2013, we enrolled 53 consecutive potential donors for living liver donor transplantation and 40 healthy patients in our study. We enrolled 40 healthy patients who annually visited our institute for health check-up and in whom hepatic steatosis was suspected or detected by using routine US examination. Before enrollment, the electronic medical records and laboratory findings (which included liver function tests within 3 months before ASQ) of all individuals were checked to confirm that
Published online before print 10.1148/radiol.2015141779 Content code: Radiology 2015; 000:1–8 Abbreviations: ASQ = acoustic structure quantification FD = focal disturbance HFF = hepatic fat fraction ROI = region of interest Author contributions: Guarantors of integrity of entire study, J.Y.S., J.Y.L., K.W.L., K.G.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, J.K.H.; clinical studies, J.Y.S., J.Y.L., N.J.Y., K.W.L., K.S.S., J.M.L., J.K.H.; experimental studies, K.G.K.; statistical analysis, J.Y.L., J.Y.S.; and manuscript editing, K.W.L., B.I.C. Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
they did not have any past medical history, clinical symptoms, or signs of other liver or renal parenchymal disease or injury, or systemic or malignant disease. During this process, three invited participants were excluded because of late discovery of past medical history of tongue cancer (n = 1), breast cancer (n = 1), and liver cirrhosis (n = 1). All individuals underwent both US and 1H MR spectroscopy within 1 week of each other. One invited participant was excluded from our study because of claustrophobia. The remaining 89 participants were finally included in this study.
US Examination All US examinations were performed by using a US imager (Aplio XG; Toshiba Medical Systems, Otawara, Japan) equipped with a 5-MHz convex transducer. The examinations were performed by one of two radiologists (J.Y.S. and J.Y.L., with 6 and 20 years of experience in liver US imaging, respectively). First, a routine liver US examination that included five different sagittal liver and right kidney views was performed. Then, US images on ASQ mode were obtained three times each from right intercostal view and right subcostal view. Display depth and transmit focus were fixed at 10 cm and 6 cm, respectively. As many regions of interest (ROIs) as possible that were as large as possible were placed on the liver, with care taken to avoid large hepatic vessels or artifacts. FD ratio automatically appeared on a monitor that displayed US images in ASQ mode (Fig 1a). The mean ROIs per US image on ASQ mode was 2.8 6 0.4 (standard deviation; range, 1–4). For comparison, hepatorenal ratio was obtained by using an echo intensity analysis of a digitized liver and right kidney image (K.K.G. and K.Y.J., division of convergence technology, National Cancer Center, with 16 and 12 years of’ experience in the development of medical imaging quantitative analysis software, respectively), which was described for the quantification of hepatic steatosis (15,23). Two rectangular ROIs (1.5 3 1.5 cm) were placed on the hepatic parenchyma and renal cortex, with
Son et al
Figure 1
Figure 1: US images show ASQ and hepatorenal ratio. (a) Multiple ROIs are placed on the liver, avoiding large intrahepatic vascular structures. Automatic display of Cm2 histogram (Cm2 is a modified adjusted x2 value calculated on the basis of the percentage of real distribution variance relative to the theoretical distribution variance of echo amplitude in ROIs) is shown in the lower right corner and focal disturbance ratio is at the bottom. (b) ROIs were placed on the hepatic parenchyma and renal cortex at a depth similar to a probe near the central part of the US image, with care taken to avoid large vessels, the renal sinus, or the renal medulla. Hepatorenal ratio is calculated by using 1.5 3 1.5 cm rectangular ROIs.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
3
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
Table 1
Son et al
Figure 2
Demographics According to Clinical Characteristics and HFF Measured by Using MR Spectroscopy Parameter No. of patients Men Women Age (y) Body mass index (kg/m2) .25 kg/m2 No. of patients .28 kg/m2 No. of patients .30 kg/m2 No. of patients Time interval between ASQ and MR spectroscopy (d) HFF (%)
Liver Donor Candidates
Invited Participants
Total
53 33 20 32.85 6 10.12 (17–53) 23.59 6 3.55 (17.19–33.64) 27.86 6 2.33 17 30.10 6 1.83 7 32.55 6 1.54 2 0.98 6 3.10 (0–7)
36 21 15 51.72 6 11.58 (27–67) 24.69 6 4.10 (18.83–36.34) 28.59 6 3.64 14 31.70 6 3.55 6 34.59 6 2.45 3 0 6 0 (0–0)
89 54 35 40.48 6 14.16 (17–67) 24.05 6 3.80 (17.10–36.34) 28.19 6 2.96 31 30.84 6 2.77 13 33.77 6 2.20 5 0.58 6 2.43 (0–7)
4.20 6 3.85 (0.70–17.30)
8.67 6 7.05 (0.96–26.7)
6.01 6 5.78 (0.70–26.7)
Figure 2: Box-and-whisker plot of HFFs of liver donor candidates and invited participants. The lines within the boxes indicate medians. Black squares represent outliers and dots represent plotting of all data.
Note.—Except where indicated, data are mean 6 standard deviation and data in parentheses are range.
the similar depth from the probe near the central part of the US image, and with care taken to avoid large vessels, the renal sinus, or the renal medulla (Fig 1b). The hepatorenal ratio was automatically calculated by dividing the mean value of the liver ROI by the mean value of the kidney ROI. The hepatorenal ratio was obtained in each of the five liver and right kidney images and averaged.
MR Imaging Technique for Hepatic Fat Quantification High-speed, T2-corrected, multiecho, single-voxel 1H MR spectroscopy was performed with a 3-T MR imager (Magnetom Verio; Siemens Healthcare, Erlangen, Germany) by using a 32-channel phased-array surface coil. A modified stimulated-echo acquisition sequence was used (echo times msec/repetition time msec, 12, 24, 36, 48, and 72/3000). Other details in the MR spectroscopy protocol were the same as those used previously (10). Sagittal, coronal, and axial sections that covered the whole liver were preliminarily acquired to position the spectroscopy acquisition voxel. A single voxel (3 3 3 3 3 cm3) was placed within the dome of the right hepatic 4
lobe, avoiding major vascular and biliary structures and the periphery of the liver. The hepatic fat fraction (HFF) was calculated automatically. An HFF cutoff value of 5.79% was used for hepatic steatosis, which was validated for hepatic steatosis that were 10% or greater by using the same protocol and same machine as previously used (10). Ten percent hepatic steatosis is known as a critical margin for liver donor safety (8–10).
Statistical Analysis All statistical analyses were performed with statistical software (MedCalc version 12.4.0.0; MedCalc Software, Mariakerke, Belgium). We used the Pearson correlation coefficient to calculate the correlation between the FD ratio and HFF and between the hepatorenal ratio and HFF. We analyzed Pearson correlation coefficient according to body mass index (.25 kg/m2, .28 kg/m2, and .30 kg/m2). Receiver operating characteristic curve analysis with binomial exact confidence interval was performed to obtain the area under the curve and to determine the cutoff value of the FD ratio and hepatorenal ratio for the diagnosis of hepatic steatosis 10% or greater.
Sensitivity, specificity, positive predictive value, and negative predictive value were also obtained. Intraclass correlation coefficient was calculated to estimate the degree of FD ratio agreement between subcostal imaging and intercostal imaging. We used a Student t test that assumed equal variances. A P value of less than .05 was considered to indicate statistical significance.
Results Patient Characteristics This study included 54 men (mean age, 40.7 years 6 13.4) and 35 women (mean age, 40.1 years 6 15.5; Table 1). The HFF of the invited participants (mean HFF, 8.67% 6 7.05 [95% confidence interval: 6.28%, 11.05%]) was significantly higher than that of the liver donor candidates (mean HFF, 4.20% 6 3.85 [95% CI: 3.14%, 5.26%]; P , .001; Fig 2). Twenty-nine of 89 participants (32.6%) were categorized into the group that had hepatic steatosis of 10% or greater (mean HFF, 13.18% 6 4.89). The remaining 60 participants (67.4%) were categorized into the group that had less than 10% hepatic steatosis (mean HFF, 2.54% 6
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
0.98). Of the 53 liver donor candidates, 10 (18.9%) were categorized into the group that had hepatic steatosis of 10% or greater (mean HFF, 11.12% 6 3.99) and 43 (81.1%) were categorized into the group that had hepatic steatosis of less than 10% (mean HFF, 2.59% 6 0.91).
FD Ratio and Hepatorenal Ratio in Total Population The correlation coefficient (r) between FD ratio and HFF was 20.62 (95% confidence interval: 20.76, 20.48). However, a scatter plot between these two
variables showed an exponential relationship (Fig 3a). When a logarithm of these variables was applied, the two showed a strong linear relationship (r = 20.87 [95% confidence interval: 20.91, 20.80]; P , .001) (Fig 3b); in patients with body mass index greater than 25 kg/m2, r was 20.83 (95% confidence interval: 20.91, 20.67; P , .001); in patients with body mass index greater than 28 kg/m2, r was 20.80 (95% confidence interval: 20.94, 20.44; P , .001); and in patients with body mass index greater than 30 kg/m2, r was 20.87 (95% confidence interval:
Son et al
20.99, 20.07; P = .058). The correlation coefficient between hepatorenal ratio and HFF was 0.49 (95% confidence interval: 0.32, 0.64; P , .001) and showed a linear relationship (Fig 3c). The area under the curve of the FD ratio for detection of hepatic steatosis 10% or greater was 0.959 (P , .001) (Table 2). The optimal cutoff value of the FD ratio was 0.1 with sensitivity of 86.2% and specificity of 100%. The area under the curve of hepatorenal ratio for detection of hepatic steatosis 10% or greater was 0.772 (95% confidence interval: 0.671, 0.854; P , .001),
Figure 3
Figure 3: Correlation between HFF and FD ratio or hepatorenal ratio. (a) Scatter plot shows an exponential correlation between HFF and FD ratio. Exponential relationship is shown. (b) Regression plot shows a strong negative linear correlation between HFF and FD ratio after logarithmic transformation of both variables. The solid line indicates regression line; the lines with small dots are 95% confidence intervals; and lines with long dashes are 95% prediction intervals. (r = 20.8669). (c) Regression plot shows a positive linear correlation between HFF and hepatorenal ratio. The solid line indicates regression line; the lines with small dots are 95% confidence intervals; and lines with long dashes are 95% prediction intervals (r = 0.4935).
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
5
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
Son et al
Table 2 Diagnostic Performance of FD Ratio and Hepatorenal Ratio in the Total Population and Liver Donor Candidates Sensitivity Parameter Total population FD ratio Hepatorenal ratio Liver donor candidates FD ratio Hepatorenal ratio
Area Under the Curve
Cutoff Value
0.959 (0.895, 0.990) 0.772 (0.671, 0.854) 0.895 (0.780, 0.962) 0.660 (0.517, 0.785)
Specificity
Percentage
Numerator/ Denominator
Percentage
Numerator/ Denominator
Positive Predictive Value (%)
Negative Predictive Value (%)
0.1 1.326
86.2 (68.3, 96.0) 72.4 (52.8, 87.3)
25/29 21/29
100 (94.0, 100) 73.3 (60.3, 83.9)
60/60 44/60
100 (74.0, 100) 34.1 (17.9, 53.5)
97.4 (90.9, 99.7) 93.3 (83.6, 98.2)
0.1 1.312
70.0 (34.8, 93.3) 60.0 (26.2, 87.8)
7/10 6/10
100 (91.8, 100) 74.4 (58.8, 86.5)
43/43 32/43
100 (53.2, 100) 30.9 (11.0, 57.9)
94.6 (83.8, 99.1) 90.7 (76.4, 97.8)
Note.—Data in parentheses indicate 95% confidence intervals. Positive predictive value and negative predictive value were calculated with disease prevalence of 16% (8). All data met the HFF cutoff value of .5.79%.
which is significantly lower than that of the FD ratio (P = .001). The cutoff value of hepatorenal ratio was 1.326, with sensitivity of 72.4% and specificity of 73.3%. The specificity of hepatorenal ratio was significantly lower than that of the FD ratio (P , .001).
FD Ratio and Hepatorenal Ratio in Liver Donor Candidates The correlation coefficient between logarithms of both FD ratio and HFF was 20.77 (95% confidence interval: 20.86, 20.63; P , .001) and showed a strong linear relationship. The correlation coefficient r between the hepatorenal ratio and HFF was 0.41 (95% confidence interval: 0.16, 0.61) with a linear relationship. The areas under the curve, cutoff values, and diagnostic performances of FD ratio for detection of hepatic steatosis 10% or greater are summarized in Table 2. The area under the curve of the hepatorenal ratio for detection of hepatic steatosis 10% or greater was 0.660 (95% confidence interval: 0.517, 0.785), which is significantly lower than that of FD ratio (P = .011). In this regard, the cutoff value of hepatorenal ratio was 1.312 (sensitivity, 60.0%; specificity, 74.4%). The specificity of hepatorenal ratio was significantly lower than that of the FD ratio (P , .001). FD Ratio according to Imaging Approach of ASQ The intraclass correlation coefficient of the FD ratio between subcostal and 6
intercostal approaches was 0.896 (95% confidence interval: 0.841, 0.932). The intraclass correlation coefficients of the FD ratios in subcostal and intercostal liver scans were 0.896 (95% confidence interval: 0.841, 0.932) and 0.893 (95% confidence interval: 0.848, 0.927), respectively. The areas under the curve of the FD ratio found by using intercostal and subcostal approaches for the diagnosis of hepatic steatosis 10% or greater were 0.970 and 0.945, respectively. For comparison, the area under the curve of the FD ratio found by using both imaging approaches was 0.959. There were no significant differences in area under the curve between the intercostal and subcostal approaches (P , .302).
Discussion In this study, the FD ratio was excellent for the diagnosis of hepatic steatosis 10% or greater with high diagnostic accuracy. This result suggests that ASQ could be valuable for the preoperative detection of hepatic steatosis 10% or greater, which is critical in patients who undergo living donor liver transplantation because of the risks of initial graft dysfunction, poor graft survival, and other complications (8,9). The FD ratio was correlated with HFF (R2 = 0.7515). Even in obese patients with body mass index greater than 25 kg/m2 or greater than 28 kg/ m2, the correlation coefficient was high. This implies that ASQ could potentially
be used to quantitatively monitor liver fat change. It is well known that hepatorenal ratio is significantly higher in patients with hepatic steatosis than in healthy patients (18,23,24). However, it is still questionable whether hepatorenal ratio is a sufficient biomarker that can be used as an alternative to MR spectroscopy for hepatic steatosis quantification. Even though a previous study reported that hepatorenal ratio had an excellent correlation with the degree of steatosis at MR spectroscopy (sensitivity, 100%; specificity, 95%) (15), other studies reported the need for an additional process to enhance the diagnostic performance of hepatorenal ratio values, such as standardization by a tissue-mimicking phantom or artificial neural network (18,25). Differences between this study and previous studies regarding diagnostic performance of hepatorenal ratio may be caused by factors such as use of different machines and different image analysis software, different study population, and different criteria to define a cohort of patients with hepatic steatosis. Our results revealed that the FD ratio is independent of the subcostal or intercostal method of liver imaging. Our study has several limitations. First, our study used MR spectroscopy rather than histologic examination as the reference standard. However, many clinical trials (1,2,11–15) used MR spectroscopy as the reference standard because of the method’s accuracy,
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
quantitative nature, and noninvasiveness when quantifying liver fat. We used our own population to set thresholds that were used to determine the accuracy. This can overestimate study results. Second, the population of this study was heterogeneous: both consecutive liver donor candidates and invited participants were included in our study. Although the disease prevalence of hepatic steatosis 10% or greater in the group of liver donor candidates was similar to that of previous studies (8,26), HFFs of included liver donor candidates (95% confidence interval: 3.14, 5.26) were too low to represent the range of HFF expected in clinical practice. To include larger HFFs, individuals suspected of having or who had evidence of hepatic steatosis were additionally included (95% confidence interval: 6.28, 11.05). Nevertheless, the number of patients with an HFF of 13%–17% (which was reported to most likely represent hepatic steatosis 30% or greater) was not enough in our study to verify the value of ASQ for detection of 30% or greater (27–29). To prove this, a further study that includes a sufficient number of patients with hepatic steatosis 30% or greater is needed. Third, our study only included simple hepatic steatosis. However, in clinical situations, hepatic steatosis presents on a diverse spectrum, from simple hepatic steatosis through steatohepatitis to cirrhosis. It is unclear how this FD ratio will perform in patients with both steatosis and fibrosis. Fourth, the size of ROIs to measure ASQ in the liver was not recorded because it was not provided by the US machine that we used. In conclusion, the results of our study suggest that FD ratio calculated by using ASQ is comparable to HFF calculated by using MR spectroscopy to quantify hepatic steatosis and detect hepatic steatosis greater than 10% in living donor liver patients. Disclosures of Conflicts of Interest: J.Y.S. disclosed no relevant relationships. J.Y.L. disclosed no relevant relationships. N.J.Y. disclosed no relevant relationships. K.W.L.
disclosed no relevant relationships. K.S.S. disclosed no relevant relationships. K.G.K. disclosed no relevant relationships. J.M.L. Activities related to the present article: Siemens Healthcare provided nonfinancial support by providing the research software and technical support for installing the sequence of chemicalshift imaging and breath-hold MR spectroscopy. Activities not related to the present article: author disclosed a grant and personal fees from Bayer Healthcare; author disclosed grants from GE Healthcare, Guerbet, Dong Seo Medical, RF Medical, Starmed, and Toshiba Healthcare; author disclosed nonfinancial support from Philips for technical support for research software. Other relationships: disclosed no relevant relationships. J.K.H. disclosed no relevant relationships. B.I.C. disclosed no relevant relationships.
References 1. Ma X, Holalkere NS, Kambadakone R A, Mino-Kenudson M, Hahn PF, Sahani DV. Imaging-based quantification of hepatic fat: methods and clinical applications. RadioGraphics 2009;29(5):1253–1277. 2. Schwenzer NF, Springer F, Schraml C, Stefan N, Machann J, Schick F. Non-invasive assessment and quantification of liver steatosis by ultrasound, computed tomography and magnetic resonance. J Hepatol 2009; 51(3):433–445.
Son et al
9. Trotter JF, Wisniewski KA, Terrault NA, et al. Outcomes of donor evaluation in adultto-adult living donor liver transplantation. Hepatology 2007;46(5):1476–1484. 10. Hwang I, Lee JM, Lee KB, et al. Hepatic steatosis in living liver donor candidates: preoperative assessment by using breath-hold triple-echo MR imaging and 1H MR spectroscopy. Radiology 2014;271(3):730–738. 11. Pineda N, Sharma P, Xu Q, Hu X, Vos M, Martin DR. Measurement of hepatic lipid: high-speed T2-corrected multiecho acquisition at 1H MR spectroscopy—a rapid and accurate technique. Radiology 2009; 252(2):568–576. 12. Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic resonance spectroscopy to measure hepatic triglyceride content: prevalence of hepatic steatosis in the general population. Am J Physiol Endocrinol Metab 2005; 288(2):E462–E468. 13. Bohte AE, Koot BG, van der Baan-Slootweg OH, et al. US cannot be used to predict the presence or severity of hepatic steatosis in severely obese adolescents. Radiology 2012; 262(1):327–334. 14. Edens MA, van Ooijen PM, Post WJ, et al. Ultrasonography to quantify hepatic fat content: validation by 1H magnetic resonance spectroscopy. Obesity (Silver Spring) 2009; 17(12):2239–2244.
3. McCormack L, Petrowsky H, Jochum W, Furrer K, Clavien PA. Hepatic steatosis is a risk factor for postoperative complications after major hepatectomy: a matched case-control study. Ann Surg 2007;245(6):923–930.
15. Mancini M, Prinster A, Annuzzi G, et al. Sonographic hepatic-renal ratio as indicator of hepatic steatosis: comparison with (1)H magnetic resonance spectroscopy. Metabolism 2009;58(12):1724–1730.
4. Kooby DA, Fong Y, Suriawinata A, et al. Impact of steatosis on perioperative outcome following hepatic resection. J Gastrointest Surg 2003;7(8):1034–1044.
16. Strauss S, Gavish E, Gottlieb P, Katsnel son L. Interobserver and intraobserver variability in the sonographic assessment of fatty liver. AJR Am J Roentgenol 2007; 189(6):W320–W323.
5. Zamboni F, Franchello A, David E, et al. Effect of macrovescicular steatosis and other donor and recipient characteristics on the outcome of liver transplantation. Clin Transplant 2001;15(1):53–57. 6. Urena MA, Moreno Gonzalez E, Romero CJ, Ruiz-Delgado FC, Moreno Sanz C. An approach to the rational use of steatotic donor livers in liver transplantation. Hepatogastroenterology 1999;46(26):1164–1173. 7. McCormack L, Clavien PA. Understanding the meaning of fat in the liver. Liver Transpl 2005;11(2):137–139. 8. Sharma A, Ashworth A, Behnke M, Cotterell A, Posner M, Fisher RA. Donor selection for adult-to-adult living donor liver transplantation: well begun is half done. Transplantation 2013;95(3):501–506.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
17. Graif M, Yanuka M, Baraz M, et al. Quantitative estimation of attenuation in ultrasound video images: correlation with histology in diffuse liver disease. Invest Radiol 2000; 35(5):319–324. 18. Kim SH, Lee JM, Kim JH, et al. Appropriateness of a donor liver with respect to macrosteatosis: application of artificial neural networks to US images—initial experience. Radiology 2005;234(3):793–803. 19. Acharya UR, Sree SV, Ribeiro R, et al. Data mining framework for fatty liver disease classification in ultrasound: a hybrid feature extraction paradigm. Med Phys 2012;39 (7):4255–4264. 20. Mihăilescu DM, Gui V, Toma CI, Popescu A, Sporea I. Computer aided diagnosis method for steatosis rating in ultrasound images using
7
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
random forests. Med Ultrason 2013;15(3): 184–190. 21. Kuroda H, Kakisaka K, Kamiyama N, et al. Non-invasive determination of hepatic steatosis by acoustic structure quantification from ultrasound echo amplitude. World J Gastroenterol 2012;18(29):3889– 3895. 22. Toyoda H, Kumada T, Kamiyama N, et al. B-mode ultrasound with algorithm based on statistical analysis of signals: evaluation of liver fibrosis in patients with chronic hepatitis C. AJR Am J Roentgenol 2009; 193(4):1037–1043. 23. von Volkmann HL, Havre RF, Løberg EM, et al. Quantitative measurement of ultrasound
8
attenuation and Hepato-Renal Index in NonAlcoholic Fatty Liver Disease. Med Ultrason 2013;15(1):16–22. 24. Osawa H, Mori Y. Sonographic diagnosis of fatty liver using a histogram technique that compares liver and renal cortical echo amplitudes. J Clin Ultrasound 1996;24(1):25– 29. 25. Xia MF, Yan HM, He WY, et al. Standardized ultrasound hepatic/renal ratio and hepatic attenuation rate to quantify liver fat content: an improvement method. Obesity (Silver Spring) 2012;20(2):444–452. 26. Pascher A, Sauer IM, Walter M, et al. Donor evaluation, donor risks, donor outcome, and donor quality of life in adult-to-adult
Son et al
living donor liver transplantation. Liver Transpl 2002;8(9):829–837. 27. Idilman IS, Aniktar H, Idilman R, et al. Hepatic steatosis: quantification by proton density fat fraction with MR imaging versus liver biopsy. Radiology 2013;267(3):767–775. 28. Noworolski SM, Lam MM, Merriman RB, Ferrell L, Qayyum A. Liver steatosis: concordance of MR imaging and MR spectroscopic data with histologic grade. Radiology 2012; 264(1):88–96. 29. Tang A, Tan J, Sun M, et al. Nonalcoholic fatty liver disease: MR imaging of liver proton density fat fraction to assess hepatic steatosis. Radiology 2013;267(2):422–431.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
Jee-Young Son, MD Jae Young Lee, MD Nam-Joon Yi, MD Kwang-Woong Lee, MD Kyung-Suk Suh, MD Kwang Gi Kim, PhD Jeong Min Lee, MD Joon Koo Han, MD Byung Ihn Choi, MD
Purpose:
To investigate the diagnostic performance of acoustic structure quantification (ASQ) for the assessment of hepatic steatosis by using hydrogen 1 (1H) magnetic resonance (MR) spectroscopy as the reference standard and to compare ASQ with hepatorenal ratio.
Materials and Methods:
This prospective study was approved by an institutional review board, and informed written consent was obtained from all participants. ASQ and MR spectroscopy were performed in 89 participants (mean age, 41.48 years 6 14.16; 35 men, 54 women) without history of chronic liver disease. Obtained were focal disturbance (FD) ratio by using ASQ, hepatic fat fraction (HFF) by using MR spectroscopy, and hepatorenal ratio by using a histogram. Correlation coefficient, intraclass correlation coefficient, and receiver operating curve analyses were performed.
Results:
FD ratio measured with ASQ had a strong linear correlation with HFF measured with MR spectroscopy after logarithmic transformation of both variables (r = 20.87; P , .001). By using HFF of 5.79% as a cutoff value of 10% hepatic steatosis, 29 of 89 participants (32.6%) were categorized into the group with hepatic steatosis of 10% or greater (mean HFF, 13.18% 6 4.89). The area under curve of the FD ratio for diagnosing hepatic steatosis 10% or greater was 0.959 (95% confidence interval: 0.895, 0.990) with sensitivity of 86.2% (95% confidence interval: 68.3%, 96.0%) and specificity of 100% (95% confidence interval: 94.0%, 100.0%) by using a cutoff value of 0.1; the area under curve and specificity of the FD ratio were significantly higher than those of the hepatorenal ratio (respectively, 0.772 and 73.3%; respective P values, .001 and ,.001).
Conclusion:
This pilot study in a cohort of patients with hepatic steatosis without other parenchymal disease suggested ASQ may be valuable for the quantification of hepatic steatosis and detection of hepatic steatosis 10% or greater in living liver donors.
1
From the Department of Radiology, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Korea (J.Y.S.); Department of Radiology, Institute of Radiation Medicine (J.Y.L., J.M.L., J.K.H., B.I.C.), and Department of Surgery (N.J.Y., K.W.L., K.S.S.), Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Republic of Korea; and Biomedical Engineering Branch, Division of Convergence Technology, National Cancer Center, Goyang, Gyeonggi-do, Korea (K.G.K.). Received August 18, 2014; revision requested October 14; revision received April 1, 2015; accepted April 21; final version accepted May 1. Supported by grant no. 04-2012-0590 of the SNUH Research Fund and by the Ministry of Science, ICT and Future Planning (MSIP), Korea, under the Convergence Information Technology Research Center (C-ITRC) support program (NIPA-2014-H0401-14-1002) supervised by the National IT Industry Promotion Agency (NIPA). Address correspondence to J.Y.L. (e-mail: [email protected]).
RSNA, 2015
q
RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
1
Original Research n Ultrasonography
Hepatic Steatosis: Assessment with Acoustic Structure Quantification of US Imaging1
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
H
epatic steatosis, which indicates the accumulation of fat in hepatocytes, is common and has a broad disease spectrum according to its pathogenesis and its severity (1). Clinically, the occurrence of nonalcoholic fatty liver disease, the most common type of hepatic steatosis, is strongly correlated with metabolic disease, including type 2 diabetes mellitus and atherosclerotic cardiovascular disease (2). Simple hepatic steatosis can evolve into a more severe stage, such as steatohepatitis or cirrhosis. Furthermore, hepatic steatosis is an important risk factor for postoperative complications after major liver resection and living donor liver transplantation (3–7). Hepatic steatosis of 10% or greater is known to be critical in patients who undergo living donor liver transplantation because of the risks of initial graft dysfunction, poor graft survival, and other complications (8–10). The widespread use of liver biopsy (which is a reference standard for quantification of liver fat) as a screening tool and to monitor treatment response is limited mainly because of invasiveness. Therefore, noninvasive imaging methods that can quantitatively measure liver fat were actively investigated. Among them, hydrogen 1 (1H) magnetic resonance (MR) spectroscopy gained ground in the past decade as an alternative noninvasive reference
Advances in Knowledge nn In a population of living donors for liver transplantation, acoustic structure quantification (ASQ) provides an excellent quantitative tool for hepatic steatosis, and there is a strong linear correlation (r = 20.87; P , .001) between focal disturbance ratio measured with ASQ and hepatic fat fraction measured with MR spectroscopy. nn ASQ provides a tool with a diagnostic accuracy (area under curve, 0.959) by using MR spectroscopy as the reference standard for diagnosing hepatic steatosis that is 10% or greater for living donor liver transplantation. 2
standard to evaluate liver fat content because of its high diagnostic accuracy and reproducibility (1,2,11–15). However, MR spectroscopy is expensive to perform as a screening and monitoring tool for hepatic steatosis. Ultrasonographic (US) imaging is a modality that is simple and inexpensive to perform and is safe for the patient. To overcome the intrinsic subjectivity associated with the use of US imaging to grade fatty liver, several methods that use US image data were proposed for the quantification of hepatic steatosis (14–20). However, these methods are not widely used in clinical practice mainly because of their complexity or their use of noncommercial software. A commercially available quantification method, acoustic structure quantification (ASQ), was introduced into clinical practice for the evaluation of diffuse liver disease. The method is based on statistical analysis of the difference between theoretical and real echo amplitude distribution (21,22). Theoretical echo amplitude distribution indicates that a speckle pattern in a certain liver region is approximated by a function of a Reyleigh distribution based on the assumption that the speckle pattern is generated only by ultrasonic interference of very small scattering objects that are located closer than the wavelength of US. However, real echo amplitude distribution of normal liver does not fit a Reyleigh distribution mainly because of the presence of small vessel walls. As hepatic steatosis progresses, the real echo amplitude distribution more likely approaches the theoretical echo amplitude distribution with masking of the small structures, such as small vessel walls (21). In an animal model of hepatic steatosis, a significant correlation was observed between focal disturbance (FD) ratio, calculated by using ASQ, and either fat droplet area or fat droplet size (21). Therefore, it was suggested that the FD ratio could serve as a quantitative
Implication for Patient Care nn ASQ with a statistical model of echo amplitudes distribution may be beneficial to detect substantial hepatic steatosis in liver donors.
Son et al
biomarker of hepatic steatosis. However, to our knowledge, there has been no clinical study that investigated the value of ASQ for the quantification of hepatic steatosis. Therefore, the purpose of our study was to investigate the diagnostic performance of ASQ for the assessment of hepatic steatosis by using 1H MR spectroscopy as the reference standard and to compare ASQ with hepatorenal ratio.
Materials and Methods Study Population This prospective study was approved by our institutional review board and all participants gave written informed consent. Between September 2012 and September 2013, we enrolled 53 consecutive potential donors for living liver donor transplantation and 40 healthy patients in our study. We enrolled 40 healthy patients who annually visited our institute for health check-up and in whom hepatic steatosis was suspected or detected by using routine US examination. Before enrollment, the electronic medical records and laboratory findings (which included liver function tests within 3 months before ASQ) of all individuals were checked to confirm that
Published online before print 10.1148/radiol.2015141779 Content code: Radiology 2015; 000:1–8 Abbreviations: ASQ = acoustic structure quantification FD = focal disturbance HFF = hepatic fat fraction ROI = region of interest Author contributions: Guarantors of integrity of entire study, J.Y.S., J.Y.L., K.W.L., K.G.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, J.K.H.; clinical studies, J.Y.S., J.Y.L., N.J.Y., K.W.L., K.S.S., J.M.L., J.K.H.; experimental studies, K.G.K.; statistical analysis, J.Y.L., J.Y.S.; and manuscript editing, K.W.L., B.I.C. Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
they did not have any past medical history, clinical symptoms, or signs of other liver or renal parenchymal disease or injury, or systemic or malignant disease. During this process, three invited participants were excluded because of late discovery of past medical history of tongue cancer (n = 1), breast cancer (n = 1), and liver cirrhosis (n = 1). All individuals underwent both US and 1H MR spectroscopy within 1 week of each other. One invited participant was excluded from our study because of claustrophobia. The remaining 89 participants were finally included in this study.
US Examination All US examinations were performed by using a US imager (Aplio XG; Toshiba Medical Systems, Otawara, Japan) equipped with a 5-MHz convex transducer. The examinations were performed by one of two radiologists (J.Y.S. and J.Y.L., with 6 and 20 years of experience in liver US imaging, respectively). First, a routine liver US examination that included five different sagittal liver and right kidney views was performed. Then, US images on ASQ mode were obtained three times each from right intercostal view and right subcostal view. Display depth and transmit focus were fixed at 10 cm and 6 cm, respectively. As many regions of interest (ROIs) as possible that were as large as possible were placed on the liver, with care taken to avoid large hepatic vessels or artifacts. FD ratio automatically appeared on a monitor that displayed US images in ASQ mode (Fig 1a). The mean ROIs per US image on ASQ mode was 2.8 6 0.4 (standard deviation; range, 1–4). For comparison, hepatorenal ratio was obtained by using an echo intensity analysis of a digitized liver and right kidney image (K.K.G. and K.Y.J., division of convergence technology, National Cancer Center, with 16 and 12 years of’ experience in the development of medical imaging quantitative analysis software, respectively), which was described for the quantification of hepatic steatosis (15,23). Two rectangular ROIs (1.5 3 1.5 cm) were placed on the hepatic parenchyma and renal cortex, with
Son et al
Figure 1
Figure 1: US images show ASQ and hepatorenal ratio. (a) Multiple ROIs are placed on the liver, avoiding large intrahepatic vascular structures. Automatic display of Cm2 histogram (Cm2 is a modified adjusted x2 value calculated on the basis of the percentage of real distribution variance relative to the theoretical distribution variance of echo amplitude in ROIs) is shown in the lower right corner and focal disturbance ratio is at the bottom. (b) ROIs were placed on the hepatic parenchyma and renal cortex at a depth similar to a probe near the central part of the US image, with care taken to avoid large vessels, the renal sinus, or the renal medulla. Hepatorenal ratio is calculated by using 1.5 3 1.5 cm rectangular ROIs.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
3
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
Table 1
Son et al
Figure 2
Demographics According to Clinical Characteristics and HFF Measured by Using MR Spectroscopy Parameter No. of patients Men Women Age (y) Body mass index (kg/m2) .25 kg/m2 No. of patients .28 kg/m2 No. of patients .30 kg/m2 No. of patients Time interval between ASQ and MR spectroscopy (d) HFF (%)
Liver Donor Candidates
Invited Participants
Total
53 33 20 32.85 6 10.12 (17–53) 23.59 6 3.55 (17.19–33.64) 27.86 6 2.33 17 30.10 6 1.83 7 32.55 6 1.54 2 0.98 6 3.10 (0–7)
36 21 15 51.72 6 11.58 (27–67) 24.69 6 4.10 (18.83–36.34) 28.59 6 3.64 14 31.70 6 3.55 6 34.59 6 2.45 3 0 6 0 (0–0)
89 54 35 40.48 6 14.16 (17–67) 24.05 6 3.80 (17.10–36.34) 28.19 6 2.96 31 30.84 6 2.77 13 33.77 6 2.20 5 0.58 6 2.43 (0–7)
4.20 6 3.85 (0.70–17.30)
8.67 6 7.05 (0.96–26.7)
6.01 6 5.78 (0.70–26.7)
Figure 2: Box-and-whisker plot of HFFs of liver donor candidates and invited participants. The lines within the boxes indicate medians. Black squares represent outliers and dots represent plotting of all data.
Note.—Except where indicated, data are mean 6 standard deviation and data in parentheses are range.
the similar depth from the probe near the central part of the US image, and with care taken to avoid large vessels, the renal sinus, or the renal medulla (Fig 1b). The hepatorenal ratio was automatically calculated by dividing the mean value of the liver ROI by the mean value of the kidney ROI. The hepatorenal ratio was obtained in each of the five liver and right kidney images and averaged.
MR Imaging Technique for Hepatic Fat Quantification High-speed, T2-corrected, multiecho, single-voxel 1H MR spectroscopy was performed with a 3-T MR imager (Magnetom Verio; Siemens Healthcare, Erlangen, Germany) by using a 32-channel phased-array surface coil. A modified stimulated-echo acquisition sequence was used (echo times msec/repetition time msec, 12, 24, 36, 48, and 72/3000). Other details in the MR spectroscopy protocol were the same as those used previously (10). Sagittal, coronal, and axial sections that covered the whole liver were preliminarily acquired to position the spectroscopy acquisition voxel. A single voxel (3 3 3 3 3 cm3) was placed within the dome of the right hepatic 4
lobe, avoiding major vascular and biliary structures and the periphery of the liver. The hepatic fat fraction (HFF) was calculated automatically. An HFF cutoff value of 5.79% was used for hepatic steatosis, which was validated for hepatic steatosis that were 10% or greater by using the same protocol and same machine as previously used (10). Ten percent hepatic steatosis is known as a critical margin for liver donor safety (8–10).
Statistical Analysis All statistical analyses were performed with statistical software (MedCalc version 12.4.0.0; MedCalc Software, Mariakerke, Belgium). We used the Pearson correlation coefficient to calculate the correlation between the FD ratio and HFF and between the hepatorenal ratio and HFF. We analyzed Pearson correlation coefficient according to body mass index (.25 kg/m2, .28 kg/m2, and .30 kg/m2). Receiver operating characteristic curve analysis with binomial exact confidence interval was performed to obtain the area under the curve and to determine the cutoff value of the FD ratio and hepatorenal ratio for the diagnosis of hepatic steatosis 10% or greater.
Sensitivity, specificity, positive predictive value, and negative predictive value were also obtained. Intraclass correlation coefficient was calculated to estimate the degree of FD ratio agreement between subcostal imaging and intercostal imaging. We used a Student t test that assumed equal variances. A P value of less than .05 was considered to indicate statistical significance.
Results Patient Characteristics This study included 54 men (mean age, 40.7 years 6 13.4) and 35 women (mean age, 40.1 years 6 15.5; Table 1). The HFF of the invited participants (mean HFF, 8.67% 6 7.05 [95% confidence interval: 6.28%, 11.05%]) was significantly higher than that of the liver donor candidates (mean HFF, 4.20% 6 3.85 [95% CI: 3.14%, 5.26%]; P , .001; Fig 2). Twenty-nine of 89 participants (32.6%) were categorized into the group that had hepatic steatosis of 10% or greater (mean HFF, 13.18% 6 4.89). The remaining 60 participants (67.4%) were categorized into the group that had less than 10% hepatic steatosis (mean HFF, 2.54% 6
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
0.98). Of the 53 liver donor candidates, 10 (18.9%) were categorized into the group that had hepatic steatosis of 10% or greater (mean HFF, 11.12% 6 3.99) and 43 (81.1%) were categorized into the group that had hepatic steatosis of less than 10% (mean HFF, 2.59% 6 0.91).
FD Ratio and Hepatorenal Ratio in Total Population The correlation coefficient (r) between FD ratio and HFF was 20.62 (95% confidence interval: 20.76, 20.48). However, a scatter plot between these two
variables showed an exponential relationship (Fig 3a). When a logarithm of these variables was applied, the two showed a strong linear relationship (r = 20.87 [95% confidence interval: 20.91, 20.80]; P , .001) (Fig 3b); in patients with body mass index greater than 25 kg/m2, r was 20.83 (95% confidence interval: 20.91, 20.67; P , .001); in patients with body mass index greater than 28 kg/m2, r was 20.80 (95% confidence interval: 20.94, 20.44; P , .001); and in patients with body mass index greater than 30 kg/m2, r was 20.87 (95% confidence interval:
Son et al
20.99, 20.07; P = .058). The correlation coefficient between hepatorenal ratio and HFF was 0.49 (95% confidence interval: 0.32, 0.64; P , .001) and showed a linear relationship (Fig 3c). The area under the curve of the FD ratio for detection of hepatic steatosis 10% or greater was 0.959 (P , .001) (Table 2). The optimal cutoff value of the FD ratio was 0.1 with sensitivity of 86.2% and specificity of 100%. The area under the curve of hepatorenal ratio for detection of hepatic steatosis 10% or greater was 0.772 (95% confidence interval: 0.671, 0.854; P , .001),
Figure 3
Figure 3: Correlation between HFF and FD ratio or hepatorenal ratio. (a) Scatter plot shows an exponential correlation between HFF and FD ratio. Exponential relationship is shown. (b) Regression plot shows a strong negative linear correlation between HFF and FD ratio after logarithmic transformation of both variables. The solid line indicates regression line; the lines with small dots are 95% confidence intervals; and lines with long dashes are 95% prediction intervals. (r = 20.8669). (c) Regression plot shows a positive linear correlation between HFF and hepatorenal ratio. The solid line indicates regression line; the lines with small dots are 95% confidence intervals; and lines with long dashes are 95% prediction intervals (r = 0.4935).
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
5
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
Son et al
Table 2 Diagnostic Performance of FD Ratio and Hepatorenal Ratio in the Total Population and Liver Donor Candidates Sensitivity Parameter Total population FD ratio Hepatorenal ratio Liver donor candidates FD ratio Hepatorenal ratio
Area Under the Curve
Cutoff Value
0.959 (0.895, 0.990) 0.772 (0.671, 0.854) 0.895 (0.780, 0.962) 0.660 (0.517, 0.785)
Specificity
Percentage
Numerator/ Denominator
Percentage
Numerator/ Denominator
Positive Predictive Value (%)
Negative Predictive Value (%)
0.1 1.326
86.2 (68.3, 96.0) 72.4 (52.8, 87.3)
25/29 21/29
100 (94.0, 100) 73.3 (60.3, 83.9)
60/60 44/60
100 (74.0, 100) 34.1 (17.9, 53.5)
97.4 (90.9, 99.7) 93.3 (83.6, 98.2)
0.1 1.312
70.0 (34.8, 93.3) 60.0 (26.2, 87.8)
7/10 6/10
100 (91.8, 100) 74.4 (58.8, 86.5)
43/43 32/43
100 (53.2, 100) 30.9 (11.0, 57.9)
94.6 (83.8, 99.1) 90.7 (76.4, 97.8)
Note.—Data in parentheses indicate 95% confidence intervals. Positive predictive value and negative predictive value were calculated with disease prevalence of 16% (8). All data met the HFF cutoff value of .5.79%.
which is significantly lower than that of the FD ratio (P = .001). The cutoff value of hepatorenal ratio was 1.326, with sensitivity of 72.4% and specificity of 73.3%. The specificity of hepatorenal ratio was significantly lower than that of the FD ratio (P , .001).
FD Ratio and Hepatorenal Ratio in Liver Donor Candidates The correlation coefficient between logarithms of both FD ratio and HFF was 20.77 (95% confidence interval: 20.86, 20.63; P , .001) and showed a strong linear relationship. The correlation coefficient r between the hepatorenal ratio and HFF was 0.41 (95% confidence interval: 0.16, 0.61) with a linear relationship. The areas under the curve, cutoff values, and diagnostic performances of FD ratio for detection of hepatic steatosis 10% or greater are summarized in Table 2. The area under the curve of the hepatorenal ratio for detection of hepatic steatosis 10% or greater was 0.660 (95% confidence interval: 0.517, 0.785), which is significantly lower than that of FD ratio (P = .011). In this regard, the cutoff value of hepatorenal ratio was 1.312 (sensitivity, 60.0%; specificity, 74.4%). The specificity of hepatorenal ratio was significantly lower than that of the FD ratio (P , .001). FD Ratio according to Imaging Approach of ASQ The intraclass correlation coefficient of the FD ratio between subcostal and 6
intercostal approaches was 0.896 (95% confidence interval: 0.841, 0.932). The intraclass correlation coefficients of the FD ratios in subcostal and intercostal liver scans were 0.896 (95% confidence interval: 0.841, 0.932) and 0.893 (95% confidence interval: 0.848, 0.927), respectively. The areas under the curve of the FD ratio found by using intercostal and subcostal approaches for the diagnosis of hepatic steatosis 10% or greater were 0.970 and 0.945, respectively. For comparison, the area under the curve of the FD ratio found by using both imaging approaches was 0.959. There were no significant differences in area under the curve between the intercostal and subcostal approaches (P , .302).
Discussion In this study, the FD ratio was excellent for the diagnosis of hepatic steatosis 10% or greater with high diagnostic accuracy. This result suggests that ASQ could be valuable for the preoperative detection of hepatic steatosis 10% or greater, which is critical in patients who undergo living donor liver transplantation because of the risks of initial graft dysfunction, poor graft survival, and other complications (8,9). The FD ratio was correlated with HFF (R2 = 0.7515). Even in obese patients with body mass index greater than 25 kg/m2 or greater than 28 kg/ m2, the correlation coefficient was high. This implies that ASQ could potentially
be used to quantitatively monitor liver fat change. It is well known that hepatorenal ratio is significantly higher in patients with hepatic steatosis than in healthy patients (18,23,24). However, it is still questionable whether hepatorenal ratio is a sufficient biomarker that can be used as an alternative to MR spectroscopy for hepatic steatosis quantification. Even though a previous study reported that hepatorenal ratio had an excellent correlation with the degree of steatosis at MR spectroscopy (sensitivity, 100%; specificity, 95%) (15), other studies reported the need for an additional process to enhance the diagnostic performance of hepatorenal ratio values, such as standardization by a tissue-mimicking phantom or artificial neural network (18,25). Differences between this study and previous studies regarding diagnostic performance of hepatorenal ratio may be caused by factors such as use of different machines and different image analysis software, different study population, and different criteria to define a cohort of patients with hepatic steatosis. Our results revealed that the FD ratio is independent of the subcostal or intercostal method of liver imaging. Our study has several limitations. First, our study used MR spectroscopy rather than histologic examination as the reference standard. However, many clinical trials (1,2,11–15) used MR spectroscopy as the reference standard because of the method’s accuracy,
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
quantitative nature, and noninvasiveness when quantifying liver fat. We used our own population to set thresholds that were used to determine the accuracy. This can overestimate study results. Second, the population of this study was heterogeneous: both consecutive liver donor candidates and invited participants were included in our study. Although the disease prevalence of hepatic steatosis 10% or greater in the group of liver donor candidates was similar to that of previous studies (8,26), HFFs of included liver donor candidates (95% confidence interval: 3.14, 5.26) were too low to represent the range of HFF expected in clinical practice. To include larger HFFs, individuals suspected of having or who had evidence of hepatic steatosis were additionally included (95% confidence interval: 6.28, 11.05). Nevertheless, the number of patients with an HFF of 13%–17% (which was reported to most likely represent hepatic steatosis 30% or greater) was not enough in our study to verify the value of ASQ for detection of 30% or greater (27–29). To prove this, a further study that includes a sufficient number of patients with hepatic steatosis 30% or greater is needed. Third, our study only included simple hepatic steatosis. However, in clinical situations, hepatic steatosis presents on a diverse spectrum, from simple hepatic steatosis through steatohepatitis to cirrhosis. It is unclear how this FD ratio will perform in patients with both steatosis and fibrosis. Fourth, the size of ROIs to measure ASQ in the liver was not recorded because it was not provided by the US machine that we used. In conclusion, the results of our study suggest that FD ratio calculated by using ASQ is comparable to HFF calculated by using MR spectroscopy to quantify hepatic steatosis and detect hepatic steatosis greater than 10% in living donor liver patients. Disclosures of Conflicts of Interest: J.Y.S. disclosed no relevant relationships. J.Y.L. disclosed no relevant relationships. N.J.Y. disclosed no relevant relationships. K.W.L.
disclosed no relevant relationships. K.S.S. disclosed no relevant relationships. K.G.K. disclosed no relevant relationships. J.M.L. Activities related to the present article: Siemens Healthcare provided nonfinancial support by providing the research software and technical support for installing the sequence of chemicalshift imaging and breath-hold MR spectroscopy. Activities not related to the present article: author disclosed a grant and personal fees from Bayer Healthcare; author disclosed grants from GE Healthcare, Guerbet, Dong Seo Medical, RF Medical, Starmed, and Toshiba Healthcare; author disclosed nonfinancial support from Philips for technical support for research software. Other relationships: disclosed no relevant relationships. J.K.H. disclosed no relevant relationships. B.I.C. disclosed no relevant relationships.
References 1. Ma X, Holalkere NS, Kambadakone R A, Mino-Kenudson M, Hahn PF, Sahani DV. Imaging-based quantification of hepatic fat: methods and clinical applications. RadioGraphics 2009;29(5):1253–1277. 2. Schwenzer NF, Springer F, Schraml C, Stefan N, Machann J, Schick F. Non-invasive assessment and quantification of liver steatosis by ultrasound, computed tomography and magnetic resonance. J Hepatol 2009; 51(3):433–445.
Son et al
9. Trotter JF, Wisniewski KA, Terrault NA, et al. Outcomes of donor evaluation in adultto-adult living donor liver transplantation. Hepatology 2007;46(5):1476–1484. 10. Hwang I, Lee JM, Lee KB, et al. Hepatic steatosis in living liver donor candidates: preoperative assessment by using breath-hold triple-echo MR imaging and 1H MR spectroscopy. Radiology 2014;271(3):730–738. 11. Pineda N, Sharma P, Xu Q, Hu X, Vos M, Martin DR. Measurement of hepatic lipid: high-speed T2-corrected multiecho acquisition at 1H MR spectroscopy—a rapid and accurate technique. Radiology 2009; 252(2):568–576. 12. Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic resonance spectroscopy to measure hepatic triglyceride content: prevalence of hepatic steatosis in the general population. Am J Physiol Endocrinol Metab 2005; 288(2):E462–E468. 13. Bohte AE, Koot BG, van der Baan-Slootweg OH, et al. US cannot be used to predict the presence or severity of hepatic steatosis in severely obese adolescents. Radiology 2012; 262(1):327–334. 14. Edens MA, van Ooijen PM, Post WJ, et al. Ultrasonography to quantify hepatic fat content: validation by 1H magnetic resonance spectroscopy. Obesity (Silver Spring) 2009; 17(12):2239–2244.
3. McCormack L, Petrowsky H, Jochum W, Furrer K, Clavien PA. Hepatic steatosis is a risk factor for postoperative complications after major hepatectomy: a matched case-control study. Ann Surg 2007;245(6):923–930.
15. Mancini M, Prinster A, Annuzzi G, et al. Sonographic hepatic-renal ratio as indicator of hepatic steatosis: comparison with (1)H magnetic resonance spectroscopy. Metabolism 2009;58(12):1724–1730.
4. Kooby DA, Fong Y, Suriawinata A, et al. Impact of steatosis on perioperative outcome following hepatic resection. J Gastrointest Surg 2003;7(8):1034–1044.
16. Strauss S, Gavish E, Gottlieb P, Katsnel son L. Interobserver and intraobserver variability in the sonographic assessment of fatty liver. AJR Am J Roentgenol 2007; 189(6):W320–W323.
5. Zamboni F, Franchello A, David E, et al. Effect of macrovescicular steatosis and other donor and recipient characteristics on the outcome of liver transplantation. Clin Transplant 2001;15(1):53–57. 6. Urena MA, Moreno Gonzalez E, Romero CJ, Ruiz-Delgado FC, Moreno Sanz C. An approach to the rational use of steatotic donor livers in liver transplantation. Hepatogastroenterology 1999;46(26):1164–1173. 7. McCormack L, Clavien PA. Understanding the meaning of fat in the liver. Liver Transpl 2005;11(2):137–139. 8. Sharma A, Ashworth A, Behnke M, Cotterell A, Posner M, Fisher RA. Donor selection for adult-to-adult living donor liver transplantation: well begun is half done. Transplantation 2013;95(3):501–506.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
17. Graif M, Yanuka M, Baraz M, et al. Quantitative estimation of attenuation in ultrasound video images: correlation with histology in diffuse liver disease. Invest Radiol 2000; 35(5):319–324. 18. Kim SH, Lee JM, Kim JH, et al. Appropriateness of a donor liver with respect to macrosteatosis: application of artificial neural networks to US images—initial experience. Radiology 2005;234(3):793–803. 19. Acharya UR, Sree SV, Ribeiro R, et al. Data mining framework for fatty liver disease classification in ultrasound: a hybrid feature extraction paradigm. Med Phys 2012;39 (7):4255–4264. 20. Mihăilescu DM, Gui V, Toma CI, Popescu A, Sporea I. Computer aided diagnosis method for steatosis rating in ultrasound images using
7
ULTRASONOGRAPHY: Assessment of Hepatic Steatosis with Acoustic Structure Quantification of US Imaging
random forests. Med Ultrason 2013;15(3): 184–190. 21. Kuroda H, Kakisaka K, Kamiyama N, et al. Non-invasive determination of hepatic steatosis by acoustic structure quantification from ultrasound echo amplitude. World J Gastroenterol 2012;18(29):3889– 3895. 22. Toyoda H, Kumada T, Kamiyama N, et al. B-mode ultrasound with algorithm based on statistical analysis of signals: evaluation of liver fibrosis in patients with chronic hepatitis C. AJR Am J Roentgenol 2009; 193(4):1037–1043. 23. von Volkmann HL, Havre RF, Løberg EM, et al. Quantitative measurement of ultrasound
8
attenuation and Hepato-Renal Index in NonAlcoholic Fatty Liver Disease. Med Ultrason 2013;15(1):16–22. 24. Osawa H, Mori Y. Sonographic diagnosis of fatty liver using a histogram technique that compares liver and renal cortical echo amplitudes. J Clin Ultrasound 1996;24(1):25– 29. 25. Xia MF, Yan HM, He WY, et al. Standardized ultrasound hepatic/renal ratio and hepatic attenuation rate to quantify liver fat content: an improvement method. Obesity (Silver Spring) 2012;20(2):444–452. 26. Pascher A, Sauer IM, Walter M, et al. Donor evaluation, donor risks, donor outcome, and donor quality of life in adult-to-adult
Son et al
living donor liver transplantation. Liver Transpl 2002;8(9):829–837. 27. Idilman IS, Aniktar H, Idilman R, et al. Hepatic steatosis: quantification by proton density fat fraction with MR imaging versus liver biopsy. Radiology 2013;267(3):767–775. 28. Noworolski SM, Lam MM, Merriman RB, Ferrell L, Qayyum A. Liver steatosis: concordance of MR imaging and MR spectroscopic data with histologic grade. Radiology 2012; 264(1):88–96. 29. Tang A, Tan J, Sun M, et al. Nonalcoholic fatty liver disease: MR imaging of liver proton density fat fraction to assess hepatic steatosis. Radiology 2013;267(2):422–431.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
